Test: morsel splitting & merging in SharedWorkSource by Dandandan · Pull Request #21640 · apache/datafusion

Dandandan · 2026-04-15T12:24:14Z

Which issue does this PR close?

Follow-up to #21351 (dynamic work scheduling in FileStream).

Rationale for this change

When a scan has few large files, idle sibling streams have nothing to steal even after #21351 enables work sharing. This PR splits large files by byte range so work can be distributed more evenly across partitions.

What changes are included in this PR?

Adds morsel splitting to SharedWorkSource::pop_front():

Splitting: when queue depth < 2 * target_partitions and a file's projected size >= 2 MiB, splits the file in half by byte range and pushes the second half back onto the shared queue
Projected size estimation: uses per-column byte_size from PartitionedFile.statistics when available (e.g. from Parquet column stats), otherwise falls back to raw_file_size * (projected_cols / total_cols)
Minimum morsel size of 1 MiB — files are never split below this threshold

Are these changes tested?

Existing tests pass. Additional tests for the splitting logic to be added.

Are there any user-facing changes?

Faster performance for queries scanning few large files with multiple partitions.

🤖 Generated with Claude Code

Dandandan · 2026-04-15T12:25:36Z

run benchmarks

adriangbot · 2026-04-15T12:27:45Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4251958591-1290-bqxbk 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (76a1f58) to 961c5fc (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T12:28:51Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4251958591-1288-lsrb5 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (76a1f58) to 961c5fc (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T12:28:54Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4251958591-1289-xhg2n 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (76a1f58) to 961c5fc (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

Dandandan · 2026-04-15T12:41:02Z

run benchmarks

adriangbot · 2026-04-15T12:43:14Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4252054949-1293-88ndt 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (ea3834e) to 961c5fc (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T12:43:17Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4252054949-1291-w8c86 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (ea3834e) to 961c5fc (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T12:43:22Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4252054949-1292-c86n9 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (ea3834e) to 961c5fc (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

Dandandan · 2026-04-15T13:49:26Z

run benchmarks

adriangbot · 2026-04-15T13:52:17Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4252614535-1294-h7qdm 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (b852406) to 961c5fc (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T13:52:17Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4252614535-1296-mzzdd 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (b852406) to 961c5fc (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T13:52:23Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4252614535-1295-pxs9d 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (b852406) to 961c5fc (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T14:08:20Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and morsel-splitting
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                     HEAD ┃                         morsel-splitting ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │              6.59 / 7.02 ±0.80 / 8.62 ms │              6.55 / 6.97 ±0.79 / 8.55 ms │     no change │
│ QQuery 2  │        148.49 / 149.16 ±0.73 / 150.54 ms │        146.86 / 148.63 ±1.38 / 150.57 ms │     no change │
│ QQuery 3  │        113.57 / 114.83 ±1.21 / 116.54 ms │        114.45 / 115.71 ±0.68 / 116.50 ms │     no change │
│ QQuery 4  │     1360.77 / 1366.27 ±4.77 / 1373.04 ms │     1344.62 / 1356.41 ±7.55 / 1366.32 ms │     no change │
│ QQuery 5  │        171.87 / 174.42 ±1.58 / 176.05 ms │        176.12 / 177.83 ±1.33 / 180.09 ms │     no change │
│ QQuery 6  │       856.75 / 886.23 ±17.33 / 900.51 ms │        846.25 / 859.98 ±8.63 / 872.96 ms │     no change │
│ QQuery 7  │        340.70 / 345.45 ±4.29 / 351.68 ms │        343.22 / 345.98 ±2.13 / 348.65 ms │     no change │
│ QQuery 8  │        116.11 / 117.69 ±1.31 / 119.70 ms │        115.21 / 116.34 ±1.33 / 118.47 ms │     no change │
│ QQuery 9  │         99.99 / 102.70 ±2.57 / 107.49 ms │        118.32 / 124.86 ±5.28 / 132.56 ms │  1.22x slower │
│ QQuery 10 │        107.01 / 108.18 ±0.95 / 109.38 ms │        107.38 / 110.57 ±1.68 / 112.00 ms │     no change │
│ QQuery 11 │        946.78 / 954.41 ±7.61 / 964.61 ms │       944.33 / 959.74 ±10.55 / 974.35 ms │     no change │
│ QQuery 12 │           46.77 / 47.43 ±0.67 / 48.49 ms │           45.21 / 46.76 ±1.21 / 48.04 ms │     no change │
│ QQuery 13 │        396.84 / 403.60 ±6.17 / 412.14 ms │        397.15 / 402.10 ±3.54 / 406.94 ms │     no change │
│ QQuery 14 │     997.75 / 1014.54 ±10.89 / 1028.96 ms │     1027.40 / 1034.99 ±4.97 / 1041.94 ms │     no change │
│ QQuery 15 │           15.13 / 16.07 ±0.88 / 17.71 ms │           16.29 / 17.60 ±1.60 / 20.58 ms │  1.10x slower │
│ QQuery 16 │              6.97 / 7.85 ±0.78 / 9.02 ms │              6.84 / 7.22 ±0.38 / 7.89 ms │ +1.09x faster │
│ QQuery 17 │        229.12 / 230.48 ±1.07 / 231.43 ms │        230.89 / 232.34 ±1.21 / 234.49 ms │     no change │
│ QQuery 18 │        127.71 / 130.49 ±1.73 / 132.03 ms │        136.85 / 138.68 ±1.00 / 139.46 ms │  1.06x slower │
│ QQuery 19 │        156.24 / 157.20 ±0.63 / 157.91 ms │        154.11 / 156.18 ±1.28 / 157.56 ms │     no change │
│ QQuery 20 │           13.69 / 14.75 ±0.64 / 15.68 ms │           14.39 / 15.11 ±0.39 / 15.54 ms │     no change │
│ QQuery 21 │           19.56 / 19.94 ±0.32 / 20.35 ms │           20.33 / 21.06 ±0.44 / 21.59 ms │  1.06x slower │
│ QQuery 22 │        484.85 / 490.73 ±7.43 / 505.16 ms │        486.75 / 492.75 ±5.80 / 502.11 ms │     no change │
│ QQuery 23 │       867.43 / 891.67 ±14.45 / 910.04 ms │        893.63 / 901.93 ±5.63 / 910.19 ms │     no change │
│ QQuery 24 │        383.52 / 387.61 ±4.51 / 395.97 ms │        389.47 / 394.73 ±3.42 / 398.44 ms │     no change │
│ QQuery 25 │        342.03 / 346.14 ±2.90 / 349.75 ms │        344.94 / 346.56 ±0.91 / 347.73 ms │     no change │
│ QQuery 26 │           80.97 / 82.85 ±1.32 / 85.02 ms │           83.08 / 85.02 ±1.37 / 87.04 ms │     no change │
│ QQuery 27 │              6.83 / 7.31 ±0.50 / 8.23 ms │              6.81 / 7.59 ±0.59 / 8.59 ms │     no change │
│ QQuery 28 │        149.06 / 150.51 ±1.38 / 152.74 ms │        157.11 / 159.08 ±2.38 / 163.72 ms │  1.06x slower │
│ QQuery 29 │        282.14 / 283.82 ±1.91 / 287.48 ms │        283.33 / 285.01 ±1.63 / 287.42 ms │     no change │
│ QQuery 30 │           42.88 / 45.29 ±1.53 / 46.61 ms │           43.97 / 47.15 ±2.31 / 50.57 ms │     no change │
│ QQuery 31 │        169.96 / 173.04 ±2.29 / 176.77 ms │        176.25 / 179.02 ±2.22 / 182.49 ms │     no change │
│ QQuery 32 │           57.30 / 59.28 ±2.58 / 64.38 ms │         57.22 / 67.22 ±16.48 / 100.05 ms │  1.13x slower │
│ QQuery 33 │        139.43 / 141.65 ±1.19 / 142.70 ms │        143.54 / 144.91 ±1.23 / 146.73 ms │     no change │
│ QQuery 34 │              6.94 / 7.13 ±0.23 / 7.54 ms │              6.80 / 7.16 ±0.31 / 7.70 ms │     no change │
│ QQuery 35 │        107.31 / 109.84 ±1.47 / 111.38 ms │        109.79 / 110.75 ±0.79 / 112.11 ms │     no change │
│ QQuery 36 │              6.70 / 7.30 ±0.89 / 9.06 ms │              6.53 / 6.85 ±0.22 / 7.19 ms │ +1.07x faster │
│ QQuery 37 │              8.65 / 8.83 ±0.12 / 9.01 ms │            9.95 / 10.69 ±0.78 / 12.15 ms │  1.21x slower │
│ QQuery 38 │           86.50 / 90.21 ±3.43 / 96.30 ms │           89.02 / 91.54 ±2.71 / 96.31 ms │     no change │
│ QQuery 39 │        126.06 / 128.89 ±1.79 / 131.37 ms │        131.35 / 134.16 ±1.55 / 136.08 ms │     no change │
│ QQuery 40 │        112.35 / 117.27 ±5.59 / 127.38 ms │        113.27 / 119.47 ±5.56 / 129.41 ms │     no change │
│ QQuery 41 │           14.24 / 14.75 ±0.39 / 15.19 ms │           14.62 / 15.85 ±1.27 / 17.64 ms │  1.07x slower │
│ QQuery 42 │        108.80 / 110.15 ±1.20 / 111.82 ms │        109.82 / 110.35 ±0.45 / 111.15 ms │     no change │
│ QQuery 43 │              6.01 / 6.34 ±0.26 / 6.68 ms │              5.85 / 6.04 ±0.19 / 6.37 ms │     no change │
│ QQuery 44 │           11.86 / 12.71 ±0.75 / 14.11 ms │           17.80 / 19.30 ±1.35 / 21.67 ms │  1.52x slower │
│ QQuery 45 │           50.03 / 51.28 ±1.16 / 53.37 ms │           51.33 / 52.11 ±0.55 / 52.79 ms │     no change │
│ QQuery 46 │             8.56 / 9.51 ±0.69 / 10.36 ms │              8.38 / 8.64 ±0.28 / 9.01 ms │ +1.10x faster │
│ QQuery 47 │        711.92 / 719.13 ±5.36 / 727.65 ms │        705.40 / 717.54 ±6.90 / 725.71 ms │     no change │
│ QQuery 48 │        292.87 / 296.73 ±2.40 / 300.45 ms │        292.42 / 296.49 ±3.62 / 302.22 ms │     no change │
│ QQuery 49 │        251.59 / 255.87 ±2.40 / 258.26 ms │        254.63 / 257.05 ±2.10 / 260.30 ms │     no change │
│ QQuery 50 │        221.74 / 231.48 ±6.42 / 238.97 ms │        225.71 / 229.06 ±2.43 / 232.38 ms │     no change │
│ QQuery 51 │        182.03 / 184.45 ±2.39 / 188.57 ms │        181.04 / 185.52 ±2.97 / 189.86 ms │     no change │
│ QQuery 52 │        107.62 / 109.17 ±0.86 / 110.19 ms │        108.21 / 110.39 ±1.44 / 112.43 ms │     no change │
│ QQuery 53 │        101.90 / 103.44 ±0.90 / 104.51 ms │        104.17 / 104.64 ±0.52 / 105.62 ms │     no change │
│ QQuery 54 │        144.68 / 148.92 ±2.88 / 153.73 ms │        148.48 / 150.12 ±1.75 / 153.38 ms │     no change │
│ QQuery 55 │        106.63 / 108.57 ±1.04 / 109.39 ms │        107.09 / 108.53 ±1.15 / 110.04 ms │     no change │
│ QQuery 56 │        141.23 / 143.31 ±1.75 / 145.96 ms │        143.26 / 144.87 ±1.93 / 148.49 ms │     no change │
│ QQuery 57 │        174.34 / 177.86 ±2.32 / 180.55 ms │        175.82 / 177.35 ±1.02 / 178.56 ms │     no change │
│ QQuery 58 │        287.43 / 297.54 ±5.53 / 302.80 ms │        293.38 / 301.08 ±6.33 / 309.29 ms │     no change │
│ QQuery 59 │        200.80 / 203.35 ±2.11 / 207.23 ms │        199.84 / 202.52 ±1.74 / 204.26 ms │     no change │
│ QQuery 60 │        143.37 / 145.65 ±1.32 / 147.11 ms │        146.00 / 147.70 ±1.12 / 148.95 ms │     no change │
│ QQuery 61 │           13.04 / 13.41 ±0.31 / 13.93 ms │           12.87 / 13.15 ±0.28 / 13.67 ms │     no change │
│ QQuery 62 │       891.96 / 908.08 ±13.98 / 931.85 ms │      888.82 / 950.31 ±49.95 / 1030.85 ms │     no change │
│ QQuery 63 │        104.09 / 107.38 ±2.01 / 110.32 ms │        103.38 / 107.00 ±1.93 / 108.92 ms │     no change │
│ QQuery 64 │        682.65 / 692.09 ±5.33 / 697.73 ms │        689.79 / 697.22 ±4.84 / 703.41 ms │     no change │
│ QQuery 65 │        251.18 / 254.98 ±3.28 / 260.24 ms │        251.84 / 257.80 ±3.56 / 262.77 ms │     no change │
│ QQuery 66 │       239.88 / 258.86 ±14.78 / 283.70 ms │       239.44 / 256.90 ±10.92 / 268.42 ms │     no change │
│ QQuery 67 │       309.02 / 316.69 ±10.12 / 334.69 ms │        309.94 / 318.15 ±5.51 / 326.62 ms │     no change │
│ QQuery 68 │            8.91 / 10.29 ±1.02 / 11.87 ms │            9.19 / 11.11 ±2.23 / 15.30 ms │  1.08x slower │
│ QQuery 69 │        102.64 / 105.05 ±2.33 / 109.44 ms │        105.85 / 106.61 ±1.13 / 108.86 ms │     no change │
│ QQuery 70 │       332.34 / 353.61 ±13.90 / 370.72 ms │       341.45 / 356.32 ±13.64 / 378.04 ms │     no change │
│ QQuery 71 │        134.70 / 137.07 ±1.99 / 139.81 ms │        136.88 / 142.05 ±6.69 / 154.79 ms │     no change │
│ QQuery 72 │        618.51 / 632.69 ±9.94 / 647.05 ms │        623.76 / 631.05 ±5.92 / 639.47 ms │     no change │
│ QQuery 73 │             7.31 / 9.10 ±2.21 / 13.37 ms │              6.84 / 8.29 ±0.96 / 9.45 ms │ +1.10x faster │
│ QQuery 74 │       575.75 / 593.81 ±10.06 / 604.14 ms │        585.94 / 596.87 ±8.10 / 606.87 ms │     no change │
│ QQuery 75 │        273.94 / 278.24 ±2.38 / 281.19 ms │        279.40 / 281.16 ±1.66 / 283.75 ms │     no change │
│ QQuery 76 │        131.10 / 134.64 ±1.78 / 135.90 ms │        134.26 / 136.06 ±1.40 / 138.02 ms │     no change │
│ QQuery 77 │        186.67 / 189.92 ±1.97 / 192.71 ms │        194.73 / 196.17 ±1.16 / 197.33 ms │     no change │
│ QQuery 78 │        343.46 / 349.19 ±4.64 / 354.50 ms │        344.52 / 349.26 ±4.53 / 357.79 ms │     no change │
│ QQuery 79 │        234.39 / 236.29 ±1.29 / 237.60 ms │        233.55 / 235.84 ±1.42 / 237.66 ms │     no change │
│ QQuery 80 │        322.58 / 327.41 ±3.74 / 331.15 ms │        326.10 / 328.00 ±1.03 / 328.84 ms │     no change │
│ QQuery 81 │           25.72 / 26.81 ±1.19 / 28.48 ms │           26.39 / 27.42 ±0.95 / 28.83 ms │     no change │
│ QQuery 82 │        199.04 / 201.08 ±2.02 / 204.57 ms │        202.48 / 204.47 ±1.70 / 206.45 ms │     no change │
│ QQuery 83 │           39.17 / 40.15 ±1.29 / 42.63 ms │           39.54 / 40.63 ±0.86 / 41.64 ms │     no change │
│ QQuery 84 │           49.44 / 50.45 ±0.85 / 51.97 ms │           49.98 / 51.33 ±0.98 / 52.74 ms │     no change │
│ QQuery 85 │        147.75 / 149.49 ±1.20 / 151.32 ms │        152.02 / 154.41 ±2.29 / 158.68 ms │     no change │
│ QQuery 86 │           39.28 / 40.91 ±1.71 / 43.23 ms │           40.00 / 40.53 ±0.53 / 41.48 ms │     no change │
│ QQuery 87 │           86.50 / 89.87 ±3.61 / 96.35 ms │           88.62 / 92.26 ±3.49 / 97.91 ms │     no change │
│ QQuery 88 │        100.70 / 101.44 ±0.67 / 102.66 ms │        111.95 / 113.05 ±0.78 / 114.03 ms │  1.11x slower │
│ QQuery 89 │        118.01 / 119.26 ±1.42 / 122.00 ms │        119.31 / 119.92 ±0.91 / 121.69 ms │     no change │
│ QQuery 90 │           23.29 / 24.07 ±0.58 / 24.79 ms │           25.73 / 27.06 ±0.92 / 28.32 ms │  1.12x slower │
│ QQuery 91 │           65.32 / 66.98 ±1.23 / 68.71 ms │           63.66 / 65.80 ±1.37 / 67.70 ms │     no change │
│ QQuery 92 │           58.67 / 59.24 ±0.47 / 60.03 ms │           57.89 / 59.88 ±1.28 / 61.89 ms │     no change │
│ QQuery 93 │        185.34 / 188.92 ±1.89 / 190.66 ms │        187.71 / 189.40 ±1.41 / 190.98 ms │     no change │
│ QQuery 94 │           61.44 / 62.08 ±0.50 / 62.93 ms │           63.03 / 63.43 ±0.39 / 64.00 ms │     no change │
│ QQuery 95 │        128.14 / 129.75 ±1.21 / 131.41 ms │        130.00 / 131.60 ±1.35 / 133.63 ms │     no change │
│ QQuery 96 │           74.01 / 75.27 ±1.18 / 77.33 ms │           71.26 / 75.31 ±2.13 / 77.25 ms │     no change │
│ QQuery 97 │        125.38 / 127.26 ±1.59 / 129.93 ms │        126.85 / 129.87 ±1.77 / 132.31 ms │     no change │
│ QQuery 98 │        155.84 / 158.09 ±1.31 / 159.95 ms │        150.61 / 155.15 ±2.44 / 157.37 ms │     no change │
│ QQuery 99 │ 10787.94 / 10819.62 ±31.73 / 10860.41 ms │ 10780.46 / 10829.24 ±31.61 / 10871.35 ms │     no change │
└───────────┴──────────────────────────────────────────┴──────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary               ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)               │ 31667.83ms │
│ Total Time (morsel-splitting)   │ 31881.87ms │
│ Average Time (HEAD)             │   319.88ms │
│ Average Time (morsel-splitting) │   322.04ms │
│ Queries Faster                  │          4 │
│ Queries Slower                  │         12 │
│ Queries with No Change          │         83 │
│ Queries with Failure            │          0 │
└─────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	158.7s
Peak memory	5.5 GiB
Avg memory	4.6 GiB
CPU user	262.0s
CPU sys	17.8s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	159.7s
Peak memory	4.9 GiB
Avg memory	4.3 GiB
CPU user	281.2s
CPU sys	18.9s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-04-15T14:08:44Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and morsel-splitting
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃                      morsel-splitting ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.22 / 4.51 ±6.40 / 17.30 ms │          1.22 / 4.51 ±6.43 / 17.37 ms │     no change │
│ QQuery 1  │        14.28 / 15.39 ±0.61 / 16.03 ms │        12.58 / 13.03 ±0.34 / 13.57 ms │ +1.18x faster │
│ QQuery 2  │        43.74 / 44.47 ±1.10 / 46.65 ms │        36.35 / 36.48 ±0.09 / 36.61 ms │ +1.22x faster │
│ QQuery 3  │        41.53 / 44.29 ±2.32 / 47.85 ms │        34.40 / 35.42 ±0.81 / 36.16 ms │ +1.25x faster │
│ QQuery 4  │     288.94 / 300.25 ±8.31 / 309.08 ms │     263.87 / 268.57 ±3.76 / 274.32 ms │ +1.12x faster │
│ QQuery 5  │     346.10 / 349.50 ±2.64 / 353.12 ms │     308.37 / 310.08 ±1.62 / 312.44 ms │ +1.13x faster │
│ QQuery 6  │           6.03 / 7.21 ±0.99 / 8.72 ms │           7.14 / 7.68 ±0.62 / 8.88 ms │  1.07x slower │
│ QQuery 7  │        16.86 / 17.16 ±0.35 / 17.83 ms │        14.39 / 14.54 ±0.16 / 14.85 ms │ +1.18x faster │
│ QQuery 8  │     425.83 / 439.31 ±7.38 / 448.12 ms │     376.94 / 380.10 ±2.47 / 384.05 ms │ +1.16x faster │
│ QQuery 9  │     660.79 / 678.36 ±9.95 / 691.49 ms │     517.91 / 526.14 ±7.77 / 538.62 ms │ +1.29x faster │
│ QQuery 10 │        93.43 / 94.63 ±0.78 / 95.89 ms │        75.59 / 77.66 ±2.12 / 81.72 ms │ +1.22x faster │
│ QQuery 11 │     105.28 / 106.92 ±1.45 / 109.23 ms │        85.42 / 86.30 ±0.50 / 86.93 ms │ +1.24x faster │
│ QQuery 12 │     343.91 / 349.68 ±3.48 / 354.58 ms │     297.80 / 300.55 ±1.52 / 302.30 ms │ +1.16x faster │
│ QQuery 13 │     461.33 / 476.75 ±9.00 / 487.99 ms │    420.58 / 436.65 ±12.18 / 451.06 ms │ +1.09x faster │
│ QQuery 14 │     357.87 / 361.69 ±2.56 / 364.74 ms │     308.02 / 311.41 ±2.93 / 316.74 ms │ +1.16x faster │
│ QQuery 15 │    361.64 / 380.57 ±16.51 / 407.75 ms │     330.60 / 342.37 ±7.72 / 353.03 ms │ +1.11x faster │
│ QQuery 16 │    727.53 / 763.19 ±33.25 / 821.81 ms │     686.71 / 691.87 ±2.92 / 695.25 ms │ +1.10x faster │
│ QQuery 17 │    737.64 / 748.68 ±11.79 / 768.91 ms │     685.16 / 690.99 ±4.68 / 699.28 ms │ +1.08x faster │
│ QQuery 18 │  1403.56 / 1409.81 ±5.40 / 1416.77 ms │ 1373.89 / 1408.17 ±43.95 / 1493.15 ms │     no change │
│ QQuery 19 │        36.69 / 39.11 ±2.00 / 42.00 ms │        30.72 / 31.48 ±0.72 / 32.78 ms │ +1.24x faster │
│ QQuery 20 │    715.31 / 731.63 ±15.98 / 758.06 ms │     480.29 / 483.44 ±3.35 / 489.20 ms │ +1.51x faster │
│ QQuery 21 │     763.01 / 767.43 ±2.87 / 770.27 ms │     565.11 / 567.81 ±1.71 / 570.39 ms │ +1.35x faster │
│ QQuery 22 │  1148.91 / 1157.22 ±5.66 / 1164.57 ms │  1008.99 / 1014.34 ±4.55 / 1020.18 ms │ +1.14x faster │
│ QQuery 23 │ 3141.40 / 3191.64 ±48.12 / 3275.97 ms │  3225.21 / 3240.71 ±9.03 / 3249.50 ms │     no change │
│ QQuery 24 │     101.82 / 104.76 ±2.13 / 108.31 ms │        44.74 / 47.06 ±2.12 / 49.82 ms │ +2.23x faster │
│ QQuery 25 │     139.33 / 142.12 ±2.07 / 145.03 ms │     111.86 / 112.66 ±0.66 / 113.60 ms │ +1.26x faster │
│ QQuery 26 │     102.85 / 104.36 ±1.46 / 106.92 ms │        44.25 / 44.80 ±0.43 / 45.56 ms │ +2.33x faster │
│ QQuery 27 │     855.07 / 860.28 ±5.92 / 869.24 ms │     635.63 / 638.09 ±2.69 / 642.98 ms │ +1.35x faster │
│ QQuery 28 │ 3294.72 / 3319.55 ±15.87 / 3338.50 ms │  2880.01 / 2892.83 ±8.52 / 2901.50 ms │ +1.15x faster │
│ QQuery 29 │        51.36 / 55.16 ±4.19 / 63.00 ms │        44.31 / 48.25 ±3.23 / 53.94 ms │ +1.14x faster │
│ QQuery 30 │     364.87 / 372.61 ±6.21 / 383.68 ms │     327.26 / 332.61 ±5.71 / 343.22 ms │ +1.12x faster │
│ QQuery 31 │     376.21 / 388.83 ±8.13 / 400.05 ms │     318.52 / 333.08 ±7.83 / 341.33 ms │ +1.17x faster │
│ QQuery 32 │  1060.54 / 1066.45 ±6.51 / 1078.26 ms │  1024.20 / 1033.00 ±6.20 / 1043.63 ms │     no change │
│ QQuery 33 │  1499.21 / 1508.24 ±6.43 / 1517.43 ms │ 1472.33 / 1495.21 ±36.59 / 1568.14 ms │     no change │
│ QQuery 34 │ 1500.53 / 1520.34 ±16.63 / 1544.51 ms │ 1517.11 / 1572.75 ±50.18 / 1667.31 ms │     no change │
│ QQuery 35 │     396.28 / 401.72 ±5.56 / 411.12 ms │    333.54 / 362.43 ±26.23 / 407.99 ms │ +1.11x faster │
│ QQuery 36 │     122.53 / 125.77 ±1.79 / 127.67 ms │      72.09 / 92.06 ±26.10 / 141.75 ms │ +1.37x faster │
│ QQuery 37 │        49.92 / 51.62 ±1.08 / 53.12 ms │        39.65 / 41.72 ±1.22 / 43.29 ms │ +1.24x faster │
│ QQuery 38 │        76.18 / 79.10 ±2.65 / 83.91 ms │        46.99 / 50.29 ±3.50 / 56.79 ms │ +1.57x faster │
│ QQuery 39 │     219.14 / 224.20 ±3.29 / 229.08 ms │     133.55 / 145.28 ±5.95 / 150.12 ms │ +1.54x faster │
│ QQuery 40 │        24.60 / 27.54 ±2.43 / 31.83 ms │        16.77 / 18.51 ±1.14 / 19.83 ms │ +1.49x faster │
│ QQuery 41 │        20.02 / 21.87 ±1.17 / 23.52 ms │       16.58 / 29.17 ±22.07 / 73.27 ms │  1.33x slower │
│ QQuery 42 │        20.30 / 21.03 ±0.50 / 21.57 ms │        16.84 / 18.68 ±1.52 / 20.67 ms │ +1.13x faster │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary               ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)               │ 22874.95ms │
│ Total Time (morsel-splitting)   │ 20588.79ms │
│ Average Time (HEAD)             │   531.98ms │
│ Average Time (morsel-splitting) │   478.81ms │
│ Queries Faster                  │         35 │
│ Queries Slower                  │          2 │
│ Queries with No Change          │          6 │
│ Queries with Failure            │          0 │
└─────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	115.4s
Peak memory	38.1 GiB
Avg memory	32.0 GiB
CPU user	1096.1s
CPU sys	82.4s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	103.5s
Peak memory	40.4 GiB
Avg memory	32.5 GiB
CPU user	1105.5s
CPU sys	80.9s
Peak spill	0 B

File an issue against this benchmark runner

Split large files and merge small same-file ranges in the shared work queue, so idle sibling streams always have right-sized work to steal. - Target morsel size of at least 1 MiB projected size - Split: when queue depth < 2 * target_partitions and file >= 2 MiB, split in half and push the second half back - Merge: when popped file < 1 MiB, absorb adjacent same-file entries until reaching 1 MiB - Estimate projected size using per-column byte_size stats from PartitionedFile.statistics when available, otherwise fall back to file_size * (projected_cols / total_cols) - For Parquet: split the ParquetAccessPlan (row group selection) via SplittableExt instead of byte-range splitting - Statistics are scaled by 0.5 (inexact) on each split half Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Dandandan · 2026-04-15T18:08:52Z

run benchmarks

adriangbot · 2026-04-15T18:10:56Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4254372730-1309-dkp7h 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (106f736) to 961c5fc (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T18:10:56Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4254372730-1310-96wn7 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (106f736) to 961c5fc (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T18:11:03Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4254372730-1311-nxr92 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (106f736) to 961c5fc (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T18:21:47Z

Benchmark for this request failed.

Last 20 lines of output:

Click to expand

BENCHMARK: tpch
QUERY: All
DATAFUSION_DIR: /workspace/datafusion-base
BRANCH_NAME: HEAD
DATA_DIR: /workspace/datafusion-bench/benchmarks/data
RESULTS_DIR: /workspace/datafusion-bench/benchmarks/results/HEAD
CARGO_COMMAND: cargo run --release
PREFER_HASH_JOIN: true
SIMULATE_LATENCY: false
***************************
RESULTS_FILE: /workspace/datafusion-bench/benchmarks/results/HEAD/tpch_sf1.json
Running tpch benchmark...
+ cargo run --release --bin dfbench -- tpch --iterations 5 --path /workspace/datafusion-bench/benchmarks/data/tpch_sf1 --scale-factor 1 --prefer_hash_join true --format parquet -o /workspace/datafusion-bench/benchmarks/results/HEAD/tpch_sf1.json
    Finished `release` profile [optimized] target(s) in 0.22s
     Running `/workspace/datafusion-base/target/release/dfbench tpch --iterations 5 --path /workspace/datafusion-bench/benchmarks/data/tpch_sf1 --scale-factor 1 --prefer_hash_join true --format parquet -o /workspace/datafusion-bench/benchmarks/results/HEAD/tpch_sf1.json`
error: unexpected argument '--scale-factor' found

Usage: dfbench tpch --path <PATH> --iterations <ITERATIONS>

For more information, try '--help'.

File an issue against this benchmark runner

Dandandan · 2026-04-15T18:25:05Z

run benchmarks

adriangbot · 2026-04-15T18:26:21Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and morsel-splitting
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃                      morsel-splitting ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.18 / 4.34 ±6.24 / 16.82 ms │          1.18 / 4.34 ±6.23 / 16.79 ms │     no change │
│ QQuery 1  │        14.23 / 14.32 ±0.10 / 14.46 ms │        11.86 / 12.09 ±0.17 / 12.35 ms │ +1.18x faster │
│ QQuery 2  │        43.28 / 43.62 ±0.30 / 44.13 ms │        34.98 / 35.54 ±0.59 / 36.60 ms │ +1.23x faster │
│ QQuery 3  │        42.70 / 45.28 ±1.63 / 46.86 ms │        31.02 / 33.49 ±2.37 / 36.64 ms │ +1.35x faster │
│ QQuery 4  │     279.24 / 286.60 ±6.68 / 296.88 ms │     254.37 / 260.46 ±3.60 / 264.41 ms │ +1.10x faster │
│ QQuery 5  │     335.96 / 339.17 ±1.87 / 341.36 ms │     297.49 / 302.76 ±3.01 / 306.50 ms │ +1.12x faster │
│ QQuery 6  │           5.00 / 5.74 ±0.64 / 6.74 ms │           4.20 / 5.25 ±1.00 / 6.70 ms │ +1.09x faster │
│ QQuery 7  │        16.92 / 17.37 ±0.48 / 18.24 ms │        13.05 / 13.41 ±0.40 / 14.13 ms │ +1.30x faster │
│ QQuery 8  │     413.12 / 418.63 ±5.66 / 429.03 ms │     365.29 / 367.26 ±2.74 / 372.59 ms │ +1.14x faster │
│ QQuery 9  │     616.09 / 626.68 ±9.69 / 642.42 ms │    469.04 / 492.05 ±16.27 / 519.86 ms │ +1.27x faster │
│ QQuery 10 │        89.96 / 91.19 ±1.01 / 92.42 ms │        72.83 / 75.89 ±2.90 / 80.68 ms │ +1.20x faster │
│ QQuery 11 │     101.88 / 102.99 ±0.84 / 104.38 ms │        83.18 / 83.43 ±0.18 / 83.66 ms │ +1.23x faster │
│ QQuery 12 │     333.46 / 339.04 ±3.16 / 342.17 ms │     287.71 / 293.76 ±4.77 / 300.12 ms │ +1.15x faster │
│ QQuery 13 │    441.23 / 461.03 ±11.62 / 475.75 ms │     416.08 / 423.19 ±6.65 / 435.10 ms │ +1.09x faster │
│ QQuery 14 │     337.62 / 340.76 ±2.88 / 345.48 ms │     295.54 / 299.54 ±2.26 / 301.62 ms │ +1.14x faster │
│ QQuery 15 │    338.35 / 351.48 ±13.97 / 373.18 ms │     317.20 / 327.26 ±8.68 / 342.26 ms │ +1.07x faster │
│ QQuery 16 │    695.22 / 711.45 ±16.40 / 738.57 ms │     656.60 / 666.61 ±6.99 / 674.94 ms │ +1.07x faster │
│ QQuery 17 │     683.97 / 690.99 ±5.82 / 697.56 ms │    664.32 / 682.90 ±24.17 / 729.26 ms │     no change │
│ QQuery 18 │ 1371.32 / 1442.04 ±45.51 / 1502.42 ms │  1311.25 / 1318.56 ±5.18 / 1324.10 ms │ +1.09x faster │
│ QQuery 19 │        36.65 / 37.65 ±0.89 / 39.32 ms │        29.15 / 31.12 ±2.71 / 36.45 ms │ +1.21x faster │
│ QQuery 20 │    715.12 / 735.04 ±21.37 / 761.83 ms │     472.78 / 476.27 ±3.80 / 483.58 ms │ +1.54x faster │
│ QQuery 21 │     757.84 / 762.12 ±2.64 / 765.52 ms │     557.22 / 558.93 ±1.86 / 562.33 ms │ +1.36x faster │
│ QQuery 22 │  1120.02 / 1126.17 ±5.89 / 1137.13 ms │     988.63 / 993.20 ±2.95 / 997.90 ms │ +1.13x faster │
│ QQuery 23 │ 3026.17 / 3054.15 ±17.84 / 3080.05 ms │  3120.73 / 3127.58 ±6.99 / 3136.57 ms │     no change │
│ QQuery 24 │      98.13 / 100.96 ±2.52 / 105.65 ms │        43.50 / 46.02 ±2.00 / 48.85 ms │ +2.19x faster │
│ QQuery 25 │     138.39 / 140.71 ±2.17 / 144.50 ms │     108.36 / 109.23 ±0.85 / 110.65 ms │ +1.29x faster │
│ QQuery 26 │       95.62 / 99.35 ±2.29 / 102.76 ms │        42.35 / 43.04 ±0.48 / 43.64 ms │ +2.31x faster │
│ QQuery 27 │     847.12 / 850.35 ±4.22 / 858.55 ms │     626.04 / 628.12 ±1.42 / 630.35 ms │ +1.35x faster │
│ QQuery 28 │ 3237.07 / 3266.83 ±15.54 / 3279.63 ms │ 2822.60 / 2847.27 ±14.87 / 2863.74 ms │ +1.15x faster │
│ QQuery 29 │        49.94 / 53.72 ±4.29 / 61.75 ms │        42.34 / 46.65 ±3.56 / 51.87 ms │ +1.15x faster │
│ QQuery 30 │     350.54 / 359.24 ±8.24 / 372.83 ms │     316.36 / 322.87 ±6.67 / 332.03 ms │ +1.11x faster │
│ QQuery 31 │    349.09 / 381.38 ±16.75 / 397.66 ms │     317.07 / 325.51 ±5.42 / 332.16 ms │ +1.17x faster │
│ QQuery 32 │ 1031.46 / 1047.38 ±19.55 / 1080.57 ms │   997.61 / 1004.78 ±9.34 / 1022.55 ms │     no change │
│ QQuery 33 │  1408.85 / 1420.19 ±7.74 / 1429.85 ms │ 1417.36 / 1427.03 ±15.01 / 1456.82 ms │     no change │
│ QQuery 34 │  1428.59 / 1434.52 ±3.98 / 1438.25 ms │ 1413.43 / 1438.55 ±16.09 / 1454.84 ms │     no change │
│ QQuery 35 │     371.17 / 375.87 ±4.26 / 381.49 ms │     305.77 / 308.36 ±2.53 / 312.88 ms │ +1.22x faster │
│ QQuery 36 │     115.83 / 118.34 ±1.49 / 120.10 ms │        66.18 / 71.29 ±3.26 / 75.51 ms │ +1.66x faster │
│ QQuery 37 │        46.86 / 48.83 ±1.59 / 50.77 ms │        37.81 / 40.91 ±2.23 / 43.48 ms │ +1.19x faster │
│ QQuery 38 │        73.97 / 76.25 ±2.46 / 80.74 ms │        43.33 / 47.34 ±2.72 / 51.84 ms │ +1.61x faster │
│ QQuery 39 │     201.79 / 210.87 ±7.69 / 222.19 ms │     127.34 / 132.57 ±6.10 / 144.01 ms │ +1.59x faster │
│ QQuery 40 │        21.06 / 23.52 ±2.34 / 27.31 ms │        16.27 / 17.39 ±0.86 / 18.40 ms │ +1.35x faster │
│ QQuery 41 │        19.01 / 21.41 ±2.36 / 25.42 ms │        15.27 / 16.59 ±1.22 / 18.24 ms │ +1.29x faster │
│ QQuery 42 │        18.65 / 19.29 ±0.41 / 19.86 ms │        14.05 / 14.23 ±0.19 / 14.60 ms │ +1.36x faster │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary               ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)               │ 22096.86ms │
│ Total Time (morsel-splitting)   │ 19772.64ms │
│ Average Time (HEAD)             │   513.88ms │
│ Average Time (morsel-splitting) │   459.83ms │
│ Queries Faster                  │         37 │
│ Queries Slower                  │          0 │
│ Queries with No Change          │          6 │
│ Queries with Failure            │          0 │
└─────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	111.5s
Peak memory	41.2 GiB
Avg memory	30.6 GiB
CPU user	1056.9s
CPU sys	80.0s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	99.8s
Peak memory	38.0 GiB
Avg memory	32.6 GiB
CPU user	1071.6s
CPU sys	72.7s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-04-15T18:27:09Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4254461570-1314-7ddmm 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (40ca776) to 5c653be (merge-base) diff using: tpch
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T18:27:13Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4254461570-1313-9jqjr 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (40ca776) to 5c653be (merge-base) diff using: tpcds
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T18:27:13Z

🤖 Benchmark running (GKE) | trigger
Instance: c4a-highmem-16 (12 vCPU / 65 GiB) | Linux bench-c4254461570-1312-h2gl2 6.12.55+ #1 SMP Sun Feb 1 08:59:41 UTC 2026 aarch64 GNU/Linux

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Comparing morsel-splitting (40ca776) to 5c653be (merge-base) diff using: clickbench_partitioned
Results will be posted here when complete

File an issue against this benchmark runner

adriangbot · 2026-04-15T18:27:25Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and morsel-splitting
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                     HEAD ┃                         morsel-splitting ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │              6.75 / 7.33 ±0.77 / 8.85 ms │              6.81 / 7.27 ±0.75 / 8.77 ms │     no change │
│ QQuery 2  │        146.44 / 147.50 ±0.88 / 149.10 ms │        147.20 / 148.83 ±1.14 / 150.40 ms │     no change │
│ QQuery 3  │        115.11 / 115.72 ±0.59 / 116.80 ms │        115.74 / 116.78 ±0.94 / 118.28 ms │     no change │
│ QQuery 4  │    1416.36 / 1447.60 ±15.92 / 1460.62 ms │    1416.56 / 1439.94 ±17.29 / 1464.04 ms │     no change │
│ QQuery 5  │        174.15 / 175.76 ±1.36 / 178.27 ms │        178.31 / 181.57 ±3.70 / 188.28 ms │     no change │
│ QQuery 6  │       872.80 / 887.21 ±13.35 / 906.38 ms │       847.13 / 892.08 ±25.59 / 924.88 ms │     no change │
│ QQuery 7  │        346.78 / 350.78 ±2.66 / 354.56 ms │        343.86 / 347.88 ±2.82 / 352.00 ms │     no change │
│ QQuery 8  │        116.40 / 118.57 ±1.69 / 120.74 ms │        118.06 / 118.85 ±0.56 / 119.80 ms │     no change │
│ QQuery 9  │        102.53 / 104.88 ±2.65 / 109.33 ms │        112.35 / 115.97 ±1.81 / 116.92 ms │  1.11x slower │
│ QQuery 10 │        108.18 / 108.65 ±0.34 / 109.03 ms │        109.91 / 111.64 ±1.05 / 112.96 ms │     no change │
│ QQuery 11 │     1014.37 / 1027.72 ±8.03 / 1037.56 ms │     1032.98 / 1038.74 ±6.61 / 1050.47 ms │     no change │
│ QQuery 12 │           46.58 / 48.69 ±1.21 / 50.12 ms │           47.19 / 48.51 ±0.78 / 49.35 ms │     no change │
│ QQuery 13 │        408.43 / 410.96 ±2.11 / 414.07 ms │        409.86 / 413.11 ±2.96 / 418.58 ms │     no change │
│ QQuery 14 │     1016.81 / 1025.48 ±6.61 / 1033.70 ms │     1025.00 / 1034.31 ±6.02 / 1043.15 ms │     no change │
│ QQuery 15 │           16.96 / 17.88 ±0.76 / 18.87 ms │           16.80 / 17.43 ±0.93 / 19.25 ms │     no change │
│ QQuery 16 │              7.42 / 7.74 ±0.26 / 8.19 ms │              7.23 / 8.17 ±0.82 / 9.55 ms │  1.06x slower │
│ QQuery 17 │        233.06 / 236.98 ±5.58 / 247.67 ms │        233.64 / 234.66 ±1.02 / 236.59 ms │     no change │
│ QQuery 18 │        130.98 / 131.66 ±0.53 / 132.58 ms │        137.42 / 138.25 ±0.66 / 139.25 ms │  1.05x slower │
│ QQuery 19 │        158.79 / 161.06 ±1.80 / 163.08 ms │        158.05 / 159.11 ±0.84 / 160.59 ms │     no change │
│ QQuery 20 │           14.35 / 14.82 ±0.42 / 15.57 ms │           15.60 / 15.99 ±0.25 / 16.32 ms │  1.08x slower │
│ QQuery 21 │           19.83 / 20.43 ±0.45 / 21.15 ms │           21.73 / 21.99 ±0.32 / 22.57 ms │  1.08x slower │
│ QQuery 22 │        503.19 / 507.21 ±2.72 / 511.13 ms │        496.06 / 500.64 ±4.27 / 506.20 ms │     no change │
│ QQuery 23 │        913.75 / 920.21 ±7.24 / 933.36 ms │        927.65 / 933.50 ±5.57 / 943.38 ms │     no change │
│ QQuery 24 │        390.73 / 393.67 ±2.98 / 398.21 ms │        397.76 / 399.97 ±1.77 / 402.96 ms │     no change │
│ QQuery 25 │        345.66 / 348.65 ±1.84 / 350.69 ms │        346.99 / 348.48 ±1.05 / 349.68 ms │     no change │
│ QQuery 26 │           81.95 / 83.60 ±0.95 / 84.60 ms │           84.30 / 85.93 ±1.19 / 87.67 ms │     no change │
│ QQuery 27 │              7.23 / 7.56 ±0.39 / 8.32 ms │              6.99 / 7.51 ±0.70 / 8.90 ms │     no change │
│ QQuery 28 │        150.60 / 150.90 ±0.30 / 151.47 ms │        160.72 / 162.10 ±1.29 / 163.69 ms │  1.07x slower │
│ QQuery 29 │        283.95 / 288.18 ±2.68 / 291.47 ms │        287.90 / 288.99 ±0.70 / 290.00 ms │     no change │
│ QQuery 30 │           44.67 / 46.58 ±1.27 / 48.40 ms │           43.65 / 45.73 ±1.53 / 47.84 ms │     no change │
│ QQuery 31 │        172.72 / 174.33 ±1.34 / 176.23 ms │        175.77 / 177.92 ±1.41 / 179.95 ms │     no change │
│ QQuery 32 │           58.27 / 59.19 ±0.57 / 60.02 ms │           58.87 / 59.99 ±0.78 / 61.18 ms │     no change │
│ QQuery 33 │        144.38 / 145.39 ±1.10 / 147.47 ms │        143.98 / 145.57 ±1.24 / 147.07 ms │     no change │
│ QQuery 34 │              7.28 / 7.56 ±0.29 / 8.07 ms │              7.20 / 7.44 ±0.19 / 7.77 ms │     no change │
│ QQuery 35 │        110.67 / 111.58 ±0.64 / 112.67 ms │        113.57 / 115.00 ±1.22 / 116.69 ms │     no change │
│ QQuery 36 │              7.00 / 7.12 ±0.12 / 7.32 ms │              6.88 / 7.00 ±0.13 / 7.24 ms │     no change │
│ QQuery 37 │              8.49 / 8.83 ±0.31 / 9.38 ms │             9.08 / 9.83 ±0.98 / 11.73 ms │  1.11x slower │
│ QQuery 38 │           88.08 / 92.09 ±3.84 / 97.89 ms │          88.48 / 94.17 ±3.94 / 100.87 ms │     no change │
│ QQuery 39 │        127.82 / 131.00 ±3.25 / 136.35 ms │        135.23 / 136.19 ±0.74 / 137.43 ms │     no change │
│ QQuery 40 │        111.03 / 117.30 ±5.67 / 127.30 ms │        111.47 / 119.24 ±6.27 / 130.10 ms │     no change │
│ QQuery 41 │           14.62 / 15.84 ±1.09 / 17.87 ms │           15.62 / 16.75 ±0.89 / 18.14 ms │  1.06x slower │
│ QQuery 42 │        108.85 / 110.26 ±1.07 / 111.35 ms │        110.00 / 112.49 ±1.29 / 113.47 ms │     no change │
│ QQuery 43 │              6.27 / 6.36 ±0.10 / 6.55 ms │              6.32 / 6.39 ±0.05 / 6.46 ms │     no change │
│ QQuery 44 │           12.19 / 12.67 ±0.48 / 13.54 ms │           18.72 / 21.02 ±1.74 / 24.10 ms │  1.66x slower │
│ QQuery 45 │           51.40 / 53.12 ±1.44 / 55.67 ms │           51.47 / 53.53 ±1.39 / 55.83 ms │     no change │
│ QQuery 46 │             9.10 / 9.62 ±0.61 / 10.56 ms │              8.52 / 8.99 ±0.44 / 9.81 ms │ +1.07x faster │
│ QQuery 47 │        754.24 / 760.69 ±5.87 / 770.02 ms │        766.85 / 773.59 ±3.93 / 777.25 ms │     no change │
│ QQuery 48 │        288.52 / 296.27 ±5.50 / 304.50 ms │        293.81 / 300.41 ±3.62 / 304.73 ms │     no change │
│ QQuery 49 │        253.17 / 256.26 ±2.12 / 259.33 ms │        258.51 / 260.16 ±2.12 / 264.26 ms │     no change │
│ QQuery 50 │        229.37 / 232.99 ±3.25 / 238.64 ms │        228.28 / 232.75 ±2.91 / 235.58 ms │     no change │
│ QQuery 51 │        182.38 / 185.74 ±1.99 / 188.66 ms │        183.01 / 186.46 ±2.22 / 189.23 ms │     no change │
│ QQuery 52 │        108.77 / 109.68 ±0.87 / 111.17 ms │        110.41 / 112.06 ±1.06 / 113.33 ms │     no change │
│ QQuery 53 │        104.08 / 104.79 ±0.94 / 106.63 ms │        104.19 / 105.60 ±1.05 / 106.75 ms │     no change │
│ QQuery 54 │        148.72 / 150.34 ±1.32 / 151.62 ms │        151.10 / 152.09 ±0.84 / 153.41 ms │     no change │
│ QQuery 55 │        107.67 / 109.95 ±1.59 / 112.20 ms │        108.55 / 109.86 ±0.69 / 110.42 ms │     no change │
│ QQuery 56 │        143.18 / 144.51 ±1.17 / 146.71 ms │        144.45 / 145.32 ±0.87 / 146.75 ms │     no change │
│ QQuery 57 │        177.46 / 179.29 ±1.17 / 180.47 ms │        177.65 / 179.71 ±1.81 / 182.93 ms │     no change │
│ QQuery 58 │       289.75 / 304.80 ±12.39 / 325.46 ms │        299.32 / 307.46 ±8.39 / 322.65 ms │     no change │
│ QQuery 59 │        201.30 / 203.11 ±1.55 / 205.01 ms │        203.04 / 205.15 ±1.49 / 206.95 ms │     no change │
│ QQuery 60 │        145.58 / 147.85 ±1.34 / 149.20 ms │        147.33 / 149.01 ±1.01 / 150.48 ms │     no change │
│ QQuery 61 │           13.34 / 13.59 ±0.17 / 13.87 ms │           13.44 / 13.58 ±0.19 / 13.96 ms │     no change │
│ QQuery 62 │     935.16 / 1016.15 ±63.73 / 1103.21 ms │     937.19 / 1005.35 ±43.26 / 1067.83 ms │     no change │
│ QQuery 63 │        104.96 / 108.28 ±2.31 / 111.04 ms │        105.82 / 108.62 ±2.88 / 113.94 ms │     no change │
│ QQuery 64 │        700.03 / 705.23 ±3.30 / 710.01 ms │        703.87 / 709.02 ±5.36 / 718.84 ms │     no change │
│ QQuery 65 │        262.83 / 267.67 ±3.48 / 273.69 ms │        267.77 / 272.57 ±3.13 / 275.66 ms │     no change │
│ QQuery 66 │       239.57 / 260.32 ±12.20 / 272.73 ms │       253.44 / 266.78 ±14.91 / 294.87 ms │     no change │
│ QQuery 67 │        320.62 / 329.20 ±9.79 / 347.92 ms │        319.78 / 326.40 ±5.71 / 334.23 ms │     no change │
│ QQuery 68 │           10.60 / 12.58 ±1.31 / 14.01 ms │            8.81 / 10.12 ±0.91 / 11.18 ms │ +1.24x faster │
│ QQuery 69 │        103.14 / 105.82 ±1.65 / 107.62 ms │        106.39 / 107.40 ±0.96 / 108.88 ms │     no change │
│ QQuery 70 │       340.48 / 366.67 ±17.71 / 391.28 ms │        331.13 / 349.18 ±9.92 / 358.59 ms │     no change │
│ QQuery 71 │        136.98 / 137.91 ±1.20 / 140.28 ms │        138.56 / 139.86 ±1.01 / 141.41 ms │     no change │
│ QQuery 72 │        629.00 / 640.08 ±8.90 / 651.18 ms │        632.73 / 642.96 ±6.73 / 653.55 ms │     no change │
│ QQuery 73 │              7.16 / 8.30 ±0.99 / 9.98 ms │              6.91 / 8.63 ±0.96 / 9.69 ms │     no change │
│ QQuery 74 │        638.28 / 646.12 ±8.44 / 660.62 ms │        645.38 / 655.60 ±6.45 / 665.08 ms │     no change │
│ QQuery 75 │        277.94 / 280.42 ±1.32 / 281.65 ms │        283.86 / 285.46 ±1.59 / 287.97 ms │     no change │
│ QQuery 76 │        134.75 / 137.11 ±1.92 / 140.54 ms │        137.12 / 138.19 ±1.27 / 140.42 ms │     no change │
│ QQuery 77 │        190.49 / 192.16 ±1.29 / 194.40 ms │        192.48 / 196.23 ±2.56 / 199.73 ms │     no change │
│ QQuery 78 │        347.35 / 354.15 ±4.61 / 361.04 ms │        353.61 / 357.93 ±3.37 / 363.88 ms │     no change │
│ QQuery 79 │        241.16 / 243.51 ±2.01 / 246.23 ms │        245.61 / 249.15 ±2.15 / 252.25 ms │     no change │
│ QQuery 80 │        326.24 / 328.15 ±2.16 / 332.20 ms │        326.86 / 330.88 ±3.53 / 335.77 ms │     no change │
│ QQuery 81 │           26.24 / 27.35 ±0.64 / 28.12 ms │           27.51 / 28.66 ±1.10 / 30.76 ms │     no change │
│ QQuery 82 │        200.35 / 201.63 ±1.27 / 203.46 ms │        202.01 / 203.96 ±1.68 / 206.79 ms │     no change │
│ QQuery 83 │           39.42 / 41.07 ±2.22 / 45.44 ms │           39.86 / 41.07 ±0.67 / 41.90 ms │     no change │
│ QQuery 84 │           48.94 / 50.44 ±1.14 / 52.24 ms │           50.00 / 51.50 ±1.44 / 53.71 ms │     no change │
│ QQuery 85 │        148.60 / 150.10 ±0.96 / 151.55 ms │        151.51 / 153.68 ±2.01 / 156.84 ms │     no change │
│ QQuery 86 │           39.26 / 40.83 ±1.01 / 42.37 ms │           41.06 / 42.40 ±1.77 / 45.85 ms │     no change │
│ QQuery 87 │           87.60 / 90.56 ±2.27 / 94.42 ms │          90.46 / 96.28 ±4.12 / 102.79 ms │  1.06x slower │
│ QQuery 88 │        100.95 / 101.51 ±0.52 / 102.41 ms │        114.79 / 116.36 ±1.34 / 118.80 ms │  1.15x slower │
│ QQuery 89 │        119.42 / 121.24 ±0.93 / 121.88 ms │        120.77 / 121.58 ±0.82 / 123.08 ms │     no change │
│ QQuery 90 │           24.12 / 24.46 ±0.30 / 25.02 ms │           25.86 / 27.36 ±0.94 / 28.51 ms │  1.12x slower │
│ QQuery 91 │           64.79 / 65.70 ±0.51 / 66.28 ms │           65.11 / 67.75 ±1.45 / 69.07 ms │     no change │
│ QQuery 92 │           58.76 / 60.10 ±1.53 / 63.05 ms │           59.23 / 60.50 ±0.64 / 61.02 ms │     no change │
│ QQuery 93 │        189.89 / 192.74 ±1.63 / 194.56 ms │        193.17 / 194.13 ±0.80 / 195.27 ms │     no change │
│ QQuery 94 │           62.94 / 64.27 ±1.14 / 66.09 ms │           62.77 / 64.26 ±0.85 / 65.11 ms │     no change │
│ QQuery 95 │        130.40 / 132.01 ±1.12 / 133.39 ms │        130.82 / 132.40 ±1.33 / 133.99 ms │     no change │
│ QQuery 96 │           70.08 / 73.76 ±2.71 / 76.79 ms │           75.39 / 76.45 ±1.07 / 78.11 ms │     no change │
│ QQuery 97 │        128.74 / 130.13 ±1.30 / 132.24 ms │        129.72 / 131.19 ±1.17 / 133.22 ms │     no change │
│ QQuery 98 │        153.20 / 158.30 ±3.21 / 162.37 ms │        159.65 / 161.17 ±1.95 / 165.03 ms │     no change │
│ QQuery 99 │ 10907.06 / 10956.14 ±43.60 / 11025.31 ms │ 10906.51 / 10951.28 ±51.20 / 11041.56 ms │     no change │
└───────────┴──────────────────────────────────────────┴──────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary               ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)               │ 32398.32ms │
│ Total Time (morsel-splitting)   │ 32590.95ms │
│ Average Time (HEAD)             │   327.26ms │
│ Average Time (morsel-splitting) │   329.20ms │
│ Queries Faster                  │          2 │
│ Queries Slower                  │         12 │
│ Queries with No Change          │         85 │
│ Queries with Failure            │          0 │
└─────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	162.3s
Peak memory	5.0 GiB
Avg memory	4.4 GiB
CPU user	268.0s
CPU sys	18.0s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	163.2s
Peak memory	5.0 GiB
Avg memory	4.3 GiB
CPU user	288.4s
CPU sys	19.2s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-04-15T18:43:13Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and morsel-splitting
--------------------
Benchmark tpcds_sf1.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                     HEAD ┃                         morsel-splitting ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1  │              6.62 / 7.11 ±0.85 / 8.80 ms │              6.54 / 6.96 ±0.79 / 8.55 ms │     no change │
│ QQuery 2  │        143.88 / 144.90 ±0.58 / 145.48 ms │        144.59 / 146.01 ±0.95 / 147.51 ms │     no change │
│ QQuery 3  │        113.11 / 113.93 ±0.87 / 115.53 ms │        113.25 / 114.51 ±0.91 / 115.60 ms │     no change │
│ QQuery 4  │    1301.21 / 1325.12 ±18.98 / 1350.62 ms │    1303.98 / 1333.72 ±19.50 / 1361.03 ms │     no change │
│ QQuery 5  │        169.66 / 171.06 ±0.86 / 171.98 ms │        176.07 / 179.20 ±2.74 / 183.98 ms │     no change │
│ QQuery 6  │       829.03 / 842.77 ±12.38 / 861.05 ms │       852.91 / 866.79 ±12.40 / 888.42 ms │     no change │
│ QQuery 7  │        340.03 / 342.65 ±1.66 / 344.65 ms │        343.76 / 347.80 ±3.57 / 354.33 ms │     no change │
│ QQuery 8  │        115.64 / 117.46 ±1.28 / 119.00 ms │        117.31 / 118.56 ±1.00 / 119.59 ms │     no change │
│ QQuery 9  │        100.68 / 104.06 ±3.40 / 108.25 ms │        111.64 / 119.37 ±5.87 / 129.13 ms │  1.15x slower │
│ QQuery 10 │        105.73 / 106.73 ±1.06 / 108.35 ms │        111.80 / 113.05 ±1.27 / 115.47 ms │  1.06x slower │
│ QQuery 11 │       905.88 / 915.74 ±10.80 / 936.19 ms │        930.51 / 940.08 ±6.37 / 949.71 ms │     no change │
│ QQuery 12 │           43.37 / 45.75 ±1.39 / 47.13 ms │           45.16 / 46.87 ±1.34 / 48.55 ms │     no change │
│ QQuery 13 │        397.93 / 401.06 ±1.85 / 403.72 ms │        397.41 / 400.94 ±3.85 / 408.01 ms │     no change │
│ QQuery 14 │     988.55 / 1004.96 ±11.50 / 1017.80 ms │     1011.35 / 1024.02 ±7.04 / 1030.72 ms │     no change │
│ QQuery 15 │           15.61 / 16.17 ±0.61 / 17.28 ms │           15.81 / 16.90 ±1.64 / 20.17 ms │     no change │
│ QQuery 16 │              7.10 / 7.45 ±0.33 / 8.01 ms │              7.45 / 8.13 ±0.76 / 9.16 ms │  1.09x slower │
│ QQuery 17 │        227.78 / 228.58 ±0.52 / 229.36 ms │        228.93 / 231.06 ±1.90 / 233.76 ms │     no change │
│ QQuery 18 │        127.47 / 127.84 ±0.25 / 128.21 ms │        131.34 / 134.98 ±2.04 / 137.09 ms │  1.06x slower │
│ QQuery 19 │        153.15 / 154.25 ±0.86 / 155.39 ms │        155.34 / 156.76 ±1.01 / 158.07 ms │     no change │
│ QQuery 20 │           14.03 / 14.35 ±0.27 / 14.80 ms │           14.23 / 15.04 ±0.45 / 15.55 ms │     no change │
│ QQuery 21 │           18.94 / 19.46 ±0.32 / 19.75 ms │           20.52 / 20.83 ±0.33 / 21.42 ms │  1.07x slower │
│ QQuery 22 │        485.77 / 492.79 ±9.64 / 511.08 ms │        484.09 / 487.80 ±2.74 / 492.43 ms │     no change │
│ QQuery 23 │        869.51 / 878.22 ±8.62 / 894.29 ms │        882.72 / 888.48 ±4.90 / 894.09 ms │     no change │
│ QQuery 24 │       382.92 / 389.14 ±11.10 / 411.33 ms │        382.92 / 387.92 ±4.23 / 393.36 ms │     no change │
│ QQuery 25 │        339.59 / 341.14 ±1.37 / 343.26 ms │        338.58 / 342.10 ±1.96 / 344.38 ms │     no change │
│ QQuery 26 │           80.50 / 82.24 ±2.09 / 86.07 ms │           81.11 / 83.10 ±1.80 / 85.87 ms │     no change │
│ QQuery 27 │              6.80 / 7.04 ±0.17 / 7.27 ms │              6.73 / 7.43 ±0.69 / 8.61 ms │  1.06x slower │
│ QQuery 28 │        148.37 / 149.80 ±1.50 / 152.65 ms │        158.21 / 159.71 ±0.93 / 160.92 ms │  1.07x slower │
│ QQuery 29 │        278.22 / 280.70 ±2.68 / 285.47 ms │        280.67 / 282.64 ±2.24 / 286.27 ms │     no change │
│ QQuery 30 │           44.19 / 45.16 ±0.89 / 46.78 ms │           42.10 / 44.88 ±1.50 / 46.33 ms │     no change │
│ QQuery 31 │        168.43 / 170.51 ±1.22 / 171.86 ms │        171.94 / 175.75 ±1.94 / 177.12 ms │     no change │
│ QQuery 32 │           56.37 / 57.28 ±0.73 / 58.49 ms │           56.83 / 57.68 ±0.49 / 58.17 ms │     no change │
│ QQuery 33 │        138.53 / 140.55 ±1.32 / 142.12 ms │        140.80 / 143.66 ±2.11 / 146.53 ms │     no change │
│ QQuery 34 │              7.10 / 7.48 ±0.38 / 7.98 ms │              6.74 / 7.01 ±0.25 / 7.43 ms │ +1.07x faster │
│ QQuery 35 │        108.60 / 109.58 ±0.93 / 111.27 ms │        108.22 / 111.29 ±1.71 / 112.95 ms │     no change │
│ QQuery 36 │              6.63 / 7.08 ±0.70 / 8.46 ms │              6.42 / 6.62 ±0.18 / 6.84 ms │ +1.07x faster │
│ QQuery 37 │              8.43 / 8.76 ±0.24 / 9.05 ms │             9.08 / 9.95 ±0.91 / 11.48 ms │  1.14x slower │
│ QQuery 38 │           86.76 / 89.34 ±3.20 / 95.38 ms │           87.19 / 91.88 ±3.86 / 98.72 ms │     no change │
│ QQuery 39 │        126.39 / 127.66 ±0.99 / 128.86 ms │        129.75 / 131.90 ±1.42 / 133.65 ms │     no change │
│ QQuery 40 │        108.59 / 117.72 ±8.35 / 130.95 ms │        108.68 / 115.06 ±5.09 / 122.57 ms │     no change │
│ QQuery 41 │           14.29 / 16.50 ±1.64 / 19.04 ms │           14.46 / 14.98 ±0.49 / 15.58 ms │ +1.10x faster │
│ QQuery 42 │        107.44 / 108.95 ±1.72 / 112.05 ms │        109.50 / 110.24 ±0.54 / 110.79 ms │     no change │
│ QQuery 43 │              6.11 / 6.25 ±0.13 / 6.50 ms │              5.85 / 5.99 ±0.20 / 6.38 ms │     no change │
│ QQuery 44 │           11.66 / 12.37 ±0.81 / 13.96 ms │           17.30 / 20.07 ±1.98 / 23.05 ms │  1.62x slower │
│ QQuery 45 │           50.63 / 52.48 ±1.55 / 55.28 ms │           50.57 / 51.30 ±0.52 / 52.16 ms │     no change │
│ QQuery 46 │              8.68 / 9.01 ±0.33 / 9.54 ms │              8.09 / 8.80 ±0.62 / 9.72 ms │     no change │
│ QQuery 47 │        692.05 / 701.56 ±5.76 / 707.80 ms │        698.31 / 699.36 ±0.76 / 700.65 ms │     no change │
│ QQuery 48 │        286.81 / 291.80 ±2.82 / 295.05 ms │        282.33 / 286.93 ±3.74 / 291.55 ms │     no change │
│ QQuery 49 │        251.96 / 255.59 ±1.86 / 257.07 ms │        256.46 / 259.38 ±3.08 / 264.48 ms │     no change │
│ QQuery 50 │        218.42 / 226.25 ±4.87 / 230.31 ms │        218.17 / 222.29 ±3.34 / 228.00 ms │     no change │
│ QQuery 51 │        179.65 / 181.62 ±1.53 / 183.29 ms │        180.89 / 183.12 ±1.67 / 185.63 ms │     no change │
│ QQuery 52 │        106.32 / 108.04 ±1.39 / 110.11 ms │        106.76 / 110.38 ±3.52 / 115.39 ms │     no change │
│ QQuery 53 │        101.51 / 103.73 ±3.04 / 109.70 ms │        103.10 / 103.58 ±0.37 / 104.21 ms │     no change │
│ QQuery 54 │        146.06 / 147.91 ±1.48 / 149.38 ms │        148.08 / 149.90 ±1.01 / 151.17 ms │     no change │
│ QQuery 55 │        105.71 / 108.18 ±1.62 / 110.62 ms │        106.89 / 108.23 ±1.16 / 110.24 ms │     no change │
│ QQuery 56 │        140.05 / 141.95 ±1.50 / 143.77 ms │        141.57 / 143.20 ±1.26 / 144.76 ms │     no change │
│ QQuery 57 │        170.60 / 173.38 ±1.70 / 175.90 ms │        171.20 / 175.35 ±2.56 / 178.17 ms │     no change │
│ QQuery 58 │        295.99 / 299.91 ±3.30 / 305.43 ms │        287.60 / 296.34 ±6.20 / 306.15 ms │     no change │
│ QQuery 59 │        195.65 / 197.41 ±1.15 / 199.05 ms │        197.83 / 200.85 ±3.22 / 206.75 ms │     no change │
│ QQuery 60 │        141.97 / 143.83 ±1.09 / 145.06 ms │        143.41 / 145.69 ±1.29 / 147.02 ms │     no change │
│ QQuery 61 │           13.14 / 13.30 ±0.15 / 13.53 ms │           12.83 / 13.08 ±0.19 / 13.34 ms │     no change │
│ QQuery 62 │      902.10 / 960.67 ±75.42 / 1109.08 ms │       898.01 / 917.73 ±23.60 / 953.33 ms │     no change │
│ QQuery 63 │        102.59 / 106.74 ±3.00 / 111.24 ms │        104.12 / 105.78 ±1.26 / 107.41 ms │     no change │
│ QQuery 64 │        680.01 / 685.19 ±4.80 / 693.67 ms │        686.03 / 691.05 ±4.34 / 696.93 ms │     no change │
│ QQuery 65 │        248.24 / 250.93 ±1.94 / 252.99 ms │        253.26 / 256.66 ±2.22 / 260.18 ms │     no change │
│ QQuery 66 │       230.79 / 244.39 ±13.32 / 265.71 ms │        244.07 / 256.67 ±8.67 / 266.51 ms │  1.05x slower │
│ QQuery 67 │        314.55 / 318.38 ±3.74 / 325.42 ms │        309.34 / 315.28 ±5.70 / 324.76 ms │     no change │
│ QQuery 68 │            9.38 / 11.32 ±1.18 / 12.67 ms │            9.05 / 10.63 ±0.92 / 11.63 ms │ +1.07x faster │
│ QQuery 69 │        101.57 / 102.89 ±1.04 / 104.28 ms │        103.43 / 106.37 ±2.21 / 109.25 ms │     no change │
│ QQuery 70 │        341.81 / 351.53 ±8.17 / 364.99 ms │       324.05 / 349.59 ±17.32 / 371.15 ms │     no change │
│ QQuery 71 │        133.42 / 136.39 ±2.40 / 140.74 ms │        134.36 / 135.99 ±1.26 / 137.89 ms │     no change │
│ QQuery 72 │        614.21 / 623.00 ±8.60 / 637.84 ms │        620.46 / 625.19 ±3.90 / 632.01 ms │     no change │
│ QQuery 73 │              6.59 / 7.75 ±0.73 / 8.63 ms │              6.63 / 7.82 ±0.70 / 8.82 ms │     no change │
│ QQuery 74 │        558.34 / 566.09 ±4.36 / 571.43 ms │        559.92 / 575.39 ±8.44 / 583.10 ms │     no change │
│ QQuery 75 │        274.48 / 276.81 ±1.44 / 278.64 ms │        276.79 / 281.34 ±2.30 / 282.98 ms │     no change │
│ QQuery 76 │        130.42 / 132.57 ±1.75 / 135.33 ms │        133.21 / 134.74 ±1.45 / 136.89 ms │     no change │
│ QQuery 77 │        185.64 / 187.03 ±1.20 / 188.60 ms │        193.17 / 193.65 ±0.26 / 193.90 ms │     no change │
│ QQuery 78 │        338.15 / 342.74 ±3.33 / 347.76 ms │        342.25 / 344.99 ±2.28 / 348.35 ms │     no change │
│ QQuery 79 │        228.22 / 230.79 ±3.01 / 236.55 ms │        231.33 / 233.38 ±1.62 / 235.77 ms │     no change │
│ QQuery 80 │        319.28 / 322.99 ±3.19 / 326.87 ms │        321.49 / 325.69 ±4.79 / 334.94 ms │     no change │
│ QQuery 81 │           26.10 / 26.82 ±0.56 / 27.73 ms │           26.13 / 27.76 ±2.26 / 32.25 ms │     no change │
│ QQuery 82 │        196.94 / 199.68 ±3.02 / 203.93 ms │        197.63 / 200.41 ±1.58 / 201.86 ms │     no change │
│ QQuery 83 │           38.51 / 39.26 ±0.60 / 40.06 ms │           39.35 / 40.25 ±0.69 / 41.32 ms │     no change │
│ QQuery 84 │           48.35 / 49.73 ±1.39 / 52.03 ms │           49.40 / 50.89 ±1.11 / 52.19 ms │     no change │
│ QQuery 85 │        146.01 / 148.61 ±1.59 / 150.71 ms │        145.75 / 150.79 ±2.77 / 153.42 ms │     no change │
│ QQuery 86 │           39.49 / 40.90 ±1.13 / 42.91 ms │           39.78 / 40.29 ±0.58 / 41.34 ms │     no change │
│ QQuery 87 │           85.37 / 88.25 ±2.39 / 92.44 ms │           87.84 / 93.23 ±3.63 / 98.33 ms │  1.06x slower │
│ QQuery 88 │         99.30 / 100.68 ±0.83 / 101.58 ms │        111.68 / 113.57 ±0.98 / 114.37 ms │  1.13x slower │
│ QQuery 89 │        118.00 / 120.02 ±2.57 / 125.03 ms │        119.97 / 121.08 ±0.70 / 121.90 ms │     no change │
│ QQuery 90 │           23.55 / 24.31 ±0.67 / 25.54 ms │           25.43 / 26.23 ±0.60 / 27.26 ms │  1.08x slower │
│ QQuery 91 │           63.33 / 64.85 ±0.87 / 65.84 ms │           64.43 / 66.26 ±1.25 / 68.07 ms │     no change │
│ QQuery 92 │           57.13 / 58.28 ±0.90 / 59.85 ms │           58.24 / 58.90 ±0.54 / 59.56 ms │     no change │
│ QQuery 93 │        183.61 / 186.61 ±1.57 / 188.24 ms │        186.19 / 187.72 ±1.27 / 189.78 ms │     no change │
│ QQuery 94 │           61.03 / 61.98 ±0.91 / 63.53 ms │           61.90 / 62.43 ±0.41 / 63.07 ms │     no change │
│ QQuery 95 │        127.74 / 129.37 ±1.52 / 132.11 ms │        129.59 / 132.05 ±2.33 / 135.78 ms │     no change │
│ QQuery 96 │           72.20 / 73.52 ±0.98 / 74.67 ms │           75.35 / 75.94 ±0.41 / 76.57 ms │     no change │
│ QQuery 97 │        125.26 / 127.43 ±1.68 / 128.98 ms │        128.31 / 129.23 ±0.66 / 130.29 ms │     no change │
│ QQuery 98 │        148.64 / 153.05 ±2.25 / 154.73 ms │        152.49 / 154.63 ±1.62 / 156.58 ms │     no change │
│ QQuery 99 │ 10732.73 / 10763.22 ±28.55 / 10800.53 ms │ 10754.23 / 10789.51 ±27.91 / 10826.30 ms │     no change │
└───────────┴──────────────────────────────────────────┴──────────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary               ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)               │ 31328.37ms │
│ Total Time (morsel-splitting)   │ 31574.60ms │
│ Average Time (HEAD)             │   316.45ms │
│ Average Time (morsel-splitting) │   318.94ms │
│ Queries Faster                  │          4 │
│ Queries Slower                  │         13 │
│ Queries with No Change          │         82 │
│ Queries with Failure            │          0 │
└─────────────────────────────────┴────────────┘

Resource Usage

tpcds — base (merge-base)

Metric	Value
Wall time	157.0s
Peak memory	5.5 GiB
Avg memory	4.6 GiB
CPU user	258.3s
CPU sys	17.4s
Peak spill	0 B

tpcds — branch

Metric	Value
Wall time	158.1s
Peak memory	4.8 GiB
Avg memory	4.3 GiB
CPU user	277.8s
CPU sys	18.4s
Peak spill	0 B

File an issue against this benchmark runner

adriangbot · 2026-04-15T18:43:33Z

🤖 Benchmark completed (GKE) | trigger

Instance: c4a-highmem-16 (12 vCPU / 65 GiB)

CPU Details (lscpu)

Architecture:                            aarch64
CPU op-mode(s):                          64-bit
Byte Order:                              Little Endian
CPU(s):                                  16
On-line CPU(s) list:                     0-15
Vendor ID:                               ARM
Model name:                              Neoverse-V2
Model:                                   1
Thread(s) per core:                      1
Core(s) per cluster:                     16
Socket(s):                               -
Cluster(s):                              1
Stepping:                                r0p1
BogoMIPS:                                2000.00
Flags:                                   fp asimd evtstrm aes pmull sha1 sha2 crc32 atomics fphp asimdhp cpuid asimdrdm jscvt fcma lrcpc dcpop sha3 sm3 sm4 asimddp sha512 sve asimdfhm dit uscat ilrcpc flagm sb paca pacg dcpodp sve2 sveaes svepmull svebitperm svesha3 svesm4 flagm2 frint svei8mm svebf16 i8mm bf16 dgh rng bti
L1d cache:                               1 MiB (16 instances)
L1i cache:                               1 MiB (16 instances)
L2 cache:                                32 MiB (16 instances)
L3 cache:                                80 MiB (1 instance)
NUMA node(s):                            1
NUMA node0 CPU(s):                       0-15
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; CSV2, BHB
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

Details

Comparing HEAD and morsel-splitting
--------------------
Benchmark clickbench_partitioned.json
--------------------
┏━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query     ┃                                  HEAD ┃                      morsel-splitting ┃        Change ┃
┡━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 0  │          1.19 / 4.40 ±6.31 / 17.03 ms │          1.20 / 4.38 ±6.29 / 16.95 ms │     no change │
│ QQuery 1  │        14.39 / 14.61 ±0.16 / 14.81 ms │        12.25 / 12.63 ±0.22 / 12.86 ms │ +1.16x faster │
│ QQuery 2  │        43.75 / 44.22 ±0.33 / 44.66 ms │        35.92 / 36.07 ±0.13 / 36.32 ms │ +1.23x faster │
│ QQuery 3  │        42.71 / 45.18 ±1.63 / 47.24 ms │        33.16 / 34.17 ±0.98 / 35.98 ms │ +1.32x faster │
│ QQuery 4  │     281.73 / 289.90 ±4.90 / 297.06 ms │     250.08 / 256.64 ±4.31 / 262.45 ms │ +1.13x faster │
│ QQuery 5  │     343.24 / 346.25 ±2.97 / 351.71 ms │     299.46 / 301.70 ±2.14 / 305.43 ms │ +1.15x faster │
│ QQuery 6  │           5.30 / 6.59 ±0.81 / 7.72 ms │           4.80 / 5.84 ±0.76 / 6.78 ms │ +1.13x faster │
│ QQuery 7  │        16.66 / 17.97 ±1.01 / 19.40 ms │        13.35 / 13.61 ±0.17 / 13.88 ms │ +1.32x faster │
│ QQuery 8  │     412.54 / 422.19 ±8.63 / 436.39 ms │     356.91 / 360.45 ±1.98 / 362.62 ms │ +1.17x faster │
│ QQuery 9  │     629.10 / 643.00 ±7.21 / 649.06 ms │    485.66 / 502.84 ±14.59 / 524.20 ms │ +1.28x faster │
│ QQuery 10 │       93.49 / 95.65 ±2.81 / 100.79 ms │        74.74 / 76.35 ±2.11 / 80.25 ms │ +1.25x faster │
│ QQuery 11 │     105.19 / 105.74 ±0.36 / 106.15 ms │        85.22 / 85.73 ±0.52 / 86.40 ms │ +1.23x faster │
│ QQuery 12 │     335.02 / 339.83 ±3.32 / 343.59 ms │     290.83 / 292.74 ±2.06 / 296.50 ms │ +1.16x faster │
│ QQuery 13 │    453.82 / 470.59 ±12.18 / 487.21 ms │    422.34 / 431.52 ±11.97 / 452.07 ms │ +1.09x faster │
│ QQuery 14 │     335.72 / 342.91 ±4.19 / 347.06 ms │     303.19 / 306.75 ±3.15 / 312.60 ms │ +1.12x faster │
│ QQuery 15 │    335.47 / 351.66 ±13.78 / 368.65 ms │    323.84 / 347.19 ±24.11 / 383.33 ms │     no change │
│ QQuery 16 │     712.02 / 722.32 ±6.27 / 731.81 ms │     675.72 / 683.87 ±5.20 / 691.32 ms │ +1.06x faster │
│ QQuery 17 │     701.58 / 709.03 ±4.90 / 715.02 ms │     679.32 / 684.83 ±3.82 / 690.57 ms │     no change │
│ QQuery 18 │ 1407.42 / 1473.80 ±38.45 / 1517.36 ms │ 1332.93 / 1342.86 ±10.81 / 1357.08 ms │ +1.10x faster │
│ QQuery 19 │        35.27 / 38.91 ±5.17 / 48.98 ms │        29.46 / 31.84 ±2.35 / 35.59 ms │ +1.22x faster │
│ QQuery 20 │    711.70 / 732.41 ±22.54 / 761.49 ms │     477.15 / 480.45 ±3.34 / 486.71 ms │ +1.52x faster │
│ QQuery 21 │     764.65 / 774.75 ±7.98 / 784.32 ms │     564.69 / 566.07 ±1.57 / 569.12 ms │ +1.37x faster │
│ QQuery 22 │  1131.27 / 1137.91 ±4.39 / 1144.53 ms │  1006.68 / 1008.28 ±2.00 / 1012.20 ms │ +1.13x faster │
│ QQuery 23 │ 3052.13 / 3098.35 ±25.53 / 3124.66 ms │  3160.94 / 3174.71 ±9.29 / 3186.31 ms │     no change │
│ QQuery 24 │      98.92 / 103.89 ±3.43 / 109.71 ms │        43.50 / 47.44 ±2.46 / 50.78 ms │ +2.19x faster │
│ QQuery 25 │     139.06 / 139.88 ±0.78 / 140.98 ms │     110.44 / 111.73 ±0.81 / 112.68 ms │ +1.25x faster │
│ QQuery 26 │     100.45 / 102.41 ±1.62 / 104.36 ms │        45.15 / 45.93 ±0.87 / 47.16 ms │ +2.23x faster │
│ QQuery 27 │     853.08 / 860.51 ±7.33 / 870.89 ms │     631.87 / 636.18 ±3.91 / 640.89 ms │ +1.35x faster │
│ QQuery 28 │  3252.33 / 3269.16 ±9.39 / 3280.43 ms │ 2796.91 / 2833.19 ±20.09 / 2853.29 ms │ +1.15x faster │
│ QQuery 29 │        50.42 / 52.82 ±2.81 / 57.44 ms │        43.76 / 47.33 ±3.56 / 52.93 ms │ +1.12x faster │
│ QQuery 30 │     358.55 / 362.78 ±3.79 / 368.88 ms │     323.30 / 331.65 ±4.89 / 337.34 ms │ +1.09x faster │
│ QQuery 31 │    373.31 / 387.95 ±10.10 / 402.53 ms │     324.52 / 336.15 ±5.90 / 340.49 ms │ +1.15x faster │
│ QQuery 32 │ 1250.41 / 1285.78 ±24.59 / 1327.39 ms │ 1016.30 / 1032.13 ±15.52 / 1060.44 ms │ +1.25x faster │
│ QQuery 33 │ 1482.78 / 1526.60 ±33.05 / 1585.31 ms │  1457.39 / 1469.84 ±8.78 / 1484.87 ms │     no change │
│ QQuery 34 │ 1462.30 / 1501.58 ±24.85 / 1533.12 ms │  1450.78 / 1460.34 ±7.35 / 1472.65 ms │     no change │
│ QQuery 35 │     377.14 / 382.14 ±6.26 / 394.39 ms │     315.20 / 318.35 ±3.98 / 326.00 ms │ +1.20x faster │
│ QQuery 36 │     113.48 / 120.74 ±3.77 / 123.83 ms │        69.35 / 73.54 ±2.34 / 76.39 ms │ +1.64x faster │
│ QQuery 37 │        49.16 / 51.02 ±1.35 / 52.37 ms │        40.34 / 41.18 ±1.08 / 43.25 ms │ +1.24x faster │
│ QQuery 38 │        74.98 / 77.19 ±1.83 / 79.29 ms │        44.09 / 44.96 ±0.81 / 46.19 ms │ +1.72x faster │
│ QQuery 39 │     208.46 / 215.72 ±6.26 / 225.53 ms │     127.53 / 133.75 ±3.35 / 137.60 ms │ +1.61x faster │
│ QQuery 40 │        23.95 / 25.13 ±0.92 / 26.22 ms │        16.50 / 17.12 ±0.80 / 18.67 ms │ +1.47x faster │
│ QQuery 41 │        19.29 / 21.33 ±1.54 / 23.98 ms │        15.30 / 15.90 ±0.58 / 16.99 ms │ +1.34x faster │
│ QQuery 42 │        19.51 / 19.97 ±0.30 / 20.29 ms │        14.78 / 15.51 ±0.96 / 17.32 ms │ +1.29x faster │
└───────────┴───────────────────────────────────────┴───────────────────────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┓
┃ Benchmark Summary               ┃            ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━┩
│ Total Time (HEAD)               │ 22734.80ms │
│ Total Time (morsel-splitting)   │ 20053.70ms │
│ Average Time (HEAD)             │   528.72ms │
│ Average Time (morsel-splitting) │   466.37ms │
│ Queries Faster                  │         37 │
│ Queries Slower                  │          0 │
│ Queries with No Change          │          6 │
│ Queries with Failure            │          0 │
└─────────────────────────────────┴────────────┘

Resource Usage

clickbench_partitioned — base (merge-base)

Metric	Value
Wall time	114.8s
Peak memory	39.8 GiB
Avg memory	28.2 GiB
CPU user	1069.9s
CPU sys	98.3s
Peak spill	0 B

clickbench_partitioned — branch

Metric	Value
Wall time	101.5s
Peak memory	44.1 GiB
Avg memory	34.1 GiB
CPU user	1084.9s
CPU sys	74.9s
Peak spill	0 B

File an issue against this benchmark runner

alamb added 2 commits April 14, 2026 11:49

Add Dynamic work scheduling in FileStream

1b1b586

fix: typos

5287210

github-actions bot added the datasource Changes to the datasource crate label Apr 15, 2026

Dandandan changed the title ~~feat: morsel splitting in SharedWorkSource~~ Test: morsel splitting in SharedWorkSource Apr 15, 2026

Dandandan changed the title ~~Test: morsel splitting in SharedWorkSource~~ Test: morsel splitting & merging in SharedWorkSource Apr 15, 2026

Dandandan force-pushed the morsel-splitting branch from 76a1f58 to ea3834e Compare April 15, 2026 12:40

Dandandan force-pushed the morsel-splitting branch 2 times, most recently from 3a13d4d to b852406 Compare April 15, 2026 13:12

Dandandan force-pushed the morsel-splitting branch from b852406 to 106f736 Compare April 15, 2026 18:04

Merge branch 'main' into morsel-splitting

40ca776

Conversation

Dandandan commented Apr 15, 2026

Which issue does this PR close?

Rationale for this change

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?

Uh oh!

Dandandan commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

Dandandan commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

Dandandan commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

Dandandan commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

Dandandan commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

adriangbot commented Apr 15, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants