[JBTM-4014] Reduce allocation pressure on Uid

[franz1981]

These are the numbers of 0c18afd71b142a69055fe196592d131e91d8860e:

Benchmark                                      Mode  Cnt     Score    Error   Units
UidBenchmark.fromByteForm                      avgt   20     9.487 ±  0.043   ns/op
UidBenchmark.fromByteForm:·gc.alloc.rate       avgt   20  8040.524 ± 36.686  MB/sec
UidBenchmark.fromByteForm:·gc.alloc.rate.norm  avgt   20    80.000 ±  0.001    B/op
UidBenchmark.fromByteForm:·gc.count            avgt   20   620.000           counts
UidBenchmark.fromByteForm:·gc.time             avgt   20   126.000               ms
UidBenchmark.getBytes                          avgt   20    11.596 ±  0.056   ns/op
UidBenchmark.getBytes:·gc.alloc.rate           avgt   20  7894.044 ± 38.134  MB/sec
UidBenchmark.getBytes:·gc.alloc.rate.norm      avgt   20    96.000 ±  0.001    B/op
UidBenchmark.getBytes:·gc.count                avgt   20   597.000           counts
UidBenchmark.getBytes:·gc.time                 avgt   20   122.000               ms
UidBenchmark.stringForm                        avgt   20    37.016 ±  0.200   ns/op
UidBenchmark.stringForm:·gc.alloc.rate         avgt   20  4739.970 ± 25.619  MB/sec
UidBenchmark.stringForm:·gc.alloc.rate.norm    avgt   20   184.000 ±  0.001    B/op
UidBenchmark.stringForm:·gc.count              avgt   20   400.000           counts
UidBenchmark.stringForm:·gc.time               avgt   20    85.000               ms

vs before 3572e121e2d5138a40a6e2b1198f6ab42ce319c8:

Benchmark                                      Mode  Cnt     Score    Error   Units
UidBenchmark.fromByteForm                      avgt   20    37.077 ±  0.147   ns/op
UidBenchmark.fromByteForm:·gc.alloc.rate       avgt   20  3497.551 ± 13.841  MB/sec
UidBenchmark.fromByteForm:·gc.alloc.rate.norm  avgt   20   136.000 ±  0.001    B/op
UidBenchmark.fromByteForm:·gc.count            avgt   20   320.000           counts
UidBenchmark.fromByteForm:·gc.time             avgt   20    76.000               ms
UidBenchmark.getBytes                          avgt   20    24.341 ±  0.080   ns/op
UidBenchmark.getBytes:·gc.alloc.rate           avgt   20  6581.201 ± 21.539  MB/sec
UidBenchmark.getBytes:·gc.alloc.rate.norm      avgt   20   168.000 ±  0.001    B/op
UidBenchmark.getBytes:·gc.count                avgt   20   520.000           counts
UidBenchmark.getBytes:·gc.time                 avgt   20   106.000               ms
UidBenchmark.stringForm                        avgt   20   101.060 ±  0.651   ns/op
UidBenchmark.stringForm:·gc.alloc.rate         avgt   20  7095.497 ± 45.711  MB/sec
UidBenchmark.stringForm:·gc.alloc.rate.norm    avgt   20   752.000 ±  0.001    B/op
UidBenchmark.stringForm:·gc.count              avgt   20   549.000           counts
UidBenchmark.stringForm:·gc.time               avgt   20   115.000               ms

which shows that there's a huge improvement in pretty much all front, although the most important factor is likely the reduction of allocation rate (normalized per op).

I've run this benchmark with:

com.hp.mwtests.ts.arjuna.performance.UidBenchmark.*  -prof gc -gc true

[jbosstm-bot]

⚠️ performance CI not started.

Author is not the 'performance' contributor, members of jbosstm can write comment of text TESTIT to permit it.

[marcosgopen]

TESTIT

[jbosstm-bot]

Started testing this pull request: https://jenkins-csb-narayana-ci.dno.corp.redhat.com/job/btny-pulls-performance/34/

[jbosstm-bot]

Tests failed (https://jenkins-csb-narayana-ci.dno.corp.redhat.com/job/btny-pulls-performance/34/): Product comparison benchmark failed

[marcosgopen]

Thanks @franz1981