SPEC2017 INTrate: Trunk LTO -Ofast - x86-64-v3 vs native
-O3 -flto vs -O3 -flto and a given -march
AMD EPYC 8635P (zen5 based Sorano)
Run-times
Benchmark
x86-64-v3
Native
500.perlbench_r
100.00%
100.67%
502.gcc_r
100.00%
100.00%
505.mcf_r
100.00%
100.00%
520.omnetpp_r
100.00%
101.26%
523.xalancbmk_r
100.00%
97.23%
525.x264_r
100.00%
100.00%
531.deepsjeng_r
100.00%
100.00%
541.leela_r
100.00%
100.00%
548.exchange2_r
100.00%
102.65%
557.xz_r
100.00%
98.96%
| Benchmark | x86-64-v3 | Native |
|---|---|---|
| 500.perlbench_r | 100.00% | 100.67% |
| 502.gcc_r | 100.00% | 100.00% |
| 505.mcf_r | 100.00% | 100.00% |
| 520.omnetpp_r | 100.00% | 101.26% |
| 523.xalancbmk_r | 100.00% | 97.23% |
| 525.x264_r | 100.00% | 100.00% |
| 531.deepsjeng_r | 100.00% | 100.00% |
| 541.leela_r | 100.00% | 100.00% |
| 548.exchange2_r | 100.00% | 102.65% |
| 557.xz_r | 100.00% | 98.96% |
Naturally, the smaller the run-times, the better.
See also continuous gcc tuning benchmarking at https://lnt.opensuse.org/db_default/v4/SPEC/spec_report/tuning.
Non-debug executable sizes
Benchmark
x86-64-v3
Native
500.perlbench_r
100.00%
101.17%
502.gcc_r
100.00%
101.31%
505.mcf_r
100.00%
104.19%
520.omnetpp_r
100.00%
101.09%
523.xalancbmk_r
100.00%
100.35%
525.x264_r
100.00%
98.60%
531.deepsjeng_r
100.00%
100.11%
541.leela_r
100.00%
100.01%
548.exchange2_r
100.00%
94.97%
557.xz_r
100.00%
99.31%
Smaller binary sizes are always better.
| Benchmark | x86-64-v3 | Native |
|---|---|---|
| 500.perlbench_r | 100.00% | 101.17% |
| 502.gcc_r | 100.00% | 101.31% |
| 505.mcf_r | 100.00% | 104.19% |
| 520.omnetpp_r | 100.00% | 101.09% |
| 523.xalancbmk_r | 100.00% | 100.35% |
| 525.x264_r | 100.00% | 98.60% |
| 531.deepsjeng_r | 100.00% | 100.11% |
| 541.leela_r | 100.00% | 100.01% |
| 548.exchange2_r | 100.00% | 94.97% |
| 557.xz_r | 100.00% | 99.31% |
AMD EPYC 75F3 (zen3 based high-frequency Milan)
Run-times
Benchmark
x86-64-v3
Native
500.perlbench_r
100.00%
101.38%
502.gcc_r
100.00%
100.00%
505.mcf_r
100.00%
98.63%
520.omnetpp_r
100.00%
100.00%
523.xalancbmk_r
100.00%
100.76%
525.x264_r
100.00%
98.92%
531.deepsjeng_r
100.00%
100.51%
541.leela_r
100.00%
100.47%
548.exchange2_r
100.00%
100.00%
557.xz_r
100.00%
99.18%
| Benchmark | x86-64-v3 | Native |
|---|---|---|
| 500.perlbench_r | 100.00% | 101.38% |
| 502.gcc_r | 100.00% | 100.00% |
| 505.mcf_r | 100.00% | 98.63% |
| 520.omnetpp_r | 100.00% | 100.00% |
| 523.xalancbmk_r | 100.00% | 100.76% |
| 525.x264_r | 100.00% | 98.92% |
| 531.deepsjeng_r | 100.00% | 100.51% |
| 541.leela_r | 100.00% | 100.47% |
| 548.exchange2_r | 100.00% | 100.00% |
| 557.xz_r | 100.00% | 99.18% |
Naturally, the smaller the run-times, the better.
See also continuous gcc tuning benchmarking at https://lnt.opensuse.org/db_default/v4/SPEC/spec_report/tuning.
Non-debug executable sizes
Benchmark
x86-64-v3
Native
500.perlbench_r
100.00%
100.83%
502.gcc_r
100.00%
101.14%
505.mcf_r
100.00%
99.52%
520.omnetpp_r
100.00%
100.85%
523.xalancbmk_r
100.00%
101.47%
525.x264_r
100.00%
100.41%
531.deepsjeng_r
100.00%
100.02%
541.leela_r
100.00%
100.35%
548.exchange2_r
100.00%
97.85%
557.xz_r
100.00%
100.03%
Smaller binary sizes are always better.
| Benchmark | x86-64-v3 | Native |
|---|---|---|
| 500.perlbench_r | 100.00% | 100.83% |
| 502.gcc_r | 100.00% | 101.14% |
| 505.mcf_r | 100.00% | 99.52% |
| 520.omnetpp_r | 100.00% | 100.85% |
| 523.xalancbmk_r | 100.00% | 101.47% |
| 525.x264_r | 100.00% | 100.41% |
| 531.deepsjeng_r | 100.00% | 100.02% |
| 541.leela_r | 100.00% | 100.35% |
| 548.exchange2_r | 100.00% | 97.85% |
| 557.xz_r | 100.00% | 100.03% |