date: 2026-09-15T09:39:38Z os: Darwin 25.2.0 arm64 macos: 26.2 (25C56) cpu: Apple M4 Pro logical cpus: 14 performance cores: 10 efficiency cores: 4 smt: none P-core L1d: 131072 B L2 (shared per cluster of 5): 16777216 B E-core L1d: 65536 B L2 (shared per cluster of 4): 4194304 B cache line: 128 B page: 16384 B memory: 24 GiB isa features: CRC32 FlagM FlagM2 FHM DotProd SHA3 RDM LSE SHA256 SHA512 SHA1 AES PMULL SB FRINTTS PACIMP LRCPC LRCPC2 FCMA JSCVT PAuth PAuth2 FPAC FPACCOMBINE DPB DPB2 BF16 I8MM WFxT RPRES ECV AFP LSE2 CSV2 CSV3 DIT FP16 BTI SME SME2 SME_F64F64 SME_I16I64 power: 'AC Power' load average at start: 10.50 8.47 8.13 frequency: not published by the vendor for this part; see estimated clock below compiler: Apple clang version 17.0.0 (clang-1700.4.4.1) uptime: 10:39 up 15 days, 20:55, 1 user, load averages: 10.50 8.47 8.13 estimated clock: 4.49 GHz (dependent 1-cycle add chain, 400000000 adds, min of 7 runs) taskpolicy -c background: estimated clock: 1.04 GHz (dependent 1-cycle add chain, 400000000 adds, min of 7 runs) bench: QUICK=0 REPS=default --- adds per clock sample: 50000000 adds per clock bracket: 5000000 (one before and one after every FMA and stream sample) fma array bytes: 16384 fma passes per sample: 2048 fma flops per sample: 268435456 stream array bytes: 67108864 reps: 15 (plus 1 warmup per kernel, discarded, after a 200 ms settling spin) efficiency core cpu ids: 0-3 (hw.perflevel1.logicalcpu) stream reference sum: 36034921960712949 RESULT stream_checksum 36034921960712949 sum fma reference (sum of 64 lanes): 83960288.0 RESULT fma_checksum 83960288.0 sum main_default: thread qos user-interactive, cpu ids seen 9,13, thread cpu share of wall 100.0% (median) RESULT main_default_qos user-interactive class RESULT main_default_cpus 9,13 ids RESULT main_default_ecore_samples 0.0 percent RESULT main_default_cpu_share 100.0 percent main_default clock, wall time min 0.222 median 0.222 max 0.225 cv 0.4% n 15 ns/add RESULT main_default_clock_ghz 4.511 GHz RESULT main_default_clock_median_ghz 4.502 GHz RESULT main_default_clock_cv 0.40 percent main_default fma, wall time min 135.699 median 144.139 max 144.349 cv 1.5% n 15 GFLOP/s RESULT main_default_fma_gflops 144.14 GFLOP/s RESULT main_default_fma_cv 1.53 percent main_default stream, wall time min 50.489 median 82.233 max 86.849 cv 14.9% n 15 GB/s RESULT main_default_stream_gbps 82.23 GB/s RESULT main_default_stream_cv 14.91 percent main_default fma per cycle, wall time min 3.838 median 3.999 max 4.034 cv 1.2% n 15 FMA/cycle RESULT main_default_fma_per_cycle 3.999 FMA/cycle RESULT main_default_fma_per_cycle_cv 1.17 percent main_default bytes per cycle, wall time min 11.190 median 18.227 max 19.291 cv 15.0% n 15 B/cycle RESULT main_default_stream_bytes_per_cycle 18.23 B/cycle RESULT main_default_stream_bytes_per_cycle_cv 15.03 percent main_default clock, thread cpu time min 0.222 median 0.222 max 0.225 cv 0.4% n 15 ns/add RESULT main_default_clock_cpu_ghz 4.511 GHz RESULT main_default_clock_cpu_median_ghz 4.502 GHz RESULT main_default_clock_cpu_cv 0.40 percent main_default fma, thread cpu time min 135.688 median 144.133 max 144.343 cv 1.5% n 15 GFLOP/s RESULT main_default_fma_cpu_gflops 144.13 GFLOP/s RESULT main_default_fma_cpu_cv 1.53 percent main_default stream, thread cpu time min 50.478 median 82.195 max 86.480 cv 14.9% n 15 GB/s RESULT main_default_stream_cpu_gbps 82.20 GB/s RESULT main_default_stream_cpu_cv 14.88 percent main_default fma per cycle, thread cpu time min 3.844 median 3.999 max 4.034 cv 1.1% n 15 FMA/cycle RESULT main_default_fma_cpu_per_cycle 3.999 FMA/cycle RESULT main_default_fma_cpu_per_cycle_cv 1.14 percent main_default bytes per cycle, thread cpu time min 11.189 median 18.220 max 19.211 cv 15.0% n 15 B/cycle RESULT main_default_stream_cpu_bytes_per_cycle 18.22 B/cycle RESULT main_default_stream_cpu_bytes_per_cycle_cv 14.98 percent thread_default: thread qos default, cpu ids seen 10,12,13, thread cpu share of wall 100.0% (median) RESULT thread_default_qos default class RESULT thread_default_cpus 10,12,13 ids RESULT thread_default_ecore_samples 0.0 percent RESULT thread_default_cpu_share 100.0 percent thread_default clock, wall time min 0.222 median 0.223 max 0.230 cv 1.1% n 15 ns/add RESULT thread_default_clock_ghz 4.503 GHz RESULT thread_default_clock_median_ghz 4.489 GHz RESULT thread_default_clock_cv 1.05 percent thread_default fma, wall time min 142.130 median 143.965 max 144.362 cv 0.5% n 15 GFLOP/s RESULT thread_default_fma_gflops 143.97 GFLOP/s RESULT thread_default_fma_cv 0.49 percent thread_default stream, wall time min 79.935 median 85.553 max 89.878 cv 2.9% n 15 GB/s RESULT thread_default_stream_gbps 85.55 GB/s RESULT thread_default_stream_cv 2.95 percent thread_default fma per cycle, wall time min 3.947 median 4.015 max 4.043 cv 0.8% n 15 FMA/cycle RESULT thread_default_fma_per_cycle 4.015 FMA/cycle RESULT thread_default_fma_per_cycle_cv 0.75 percent thread_default bytes per cycle, wall time min 17.732 median 19.137 max 20.115 cv 2.9% n 15 B/cycle RESULT thread_default_stream_bytes_per_cycle 19.14 B/cycle RESULT thread_default_stream_bytes_per_cycle_cv 2.85 percent thread_default clock, thread cpu time min 0.222 median 0.223 max 0.230 cv 1.0% n 15 ns/add RESULT thread_default_clock_cpu_ghz 4.507 GHz RESULT thread_default_clock_cpu_median_ghz 4.489 GHz RESULT thread_default_clock_cpu_cv 1.00 percent thread_default fma, thread cpu time min 142.117 median 143.953 max 144.353 cv 0.5% n 15 GFLOP/s RESULT thread_default_fma_cpu_gflops 143.95 GFLOP/s RESULT thread_default_fma_cpu_cv 0.49 percent thread_default stream, thread cpu time min 79.907 median 85.534 max 89.838 cv 2.9% n 15 GB/s RESULT thread_default_stream_cpu_gbps 85.53 GB/s RESULT thread_default_stream_cpu_cv 2.88 percent thread_default fma per cycle, thread cpu time min 3.947 median 4.015 max 4.043 cv 0.8% n 15 FMA/cycle RESULT thread_default_fma_cpu_per_cycle 4.015 FMA/cycle RESULT thread_default_fma_cpu_per_cycle_cv 0.75 percent thread_default bytes per cycle, thread cpu time min 17.729 median 19.127 max 20.108 cv 2.7% n 15 B/cycle RESULT thread_default_stream_cpu_bytes_per_cycle 19.13 B/cycle RESULT thread_default_stream_cpu_bytes_per_cycle_cv 2.73 percent thread_background: thread qos background (QOS_CLASS_BACKGROUND requested), cpu ids seen 11,12, thread cpu share of wall 100.0% (median) RESULT thread_background_qos background class RESULT thread_background_cpus 11,12 ids RESULT thread_background_ecore_samples 0.0 percent RESULT thread_background_cpu_share 100.0 percent thread_background clock, wall time min 0.222 median 0.222 max 0.225 cv 0.4% n 15 ns/add RESULT thread_background_clock_ghz 4.509 GHz RESULT thread_background_clock_median_ghz 4.497 GHz RESULT thread_background_clock_cv 0.41 percent thread_background fma, wall time min 139.825 median 143.904 max 144.362 cv 1.0% n 15 GFLOP/s RESULT thread_background_fma_gflops 143.90 GFLOP/s RESULT thread_background_fma_cv 1.04 percent thread_background stream, wall time min 82.515 median 84.114 max 86.975 cv 1.8% n 15 GB/s RESULT thread_background_stream_gbps 84.11 GB/s RESULT thread_background_stream_cv 1.75 percent thread_background fma per cycle, wall time min 3.922 median 4.000 max 4.076 cv 0.8% n 15 FMA/cycle RESULT thread_background_fma_per_cycle 4.000 FMA/cycle RESULT thread_background_fma_per_cycle_cv 0.85 percent thread_background bytes per cycle, wall time min 18.328 median 18.722 max 19.707 cv 2.1% n 15 B/cycle RESULT thread_background_stream_bytes_per_cycle 18.72 B/cycle RESULT thread_background_stream_bytes_per_cycle_cv 2.13 percent thread_background clock, thread cpu time min 0.222 median 0.222 max 0.225 cv 0.4% n 15 ns/add RESULT thread_background_clock_cpu_ghz 4.509 GHz RESULT thread_background_clock_cpu_median_ghz 4.497 GHz RESULT thread_background_clock_cpu_cv 0.41 percent thread_background fma, thread cpu time min 139.816 median 143.888 max 144.356 cv 1.0% n 15 GFLOP/s RESULT thread_background_fma_cpu_gflops 143.89 GFLOP/s RESULT thread_background_fma_cpu_cv 1.04 percent thread_background stream, thread cpu time min 82.490 median 84.053 max 86.952 cv 1.8% n 15 GB/s RESULT thread_background_stream_cpu_gbps 84.05 GB/s RESULT thread_background_stream_cpu_cv 1.76 percent thread_background fma per cycle, thread cpu time min 3.922 median 4.000 max 4.076 cv 0.8% n 15 FMA/cycle RESULT thread_background_fma_cpu_per_cycle 4.000 FMA/cycle RESULT thread_background_fma_cpu_per_cycle_cv 0.85 percent thread_background bytes per cycle, thread cpu time min 18.324 median 18.718 max 19.699 cv 2.1% n 15 B/cycle RESULT thread_background_stream_cpu_bytes_per_cycle 18.72 B/cycle RESULT thread_background_stream_cpu_bytes_per_cycle_cv 2.13 percent checksums: every sample matched its reference --- MAIN_ONLY=1 MAIN_TAG=process_background taskpolicy -c background ./bench adds per clock sample: 50000000 adds per clock bracket: 5000000 (one before and one after every FMA and stream sample) fma array bytes: 16384 fma passes per sample: 2048 fma flops per sample: 268435456 stream array bytes: 67108864 reps: 15 (plus 1 warmup per kernel, discarded, after a 200 ms settling spin) efficiency core cpu ids: 0-3 (hw.perflevel1.logicalcpu) stream reference sum: 36034921960712949 RESULT stream_checksum 36034921960712949 sum fma reference (sum of 64 lanes): 83960288.0 RESULT fma_checksum 83960288.0 sum process_background: thread qos background, cpu ids seen 0,1,2,3, thread cpu share of wall 75.8% (median) RESULT process_background_qos background class RESULT process_background_cpus 0,1,2,3 ids RESULT process_background_ecore_samples 100.0 percent RESULT process_background_cpu_share 75.8 percent process_background clock, wall time min 0.718 median 1.358 max 2.072 cv 23.2% n 15 ns/add RESULT process_background_clock_ghz 1.393 GHz RESULT process_background_clock_median_ghz 0.736 GHz RESULT process_background_clock_cv 23.19 percent process_background fma, wall time min 5.419 median 12.003 max 19.022 cv 26.2% n 15 GFLOP/s RESULT process_background_fma_gflops 12.00 GFLOP/s RESULT process_background_fma_cv 26.15 percent process_background stream, wall time min 2.518 median 8.372 max 12.777 cv 27.8% n 15 GB/s RESULT process_background_stream_gbps 8.37 GB/s RESULT process_background_stream_cv 27.84 percent process_background fma per cycle, wall time min 0.604 median 1.869 max 3.070 cv 29.3% n 15 FMA/cycle RESULT process_background_fma_per_cycle 1.869 FMA/cycle RESULT process_background_fma_per_cycle_cv 29.33 percent process_background bytes per cycle, wall time min 8.328 median 11.302 max 21.134 cv 29.4% n 15 B/cycle RESULT process_background_stream_bytes_per_cycle 11.30 B/cycle RESULT process_background_stream_bytes_per_cycle_cv 29.40 percent process_background clock, thread cpu time min 0.656 median 0.896 max 1.029 cv 9.8% n 15 ns/add RESULT process_background_clock_cpu_ghz 1.525 GHz RESULT process_background_clock_cpu_median_ghz 1.116 GHz RESULT process_background_clock_cpu_cv 9.81 percent process_background fma, thread cpu time min 15.596 median 15.997 max 25.580 cv 17.5% n 15 GFLOP/s RESULT process_background_fma_cpu_gflops 16.00 GFLOP/s RESULT process_background_fma_cpu_cv 17.46 percent process_background stream, thread cpu time min 9.834 median 10.933 max 12.860 cv 8.2% n 15 GB/s RESULT process_background_stream_cpu_gbps 10.93 GB/s RESULT process_background_stream_cpu_cv 8.18 percent process_background fma per cycle, thread cpu time min 1.605 median 1.974 max 2.993 cv 17.6% n 15 FMA/cycle RESULT process_background_fma_cpu_per_cycle 1.974 FMA/cycle RESULT process_background_fma_cpu_per_cycle_cv 17.56 percent process_background bytes per cycle, thread cpu time min 7.887 median 10.625 max 12.937 cv 13.4% n 15 B/cycle RESULT process_background_stream_cpu_bytes_per_cycle 10.63 B/cycle RESULT process_background_stream_cpu_bytes_per_cycle_cv 13.36 percent checksums: every sample matched its reference