diff --git a/speed-bench/m4_macstudio_max_128g.csv b/speed-bench/m4_macstudio_max_128g.csv new file mode 100644 index 00000000..7e57a67f --- /dev/null +++ b/speed-bench/m4_macstudio_max_128g.csv @@ -0,0 +1,34 @@ +ctx_tokens,prefill_tokens,prefill_tps,gen_tokens,gen_tps,kvcache_bytes +2048,2048,329.81,128,25.77,52184460 +4096,2048,293.32,128,25.16,80373132 +6144,2048,288.21,128,24.88,108561804 +8192,2048,289.92,128,25.48,136750476 +10240,2048,277.44,128,25.06,164939148 +12288,2048,276.85,128,24.80,193127820 +14336,2048,276.13,128,24.70,221316492 +16384,2048,273.59,128,24.62,249505164 +18432,2048,269.88,128,24.63,277693836 +20480,2048,263.41,128,24.88,305882508 +22528,2048,263.31,128,24.49,334071180 +24576,2048,261.91,128,24.44,362259852 +26624,2048,255.81,128,23.84,390448524 +28672,2048,251.71,128,24.23,418637196 +30720,2048,249.15,128,23.23,446825868 +32768,2048,238.46,128,22.90,475014540 +34816,2048,231.65,128,22.80,503203212 +36864,2048,236.48,128,23.56,531391884 +38912,2048,241.51,128,23.35,559580556 +40960,2048,236.61,128,23.47,587769228 +43008,2048,233.86,128,23.38,615957900 +45056,2048,232.21,128,23.40,644146572 +47104,2048,231.76,128,23.07,672335244 +49152,2048,225.12,128,21.92,700523916 +51200,2048,209.12,128,21.02,728712588 +53248,2048,205.83,128,21.07,756901260 +55296,2048,208.98,128,21.15,785089932 +57344,2048,203.39,128,21.18,813278604 +59392,2048,200.77,128,21.10,841467276 +61440,2048,199.05,128,20.98,869655948 +63488,2048,208.65,128,22.23,897844620 +65536,2048,210.47,128,22.34,926033292 + diff --git a/speed-bench/m4_macstudio_max_128g_ts.svg b/speed-bench/m4_macstudio_max_128g_ts.svg new file mode 100644 index 00000000..c13cd29d --- /dev/null +++ b/speed-bench/m4_macstudio_max_128g_ts.svg @@ -0,0 +1,47 @@ + + + + +M4 Max 128g macstudio t/s + +0 + +100 + +200 + +300 + +400 +0 +10 +20 +30 + +0 + +20k + +40k + +60k + + + +ctx size +prefill t/s +generation t/s + + + + +prefill + +generation +