|
|
@@ -0,0 +1,54 @@
|
|
|
+# Throughput Benchmark — deepseek-v4.1-flash
|
|
|
+
|
|
|
+- **Run ID**: `2026-11-02T22-48-46.376991Z_4528bb01`
|
|
|
+- **Date**: `2026-11-02T22:48:46.377074+00:00`
|
|
|
+- **Mode**: throughput-only
|
|
|
+- **tool-eval-bench**: `v3.0.0`
|
|
|
+- **Backend**: unknown
|
|
|
+- **Server**: http://***:30000/v1
|
|
|
+- **Model (API)**: `deepseek-v4.1-flash`
|
|
|
+- **Model (Root)**: `checkpoint-expanded-scales`
|
|
|
+
|
|
|
+## Inference Engine
|
|
|
+
|
|
|
+| Property | Value |
|
|
|
+|---|---|
|
|
|
+| Max Model Length | 1,048,576 |
|
|
|
+| Host | `pluto` |
|
|
|
+| Platform | `Linux-7.2.11-arch1-1-x86_64-with-glibc2.44` |
|
|
|
+| Python | 3.12.15 |
|
|
|
+
|
|
|
+## Results
|
|
|
+
|
|
|
+| Test | pp t/s | tg t/s | TTFT (ms) | Total (ms) | Tokens |
|
|
|
+|---|---:|---:|---:|---:|---:|
|
|
|
+| pp2048 tg128 @ d0 | 1,806 | 72.8 | 1,271 | 3,015 | 2048+128 |
|
|
|
+| pp2048 tg128 @ d0 c2 | 1,922 | 95.3 | 2,057 | 4,432 | 2048+128 |
|
|
|
+| pp2048 tg128 @ d0 c4 | 1,967 | 123.0 | 3,784 | 7,112 | 2048+128 |
|
|
|
+| pp2048 tg128 @ d0 c8 | 2,099 | 172.3 | 7,048 | 11,815 | 2048+128 |
|
|
|
+| pp2048 tg128 @ d0 c16 | 2,104 | 214.3 | 13,641 | 20,624 | 2048+128 |
|
|
|
+| pp2048 tg128 @ d0 c32 | 2,028 | 260.1 | 28,060 | 38,001 | 2048+128 |
|
|
|
+| ctx pp4096 tg128 @ d4096 | 3,056 | 70.7 | 1,423 | 3,219 | 4096+128 |
|
|
|
+| pp2048 tg128 @ d4096 | 1,016 | 70.7 | 2,093 | 3,890 | 2048+128 |
|
|
|
+| ctx pp4096 tg128 @ d4096 c2 | 2,830 | 97.4 | 2,759 | 5,115 | 4096+128 |
|
|
|
+| pp2048 tg128 @ d4096 c2 | 960 | 96.1 | 4,080 | 6,417 | 2048+128 |
|
|
|
+| ctx pp4096 tg128 @ d4096 c4 | 2,739 | 129.7 | 5,527 | 8,847 | 4096+128 |
|
|
|
+| pp2048 tg128 @ d4096 c4 | 951 | 125.9 | 8,103 | 11,554 | 2048+128 |
|
|
|
+| ctx pp4096 tg128 @ d4096 c8 | 2,513 | 150.7 | 11,225 | 16,023 | 4096+128 |
|
|
|
+| pp2048 tg128 @ d4096 c8 | 948 | 157.1 | 16,119 | 21,270 | 2048+128 |
|
|
|
+| ctx pp4096 tg128 @ d4096 c16 | 2,679 | 201.9 | 22,208 | 29,292 | 4096+128 |
|
|
|
+| pp2048 tg128 @ d4096 c16 | 925 | 188.8 | 32,771 | 40,719 | 2048+128 |
|
|
|
+| ctx pp4096 tg128 @ d4096 c32 | 2,639 | 235.5 | 44,820 | 55,168 | 4096+128 |
|
|
|
+| pp2048 tg128 @ d4096 c32 | 916 | 225.9 | 65,939 | 77,893 | 2048+128 |
|
|
|
+| ctx pp8192 tg128 @ d8192 | 3,377 | 69.9 | 2,503 | 4,320 | 8192+128 |
|
|
|
+| pp2048 tg128 @ d8192 | 656 | 72.0 | 3,190 | 4,954 | 2048+128 |
|
|
|
+| ctx pp8192 tg128 @ d8192 c2 | 3,200 | 101.7 | 4,996 | 7,278 | 8192+128 |
|
|
|
+| pp2048 tg128 @ d8192 c2 | 615 | 98.8 | 6,475 | 8,798 | 2048+128 |
|
|
|
+| ctx pp8192 tg128 @ d8192 c4 | 3,163 | 128.0 | 9,986 | 13,286 | 8192+128 |
|
|
|
+| pp2048 tg128 @ d8192 c4 | 627 | 132.9 | 12,575 | 15,808 | 2048+128 |
|
|
|
+| ctx pp8192 tg128 @ d8192 c8 | 3,141 | 170.9 | 20,012 | 24,829 | 8192+128 |
|
|
|
+| pp2048 tg128 @ d8192 c8 | 627 | 158.4 | 24,974 | 30,128 | 2048+128 |
|
|
|
+| ctx pp8192 tg128 @ d8192 c16 | 3,073 | 203.7 | 40,393 | 47,445 | 8192+128 |
|
|
|
+| pp2048 tg128 @ d8192 c16 | 612 | 190.4 | 50,877 | 58,566 | 2048+128 |
|
|
|
+| ctx pp8192 tg128 @ d8192 c32 | 3,120 | 224.7 | 78,248 | 88,790 | 8192+128 |
|
|
|
+| pp2048 tg128 @ d8192 c32 | 633 | 227.9 | 97,960 | 109,878 | 2048+128 |
|