Qwen3.8 27B Q5_K_M + 7900xtx + DeepSeek Harness实战平均 52 t/s
-
软硬件配置：
cpu： AMD Ryzen 5 5600
RAM： 32GB 3200
GPU： 7900xtx
OS： Ubuntu 26.04
软件 ： llama.cpp + vulkan
版本：version: 9737 (67e9fd3b7) built with GNU 15.2.0 for Linux x86_64测试：
- llama.cpp 自带webui 创建打飞机游戏
提示词：
"Create a complete, fully functional 2D space shooter arcade game inside a single HTML file using HTML5 Canvas, CSS, and vanilla JavaScript. **Requirements:** 1. **Canvas & Layout:** Set up a centered 800x600 black canvas with a retro arcade UI showing the current score and remaining lives at the top. 2. **Player Ship:** Draw a distinct player ship at the bottom center. Allow smooth movement left and right using the Arrow Keys or A/D keys, constrained within the canvas bounds. 3. **Shooting Mechanics:** Pressing the Spacebar should fire a laser projectile upward from the player's position. Implement a brief cooldown between shots. 4. **Enemies:** Spawn alien ships or asteroids at random X coordinates along the top, moving downward at varying speeds. 5. **Collision Detection:** Write precise collision logic for bullets hitting enemies (destroying both and adding points) and enemies hitting the player or bottom screen (losing a life). 6. **Game Loop & States:** Include smooth requestAnimationFrame logic, a 'Game Over' screen, and a 'Restart' button functionality. Clean, well-commented code only."日志
1.47.020.662 I slot print_timing: id 1 | task 0 | n_decoded = 5409, tg = 89.47 t/s, tg_3s = 71.71 t/s 1.47.476.673 I slot print_timing: id 1 | task 0 | prompt eval time = 519.21 ms / 262 tokens ( 1.98 ms per token, 504.61 tokens per second) 1.47.476.677 I slot print_timing: id 1 | task 0 | eval time = 60909.77 ms / 5440 tokens ( 11.20 ms per token, 89.31 tokens per second) 1.47.476.677 I slot print_timing: id 1 | task 0 | total time = 61428.98 ms / 5702 tokens 1.47.476.681 I slot print_timing: id 1 | task 0 | graphs reused = 1480 1.47.476.684 I slot print_timing: id 1 | task 0 | draft acceptance = 0.87497 ( 3940 accepted / 4503 generated), mean acceptance length = 3.62, acceptance rate per position = (0.943, 0.875, 0.806)- 配合 Deepseek Harness 修改QT + C++ 项目 SimulIDE， 使其能在apple silicon上运行，并修改Qt 文本框输入bug
llama.cpp 日志中的一段
32.32.489.805 I slot print_timing: id 1 | task 10430 | prompt eval time = 17284.15 ms / 4098 tokens ( 4.22 ms per token, 237.10 tokens per second) 32.32.489.807 I slot print_timing: id 1 | task 10430 | eval time = 5461.48 ms / 277 tokens ( 19.72 ms per token, 50.72 tokens per second) 32.32.489.808 I slot print_timing: id 1 | task 10430 | total time = 22745.63 ms / 4375 tokens 32.32.489.808 I slot print_timing: id 1 | task 10430 | graphs reused = 9929 32.32.489.811 I slot print_timing: id 1 | task 10430 | draft acceptance = 0.89333 ( 201 accepted / 225 generated), mean acceptance length = 3.68, acceptance rate per position = (0.960, 0.893, 0.827)DSH 摘要：
63歩ILLM 13m13S・工具週用 2m2s｜首token 平均3.8s・52 tok/s|緩存命中98%1輸入 4.4M tok・輸出 29.2K tokllama.cpp运行参数：
-t 10 -b 512 -ub 256 --spec-draft-n-max 3 --fit off --no-context-shift --metrics --kv-unified --jinja --cache-type-k q8_0 --cache-type-v q8_0 -fa on --spec-type draft-mtp --ctx-size 131072 --parallel 2 -ngl -1 --host 0.0.0.0 --port 8080 --chat-template-kwargs {"enable_thinking": true, "preserve_think": false, "reasoning_effort": "medium"} --no-mmap --temp 0.6 --top-p 0.5 --top-k 15 --repeat-penalty 1.0 --override-tensor blk\.\d+\.ffn_.*_exps\.=CPU --alias qwen3.8-27b-q5 - llama.cpp 自带webui 创建打飞机游戏
-
所有测试都是基于unsloth/Qwen3.8-27B-GGUF
相同参数跑 Q6_K效果：
3.24.516.851 I slot print_timing: id 1 | task 0 | prompt eval time = 2188.90 ms / 263 tokens ( 8.32 ms per token, 120.15 tokens per second) 3.24.516.854 I slot print_timing: id 1 | task 0 | eval time = 162164.36 ms / 4584 tokens ( 35.38 ms per token, **28.27 tokens** per second) 3.24.516.854 I slot print_timing: id 1 | task 0 | total time = 164353.27 ms / 4847 tokens 3.24.516.859 I slot print_timing: id 1 | task 0 | graphs reused = 1263 3.24.516.862 I slot print_timing: id 1 | task 0 | draft acceptance = **0.86016** ( 3303 accepted / 3840 generated), mean acceptance length = 3.58, acceptance rate per position = (0.934, 0.869, 0.778)UD_Q6_K_XL:
3.24.218.990 I slot print_timing: id 1 | task 0 | eval time = 146468.18 ms / 3978 tokens ( 36.82 ms per token, 27.16 tokens per second) 3.24.218.991 I slot print_timing: id 1 | task 0 | total time = 148616.68 ms / 4240 tokens 3.24.218.996 I slot print_timing: id 1 | task 0 | graphs reused = 1109 3.24.218.999 I slot print_timing: id 1 | task 0 | draft acceptance = 0.84698 ( 2856 accepted / 3372 generated), mean acceptance length = 3.54, acceptance rate per position = (0.924, 0.849, 0.768) -
Harness 提示词： 介绍一下“八卦”的来由
-
冷启动
-
极简模式：
Total duration: 16.2 s
TTFT: 5.20 s
Generation: 11.0 s
Throughput: 74.8 tok/s -
标准模式：
Total duration: 20.1 s
TTFT: 12.3 s
Generation: 7.78 s
Throughput: 83.8 tok/s
-
-
缓存填满新开会话
-
极简模式：
Total duration: 17.1 s
TTFT: 2.65 s
Generation: 14.4 s
Throughput: 65.1 tok/s -
标准模式：
Total duration: 28.1 s
TTFT: 12.3 s
Generation: 15.8 s
Throughput: 61.0 tok/s
-
-
-
xtx能到这个水平实属不易，你上下文能做到多少？如果上下文也够，看你参数里开到128k，能驱动dsh，那就有实用价值了。说真的 ，xtx这种显卡就是安静的很，做工不错，功耗可控。能玩上deepseek harness这一条就能值回票价。
-
是的，我上下文128k， 如我在一楼写的，修改现有的c++项目(大约 355595行代码)，效果很好。 同样的任务比Deepseek flash慢 一半， 结合Harness一次性完成，效果一样。
在opencode中， DSF 一次完美完成， 本地Qwen需要多次修改。
-
，
T terry 固定了此主题
-
，T terry 引用了 此主题
-
作业质量非常高！9700x、64G ddr5、R9700 跑QWen3.8-27b-Q4_K_M。没抄作业之前，接入Codex干活吐字只有3.1 t/s，还经常崩掉，完全不能忍。抄作业以后，接入Codex干活吐字能到50t/s，提升十多倍，感觉顺畅了。万分感谢！
-
看视频了。抄作业，已关注。请继续更新。
-
抄了作业，跑楼主参数，只有20tokens/s，比较尴尬，经过deepseek了一下，应该是我的CPU有点弱，我将--override-tensor blk.\d+.ffn_.*_exps.=CPU \注释掉了，并将--parallel改成了1，其他参数未动，现在能到60t/s了
-
@exllm 并行改为1后，预填充性能有接近5~10倍的提升（不同的任务变化很大），但生成token的性能提升不到10%。我也将“--override-tensor blk.\d+.ffn_.*_exps.=CPU \”注释掉进行过测试，性能反而有略微下降，可能跟我的硬件和跑的任务也有一定关系。
