For simplicity, unlike many other JavaScript runtimes, a lot of the operations in k6 are synchronous. That means that, for example, the let response = http.get("https://test-api.k6.io/") call from the Running k6 example script will block the VU execution until the HTTP request is completed, save the response information in the response variable and only then continue executing the rest of the script - no callbacks and promises needed.
KataGo 是目前 open source 裡最強的計算引擎了,不過先前的缺點就是得透過 OpenCL 或是 CUDA 才能跑,所以基本上得有張夠力的顯示卡才行。
如果要想要在 CPU 上跑 (不透過硬體顯示卡),一種方式是透過 OpenCL 的方式模擬,在 Linux 下可以透過 pocl 達成,效能就普普通通,但算是會動的東西,不過 Windows 下好像不太好弄... 這也是先前蠻多人還是繼續使用 Leela Zero 的原因。
最近 KataGo 在 1.5 版實做了純 CPU 版本的程式碼,是透過 Eigen 這套 library 達成的,不過大家測過以後發現慢到爆炸 XDDD
因為作者沒有提供 CPU 版本的 binary,我自己在 Linux 下抓程式碼 compile 後測試發現只會用一個 CPU (沒有 multi threading),對比於在 1080Ti 上跑 OpenCL 版本大約 150 visits/sec (40b),但 CPU 版本是 0.0x visits/sec 啊 XDDD
作者自己在 GitHub 上討論時也有提到這個版本只有確認正確性,完全沒有考慮效能...
不過就有其他人跳出來改善了,在「Optimization of Eigen backend #288」這邊可以看到 kaorahi 拋出了不少修改,可以看到從一開始的 eigen_naive_loop (對比 1.5 版有 13x 的成長) 一路到 borrow_tensorflow (1400x) 的版本,使得在 CPU 上面跑 15b 也有 10 visits/sec 了:
"borrow_tensorflow" version: x1400 speed up from 1.5.0 (70% of libtensorflow backend). Now 15b net is usable for me. I get 19 visits/s in benchmark and 10 visits/s in GUI with 15b net.
這樣看起來已經快了不少,這樣子 Leela Zero 應該會逐漸淡出了,CPU-only 算是最後一塊 Leela Zero 還可以爭的地盤...
It's written in Go, a language that compiles to native code
Parsing, printing, and source map generation are all fully parallelized
Everything is done in very few passes without expensive data transformations
Code is written with speed in mind, and tries to avoid unnecessary allocations
不過作者有提到這個專案畢竟比較新,還沒有被時間磨練過,可能會有些 bug:
This is a hobby project that I wrote over the 2019-2020 winter break. I believe that it's relatively complete and functional. However, it's brand new code and probably has a lot of bugs. It also hasn't yet been used in production by anyone. Use at your own risk.
The first serious run of KataGo ran for 7 days in Februrary 2019 on up to 35xV100 GPUs. This is the run featured in the paper. It achieved close to LZ130 strength before it was halted, or up to just barely superhuman.
Following some further improvements and much-improved hyperparameters, KataGo performed a second serious run in May-June a max of 28xV100 GPUs, surpassing the February run after just three and a half days. The run was halted after 19 days, with the final 20-block networks reaching a final strength slightly stronger than LZ-ELFv2! (This is Facebook's very strong 20-block ELF network, running on Leela Zero's search architecture). Comparing to the yet larger Leela Zero 40-block networks, KataGo's network falls somewhere around LZ200 at visit parity, despite only itself being 20 blocks.
從論文裡面可以看到,跟 Leela Zero 一樣是逐步提昇 (應該也是用 Net2Net),而不是一開始就拉到 20x256:
In KataGo’s main 19-day run, (b, c) began at (6, 96) and switched to (10, 128), (15, 192), and (20, 256), at roughly 0.75 days, 1.75 days, and 7.5 days, respectively. The final size approximately matches that of AlphaZero and ELF.
訓練速度上會有這麼大的改善,分成兩個類型,一種是一般性的 (在「Major General Improvements」這章),另外一類是特定於圍棋領域的改進 (在「Major Domain-Specific Improvements」這章)。
在 Leela Zero 的 issue tracking 裡面也可以看到很多關於 KataGo 的消息,看起來作者也在裡面一起討論,應該會有一些結果出來...
沒有太意外是使用 Leela Zero + Lizzle,畢竟這是 open source project,在軟體與資料的取得上相當方便,而且在好的硬體上已經可以超越人類頂尖棋手。
由於在 Lizzle 的介面上可以看到勝率,以及 Leela Zero 考慮的下一手 (通常會有多個選點),而且當游標移到這些選點上以後,還會有可能的變化圖可以看,所以對於棋手在熟悉操作介面後,可以很快的擺個變化圖,然後讓 Leela Zero 分析後續的發展,而棋手就可以快速判斷出「喔喔原來是這樣啊」。