root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-26 11:24:35 +00:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

vulkan: Set limit for task concurrency (#5427)

Browse Source 
A common default for the maximum number of open files is 256, which can
lead to `asyncio.gather(*tasks)` failing with Too many open files.

    $ python ggml_vk_generate_shaders.py --glslc=$ANDROID_NDK_PATH/shader-tools/darwin-x86_64/glslc
    ggml_vulkan: Generating and compiling shaders to SPIR-V
    Traceback (most recent call last):
      File "/Users/neuman/Code.noindex/github/llama.cpp/ggml_vk_generate_shaders.py", line 2326, in <module>
        asyncio.run(main())
      File "/Users/neuman/Code.noindex/miniforge3/lib/python3.10/asyncio/runners.py", line 44, in run
        return loop.run_until_complete(main)
      File "/Users/neuman/Code.noindex/miniforge3/lib/python3.10/asyncio/base_events.py", line 649, in run_until_complete
        return future.result()
      File "/Users/neuman/Code.noindex/github/llama.cpp/ggml_vk_generate_shaders.py", line 2294, in main
        await asyncio.gather(*tasks)
    [...snip...]
    OSError: [Errno 24] Too many open files

This change sets a reasonable concurrency limit for tasks (and therefore
open files), without significant impact on run time.
This commit is contained in:
Neuman Vong 2024-02-10 05:30:19 +11:00 committed by GitHub
parent e00d2a62dd
commit 4b7b38bef5
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 10 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
ggml_vk_generate_shaders.py
View File

@ -2067,6 +2067,8 @@ type_names = {
K_QUANTS_PER_ITERATION = 2 K_QUANTS_PER_ITERATION = 2
ASYNCIO_CONCURRENCY = 64
output_dir = gettempdir() output_dir = gettempdir()
lock = asyncio.Lock() lock = asyncio.Lock()
@ -2291,7 +2293,14 @@ async def main():
tasks.append(string_to_spv("rope_neox_f32", rope_neox_src, {"A_TYPE": "float", "D_TYPE": "float"})) tasks.append(string_to_spv("rope_neox_f32", rope_neox_src, {"A_TYPE": "float", "D_TYPE": "float"}))
tasks.append(string_to_spv("rope_neox_f16", rope_neox_src, {"A_TYPE": "float16_t", "D_TYPE": "float16_t"})) tasks.append(string_to_spv("rope_neox_f16", rope_neox_src, {"A_TYPE": "float16_t", "D_TYPE": "float16_t"}))
await asyncio.gather(*tasks) # Helper to decorate tasks with semaphore acquisition.
async def withSemaphore(sem, task):
async with sem:
return await task
# Run tasks concurrently guarded by a concurrency limit.
sem = asyncio.Semaphore(ASYNCIO_CONCURRENCY)
await asyncio.gather(*(withSemaphore(sem, task) for task in tasks))
with open("ggml-vulkan-shaders.hpp", "w") as f: with open("ggml-vulkan-shaders.hpp", "w") as f:
f.write("#include <cstdint>\n\n") f.write("#include <cstdint>\n\n")