xinyun/DeepSeek-V3
1
0
Fork 0
mirror of https://git.datalinker.icu/deepseek-ai/DeepSeek-V3.git synced 2025-12-09 04:44:28 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix act_quant_kernel

Browse Source 
Signed-off-by: youkaichao <youkaichao@gmail.com>
This commit is contained in:
youkaichao 2025-08-27 16:20:55 +08:00
parent b15f0dbbbe
commit 4667d669c7
1 changed files with 1 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
inference/kernel.py
View File

@ -23,8 +23,7 @@ def act_quant_kernel(x_ptr, y_ptr, s_ptr, BLOCK_SIZE: tl.constexpr, scale_fmt: t
pid = tl.program_id(axis=0)
offs = pid * BLOCK_SIZE + tl.arange(0, BLOCK_SIZE)
x = tl.load(x_ptr + offs).to(tl.float32)
amax = tl.max(tl.abs(x))
amax = tl.min(amax, 1e-4)
amax = tl.max(tl.abs(x), 1e-4)
s = amax / 448.
if scale_fmt == "ue8m0":
exp = tl.math.ceil(tl.math.log2(s))