mikhail/llama.cpp-mtp-turboquant
1
0
Fork 0
Code Issues Pull Requests Actions 32 Packages Projects Releases Wiki Activity

vulkan: restore -inf check in FA shaders (#19582)

Browse Source
This commit is contained in:
Jeff Bolz
2026-02-13 11:35:29 -08:00
committed by GitHub
parent b48e80f677
commit 05a6f0e894
4 changed files with 49 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-vulkan/ggml-vulkan.cpp
+3 -1
View File
@@ -8422,6 +8422,8 @@ static bool ggml_vk_flash_attn_coopmat_shmem_support(const vk_device& device, co
const uint32_t acctype = f32acc ? 4 : 2;
const uint32_t f16vec4 = 8;
const uint32_t tmpsh = (Bc / MatBc) * sizeof(float);
const uint32_t qstride = hsk_pad / 4 + 2;
const uint32_t Qf = Br * qstride * f16vec4;
@@ -8438,7 +8440,7 @@ static bool ggml_vk_flash_attn_coopmat_shmem_support(const vk_device& device, co
const uint32_t slope = Br * acctype;
const uint32_t total_size = Qf + Psh + sfsh + ksh + slope;
const uint32_t total_size = tmpsh + Qf + Psh + sfsh + ksh + slope;
const bool supported = total_size <= device->properties.limits.maxComputeSharedMemorySize;
VK_LOG_DEBUG("ggml_vk_flash_attn_coopmat_shmem_support(HSK=" << hsk << ", HSV=" << hsv << ", f32acc=" << f32acc << ", kv_type=" << kv_type << ", total_size=" << total_size << ", supported=" << supported);