furyhawk/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-07-21 02:05:51 +00:00
Code Issues Packages Projects Releases Wiki Activity

use minimal shmem size 8 instead of 1 to workaround cm2 compiler bug

Browse Source
This commit is contained in:
Ruben Ortlam
2026-07-17 13:21:28 +02:00
parent b3edd252cf
commit bdccb7da38
1 changed files with 7 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-vulkan/vulkan-shaders/types.glsl
+7 -7
View File
@@ -892,9 +892,9 @@ const uint32_t[2048] iq1s_grid_gpu_const = {
#endif
#ifdef MULMAT_QUANT
shared uint16_t iq1s_grid[(MmTypeA == GGML_TYPE_IQ1_S || MmTypeA == GGML_TYPE_IQ1_M) ? 2048 : 1];
shared uint16_t iq1s_grid[(MmTypeA == GGML_TYPE_IQ1_S || MmTypeA == GGML_TYPE_IQ1_M) ? 2048 : 8];
#if defined(NEEDS_IQ1S_GRID_GPU)
shared uint32_t iq1s_grid_gpu[(MmTypeA == GGML_TYPE_IQ1_S || MmTypeA == GGML_TYPE_IQ1_M) ? 2048 : 1];
shared uint32_t iq1s_grid_gpu[(MmTypeA == GGML_TYPE_IQ1_S || MmTypeA == GGML_TYPE_IQ1_M) ? 2048 : 8];
#endif
#else
shared uint16_t iq1s_grid[2048];
@@ -931,7 +931,7 @@ void init_iq_shmem(uvec3 wgsize)
#if defined(DATA_A_IQ2_XXS) || defined(DATA_A_IQ2_XS) || defined(DATA_A_IQ2_S) || defined(MULMAT_QUANT)
#ifdef MULMAT_QUANT
shared uvec2 iq2_grid[MmTypeA == GGML_TYPE_IQ2_S ? 1024 : MmTypeA == GGML_TYPE_IQ2_XS ? 512 : MmTypeA == GGML_TYPE_IQ2_XXS ? 256 : 1];
shared uvec2 iq2_grid[MmTypeA == GGML_TYPE_IQ2_S ? 1024 : MmTypeA == GGML_TYPE_IQ2_XS ? 512 : MmTypeA == GGML_TYPE_IQ2_XXS ? 256 : 8];
#elif defined(DATA_A_IQ2_S)
shared uvec2 iq2_grid[1024];
#elif defined(DATA_A_IQ2_XS)
@@ -1523,7 +1523,7 @@ void init_iq_shmem(uvec3 wgsize)
#if defined(DATA_A_IQ3_XXS) || defined(DATA_A_IQ3_S) || defined(MULMAT_QUANT)
#ifdef MULMAT_QUANT
shared uint32_t iq3_grid[MmTypeA == GGML_TYPE_IQ3_S ? 512 : MmTypeA == GGML_TYPE_IQ3_XXS ? 256 : 1];
shared uint32_t iq3_grid[MmTypeA == GGML_TYPE_IQ3_S ? 512 : MmTypeA == GGML_TYPE_IQ3_XXS ? 256 : 8];
#elif defined(DATA_A_IQ3_S)
shared uint32_t iq3_grid[512];
#else
@@ -1828,7 +1828,7 @@ const int8_t kvalues_iq4nl_const[16] = {
};
#ifdef MULMAT_QUANT
shared FLOAT_TYPE kvalues_iq4nl[(MmTypeA == GGML_TYPE_IQ4_NL || MmTypeA == GGML_TYPE_IQ4_XS) ? 16 : 1];
shared FLOAT_TYPE kvalues_iq4nl[(MmTypeA == GGML_TYPE_IQ4_NL || MmTypeA == GGML_TYPE_IQ4_XS) ? 16 : 8];
#else
shared FLOAT_TYPE kvalues_iq4nl[16];
#endif
@@ -1854,7 +1854,7 @@ const int8_t kvalues_mxfp4_const[16] = {
};
#ifdef MULMAT_QUANT
shared int8_t kvalues_mxfp4[(MmTypeA == GGML_TYPE_MXFP4 || MmTypeA == GGML_TYPE_NVFP4) ? 16 : 1];
shared int8_t kvalues_mxfp4[(MmTypeA == GGML_TYPE_MXFP4 || MmTypeA == GGML_TYPE_NVFP4) ? 16 : 8];
#else
shared int8_t kvalues_mxfp4[16];
#endif
@@ -1862,7 +1862,7 @@ shared int8_t kvalues_mxfp4[16];
#if (defined(DATA_A_NVFP4) || defined(MULMAT_QUANT)) && !defined(USE_OCP_FP4)
#ifdef MULMAT_QUANT
shared float ue4m3_fp32_lut[MmTypeA == GGML_TYPE_NVFP4 ? 128 : 1];
shared float ue4m3_fp32_lut[MmTypeA == GGML_TYPE_NVFP4 ? 128 : 8];
#else
shared float ue4m3_fp32_lut[128];
#endif