mat*mat for q4_0, q8_0

9 months ago · 64001a480a
parent 04499d1c7d
commit 64001a480a
2 changed files with 9 additions and 1 deletions
--- a/gpt4all-backend/llama.cpp-mainline
+++ b/gpt4all-backend/llama.cpp-mainline
@ -1 +1 @@
-Subproject commit 7b8f00f5ccf4fc3cc67fe1ced792b3aec1ae6c1c
+Subproject commit 6c2e230c3315697316adb2656a6309678cf092c0
--- a/gpt4all-backend/llama.cpp.cmake
+++ b/gpt4all-backend/llama.cpp.cmake
@ -238,6 +238,10 @@ if (LLAMA_KOMPUTE)
          kompute/op_norm.comp
          kompute/op_rmsnorm.comp
          kompute/op_diagmask.comp
+          kompute/op_mul_mat_mat_f32.comp
+          kompute/op_mul_mat_mat_f16.comp
+          kompute/op_mul_mat_mat_q8_0.comp
+          kompute/op_mul_mat_mat_q4_0.comp
          kompute/op_mul_mat_f16.comp
          kompute/op_mul_mat_q8_0.comp
          kompute/op_mul_mat_q4_0.comp
@ -268,6 +272,10 @@ if (LLAMA_KOMPUTE)
          shaderop_norm.h
          shaderop_rmsnorm.h
          shaderop_diagmask.h
+          shaderop_mul_mat_mat_f32.h
+          shaderop_mul_mat_mat_f16.h
+          shaderop_mul_mat_mat_q8_0.h
+          shaderop_mul_mat_mat_q4_0.h
          shaderop_mul_mat_f16.h
          shaderop_mul_mat_q8_0.h
          shaderop_mul_mat_q4_0.h