Vulkan Flash Attention Coopmat1 Refactor (#19075)

* vulkan: use coopmat for flash attention p*v matrix multiplication

* fix P loading issue

* fix barrier position

* remove reduction that is no longer needed

* move max thread reduction into loop

* remove osh padding

* add bounds checks and padding

* remove unused code

* fix shmem sizes, loop duration and accesses

* don't overwrite Qf, add new shared psh buffer instead

* add missing bounds checks

* use subgroup reductions

* optimize

* move bounds check, reduce barriers

* support other Bc values and other subgroup sizes

* remove D_split

* replace Of register array with shared memory Ofsh array

* parallelize HSV across the rowgroups

* go back to Of in registers, not shmem

* vectorize sfsh

* don't store entire K tile in shmem

* fixes

* load large k tiles to shmem on Nvidia

* adapt shared memory host check function to shader changes

* remove Bc 32 case

* remove unused variable

* fix missing mask reduction tmspsh barrier

* fix mask bounds check

* fix rowmax f16 under/overflow to inf

* fix flash_attn_cm2 BLOCK_SIZE preprocessor directives

ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_base.glsl

@@ -8,6 +8,8 @@ layout (constant_id = 3) const uint32_t HSK = 32;
 layout (constant_id = 4) const uint32_t HSV = 32;
 layout (constant_id = 5) const uint32_t Clamp = 0;
 layout (constant_id = 6) const uint32_t D_split = 16;
+layout (constant_id = 7) const uint32_t SubGroupSize = 32;
+layout (constant_id = 8) const uint32_t K_LOAD_SHMEM = 0;
 
 // Round up head sizes to a multiple of 16, for coopmat1/coopmat2 paths
 const uint32_t HSK_pad = (HSK + 15) & ~15;
@@ -74,6 +76,10 @@ layout (binding = 1) readonly buffer K_PACKED16 {A_TYPE_PACKED16 k_data_packed16
 layout (binding = 2) readonly buffer V_PACKED16 {A_TYPE_PACKED16 v_data_packed16[];} v_packed;
 #endif
 
+#ifndef BLOCK_SIZE
+#define BLOCK_SIZE 1
+#endif
+
 #if defined(DATA_A_F32)
 #undef BLOCK_SIZE
 #define BLOCK_SIZE 4