metal : support FA without mask + add asserts (llama/7278)

* ggml : fa without mask + add asserts

ggml-ci

* metal : support non-contiguous KV

ggml-ci
This commit is contained in:
Georgi Gerganov
2024-05-14 19:09:30 +03:00
parent c451080c8b
commit 1056ad762c
4 changed files with 70 additions and 65 deletions

10
ggml.c
View File

@ -2824,6 +2824,16 @@ bool ggml_are_same_shape(const struct ggml_tensor * t0, const struct ggml_tensor
(t0->ne[3] == t1->ne[3] );
}
bool ggml_are_same_stride(const struct ggml_tensor * t0, const struct ggml_tensor * t1) {
static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function");
return
(t0->nb[0] == t1->nb[0] ) &&
(t0->nb[1] == t1->nb[1] ) &&
(t0->nb[2] == t1->nb[2] ) &&
(t0->nb[3] == t1->nb[3] );
}
// check if t1 can be represented as a repeatition of t0
static inline bool ggml_can_repeat(const struct ggml_tensor * t0, const struct ggml_tensor * t1) {
static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function");