root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2025-01-13 04:00:16 +00:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

metal : opt mul_mm_id

Browse Source
This commit is contained in:
Georgi Gerganov 2024-01-02 20:50:18 +02:00
parent 21e100d6dc
commit 9f51f3e695
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
1 changed files with 0 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
ggml-metal.metal
View File

@ -3974,7 +3974,6 @@ void kernel_mul_mm_impl(device const uchar * src0,
}
// same as kernel_mul_mm_impl, but src1 and dst are accessed via indices stored in src1ids
// TODO: simplify and optimize constants
template<typename block_q, short nl, void (*dequantize_func)(device const block_q *, short, thread half4x4 &)>
void kernel_mul_mm_id_impl(
device const uchar * src0,
@ -4042,7 +4041,6 @@ void kernel_mul_mm_id_impl(
dequantize_func(x, il, temp_a);
threadgroup_barrier(mem_flags::mem_threadgroup);
#pragma unroll(16)
for (int i = 0; i < 16; i++) {
*(sa + SG_MAT_SIZE * ((tiitg / THREAD_PER_ROW / 8) \
+ (tiitg % THREAD_PER_ROW) * 16 + (i / 8) * 8) \
@ -4061,14 +4059,11 @@ void kernel_mul_mm_id_impl(
threadgroup half * lsma = (sa + THREAD_MAT_M * SG_MAT_SIZE * (sgitg % 2));
threadgroup float * lsmb = (sb + THREAD_MAT_N * SG_MAT_SIZE * (sgitg / 2));
#pragma unroll(4)
for (int ik = 0; ik < BLOCK_SIZE_K / 8; ik++) {
#pragma unroll(4)
for (int i = 0; i < 4; i++) {
simdgroup_load(ma[i],lsma + SG_MAT_SIZE * i);
}
simdgroup_barrier(mem_flags::mem_none);
#pragma unroll(2)
for (int i = 0; i < 2; i++) {
simdgroup_load(mb[i],lsmb + SG_MAT_SIZE * i);
}
@ -4076,23 +4071,13 @@ void kernel_mul_mm_id_impl(
lsma += BLOCK_SIZE_M / SG_MAT_ROW * SG_MAT_SIZE;
lsmb += BLOCK_SIZE_N / SG_MAT_ROW * SG_MAT_SIZE;
#pragma unroll(8)
for (int i = 0; i < 8; i++){
simdgroup_multiply_accumulate(c_res[i], mb[i/4], ma[i%4], c_res[i]);
}
}
}
// TODO: this branch is invalid - need to fix it
//if ((r0 + 1) * BLOCK_SIZE_M <= ne0 && (r1 + 1) * BLOCK_SIZE_N <= ne1) {
// device float * C = dst + (BLOCK_SIZE_M * r0 + 32 * (sgitg & 1)) \
// + im*ne1*ne0;
// for (int i = 0; i < 8; i++) {
// simdgroup_store(c_res[i], C + 8 * (i%4) + ne0*src1ids[8*(i/4) + BLOCK_SIZE_N * r1 + 16 * (sgitg >> 1)], ne0);
// }
//} else {
{
// block is smaller than 64x32, we should avoid writing data outside of the matrix
threadgroup_barrier(mem_flags::mem_threadgroup);
threadgroup float * temp_str = ((threadgroup float *)shared_memory) \
+ 32 * (sgitg&1) + (16 * (sgitg>>1)) * BLOCK_SIZE_M;