vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Added gqa8 kernel to allow llama-2-70B on metal

Browse source
This commit is contained in:
Matteo Boschini 2023-07-31 00:02:04 +02:00
parent 2dbf518911
commit ae58ac7dd4
2 changed files with 62 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

17
ggml-metal.m
View file

@ -65,6 +65,7 @@ struct ggml_metal_context {
GGML_METAL_DECL_KERNEL(rms_norm);
GGML_METAL_DECL_KERNEL(norm);
GGML_METAL_DECL_KERNEL(mul_mat_f16_f32);
GGML_METAL_DECL_KERNEL(mul_mat_f16_f32_gqa8);
GGML_METAL_DECL_KERNEL(mul_mat_q4_0_f32);
GGML_METAL_DECL_KERNEL(mul_mat_q4_1_f32);
GGML_METAL_DECL_KERNEL(mul_mat_q2_K_f32);
@ -182,6 +183,7 @@ struct ggml_metal_context * ggml_metal_init(int n_cb) {
GGML_METAL_ADD_KERNEL(rms_norm);
GGML_METAL_ADD_KERNEL(norm);
GGML_METAL_ADD_KERNEL(mul_mat_f16_f32);
GGML_METAL_ADD_KERNEL(mul_mat_f16_f32_gqa8);
GGML_METAL_ADD_KERNEL(mul_mat_q4_0_f32);
GGML_METAL_ADD_KERNEL(mul_mat_q4_1_f32);
GGML_METAL_ADD_KERNEL(mul_mat_q2_K_f32);
@ -718,7 +720,8 @@ void ggml_metal_graph_compute(
// TODO: needs to be updated after PR: https://github.com/ggerganov/ggml/pull/224
GGML_ASSERT(ne00 == ne10);
GGML_ASSERT(ne02 == ne12);
int llama_2_70_gqa_step = ne02 == 8 && ne12 == 64;
GGML_ASSERT(ne02 == ne12 || llama_2_70_gqa_step);
if (ggml_is_contiguous(src0) &&
ggml_is_contiguous(src1) &&
@ -749,8 +752,8 @@ void ggml_metal_graph_compute(
// we need to do ne02 multiplications
// TODO: is there a way to do this in parallel - currently very slow ..
// TODO: might be possible to offload part of the computation to ANE using Accelerate's CBLAS
for (int64_t i02 = 0; i02 < ne02; ++i02) {
size_t offs_src0_cur = offs_src0 + i02*nb02;
for (int64_t i02 = 0; i02 < ne12; ++i02) {
size_t offs_src0_cur = offs_src0 + i02/(ne12/ne02)*nb02; // gqa not used for now
size_t offs_src1_cur = offs_src1 + i02*nb12;
size_t offs_dst_cur = offs_dst + i02*nb2;
@ -772,11 +775,15 @@ void ggml_metal_graph_compute(
switch (src0t) {
case GGML_TYPE_F16:
{
GGML_ASSERT(ne02 == ne12);
GGML_ASSERT(ne02 == ne12 || llama_2_70_gqa_step);
nth0 = 64;
nth1 = 1;
[encoder setComputePipelineState:ctx->pipeline_mul_mat_f16_f32];
if (llama_2_70_gqa_step) {
[encoder setComputePipelineState:ctx->pipeline_mul_mat_f16_f32_gqa8];
} else {
[encoder setComputePipelineState:ctx->pipeline_mul_mat_f16_f32];
}
} break;
case GGML_TYPE_Q4_0:
{

50
ggml-metal.metal
View file

@ -552,6 +552,56 @@ kernel void kernel_mul_mat_f16_f32(
}
}
kernel void kernel_mul_mat_f16_f32_gqa8(
device const char * src0,
device const char * src1,
device float * dst,
constant int64_t & ne00,
constant int64_t & ne01,
constant uint64_t & nb00,
constant uint64_t & nb01,
constant uint64_t & nb02,
constant int64_t & ne10,
constant int64_t & ne11,
constant uint64_t & nb10,
constant uint64_t & nb11,
constant uint64_t & nb12,
constant int64_t & ne0,
constant int64_t & ne1,
threadgroup float * sum [[threadgroup(0)]],
uint3 tgpig[[threadgroup_position_in_grid]],
uint3 tpig[[thread_position_in_grid]],
uint3 tpitg[[thread_position_in_threadgroup]],
uint3 tptg[[threads_per_threadgroup]]) {
const int64_t r0 = tgpig.x;
const int64_t r1 = tgpig.y;
const int64_t im = tgpig.z;
device const half * x = (device const half *) (src0 + r0*nb01 + im/8*nb02);
device const float * y = (device const float *) (src1 + r1*nb11 + im*nb12);
sum[tpitg.x] = 0.0f;
for (int i = tpitg.x; i < ne00; i += tptg.x) {
sum[tpitg.x] += (float) x[i] * (float) y[i];
}
// accumulate the sum from all threads in the threadgroup
threadgroup_barrier(mem_flags::mem_threadgroup);
for (uint i = tptg.x/2; i > 0; i /= 2) {
if (tpitg.x < i) {
sum[tpitg.x] += sum[tpitg.x + i];
}
threadgroup_barrier(mem_flags::mem_threadgroup);
}
if (tpitg.x == 0) {
dst[im*ne1*ne0 + r1*ne0 + r0] = sum[0];
}
}
kernel void kernel_alibi_f32(
device const float * src0,
device float * dst,