root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2025-01-07 17:21:46 +00:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

float -> half regs

Browse Source
This commit is contained in:
Georgi Gerganov 2024-11-07 17:06:34 +02:00
parent e121d82f6a
commit 3ab47eb746
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
1 changed files with 18 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

36
ggml/src/ggml-metal.metal
View File

@ -2898,8 +2898,8 @@ kernel void kernel_flash_attn_ext(
threadgroup_barrier(mem_flags::mem_threadgroup);
{
float S[Q] = { [0 ... Q-1] = 0.0f };
float M[Q] = { [0 ... Q-1] = -FLT_MAX/2 };
half S[Q] = { [0 ... Q-1] = 0.0f };
half M[Q] = { [0 ... Q-1] = -__FLT16_MAX__/2 };
// thread indices inside the simdgroup
// TODO: see if we can utilize quad-group functions for better performance
@ -2934,14 +2934,14 @@ kernel void kernel_flash_attn_ext(
const bool has_mask = mask != q;
float slope = 1.0f;
half slope = 1.0f;
// ALiBi
if (max_bias > 0.0f) {
const uint32_t h = iq2;
const short h = iq2;
const float base = h < n_head_log2 ? m0 : m1;
const int exph = h < n_head_log2 ? h + 1 : 2*(h - n_head_log2) + 1;
const half base = h < n_head_log2 ? m0 : m1;
const short exph = h < n_head_log2 ? h + 1 : 2*(h - n_head_log2) + 1;
slope = pow(base, exph);
}
@ -3047,10 +3047,10 @@ kernel void kernel_flash_attn_ext(
// online softmax
{
for (short j = 0; j < Q; ++j) {
const float m = M[j];
const half m = M[j];
// scale and apply the logitcap / mask
float s = ss[j*TS + tiisg]*scale;
half s = ss[j*TS + tiisg]*scale;
if (logit_softcap != 0.0f) {
s = logit_softcap*precise::tanh(s);
@ -3061,8 +3061,8 @@ kernel void kernel_flash_attn_ext(
M[j] = simd_max(max(M[j], s));
const float ms = exp(m - M[j]);
const float vs = exp(s - M[j]);
const half ms = exp(m - M[j]);
const half vs = exp(s - M[j]);
S[j] = S[j]*ms + simd_sum(vs);
@ -3163,8 +3163,8 @@ kernel void kernel_flash_attn_ext(
// reduce the warps sequentially
for (short sg = 1; sg < nsg; ++sg) {
float S = { 0.0f };
float M = { -FLT_MAX/2 };
half S = { 0.0f };
half M = { -__FLT16_MAX__/2 };
threadgroup_barrier(mem_flags::mem_threadgroup);
@ -3180,16 +3180,16 @@ kernel void kernel_flash_attn_ext(
// the first simdgroup accumulates the results from the other simdgroups
if (sgitg == 0) {
for (short j = 0; j < Q; ++j) {
const float S0 = ss[j*TS + 0];
const float S1 = ss[j*TS + sg*SH + 0];
const half S0 = ss[j*TS + 0];
const half S1 = ss[j*TS + sg*SH + 0];
const float M0 = ss[j*TS + 1];
const float M1 = ss[j*TS + sg*SH + 1];
const half M0 = ss[j*TS + 1];
const half M1 = ss[j*TS + sg*SH + 1];
M = max(M0, M1);
const float ms0 = exp(M0 - M);
const float ms1 = exp(M1 - M);
const half ms0 = exp(M0 - M);
const half ms1 = exp(M1 - M);
S = S0*ms0 + S1*ms1;