holm/whisper.cpp
1
1
Fork 0
forked from extern/whisper.cpp
Code Issues Pull Requests Packages Projects Releases Wiki Activity

ggml : speed-up soft max via Accelerate + unroll

Browse Source
This commit is contained in:
Georgi Gerganov 2023-01-07 16:11:41 +02:00
parent d51fc3ee0a
commit d61d55cd4b
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
2 changed files with 114 additions and 54 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

166
ggml.c
View File

@ -81,6 +81,7 @@ typedef void* thread_ret_t;
#define GGML_DEBUG 0 #define GGML_DEBUG 0
#define GGML_GELU_FP16 #define GGML_GELU_FP16
#define GGML_SOFT_MAX_UNROLL 4
#if UINTPTR_MAX == 0xFFFFFFFF #if UINTPTR_MAX == 0xFFFFFFFF
#define GGML_MEM_ALIGN 4 #define GGML_MEM_ALIGN 4
@ -310,6 +311,7 @@ int64_t ggml_cycles_per_ms(void) {
return CLOCKS_PER_SEC/1000; return CLOCKS_PER_SEC/1000;
} }
//#define GGML_PERF
#ifdef GGML_PERF #ifdef GGML_PERF
#define ggml_perf_time_ms() ggml_time_ms() #define ggml_perf_time_ms() ggml_time_ms()
#define ggml_perf_time_us() ggml_time_us() #define ggml_perf_time_us() ggml_time_us()
@ -1316,25 +1318,25 @@ size_t ggml_element_size(const struct ggml_tensor * tensor) {
return GGML_TYPE_SIZE[tensor->type]; return GGML_TYPE_SIZE[tensor->type];
} }
bool ggml_is_scalar(const struct ggml_tensor * tensor) { static inline bool ggml_is_scalar(const struct ggml_tensor * tensor) {
static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function"); static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function");
return tensor->ne[0] == 1 && tensor->ne[1] == 1 && tensor->ne[2] == 1 && tensor->ne[3] == 1; return tensor->ne[0] == 1 && tensor->ne[1] == 1 && tensor->ne[2] == 1 && tensor->ne[3] == 1;
} }
bool ggml_is_vector(const struct ggml_tensor * tensor) { static inline bool ggml_is_vector(const struct ggml_tensor * tensor) {
static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function"); static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function");
return tensor->ne[1] == 1 && tensor->ne[2] == 1 && tensor->ne[3] == 1; return tensor->ne[1] == 1 && tensor->ne[2] == 1 && tensor->ne[3] == 1;
} }
bool ggml_is_matrix(const struct ggml_tensor * tensor) { static inline bool ggml_is_matrix(const struct ggml_tensor * tensor) {
static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function"); static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function");
return tensor->ne[2] == 1 && tensor->ne[3] == 1; return tensor->ne[2] == 1 && tensor->ne[3] == 1;
} }
bool ggml_can_mul_mat(const struct ggml_tensor * t0, const struct ggml_tensor * t1) { static inline bool ggml_can_mul_mat(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"); static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function");
return return
@ -1343,7 +1345,7 @@ bool ggml_can_mul_mat(const struct ggml_tensor * t0, const struct ggml_tensor *
(t0->ne[3] == t1->ne[3]); (t0->ne[3] == t1->ne[3]);
} }
bool ggml_is_contiguous(const struct ggml_tensor * tensor) { static inline bool ggml_is_contiguous(const struct ggml_tensor * tensor) {
static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function"); static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function");
return return
@ -1353,7 +1355,7 @@ bool ggml_is_contiguous(const struct ggml_tensor * tensor) {
tensor->nb[3] == tensor->nb[2]*tensor->ne[2]; tensor->nb[3] == tensor->nb[2]*tensor->ne[2];
} }
bool ggml_is_padded_1d(const struct ggml_tensor * tensor) { static inline bool ggml_is_padded_1d(const struct ggml_tensor * tensor) {
static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function"); static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function");
return return
@ -1362,7 +1364,7 @@ bool ggml_is_padded_1d(const struct ggml_tensor * tensor) {
tensor->nb[3] == tensor->nb[2]*tensor->ne[2]; tensor->nb[3] == tensor->nb[2]*tensor->ne[2];
} }
bool ggml_are_same_shape(const struct ggml_tensor * t0, const struct ggml_tensor * t1) { static inline bool ggml_are_same_shape(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"); static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function");
return return
@ -1373,7 +1375,7 @@ bool ggml_are_same_shape(const struct ggml_tensor * t0, const struct ggml_tensor
} }
// check if t1 can be represented as a repeatition of t0 // check if t1 can be represented as a repeatition of t0
bool ggml_can_repeat(const struct ggml_tensor * t0, const struct ggml_tensor * t1) { 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"); static_assert(GGML_MAX_DIMS == 4, "GGML_MAX_DIMS is not 4 - update this function");
return return
@ -1383,14 +1385,20 @@ bool ggml_can_repeat(const struct ggml_tensor * t0, const struct ggml_tensor * t
(t1->ne[3]%t0->ne[3] == 0); (t1->ne[3]%t0->ne[3] == 0);
} }
int ggml_up32(int n) { static inline int ggml_up32(int n) {
return (n + 31) & ~31; return (n + 31) & ~31;
} }
int ggml_up64(int n) { static inline int ggml_up64(int n) {
return (n + 63) & ~63; return (n + 63) & ~63;
} }
static inline int ggml_up(int n, int m) {
// assert m is a power of 2
GGML_ASSERT((m & (m - 1)) == 0);
return (n + m - 1) & ~(m - 1);
}
// assert that pointer is aligned to GGML_MEM_ALIGN // assert that pointer is aligned to GGML_MEM_ALIGN
#define ggml_assert_aligned(ptr) \ #define ggml_assert_aligned(ptr) \
assert(((uintptr_t) (ptr))%GGML_MEM_ALIGN == 0) assert(((uintptr_t) (ptr))%GGML_MEM_ALIGN == 0)
@ -5094,21 +5102,19 @@ static void ggml_compute_forward_soft_max_f32(
#endif #endif
float max = -INFINITY; float max = -INFINITY;
for (int i = 0; i < nc; i++) { ggml_vec_max_f32(nc, &max, p);
max = MAX(max, p[i]);
}
ggml_float sum = 0.0; ggml_float sum = 0.0;
uint16_t ss; uint16_t scvt;
for (int i = 0; i < nc; i++) { for (int i = 0; i < nc; i++) {
if (p[i] == -INFINITY) { if (p[i] == -INFINITY) {
p[i] = 0.0; p[i] = 0.0f;
} else { } else {
//const float val = (p[i] == -INFINITY) ? 0.0 : exp(p[i] - max); //const float val = (p[i] == -INFINITY) ? 0.0 : exp(p[i] - max);
ggml_fp16_t s = GGML_FP32_TO_FP16(p[i] - max); ggml_fp16_t s = GGML_FP32_TO_FP16(p[i] - max);
memcpy(&ss, &s, sizeof(ss)); memcpy(&scvt, &s, sizeof(scvt));
const float val = GGML_FP16_TO_FP32(table_exp_f16[ss]); const float val = GGML_FP16_TO_FP32(table_exp_f16[scvt]);
sum += val; sum += val;
p[i] = val; p[i] = val;
} }
@ -5820,6 +5826,8 @@ static void ggml_compute_forward_flash_attn_f32(
const int P = nek1 - N; const int P = nek1 - N;
const int M = P + N; const int M = P + N;
const int Mup = ggml_up(M, GGML_SOFT_MAX_UNROLL);
GGML_ASSERT(ne0 == D); GGML_ASSERT(ne0 == D);
GGML_ASSERT(ne1 == N); GGML_ASSERT(ne1 == N);
GGML_ASSERT(P >= 0); GGML_ASSERT(P >= 0);
@ -5872,7 +5880,11 @@ static void ggml_compute_forward_flash_attn_f32(
const int iq2 = (ir - iq3*neq2*neq1)/neq1; const int iq2 = (ir - iq3*neq2*neq1)/neq1;
const int iq1 = (ir - iq3*neq2*neq1 - iq2*neq1); const int iq1 = (ir - iq3*neq2*neq1 - iq2*neq1);
float * S = (float *) params->wdata + ith*(M + CACHE_LINE_SIZE_F32); float * S = (float *) params->wdata + ith*(Mup + CACHE_LINE_SIZE_F32);
for (int i = M; i < Mup; ++i) {
S[i] = -INFINITY;
}
for (int ic = 0; ic < nek1; ++ic) { for (int ic = 0; ic < nek1; ++ic) {
// k indices // k indices
@ -5903,30 +5915,50 @@ static void ggml_compute_forward_flash_attn_f32(
// softmax // softmax
{ {
float max = -INFINITY; float max = -INFINITY;
for (int i = 0; i < M; i++) { ggml_vec_max_f32(M, &max, S);
max = MAX(max, S[i]);
}
ggml_float sum = 0.0; float sum = 0.0f;
{
#ifndef GGML_USE_ACCELERATE
uint16_t scvt[GGML_SOFT_MAX_UNROLL];
ggml_float sump[GGML_SOFT_MAX_UNROLL] = { 0.0 };
uint16_t ss; for (int i = 0; i < Mup; i += GGML_SOFT_MAX_UNROLL) {
for (int i = 0; i < M; i++) { float * SS = S + i;
if (S[i] == -INFINITY) {
S[i] = 0.0; for (int j = 0; j < GGML_SOFT_MAX_UNROLL; ++j) {
} else { if (SS[j] == -INFINITY) {
//const float val = (S[i] == -INFINITY) ? 0.0 : exp(S[i] - max); SS[j] = 0.0f;
ggml_fp16_t s = GGML_FP32_TO_FP16(S[i] - max); } else {
memcpy(&ss, &s, sizeof(ss)); ggml_fp16_t s = GGML_FP32_TO_FP16(SS[j] - max);
const float val = GGML_FP16_TO_FP32(table_exp_f16[ss]); memcpy(&scvt[j], &s, sizeof(uint16_t));
sum += val; const float val = GGML_FP16_TO_FP32(table_exp_f16[scvt[j]]);
S[i] = val; sump[j] += val;
SS[j] = val;
}
}
} }
for (int i = 0; i < GGML_SOFT_MAX_UNROLL; i++) {
sum += sump[i];
}
#else
vvexpf(S, S, &Mup);
ggml_vec_sum_f32(Mup, &sum, S);
#endif
} }
assert(sum > 0.0f); assert(sum > 0.0f);
sum = 1.0/sum; sum = 1.0/sum;
ggml_vec_scale_f32(M, S, sum); ggml_vec_scale_f32(M, S, sum);
#ifndef NDEBUG
for (int i = 0; i < M; ++i) {
assert(!isnan(S[i]));
assert(!isinf(S[i]));
}
#endif
} }
for (int ic = 0; ic < nev1; ++ic) { for (int ic = 0; ic < nev1; ++ic) {
@ -6001,6 +6033,8 @@ static void ggml_compute_forward_flash_attn_f16(
const int P = nek1 - N; const int P = nek1 - N;
const int M = P + N; const int M = P + N;
const int Mup = ggml_up(M, GGML_SOFT_MAX_UNROLL);
GGML_ASSERT(ne0 == D); GGML_ASSERT(ne0 == D);
GGML_ASSERT(ne1 == N); GGML_ASSERT(ne1 == N);
GGML_ASSERT(P >= 0); GGML_ASSERT(P >= 0);
@ -6053,7 +6087,11 @@ static void ggml_compute_forward_flash_attn_f16(
const int iq2 = (ir - iq3*neq2*neq1)/neq1; const int iq2 = (ir - iq3*neq2*neq1)/neq1;
const int iq1 = (ir - iq3*neq2*neq1 - iq2*neq1); const int iq1 = (ir - iq3*neq2*neq1 - iq2*neq1);
float * S = (float *) params->wdata + ith*(2*M + CACHE_LINE_SIZE_F32); float * S = (float *) params->wdata + ith*(2*Mup + CACHE_LINE_SIZE_F32);
for (int i = M; i < Mup; ++i) {
S[i] = -INFINITY;
}
for (int ic = 0; ic < nek1; ++ic) { for (int ic = 0; ic < nek1; ++ic) {
// k indices // k indices
@ -6084,30 +6122,50 @@ static void ggml_compute_forward_flash_attn_f16(
// softmax // softmax
{ {
float max = -INFINITY; float max = -INFINITY;
for (int i = 0; i < M; i++) { ggml_vec_max_f32(M, &max, S);
max = MAX(max, S[i]);
}
ggml_float sum = 0.0; float sum = 0.0f;
{
#ifndef GGML_USE_ACCELERATE
uint16_t scvt[GGML_SOFT_MAX_UNROLL];
ggml_float sump[GGML_SOFT_MAX_UNROLL] = { 0.0 };
uint16_t ss; for (int i = 0; i < Mup; i += GGML_SOFT_MAX_UNROLL) {
for (int i = 0; i < M; i++) { float * SS = S + i;
if (S[i] == -INFINITY) {
S[i] = 0.0; for (int j = 0; j < GGML_SOFT_MAX_UNROLL; ++j) {
} else { if (SS[j] == -INFINITY) {
//const float val = (S[i] == -INFINITY) ? 0.0 : exp(S[i] - max); SS[j] = 0.0f;
ggml_fp16_t s = GGML_FP32_TO_FP16(S[i] - max); } else {
memcpy(&ss, &s, sizeof(ss)); ggml_fp16_t s = GGML_FP32_TO_FP16(SS[j] - max);
const float val = GGML_FP16_TO_FP32(table_exp_f16[ss]); memcpy(&scvt[j], &s, sizeof(uint16_t));
sum += val; const float val = GGML_FP16_TO_FP32(table_exp_f16[scvt[j]]);
S[i] = val; sump[j] += val;
SS[j] = val;
}
}
} }
for (int i = 0; i < GGML_SOFT_MAX_UNROLL; i++) {
sum += sump[i];
}
#else
vvexpf(S, S, &Mup);
ggml_vec_sum_f32(Mup, &sum, S);
#endif
} }
assert(sum > 0.0f); assert(sum > 0.0f);
sum = 1.0/sum; sum = 1.0/sum;
ggml_vec_scale_f32(M, S, sum); ggml_vec_scale_f32(M, S, sum);
#ifndef NDEBUG
for (int i = 0; i < M; ++i) {
assert(!isnan(S[i]));
assert(!isinf(S[i]));
}
#endif
} }
ggml_fp16_t * S16 = (ggml_fp16_t *) ((float *) params->wdata + ith*(2*M + CACHE_LINE_SIZE_F32) + M); ggml_fp16_t * S16 = (ggml_fp16_t *) ((float *) params->wdata + ith*(2*M + CACHE_LINE_SIZE_F32) + M);
@ -7188,14 +7246,16 @@ void ggml_graph_compute(struct ggml_context * ctx, struct ggml_cgraph * cgraph)
size_t cur = 0; size_t cur = 0;
const int ne11 = ggml_up(node->src1->ne[1], GGML_SOFT_MAX_UNROLL);
if (node->src1->type == GGML_TYPE_F32) { if (node->src1->type == GGML_TYPE_F32) {
cur = sizeof(float)*node->src1->ne[1]*node->n_tasks; // TODO: this can become (n_tasks-1) cur = sizeof(float)*ne11*node->n_tasks; // TODO: this can become (n_tasks-1)
cur += sizeof(float)*node->src1->ne[1]*node->n_tasks; // this is overestimated by x2 cur += sizeof(float)*ne11*node->n_tasks; // this is overestimated by x2
} }
if (node->src1->type == GGML_TYPE_F16) { if (node->src1->type == GGML_TYPE_F16) {
cur = sizeof(float)*node->src1->ne[1]*node->n_tasks; // TODO: this can become (n_tasks-1) cur = sizeof(float)*ne11*node->n_tasks; // TODO: this can become (n_tasks-1)
cur += sizeof(float)*node->src1->ne[1]*node->n_tasks; // this is overestimated by x2 cur += sizeof(float)*ne11*node->n_tasks; // this is overestimated by x2
} }
work_size = MAX(work_size, cur); work_size = MAX(work_size, cur);

2
whisper.cpp
View File

@ -131,7 +131,7 @@ static const std::map<std::string, std::pair<int, std::string>> g_lang = {
{ "su", { 98, "sundanese", } }, { "su", { 98, "sundanese", } },
}; };
static const size_t MB = 1024*1024; static const size_t MB = 3*1024*1024;
static const std::map<e_model, size_t> MEM_REQ_MODEL = { static const std::map<e_model, size_t> MEM_REQ_MODEL = {
{ MODEL_TINY, 74ull*MB }, { MODEL_TINY, 74ull*MB },