root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-26 03:14:35 +00:00
Code Issues Actions 7 Packages Projects Releases Wiki Activity

CUDA: MMQ support for iq4_nl, iq4_xs (#8278)

Browse Source
This commit is contained in:
Johannes Gäßler 2024-07-05 09:06:31 +02:00 committed by GitHub
parent 0a423800ff
commit 8e558309dc
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
7 changed files with 226 additions and 80 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
ggml/src/ggml-cuda/fattn-common.cuh
View File

@ -68,7 +68,7 @@ static __device__ __forceinline__ T vec_dot_fattn_vec_KQ_q4_0(
const int iqs4 = k_KQ % QI4_0; const int iqs4 = k_KQ % QI4_0;
const int shift = k_KQ & (QI8_1/2); const int shift = k_KQ & (QI8_1/2);
const int v = (get_int_from_uint8(K_q4_0[ib].qs, iqs4) >> shift) & 0x0F0F0F0F; const int v = (get_int_b2(K_q4_0[ib].qs, iqs4) >> shift) & 0x0F0F0F0F;
const int u = Q_q8[k_KQ_0/WARP_SIZE]; const int u = Q_q8[k_KQ_0/WARP_SIZE];
const int sumi = ggml_cuda_dp4a(v, u, 0); const int sumi = ggml_cuda_dp4a(v, u, 0);
@ -108,7 +108,7 @@ static __device__ __forceinline__ T vec_dot_fattn_vec_KQ_q4_1(
const int iqs4 = k_KQ % QI4_1; const int iqs4 = k_KQ % QI4_1;
const int shift = k_KQ & (QI8_1/2); const int shift = k_KQ & (QI8_1/2);
const int v = (get_int_from_uint8_aligned(K_q4_1[ib].qs, iqs4) >> shift) & 0x0F0F0F0F; const int v = (get_int_b4(K_q4_1[ib].qs, iqs4) >> shift) & 0x0F0F0F0F;
const int u = Q_q8[k_KQ_0/WARP_SIZE]; const int u = Q_q8[k_KQ_0/WARP_SIZE];
const int sumi = ggml_cuda_dp4a(v, u, 0); const int sumi = ggml_cuda_dp4a(v, u, 0);
@ -153,8 +153,8 @@ static __device__ __forceinline__ T vec_dot_fattn_vec_KQ_q5_0(
const int iqs8 = k_KQ % QI8_1; const int iqs8 = k_KQ % QI8_1;
const int shift = k_KQ & (QI8_1/2); const int shift = k_KQ & (QI8_1/2);
int v = (get_int_from_uint8(K_q5_0[ib].qs, iqs4) >> shift) & 0x0F0F0F0F; int v = (get_int_b2(K_q5_0[ib].qs, iqs4) >> shift) & 0x0F0F0F0F;
const int vh = get_int_from_uint8(K_q5_0[ib].qh, 0) >> (iqs8 * QI5_0); const int vh = get_int_b2(K_q5_0[ib].qh, 0) >> (iqs8 * QI5_0);
v |= (vh << 4) & 0x00000010; // 0 -> 4 v |= (vh << 4) & 0x00000010; // 0 -> 4
v |= (vh << 11) & 0x00001000; // 1 -> 12 v |= (vh << 11) & 0x00001000; // 1 -> 12
v |= (vh << 18) & 0x00100000; // 2 -> 20 v |= (vh << 18) & 0x00100000; // 2 -> 20
@ -200,8 +200,8 @@ static __device__ __forceinline__ T vec_dot_fattn_vec_KQ_q5_1(
const int iqs8 = k_KQ % QI8_1; const int iqs8 = k_KQ % QI8_1;
const int shift = k_KQ & (QI8_1/2); const int shift = k_KQ & (QI8_1/2);
int v = (get_int_from_uint8(K_q5_1[ib].qs, iqs4) >> shift) & 0x0F0F0F0F; int v = (get_int_b2(K_q5_1[ib].qs, iqs4) >> shift) & 0x0F0F0F0F;
const int vh = get_int_from_uint8(K_q5_1[ib].qh, 0) >> (iqs8 * QI5_1); const int vh = get_int_b2(K_q5_1[ib].qh, 0) >> (iqs8 * QI5_1);
v |= (vh << 4) & 0x00000010; // 0 -> 4 v |= (vh << 4) & 0x00000010; // 0 -> 4
v |= (vh << 11) & 0x00001000; // 1 -> 12 v |= (vh << 11) & 0x00001000; // 1 -> 12
v |= (vh << 18) & 0x00100000; // 2 -> 20 v |= (vh << 18) & 0x00100000; // 2 -> 20
@ -249,7 +249,7 @@ static __device__ __forceinline__ T vec_dot_fattn_vec_KQ_q8_0(
const int ib = k_KQ / QI8_0; const int ib = k_KQ / QI8_0;
const int iqs = k_KQ % QI8_0; const int iqs = k_KQ % QI8_0;
const int v = get_int_from_int8(K_q8_0[ib].qs, iqs); const int v = get_int_b2(K_q8_0[ib].qs, iqs);
T Q_d; T Q_d;
if (std::is_same<T, half>::value) { if (std::is_same<T, half>::value) {
@ -408,7 +408,7 @@ static __device__ __forceinline__ T dequantize_1_q5_0(const void * __restrict__
const T d = x[ib].d; const T d = x[ib].d;
const int ql0 = x[ib].qs[iqs]; const int ql0 = x[ib].qs[iqs];
const int qh0 = get_int_from_uint8(x[ib].qh, 0); const int qh0 = get_int_b2(x[ib].qh, 0);
const int ql = ((ql0 >> (4*shift)) & 0x0F); const int ql = ((ql0 >> (4*shift)) & 0x0F);
const int qh = ((qh0 >> idq) << 4) & 0x10; const int qh = ((qh0 >> idq) << 4) & 0x10;
const int q = (ql | qh) - 16; const int q = (ql | qh) - 16;
@ -433,7 +433,7 @@ static __device__ __forceinline__ T dequantize_1_q5_1(const void * __restrict__
const half2 dm = x[ib].dm; const half2 dm = x[ib].dm;
const int ql0 = x[ib].qs[iqs]; const int ql0 = x[ib].qs[iqs];
const int qh0 = get_int_from_uint8_aligned(x[ib].qh, 0); const int qh0 = get_int_b4(x[ib].qh, 0);
const int ql = ((ql0 >> (4*shift)) & 0x0F); const int ql = ((ql0 >> (4*shift)) & 0x0F);
const int qh = ((qh0 >> idq) << 4) & 0x10; const int qh = ((qh0 >> idq) << 4) & 0x10;
const int q = (ql | qh); const int q = (ql | qh);

8
ggml/src/ggml-cuda/mmq.cu
View File

@ -59,6 +59,12 @@ void ggml_cuda_op_mul_mat_q(
case GGML_TYPE_Q6_K: case GGML_TYPE_Q6_K:
mul_mat_q_case<GGML_TYPE_Q6_K>(ctx, args, stream); mul_mat_q_case<GGML_TYPE_Q6_K>(ctx, args, stream);
break; break;
case GGML_TYPE_IQ4_XS:
mul_mat_q_case<GGML_TYPE_IQ4_XS>(ctx, args, stream);
break;
case GGML_TYPE_IQ4_NL:
mul_mat_q_case<GGML_TYPE_IQ4_NL>(ctx, args, stream);
break;
default: default:
GGML_ASSERT(false); GGML_ASSERT(false);
break; break;
@ -87,6 +93,8 @@ bool ggml_cuda_should_use_mmq(enum ggml_type type, int cc, int64_t ne11) {
case GGML_TYPE_Q4_K: case GGML_TYPE_Q4_K:
case GGML_TYPE_Q5_K: case GGML_TYPE_Q5_K:
case GGML_TYPE_Q6_K: case GGML_TYPE_Q6_K:
case GGML_TYPE_IQ4_XS:
case GGML_TYPE_IQ4_NL:
mmq_supported = true; mmq_supported = true;
break; break;
default: default:

188
ggml/src/ggml-cuda/mmq.cuh
View File

@ -101,6 +101,8 @@ static constexpr __host__ __device__ tile_x_sizes mmq_get_dp4a_tile_x_sizes(ggml
type == GGML_TYPE_Q4_K ? MMQ_DP4A_TXS_Q4_K : type == GGML_TYPE_Q4_K ? MMQ_DP4A_TXS_Q4_K :
type == GGML_TYPE_Q5_K ? MMQ_DP4A_TXS_Q5_K : type == GGML_TYPE_Q5_K ? MMQ_DP4A_TXS_Q5_K :
type == GGML_TYPE_Q6_K ? MMQ_DP4A_TXS_Q6_K : type == GGML_TYPE_Q6_K ? MMQ_DP4A_TXS_Q6_K :
type == GGML_TYPE_IQ4_XS ? MMQ_DP4A_TXS_Q5_0 :
type == GGML_TYPE_IQ4_NL ? MMQ_DP4A_TXS_Q5_0 :
tile_x_sizes{0, 0, 0}; tile_x_sizes{0, 0, 0};
} }
@ -137,6 +139,8 @@ static constexpr __host__ __device__ int mmq_get_mma_tile_x_k(ggml_type type) {
type == GGML_TYPE_Q4_K ? MMQ_MMA_TILE_X_K_Q4_K : type == GGML_TYPE_Q4_K ? MMQ_MMA_TILE_X_K_Q4_K :
type == GGML_TYPE_Q5_K ? MMQ_MMA_TILE_X_K_Q5_K : type == GGML_TYPE_Q5_K ? MMQ_MMA_TILE_X_K_Q5_K :
type == GGML_TYPE_Q6_K ? MMQ_MMA_TILE_X_K_Q6_K : type == GGML_TYPE_Q6_K ? MMQ_MMA_TILE_X_K_Q6_K :
type == GGML_TYPE_IQ4_XS ? MMQ_MMA_TILE_X_K_Q5_0 :
type == GGML_TYPE_IQ4_NL ? MMQ_MMA_TILE_X_K_Q5_0 :
0; 0;
} }
@ -185,9 +189,9 @@ template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinlin
const block_q4_0 * bxi = (const block_q4_0 *) x + kbx0 + i*stride + kbx; const block_q4_0 * bxi = (const block_q4_0 *) x + kbx0 + i*stride + kbx;
#ifdef INT8_MMA_AVAILABLE #ifdef INT8_MMA_AVAILABLE
x_qs[i*MMQ_MMA_TILE_X_K_Q4_0 + threadIdx.x] = get_int_from_uint8(bxi->qs, kqsx); x_qs[i*MMQ_MMA_TILE_X_K_Q4_0 + threadIdx.x] = get_int_b2(bxi->qs, kqsx);
#else #else
x_qs[i*(WARP_SIZE + 1) + threadIdx.x] = get_int_from_uint8(bxi->qs, kqsx); x_qs[i*(WARP_SIZE + 1) + threadIdx.x] = get_int_b2(bxi->qs, kqsx);
#endif // INT8_MMA_AVAILABLE #endif // INT8_MMA_AVAILABLE
} }
@ -348,9 +352,9 @@ template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinlin
const block_q4_1 * bxi = (const block_q4_1 *) x + kbx0 + i*stride + kbx; const block_q4_1 * bxi = (const block_q4_1 *) x + kbx0 + i*stride + kbx;
#ifdef INT8_MMA_AVAILABLE #ifdef INT8_MMA_AVAILABLE
x_qs[i*MMQ_MMA_TILE_X_K_Q4_1 + threadIdx.x] = get_int_from_uint8_aligned(bxi->qs, kqsx); x_qs[i*MMQ_MMA_TILE_X_K_Q4_1 + threadIdx.x] = get_int_b4(bxi->qs, kqsx);
#else #else
x_qs[i*(WARP_SIZE + 1) + threadIdx.x] = get_int_from_uint8_aligned(bxi->qs, kqsx); x_qs[i*(WARP_SIZE + 1) + threadIdx.x] = get_int_b4(bxi->qs, kqsx);
#endif // INT8_MMA_AVAILABLE #endif // INT8_MMA_AVAILABLE
} }
@ -509,8 +513,8 @@ template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinlin
const block_q5_0 * bxi = (const block_q5_0 *) x + kbx0 + i*stride + kbx; const block_q5_0 * bxi = (const block_q5_0 *) x + kbx0 + i*stride + kbx;
const int ql = get_int_from_uint8(bxi->qs, kqsx); const int ql = get_int_b2(bxi->qs, kqsx);
const int qh = get_int_from_uint8(bxi->qh, 0) >> (4 * (threadIdx.x % QI5_0)); const int qh = get_int_b2(bxi->qh, 0) >> (4 * (threadIdx.x % QI5_0));
int qs0 = (ql >> 0) & 0x0F0F0F0F; int qs0 = (ql >> 0) & 0x0F0F0F0F;
qs0 |= (qh << 4) & 0x00000010; // 0 -> 4 qs0 |= (qh << 4) & 0x00000010; // 0 -> 4
@ -674,8 +678,8 @@ template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinlin
const block_q5_1 * bxi = (const block_q5_1 *) x + kbx0 + i*stride + kbx; const block_q5_1 * bxi = (const block_q5_1 *) x + kbx0 + i*stride + kbx;
const int ql = get_int_from_uint8_aligned(bxi->qs, kqsx); const int ql = get_int_b4(bxi->qs, kqsx);
const int qh = get_int_from_uint8_aligned(bxi->qh, 0) >> (4 * (threadIdx.x % QI5_1)); const int qh = get_int_b4(bxi->qh, 0) >> (4 * (threadIdx.x % QI5_1));
int qs0 = (ql >> 0) & 0x0F0F0F0F; int qs0 = (ql >> 0) & 0x0F0F0F0F;
qs0 |= (qh << 4) & 0x00000010; // 0 -> 4 qs0 |= (qh << 4) & 0x00000010; // 0 -> 4
@ -839,9 +843,9 @@ template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinlin
const block_q8_0 * bxi = (const block_q8_0 *) x + kbx0 + i*stride + kbx; const block_q8_0 * bxi = (const block_q8_0 *) x + kbx0 + i*stride + kbx;
#ifdef INT8_MMA_AVAILABLE #ifdef INT8_MMA_AVAILABLE
x_qs[i*MMQ_MMA_TILE_X_K_Q8_0 + threadIdx.x] = get_int_from_int8(bxi->qs, kqsx); x_qs[i*MMQ_MMA_TILE_X_K_Q8_0 + threadIdx.x] = get_int_b2(bxi->qs, kqsx);
#else #else
x_qs[i*(WARP_SIZE + 1) + threadIdx.x] = get_int_from_int8(bxi->qs, kqsx); x_qs[i*(WARP_SIZE + 1) + threadIdx.x] = get_int_b2(bxi->qs, kqsx);
#endif // INT8_MMA_AVAILABLE #endif // INT8_MMA_AVAILABLE
} }
@ -984,7 +988,7 @@ template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinlin
const block_q2_K * bxi = (const block_q2_K *) x + kbx0 + i*stride + kbx; const block_q2_K * bxi = (const block_q2_K *) x + kbx0 + i*stride + kbx;
const int x_ql_0 = get_int_from_uint8(bxi->qs, kqsx); const int x_ql_0 = get_int_b2(bxi->qs, kqsx);
#pragma unroll #pragma unroll
for (int l = 0; l < QR2_K; ++l) { for (int l = 0; l < QR2_K; ++l) {
@ -1166,8 +1170,8 @@ template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinlin
const block_q3_K * bxi = (const block_q3_K *) x + kbx0 + i*stride + kbx; const block_q3_K * bxi = (const block_q3_K *) x + kbx0 + i*stride + kbx;
const int x_ql_0 = get_int_from_uint8(bxi->qs, kqsx); const int x_ql_0 = get_int_b2(bxi->qs, kqsx);
const int x_qh_0 = get_int_from_uint8(bxi->hmask, kqsx % (QI3_K/2)) >> (4 * (kqsx / (QI3_K/2))); const int x_qh_0 = get_int_b2(bxi->hmask, kqsx % (QI3_K/2)) >> (4 * (kqsx / (QI3_K/2)));
#pragma unroll #pragma unroll
for (int l = 0; l < QR3_K; ++l) { for (int l = 0; l < QR3_K; ++l) {
@ -1225,11 +1229,11 @@ template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinlin
const int ksc_low = ksc % (QI3_K/8); const int ksc_low = ksc % (QI3_K/8);
const int shift_low = 4 * (ksc / (QI3_K/8)); const int shift_low = 4 * (ksc / (QI3_K/8));
const int sc_low = (get_int_from_uint8(bxi->scales, ksc_low) >> shift_low) & 0x0F0F0F0F; const int sc_low = (get_int_b2(bxi->scales, ksc_low) >> shift_low) & 0x0F0F0F0F;
const int ksc_high = QI3_K/8; const int ksc_high = QI3_K/8;
const int shift_high = 2 * ksc; const int shift_high = 2 * ksc;
const int sc_high = ((get_int_from_uint8(bxi->scales, ksc_high) >> shift_high) << 4) & 0x30303030; const int sc_high = ((get_int_b2(bxi->scales, ksc_high) >> shift_high) << 4) & 0x30303030;
const int sc = __vsubss4(sc_low | sc_high, 0x20202020); const int sc = __vsubss4(sc_low | sc_high, 0x20202020);
@ -1393,9 +1397,9 @@ template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinlin
const block_q4_K * bxi = (const block_q4_K *) x + kbx0 + i*stride + kbx; const block_q4_K * bxi = (const block_q4_K *) x + kbx0 + i*stride + kbx;
#ifdef INT8_MMA_AVAILABLE #ifdef INT8_MMA_AVAILABLE
x_qs[i*MMQ_MMA_TILE_X_K_Q4_K + threadIdx.x] = get_int_from_uint8_aligned(bxi->qs, kqsx); x_qs[i*MMQ_MMA_TILE_X_K_Q4_K + threadIdx.x] = get_int_b4(bxi->qs, kqsx);
#else #else
x_qs[i*(WARP_SIZE + 1) + threadIdx.x] = get_int_from_uint8_aligned(bxi->qs, kqsx); x_qs[i*(WARP_SIZE + 1) + threadIdx.x] = get_int_b4(bxi->qs, kqsx);
#endif // INT8_MMA_AVAILABLE #endif // INT8_MMA_AVAILABLE
} }
@ -1610,11 +1614,11 @@ template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinlin
const block_q5_K * bxi = (const block_q5_K *) x + kbx0 + i*stride + kbx; const block_q5_K * bxi = (const block_q5_K *) x + kbx0 + i*stride + kbx;
const int ky = QR5_K*kqsx; const int ky = QR5_K*kqsx;
const int ql = get_int_from_uint8_aligned(bxi->qs, kqsx); const int ql = get_int_b4(bxi->qs, kqsx);
const int ql0 = (ql >> 0) & 0x0F0F0F0F; const int ql0 = (ql >> 0) & 0x0F0F0F0F;
const int ql1 = (ql >> 4) & 0x0F0F0F0F; const int ql1 = (ql >> 4) & 0x0F0F0F0F;
const int qh = get_int_from_uint8_aligned(bxi->qh, kqsx % (QI5_K/4)); const int qh = get_int_b4(bxi->qh, kqsx % (QI5_K/4));
const int qh0 = ((qh >> (2 * (kqsx / (QI5_K/4)) + 0)) << 4) & 0x10101010; const int qh0 = ((qh >> (2 * (kqsx / (QI5_K/4)) + 0)) << 4) & 0x10101010;
const int qh1 = ((qh >> (2 * (kqsx / (QI5_K/4)) + 1)) << 4) & 0x10101010; const int qh1 = ((qh >> (2 * (kqsx / (QI5_K/4)) + 1)) << 4) & 0x10101010;
@ -1832,11 +1836,11 @@ template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinlin
const block_q6_K * bxi = (const block_q6_K *) x + kbx0 + i*stride + kbx; const block_q6_K * bxi = (const block_q6_K *) x + kbx0 + i*stride + kbx;
const int ky = QR6_K*kqsx; const int ky = QR6_K*kqsx;
const int ql = get_int_from_uint8(bxi->ql, kqsx); const int ql = get_int_b2(bxi->ql, kqsx);
const int ql0 = (ql >> 0) & 0x0F0F0F0F; const int ql0 = (ql >> 0) & 0x0F0F0F0F;
const int ql1 = (ql >> 4) & 0x0F0F0F0F; const int ql1 = (ql >> 4) & 0x0F0F0F0F;
const int qh = get_int_from_uint8(bxi->qh, (QI6_K/4) * (kqsx / (QI6_K/2)) + kqsx % (QI6_K/4)); const int qh = get_int_b2(bxi->qh, (QI6_K/4) * (kqsx / (QI6_K/2)) + kqsx % (QI6_K/4));
const int qh0 = ((qh >> (2 * ((kqsx % (QI6_K/2)) / (QI6_K/4)))) << 4) & 0x30303030; const int qh0 = ((qh >> (2 * ((kqsx % (QI6_K/2)) / (QI6_K/4)))) << 4) & 0x30303030;
const int qh1 = (qh >> (2 * ((kqsx % (QI6_K/2)) / (QI6_K/4)))) & 0x30303030; const int qh1 = (qh >> (2 * ((kqsx % (QI6_K/2)) / (QI6_K/4)))) & 0x30303030;
@ -1883,9 +1887,9 @@ template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinlin
const block_q6_K * bxi = (const block_q6_K *) x + kbx0 + i*stride + (threadIdx.x % (WARP_SIZE/8)) / 4; const block_q6_K * bxi = (const block_q6_K *) x + kbx0 + i*stride + (threadIdx.x % (WARP_SIZE/8)) / 4;
#ifdef INT8_MMA_AVAILABLE #ifdef INT8_MMA_AVAILABLE
x_sc[i*MMQ_MMA_TILE_X_K_Q6_K + threadIdx.x % (WARP_SIZE/8)] = get_int_from_int8(bxi->scales, threadIdx.x % (QI6_K/8)); x_sc[i*MMQ_MMA_TILE_X_K_Q6_K + threadIdx.x % (WARP_SIZE/8)] = get_int_b2(bxi->scales, threadIdx.x % (QI6_K/8));
#else #else
x_sc[i*(WARP_SIZE/8) + i/8 + threadIdx.x % (WARP_SIZE/8)] = get_int_from_int8(bxi->scales, threadIdx.x % (QI6_K/8)); x_sc[i*(WARP_SIZE/8) + i/8 + threadIdx.x % (WARP_SIZE/8)] = get_int_b2(bxi->scales, threadIdx.x % (QI6_K/8));
#endif // INT8_MMA_AVAILABLE #endif // INT8_MMA_AVAILABLE
} }
} }
@ -2018,6 +2022,124 @@ static __device__ __forceinline__ void vec_dot_q6_K_q8_1_mma(
#endif // INT8_MMA_AVAILABLE #endif // INT8_MMA_AVAILABLE
} }
template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinline__ void load_tiles_iq4_nl(
const char * __restrict__ x, int * __restrict__ x_tile, const int & kbx0, const int & i_max, const int & stride) {
#ifdef INT8_MMA_AVAILABLE
int * x_qs = (int *) x_tile;
float * x_df = (float *) (x_qs + WARP_SIZE*2);
#else
constexpr tile_x_sizes txs = mmq_get_dp4a_tile_x_sizes(GGML_TYPE_IQ4_NL, mmq_y);
int * x_qs = (int *) x_tile;
float * x_df = (float *) (x_qs + txs.qs);
#endif // INT8_MMA_AVAILABLE
const int kbx = threadIdx.x / QI4_NL;
const int kqsx = threadIdx.x % QI4_NL;
#pragma unroll
for (int i0 = 0; i0 < mmq_y; i0 += nwarps) {
int i = i0 + threadIdx.y;
if (need_check) {
i = min(i, i_max);
}
const block_iq4_nl * bxi = (const block_iq4_nl *) x + kbx0 + i*stride + kbx;
const int aux_q4 = get_int_b2(bxi->qs, kqsx);
const int2 v = get_int_from_table_16(aux_q4);
const int k0 = 8 * (threadIdx.x / 4) + threadIdx.x % 4;
#ifdef INT8_MMA_AVAILABLE
x_qs[i*MMQ_MMA_TILE_X_K_Q5_0 + k0 + 0] = v.x;
x_qs[i*MMQ_MMA_TILE_X_K_Q5_0 + k0 + 4] = v.y;
#else
x_qs[i*(2*WARP_SIZE + 1) + k0 + 0] = v.x;
x_qs[i*(2*WARP_SIZE + 1) + k0 + 4] = v.y;
#endif // INT8_MMA_AVAILABLE
}
const int blocks_per_tile_x_row = WARP_SIZE / QI4_NL;
const int kbxd = threadIdx.x % blocks_per_tile_x_row;
#pragma unroll
for (int i0 = 0; i0 < mmq_y; i0 += nwarps * QI4_NL) {
int i = i0 + threadIdx.y * QI4_NL + threadIdx.x / blocks_per_tile_x_row;
if (need_check) {
i = min(i, i_max);
}
const block_iq4_nl * bxi = (const block_iq4_nl *) x + kbx0 + i*stride + kbxd;
#ifdef INT8_MMA_AVAILABLE
x_df[i*MMQ_MMA_TILE_X_K_Q5_0 + kbxd] = __half2float(bxi->d);
#else
x_df[i*(WARP_SIZE/4) + i/4 + kbxd] = __half2float(bxi->d);
#endif // INT8_MMA_AVAILABLE
}
}
template <int mmq_y, int nwarps, bool need_check> static __device__ __forceinline__ void load_tiles_iq4_xs(
const char * __restrict__ x, int * __restrict__ x_tile, const int & kbx0, const int & i_max, const int & stride) {
#ifdef INT8_MMA_AVAILABLE
int * x_qs = (int *) x_tile;
float * x_df = (float *) (x_qs + WARP_SIZE*2);
#else
constexpr tile_x_sizes txs = mmq_get_dp4a_tile_x_sizes(GGML_TYPE_IQ4_XS, mmq_y);
int * x_qs = (int *) x_tile;
float * x_df = (float *) (x_qs + txs.qs);
#endif // INT8_MMA_AVAILABLE
const int kbx = 0; // threadIdx.x / QI4_XS
const int kqsx = threadIdx.x; // threadIdx.x % QI4_XS
#pragma unroll
for (int i0 = 0; i0 < mmq_y; i0 += nwarps) {
int i = i0 + threadIdx.y;
if (need_check) {
i = min(i, i_max);
}
const block_iq4_xs * bxi = (const block_iq4_xs *) x + kbx0 + i*stride + kbx;
const int aux_q4 = get_int_b4(bxi->qs, kqsx);
const int2 v = get_int_from_table_16(aux_q4);
const int k0 = 8 * (threadIdx.x / 4) + threadIdx.x % 4;
#ifdef INT8_MMA_AVAILABLE
x_qs[i*MMQ_MMA_TILE_X_K_Q5_0 + k0 + 0] = v.x;
x_qs[i*MMQ_MMA_TILE_X_K_Q5_0 + k0 + 4] = v.y;
#else
x_qs[i*(2*WARP_SIZE + 1) + k0 + 0] = v.x;
x_qs[i*(2*WARP_SIZE + 1) + k0 + 4] = v.y;
#endif // INT8_MMA_AVAILABLE
}
#pragma unroll
for (int i0 = 0; i0 < mmq_y; i0 += nwarps * 4) {
int i = i0 + threadIdx.y * 4 + threadIdx.x / (WARP_SIZE/4);
if (need_check) {
i = min(i, i_max);
}
const block_iq4_xs * bxi = (const block_iq4_xs *) x + kbx0 + i*stride;
const float d = __half2float(bxi->d);
const int ls = ((bxi->scales_l[(threadIdx.x % 8)/2] >> (4*(threadIdx.x % 2))) & 0x0F)
| (((bxi->scales_h >> (2*(threadIdx.x % 8))) & 0x03) << 4);
#ifdef INT8_MMA_AVAILABLE
x_df[i*MMQ_MMA_TILE_X_K_Q5_0 + threadIdx.x % 8] = d * (ls - 32);
#else
x_df[i*(WARP_SIZE/4) + i/4 + threadIdx.x % 8] = d * (ls - 32);
#endif // INT8_MMA_AVAILABLE
}
}
template<int mmq_x, int mmq_y, int nwarps, bool need_check> template<int mmq_x, int mmq_y, int nwarps, bool need_check>
static __device__ __forceinline__ void mmq_write_back_dp4a( static __device__ __forceinline__ void mmq_write_back_dp4a(
const float * __restrict__ sum, float * __restrict__ dst, const int & stride, const int & i_max, const int & j_max) { const float * __restrict__ sum, float * __restrict__ dst, const int & stride, const int & i_max, const int & j_max) {
@ -2167,6 +2289,22 @@ struct mmq_type_traits<mmq_x, mmq_y, nwarps, need_check, GGML_TYPE_Q6_K> {
static constexpr vec_dot_mmq_t vec_dot_dp4a = vec_dot_q6_K_q8_1_dp4a<mmq_x, mmq_y, nwarps>; static constexpr vec_dot_mmq_t vec_dot_dp4a = vec_dot_q6_K_q8_1_dp4a<mmq_x, mmq_y, nwarps>;
}; };
template <int mmq_x, int mmq_y, int nwarps, bool need_check>
struct mmq_type_traits<mmq_x, mmq_y, nwarps, need_check, GGML_TYPE_IQ4_NL> {
static constexpr int vdr = VDR_IQ4_NL_Q8_1_MMQ;
static constexpr load_tiles_mmq_t load_tiles = load_tiles_iq4_nl<mmq_y, nwarps, need_check>;
static constexpr vec_dot_mmq_t vec_dot_mma = vec_dot_q5_0_q8_1_mma<mmq_x, mmq_y, nwarps>;
static constexpr vec_dot_mmq_t vec_dot_dp4a = vec_dot_q5_0_q8_1_dp4a<mmq_x, mmq_y, nwarps>;
};
template <int mmq_x, int mmq_y, int nwarps, bool need_check>
struct mmq_type_traits<mmq_x, mmq_y, nwarps, need_check, GGML_TYPE_IQ4_XS> {
static constexpr int vdr = VDR_IQ4_XS_Q8_1_MMQ;
static constexpr load_tiles_mmq_t load_tiles = load_tiles_iq4_xs<mmq_y, nwarps, need_check>;
static constexpr vec_dot_mmq_t vec_dot_mma = vec_dot_q5_0_q8_1_mma<mmq_x, mmq_y, nwarps>;
static constexpr vec_dot_mmq_t vec_dot_dp4a = vec_dot_q5_0_q8_1_dp4a<mmq_x, mmq_y, nwarps>;
};
static bool mmq_need_sum(const ggml_type type_x) { static bool mmq_need_sum(const ggml_type type_x) {
switch (type_x) { switch (type_x) {
case GGML_TYPE_Q4_0: case GGML_TYPE_Q4_0:
@ -2184,6 +2322,8 @@ static bool mmq_need_sum(const ggml_type type_x) {
case GGML_TYPE_Q5_K: case GGML_TYPE_Q5_K:
return true; return true;
case GGML_TYPE_Q6_K: case GGML_TYPE_Q6_K:
case GGML_TYPE_IQ4_XS:
case GGML_TYPE_IQ4_NL:
return false; return false;
default: default:
GGML_ASSERT(false); GGML_ASSERT(false);
@ -2608,6 +2748,8 @@ extern DECL_MMQ_CASE(GGML_TYPE_Q3_K);
extern DECL_MMQ_CASE(GGML_TYPE_Q4_K); extern DECL_MMQ_CASE(GGML_TYPE_Q4_K);
extern DECL_MMQ_CASE(GGML_TYPE_Q5_K); extern DECL_MMQ_CASE(GGML_TYPE_Q5_K);
extern DECL_MMQ_CASE(GGML_TYPE_Q6_K); extern DECL_MMQ_CASE(GGML_TYPE_Q6_K);
extern DECL_MMQ_CASE(GGML_TYPE_IQ4_NL);
extern DECL_MMQ_CASE(GGML_TYPE_IQ4_XS);
// ------------------------------------------------------------------------------------------------------------------------- // -------------------------------------------------------------------------------------------------------------------------

3
ggml/src/ggml-cuda/template-instances/generate_cu_files.py
View File

@ -22,7 +22,8 @@ SOURCE_FATTN_WMMA_CASE = "DECL_FATTN_WMMA_F16_CASE({head_size}, {cols_per_block}
TYPES_MMQ = [ TYPES_MMQ = [
"GGML_TYPE_Q4_0", "GGML_TYPE_Q4_1", "GGML_TYPE_Q5_0", "GGML_TYPE_Q5_1", "GGML_TYPE_Q8_0", "GGML_TYPE_Q4_0", "GGML_TYPE_Q4_1", "GGML_TYPE_Q5_0", "GGML_TYPE_Q5_1", "GGML_TYPE_Q8_0",
"GGML_TYPE_Q2_K", "GGML_TYPE_Q3_K", "GGML_TYPE_Q4_K", "GGML_TYPE_Q5_K", "GGML_TYPE_Q6_K" "GGML_TYPE_Q2_K", "GGML_TYPE_Q3_K", "GGML_TYPE_Q4_K", "GGML_TYPE_Q5_K", "GGML_TYPE_Q6_K",
"GGML_TYPE_IQ4_NL", "GGML_TYPE_IQ4_XS"
] ]
SOURCE_MMQ = """// This file has been autogenerated by generate_cu_files.py, do not edit manually. SOURCE_MMQ = """// This file has been autogenerated by generate_cu_files.py, do not edit manually.

5
ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu Normal file
View File

@ -0,0 +1,5 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../mmq.cuh"
DECL_MMQ_CASE(GGML_TYPE_IQ4_NL);

5
ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu Normal file
View File

@ -0,0 +1,5 @@
// This file has been autogenerated by generate_cu_files.py, do not edit manually.
#include "../mmq.cuh"
DECL_MMQ_CASE(GGML_TYPE_IQ4_XS);

43
ggml/src/ggml-cuda/vecdotq.cuh
View File

@ -1,36 +1,8 @@
#include "common.cuh" #include "common.cuh"
#include <cstdint> #include <cstdint>
static __device__ __forceinline__ int get_int_from_int8(const int8_t * x8, const int & i32) {
const uint16_t * x16 = (const uint16_t *) (x8 + sizeof(int) * i32); // assume at least 2 byte alignment
int x32 = 0;
x32 |= x16[0] << 0;
x32 |= x16[1] << 16;
return x32;
}
static __device__ __forceinline__ int get_int_from_uint8(const uint8_t * x8, const int & i32) {
const uint16_t * x16 = (const uint16_t *) (x8 + sizeof(int) * i32); // assume at least 2 byte alignment
int x32 = 0;
x32 |= x16[0] << 0;
x32 |= x16[1] << 16;
return x32;
}
static __device__ __forceinline__ int get_int_from_int8_aligned(const int8_t * x8, const int & i32) {
return *((const int *) (x8 + sizeof(int) * i32)); // assume at least 4 byte alignment
}
static __device__ __forceinline__ int get_int_from_uint8_aligned(const uint8_t * x8, const int & i32) {
return *((const int *) (x8 + sizeof(int) * i32)); // assume at least 4 byte alignment
}
static __device__ __forceinline__ int get_int_b2(const void * x, const int & i32) { static __device__ __forceinline__ int get_int_b2(const void * x, const int & i32) {
const uint16_t * x16 = (const uint16_t *) x; const uint16_t * x16 = (const uint16_t *) x; // assume at least 2 byte alignment
int x32 = x16[2*i32 + 0] << 0; int x32 = x16[2*i32 + 0] << 0;
x32 |= x16[2*i32 + 1] << 16; x32 |= x16[2*i32 + 1] << 16;
@ -768,6 +740,7 @@ static __device__ __forceinline__ float vec_dot_q6_K_q8_1(
} }
#define VDR_IQ2_XXS_Q8_1_MMVQ 2 #define VDR_IQ2_XXS_Q8_1_MMVQ 2
#define VDR_IQ2_XXS_Q8_1_MMQ 2
static __device__ __forceinline__ float vec_dot_iq2_xxs_q8_1( static __device__ __forceinline__ float vec_dot_iq2_xxs_q8_1(
const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) { const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) {
@ -802,6 +775,7 @@ static __device__ __forceinline__ float vec_dot_iq2_xxs_q8_1(
} }
#define VDR_IQ2_XS_Q8_1_MMVQ 2 #define VDR_IQ2_XS_Q8_1_MMVQ 2
#define VDR_IQ2_XS_Q8_1_MMQ 2
static __device__ __forceinline__ float vec_dot_iq2_xs_q8_1( static __device__ __forceinline__ float vec_dot_iq2_xs_q8_1(
const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) { const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) {
@ -840,6 +814,7 @@ static __device__ __forceinline__ float vec_dot_iq2_xs_q8_1(
} }
#define VDR_IQ2_S_Q8_1_MMVQ 2 #define VDR_IQ2_S_Q8_1_MMVQ 2
#define VDR_IQ2_S_Q8_1_MMQ 2
static __device__ __forceinline__ float vec_dot_iq2_s_q8_1( static __device__ __forceinline__ float vec_dot_iq2_s_q8_1(
const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) { const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) {
@ -887,6 +862,7 @@ static __device__ __forceinline__ float vec_dot_iq2_s_q8_1(
} }
#define VDR_IQ3_XXS_Q8_1_MMVQ 2 #define VDR_IQ3_XXS_Q8_1_MMVQ 2
#define VDR_IQ3_XXS_Q8_1_MMQ 2
static __device__ __forceinline__ float vec_dot_iq3_xxs_q8_1( static __device__ __forceinline__ float vec_dot_iq3_xxs_q8_1(
const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) { const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) {
@ -921,6 +897,7 @@ static __device__ __forceinline__ float vec_dot_iq3_xxs_q8_1(
} }
#define VDR_IQ3_S_Q8_1_MMVQ 2 #define VDR_IQ3_S_Q8_1_MMVQ 2
#define VDR_IQ3_S_Q8_1_MMQ 2
// TODO: don't use lookup table for signs // TODO: don't use lookup table for signs
static __device__ __forceinline__ float vec_dot_iq3_s_q8_1( static __device__ __forceinline__ float vec_dot_iq3_s_q8_1(
@ -962,6 +939,9 @@ static __device__ __forceinline__ float vec_dot_iq3_s_q8_1(
return d * sumi; return d * sumi;
} }
#define VDR_IQ1_S_Q8_1_MMVQ 1
#define VDR_IQ1_S_Q8_1_MMQ 1
static __device__ __forceinline__ float vec_dot_iq1_s_q8_1( static __device__ __forceinline__ float vec_dot_iq1_s_q8_1(
const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) { const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) {
const block_iq1_s * bq1 = (const block_iq1_s *) vbq + kbx; const block_iq1_s * bq1 = (const block_iq1_s *) vbq + kbx;
@ -992,6 +972,9 @@ static __device__ __forceinline__ float vec_dot_iq1_s_q8_1(
return d1q * (ds.x*sumi + ds.y*delta); return d1q * (ds.x*sumi + ds.y*delta);
} }
#define VDR_IQ1_M_Q8_1_MMVQ 1
#define VDR_IQ1_M_Q8_1_MMQ 1
static __device__ __forceinline__ float vec_dot_iq1_m_q8_1( static __device__ __forceinline__ float vec_dot_iq1_m_q8_1(
const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) { const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) {
@ -1051,6 +1034,7 @@ static __device__ __forceinline__ int2 get_int_from_table_16(const int & q4) {
} }
#define VDR_IQ4_NL_Q8_1_MMVQ 2 #define VDR_IQ4_NL_Q8_1_MMVQ 2
#define VDR_IQ4_NL_Q8_1_MMQ 4
static __device__ __forceinline__ float vec_dot_iq4_nl_q8_1( static __device__ __forceinline__ float vec_dot_iq4_nl_q8_1(
const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) { const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) {
@ -1074,6 +1058,7 @@ static __device__ __forceinline__ float vec_dot_iq4_nl_q8_1(
} }
#define VDR_IQ4_XS_Q8_1_MMVQ 4 #define VDR_IQ4_XS_Q8_1_MMVQ 4
#define VDR_IQ4_XS_Q8_1_MMQ 4
static __device__ __forceinline__ float vec_dot_iq4_xs_q8_1( static __device__ __forceinline__ float vec_dot_iq4_xs_q8_1(
const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) { const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) {