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

revert cpy code

Browse source
This commit is contained in:
luoyu-intel 2024-07-11 07:51:23 +00:00
parent 92784e8059
commit d9b89eb308
1 changed files with 27 additions and 14 deletions
Download patch file Download diff file Expand all files Collapse all files

41
ggml/src/ggml-sycl/dmmv.cpp
View file

@ -130,12 +130,12 @@ static void dequantize_mul_mat_vec_q4_0(const void * __restrict__ vx, const dflo
v.x() = ((vui & 0xF) - 8) * d; v.x() = ((vui & 0xF) - 8) * d;
v.y() = ((vui >> 4) - 8) * d; v.y() = ((vui >> 4) - 8) * d;
#ifdef GGML_SYCL_F16 #ifdef GGML_SYCL_F16
dfloat2 t1{ y[iybs + ir + 0], dfloat2 t1{ y[iybs + ir * 2 + 0],
y[iybs + ir + QK4_0 / 2] }; y[iybs + ir * 2 + 1] };
tmp += v * t1; tmp += v * t1;
#else #else
tmp += v.x() * y[iybs + ir + 0]; tmp += v.x() * y[iybs + ir * 2 + 0];
tmp += v.y() * y[iybs + ir + QK4_0 / 2]; tmp += v.y() * y[iybs + ir * 2 + 1];
#endif #endif
} }
} }
@ -157,12 +157,12 @@ static void dequantize_mul_mat_vec_q4_0(const void * __restrict__ vx, const dflo
v.x() = ((vui & 0xF) - 8) * d; v.x() = ((vui & 0xF) - 8) * d;
v.y() = ((vui >> 4) - 8) * d; v.y() = ((vui >> 4) - 8) * d;
#ifdef GGML_SYCL_F16 #ifdef GGML_SYCL_F16
dfloat2 t1{ y[iybs + ir + 0], dfloat2 t1{ y[iybs + ir * 2 + 0],
y[iybs + ir + QK4_0 / 2] }; y[iybs + ir * 2 + 1] };
tmp += v * t1; tmp += v * t1;
#else #else
tmp += v.x() * y[iybs + ir + 0]; tmp += v.x() * y[iybs + ir * 2 + 0];
tmp += v.y() * y[iybs + ir + QK4_0 / 2]; tmp += v.y() * y[iybs + ir * 2 + 1];
#endif #endif
} }
} }
@ -180,12 +180,12 @@ static void dequantize_mul_mat_vec_q4_0(const void * __restrict__ vx, const dflo
v.x() = ((vui & 0xF) - 8) * d; v.x() = ((vui & 0xF) - 8) * d;
v.y() = ((vui >> 4) - 8) * d; v.y() = ((vui >> 4) - 8) * d;
#ifdef GGML_SYCL_F16 #ifdef GGML_SYCL_F16
dfloat2 t1{ y[iybs + ir * QK4_0 + iqs + 0], dfloat2 t1{ y[iybs + ir * QK4_0 + iqs * 2 + 0],
y[iybs + ir * QK4_0 + iqs + QK4_0 / 2] }; y[iybs + ir * QK4_0 + iqs * 2 + 1] };
tmp += v * t1; tmp += v * t1;
#else #else
tmp += v.x() * y[iybs + ir * QK4_0 + iqs + 0]; tmp += v.x() * y[iybs + ir * QK4_0 + iqs * 2 + 0];
tmp += v.y() * y[iybs + ir * QK4_0 + iqs + QK4_0 / 2]; tmp += v.y() * y[iybs + ir * QK4_0 + iqs * 2 + 1];
#endif #endif
} }
@ -904,11 +904,24 @@ static void dequantize_mul_mat_vec_q4_0_sycl(const void *vx, const dfloat *y,
stream->parallel_for( stream->parallel_for(
nrows * ncols / QK4_0, nrows * ncols / QK4_0,
[=](auto i) [[intel::reqd_sub_group_size(WARP_SIZE)]] { [=](auto i) [[intel::reqd_sub_group_size(WARP_SIZE)]] {
const block_q4_0 * x = (const block_q4_0 *) vx; const block_q4_0 *x = (const block_q4_0 *)vx;
int ib = i; int ib = i;
typedef sycl::vec<uint8_t, QK4_0 / 2> CT; typedef sycl::vec<uint8_t, QK4_0 / 2> CT;
CT tmp = *(CT *)x[ib].qs; CT tmp = *(CT *)x[ib].qs;
*(CT*)(vx_tmp + ib * QK4_0 / 2) = tmp; for (int j = 0; j < QK4_0 / 2; j += 2)
{
const int vui = tmp[j];
const int vui1 = tmp[j + 1];
uint8_t nv = (vui & 0xF) | (vui1 << 4);
*(uint8_t *)(vx_tmp + ib * QK4_0 / 2 + j / 2) = nv;
}
for (int j = 0; j < QK4_0 / 2; j += 2)
{
const int vui = tmp[j];
const int vui1 = tmp[j + 1];
uint8_t nv = (vui >> 4) | (vui1 & 0xf0);
*(uint8_t *)(vx_tmp + ib * QK4_0 / 2 + j / 2 + QK4_0 / 4) = nv;
}
*(sycl::half *)(vx_tmp + ncols * nrows / 2 + ib * 2) = x[ib].d; *(sycl::half *)(vx_tmp + ncols * nrows / 2 + ib * 2) = x[ib].d;
}); });