ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-01-26 17:20:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

NEON implementation for trellis quants (#471)

Browse Source 
* iq2_kt: NEON implementation

* iq3_kt: NEON implementation

* iq4_kt: not working NEON implementation

* iq4_kt: NEON implementation

Have to use f32 arithmetic else I get gibberish?
Correspondigly ridiculously slow.

* Cleanup

* iq4_kt: slightly faster TG on NEON

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
Kawrakow
2025-05-29 18:57:41 +03:00
committed by GitHub
parent ccc00a4a56
commit 0dea2e8a81
3 changed files with 409 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
ggml/src/ggml.c
View File

@@ -1583,7 +1583,11 @@ static const ggml_type_traits_t type_traits[GGML_TYPE_COUNT] = {
.from_float = quantize_row_iq2_kt,
.from_float_ref = (ggml_from_float_t)quantize_row_iq2_kt_ref,
.vec_dot = vec_dot_iq2_kt_q8_k,
.vec_dot_type = GGML_TYPE_Q8_K,
#ifdef __ARM_NEON
.vec_dot_type = GGML_TYPE_F16,
#else
.vec_dot_type = GGML_TYPE_F32,
#endif
.nrows = 1,
.row_meta_size = 4,
},
@@ -1596,7 +1600,11 @@ static const ggml_type_traits_t type_traits[GGML_TYPE_COUNT] = {
.from_float = quantize_row_iq3_kt,
.from_float_ref = (ggml_from_float_t)quantize_row_iq3_kt_ref,
.vec_dot = vec_dot_iq3_kt_q8_k,
.vec_dot_type = GGML_TYPE_Q8_K,
#ifdef __ARM_NEON
.vec_dot_type = GGML_TYPE_F16,
#else
.vec_dot_type = GGML_TYPE_F32,
#endif
.nrows = 1,
.row_meta_size = 4,
},
@@ -1609,7 +1617,12 @@ static const ggml_type_traits_t type_traits[GGML_TYPE_COUNT] = {
.from_float = quantize_row_iq4_kt,
.from_float_ref = (ggml_from_float_t)quantize_row_iq4_kt_ref,
.vec_dot = vec_dot_iq4_kt_q8_k,
.vec_dot_type = GGML_TYPE_Q8_K,
#ifdef __ARM_NEON
//.vec_dot_type = GGML_TYPE_F16,
.vec_dot_type = GGML_TYPE_F32,
#else
.vec_dot_type = GGML_TYPE_F32,
#endif
.nrows = 1,
.row_meta_size = 8,
},

416
ggml/src/iqk/iqk_gemm_ktquants.cpp
View File

@@ -1,3 +1,4 @@
#include "iqk_common.h"
#include "iqk_gemm_ktquants.h"
#include "ggml.h"
@@ -316,37 +317,13 @@ bool iqk_set_kernels_ktquants(int ne00, int typeA, int typeB, std::array<mul_mat
switch (typeA) {
case GGML_TYPE_IQ2_KT:
assert (ne00 % QK_K == 0);
kernels[0] = mul_mat_iq2_kt_F32_T<1>;
kernels[1] = mul_mat_iq2_kt_F32_T<2>;
kernels[2] = mul_mat_iq2_kt_F32_T<3>;
kernels[3] = mul_mat_iq2_kt_F32_T<4>;
kernels[4] = mul_mat_iq2_kt_F32_T<5>;
kernels[5] = mul_mat_iq2_kt_F32_T<6>;
kernels[6] = mul_mat_iq2_kt_F32_T<7>;
kernels[7] = mul_mat_iq2_kt_F32_T<8>;
IQK_SET_MUL_MAT_FUNCTIONS(mul_mat_iq2_kt_F32_T, kernels);
break;
case GGML_TYPE_IQ3_KT:
assert (ne00 % QK_K == 0);
kernels[0] = mul_mat_iq3_kt_F32_T<1>;
kernels[1] = mul_mat_iq3_kt_F32_T<2>;
kernels[2] = mul_mat_iq3_kt_F32_T<3>;
kernels[3] = mul_mat_iq3_kt_F32_T<4>;
kernels[4] = mul_mat_iq3_kt_F32_T<5>;
kernels[5] = mul_mat_iq3_kt_F32_T<6>;
kernels[6] = mul_mat_iq3_kt_F32_T<7>;
kernels[7] = mul_mat_iq3_kt_F32_T<8>;
IQK_SET_MUL_MAT_FUNCTIONS(mul_mat_iq3_kt_F32_T, kernels);
break;
case GGML_TYPE_IQ4_KT:
assert (ne00 % QK_K == 0);
kernels[0] = mul_mat_iq4_kt_F32_T<1>;
kernels[1] = mul_mat_iq4_kt_F32_T<2>;
kernels[2] = mul_mat_iq4_kt_F32_T<3>;
kernels[3] = mul_mat_iq4_kt_F32_T<4>;
kernels[4] = mul_mat_iq4_kt_F32_T<5>;
kernels[5] = mul_mat_iq4_kt_F32_T<6>;
kernels[6] = mul_mat_iq4_kt_F32_T<7>;
kernels[7] = mul_mat_iq4_kt_F32_T<8>;
IQK_SET_MUL_MAT_FUNCTIONS(mul_mat_iq4_kt_F32_T, kernels);
break;
default:
return false;
@@ -358,8 +335,391 @@ bool iqk_set_kernels_ktquants(int ne00, int typeA, int typeB, std::array<mul_mat
#else // !__x86_64__
namespace {
struct Trellis1 {
constexpr static uint32_t kmask = 0x8fff8fff;
constexpr static uint32_t km32 = 0x3b603b60;
constexpr static uint32_t ka = 89226354;
constexpr static uint32_t kb = 64248484;
constexpr static uint32_t ka1 = ka*ka;
constexpr static uint32_t kb1 = kb*ka+kb;
constexpr static uint32_t ka2 = ka1*ka;
constexpr static uint32_t kb2 = kb1*ka+kb;
constexpr static uint32_t ka3 = ka2*ka;
constexpr static uint32_t kb3 = kb2*ka+kb;
constexpr static uint32_t ka4 = ka3*ka;
constexpr static uint32_t kb4 = kb3*ka+kb;
constexpr static uint32_t ka5 = ka4*ka;
constexpr static uint32_t kb5 = kb4*ka+kb;
constexpr static uint32_t ka6 = ka5*ka;
constexpr static uint32_t kb6 = kb5*ka+kb;
constexpr static uint32_t ka7 = ka6*ka;
constexpr static uint32_t kb7 = kb6*ka+kb;
const uint32x4x2_t mka = {uint32x4_t{ka, ka1, ka2, ka3}, uint32x4_t{ka4, ka5, ka6, ka7}};
const uint32x4x2_t mkb = {uint32x4_t{kb, kb1, kb2, kb3}, uint32x4_t{kb4, kb5, kb6, kb7}};
const uint32x4_t mask1 = vdupq_n_u32(kmask);
const uint32x4_t mask2 = vdupq_n_u32(km32);
inline uint32x4x2_t next8(uint32_t val) const {
auto mval = vdupq_n_u32(val);
uint32x4x2_t mres;
// This does not seem to be faster
//mres.val[0] = vmlaq_u32(mkb.val[0], mka.val[0], mval);
//mres.val[1] = vmlaq_u32(mkb.val[1], mka.val[1], mval);
mres.val[0] = vaddq_u32(vmulq_u32(mval, mka.val[0]), mkb.val[0]);
mres.val[1] = vaddq_u32(vmulq_u32(mval, mka.val[1]), mkb.val[1]);
mres.val[0] = veorq_u32(vandq_u32(mres.val[0], mask1), mask2);
mres.val[1] = veorq_u32(vandq_u32(mres.val[1], mask1), mask2);
return mres;
}
inline uint32x4x2_t next8(uint32_t val1, uint32_t val2) const {
auto mval1 = vdupq_n_u32(val1);
auto mval2 = vdupq_n_u32(val2);
uint32x4x2_t mres;
// This does not seem to be faster
//mres.val[0] = vmlaq_u32(mkb.val[0], mka.val[0], mval1);
//mres.val[1] = vmlaq_u32(mkb.val[0], mka.val[0], mval2);
mres.val[0] = vaddq_u32(vmulq_u32(mval1, mka.val[0]), mkb.val[0]);
mres.val[1] = vaddq_u32(vmulq_u32(mval2, mka.val[0]), mkb.val[0]);
mres.val[0] = veorq_u32(vandq_u32(mres.val[0], mask1), mask2);
mres.val[1] = veorq_u32(vandq_u32(mres.val[1], mask1), mask2);
return mres;
}
static inline float16x8_t gen8(const uint32x4x2_t& i8) {
auto fv1 = vreinterpretq_f16_u32(i8.val[0]);
auto fv2 = vreinterpretq_f16_u32(i8.val[1]);
return vpaddq_f16(fv1, fv2);
}
inline float16x8_t gen8(uint32_t val) const { return gen8(next8(val)); }
inline float16x8_t gen8(uint32_t val1, uint32_t val2) const { return gen8(next8(val1, val2)); }
inline float32x4x2_t gen8_f32(uint32_t val1, uint32_t val2) const {
auto x16 = gen8(val1, val2);
return { vcvt_f32_f16(vget_low_f16(x16)), vcvt_f32_f16(vget_high_f16(x16)) };
}
inline float32x4x2_t gen8_f32(uint32_t val1, uint32_t val2, float16x8_t scale) const {
auto x16 = vmulq_f16(gen8(val1, val2), scale);
return { vcvt_f32_f16(vget_low_f16(x16)), vcvt_f32_f16(vget_high_f16(x16)) };
}
};
template <int nrc_y>
static void mul_mat_iq2_kt_F16_T(int n, const void * vx, size_t bx, const DataInfo& info, int nrc_x) {
assert(n%QK_K == 0);
const int nb = n/QK_K;
Trellis1 trellis;
auto values = vld1q_s8(iq4k_values);
union { float16x8_t vec; float16_t val[8]; } s_helper;
constexpr int k_acc = nrc_y == 1 ? 2 : nrc_y;
float16x8_t accd[k_acc];
const float16_t * y[nrc_y];
for (int iy = 0; iy < nrc_y; ++iy) y[iy] = (const float16_t *)info.src1_row(iy);
for (int ix = 0; ix < nrc_x; ++ix) {
const float * dptr = (const float *)((const char*)vx + ix*bx);
const float d = *dptr * 31.75f * 1.05f;
const block_iq2_kt * x = (const block_iq2_kt *)(dptr + 1);
for (int iy = 0; iy < k_acc; ++iy) accd[iy] = vdupq_n_f16(0);
for (int i = 0; i < nb; ++i) {
const uint16_t * ql = (const uint16_t *)x[i].ql;
auto u32 = *(const uint32_t *)x[i].scales;
auto s8_u32 = uint32x2_t{u32, u32 >> 4};
s8_u32 = vand_u8(s8_u32, vdup_n_u32(0x0f0f0f0f));
auto s8 = vqtbl1_s8(values, vreinterpret_u8_u32(s8_u32));
auto s16 = vmovl_s8(s8);
s_helper.vec = vcvtq_f16_s16(s16);
for (int ib = 0; ib < QK_K/64; ++ib) {
auto scale1 = vdupq_n_f16(s_helper.val[2*ib+0]);
auto scale2 = vdupq_n_f16(s_helper.val[2*ib+1]);
for (int j = 0; j < 4; ++j) {
auto xval1 = vmulq_f16(scale1, trellis.gen8(ql[8*ib+j+0]+4096));
auto xval2 = vmulq_f16(scale2, trellis.gen8(ql[8*ib+j+4]+4096));
if constexpr (nrc_y == 1) {
accd[0] = vfmaq_f16(accd[0], xval1, vld1q_f16(y[0] + i*QK_K + 64*ib + 8*j + 0));
accd[1] = vfmaq_f16(accd[1], xval2, vld1q_f16(y[0] + i*QK_K + 64*ib + 8*j + 32));
} else {
for (int iy = 0; iy < nrc_y; ++iy) {
accd[iy] = vfmaq_f16(accd[iy], xval1, vld1q_f16(y[iy] + i*QK_K + 64*ib + 8*j + 0));
accd[iy] = vfmaq_f16(accd[iy], xval2, vld1q_f16(y[iy] + i*QK_K + 64*ib + 8*j + 32));
}
}
}
}
}
if constexpr (nrc_y == 1) {
auto res16 = vpaddq_f16(accd[0], accd[1]);
auto res = vaddq_f32(vcvt_f32_f16(vget_low_f16(res16)), vcvt_f32_f16(vget_high_f16(res16)));
info.store(ix, 0, vaddvq_f32(res)*d);
} else {
for (int iy = 0; iy < nrc_y; ++iy) {
auto res = vaddq_f32(vcvt_f32_f16(vget_low_f16(accd[iy])), vcvt_f32_f16(vget_high_f16(accd[iy])));
info.store(ix, iy, vaddvq_f32(res)*d);
}
}
}
}
template <int nrc_y>
static void mul_mat_iq3_kt_F16_T(int n, const void * vx, size_t bx, const DataInfo& info, int nrc_x) {
assert(n%QK_K == 0);
const int nb = n/QK_K;
Trellis1 trellis;
union { float16x8_t vec; float16_t val[8]; } s_helper;
uint16x8_t all_signs[4];
auto mask1 = vdupq_n_u16(0x01);
auto mask2 = vdupq_n_u16(0x10);
float16x8_t accd[nrc_y];
const float16_t * y[nrc_y];
for (int iy = 0; iy < nrc_y; ++iy) y[iy] = (const float16_t *)info.src1_row(iy);
for (int ix = 0; ix < nrc_x; ++ix) {
const float * dptr = (const float *)((const char*)vx + ix*bx);
const float d = *dptr * 31.75f * 1.015f;
const block_iq3_kt * x = (const block_iq3_kt *)(dptr + 1);
for (int iy = 0; iy < nrc_y; ++iy) accd[iy] = vdupq_n_f16(0);
for (int i = 0; i < nb; ++i) {
const uint16_t * ql = (const uint16_t *)x[i].ql;
const uint8_t * qh = x[i].qh;
auto u32 = *(const uint32_t *)x[i].scales;
auto s8_u32 = uint32x2_t{u32, u32 >> 4};
s8_u32 = vand_u8(s8_u32, vdup_n_u32(0x0f0f0f0f));
auto s16 = vmovl_s8(vreinterpret_s8_u32(s8_u32));
s_helper.vec = vcvtq_f16_s16(s16);
for (int j = 0; j < 4; ++j) all_signs[j] = vmovl_u8(vld1_u8(qh + 8*j));
for (int ib = 0; ib < 4; ++ib) {
auto scale1 = vdupq_n_f16(s_helper.val[ib+0]);
auto scale2 = vdupq_n_f16(s_helper.val[ib+4]);
for (int j = 0; j < 4; ++j) {
uint32_t val1 = ql[4*ib+j ] + 4096;
uint32_t val2 = ql[4*ib+j+16] + 4096;
auto sign1 = vshlq_n_u16(vandq_u16(all_signs[j], mask1), 15);
auto sign2 = vshlq_n_u16(vandq_u16(all_signs[j], mask2), 11);
all_signs[j] = vshrq_n_u16(all_signs[j], 1);
auto x_val1 = vabsq_f16(trellis.gen8(val1));
auto x_val2 = vabsq_f16(trellis.gen8(val2));
x_val1 = vmulq_f16(scale1, vreinterpretq_f16_u16(vorrq_u16(vreinterpretq_u16_f16(x_val1), sign1)));
x_val2 = vmulq_f16(scale2, vreinterpretq_f16_u16(vorrq_u16(vreinterpretq_u16_f16(x_val2), sign2)));
for (int iy = 0; iy < nrc_y; ++iy) {
accd[iy] = vfmaq_f16(accd[iy], x_val1, vld1q_f16(y[iy] + i*QK_K+32*ib+8*j ));
accd[iy] = vfmaq_f16(accd[iy], x_val2, vld1q_f16(y[iy] + i*QK_K+32*ib+8*j+128));
}
}
}
}
for (int iy = 0; iy < nrc_y; ++iy) {
auto res = vaddq_f32(vcvt_f32_f16(vget_low_f16(accd[iy])), vcvt_f32_f16(vget_high_f16(accd[iy])));
info.store(ix, iy, d*vaddvq_f32(res));
}
}
}
template <int nrc_y>
static void mul_mat_iq4_kt_F16_T(int n, const void * vx, size_t bx, const DataInfo& info, int nrc_x) {
assert(n%QK_K == 0);
const int nb = n/QK_K;
constexpr int kNumGroups = 64;
Trellis1 trellis;
union { float16x8_t vec; float16_t val[8]; } s_helper;
union { uint16x8_t vec; uint16_t val[8]; } o_helper;
constexpr int k_acc = nrc_y == 1 ? 2 : nrc_y;
float16x8_t accd[k_acc];
const float16_t * y[nrc_y];
float row_sum[nrc_y];
for (int iy = 0; iy < nrc_y; ++iy) {
y[iy] = (const float16_t *)info.src1_row(iy);
auto sum = vdupq_n_f16(0);
for (int i = 0; i < n/8; ++i) sum = vaddq_f16(sum, vld1q_f16(y[iy] + 8*i));
auto sum32 = vaddq_f32(vcvt_f32_f16(vget_low_f16(sum)), vcvt_f32_f16(vget_high_f16(sum)));
//auto sum32 = vdupq_n_f32(0);
//for (int i = 0; i < n/4; ++i) sum32 = vaddq_f32(sum32, vcvt_f32_f16(vld1_f16(y[iy] + 4*i)));
row_sum[iy] = vaddvq_f32(sum32);
}
for (int ix = 0; ix < nrc_x; ++ix) {
const float * dptr = (const float *)((const char*)vx + ix*bx);
auto d = dptr[0] * 31.75f * 1.01f;
auto dav = dptr[1];
const block_iq4_kt * x = (const block_iq4_kt *)(dptr + 2);
for (int iy = 0; iy < k_acc; ++iy) accd[iy] = vdupq_n_f16(0);
for (int i = 0; i < nb; ++i) {
const uint32_t * shb = x[i].qs;
auto vshb = vld1q_u32_x2(shb);
auto vshb16 = vcombine_u16(vmovn_u32(vandq_u32(vshb.val[0], vdupq_n_u32(0xff))), vmovn_u32(vandq_u32(vshb.val[1], vdupq_n_u32(0xff))));
const uint8_t * ql = (const uint8_t *)(shb + 8);
const uint8_t * qh = ql + kNumGroups;
auto iscales = vsubq_s16(vreinterpretq_s16_u16(vshrq_n_u16(vshb16, 1)), vdupq_n_s16(64));
s_helper.vec = vcvtq_f16_s16(iscales);
o_helper.vec = vaddq_u16(vshlq_n_u16(vandq_u16(vshb16, vdupq_n_u16(1)), 15), vdupq_n_u16(4096));
for (int ib = 0; ib < 4; ++ib) {
auto scale1 = vdupq_n_f16(s_helper.val[ib+0]);
auto scale2 = vdupq_n_f16(s_helper.val[ib+4]);
for (int j = 0; j < 4; ++j) {
const uint32_t sh1 = shb[ib+0] >> (8 + 6*j);
const uint32_t sh2 = shb[ib+4] >> (8 + 6*j);
uint32_t val1 = ql[8*ib+2*j+ 0] + ((qh[8*ib+2*j+0] << 8) & 0xf00) + ((sh1 & 7) << 12) + o_helper.val[ib+0];
uint32_t val2 = ql[8*ib+2*j+32] + ((qh[8*ib+2*j+0] << 4) & 0xf00) + ((sh2 & 7) << 12) + o_helper.val[ib+4];
uint32_t val3 = ql[8*ib+2*j+ 1] + ((qh[8*ib+2*j+1] << 8) & 0xf00) + ((sh1 & 56) << 9) + o_helper.val[ib+0];
uint32_t val4 = ql[8*ib+2*j+33] + ((qh[8*ib+2*j+1] << 4) & 0xf00) + ((sh2 & 56) << 9) + o_helper.val[ib+4];
auto x_val1 = vmulq_f16(scale1, trellis.gen8(val1, val3));
auto x_val2 = vmulq_f16(scale2, trellis.gen8(val2, val4));
if constexpr (nrc_y == 1) {
auto y1 = vld1q_f16(y[0] + i*QK_K+32*ib+8*j+ 0);
auto y2 = vld1q_f16(y[0] + i*QK_K+32*ib+8*j+128);
accd[0] = vfmaq_f16(accd[0], y1, x_val1);
accd[1] = vfmaq_f16(accd[1], y2, x_val2);
} else {
for (int iy = 0; iy < nrc_y; ++iy) {
auto y1 = vld1q_f16(y[iy] + i*QK_K+32*ib+8*j+ 0);
auto y2 = vld1q_f16(y[iy] + i*QK_K+32*ib+8*j+128);
accd[iy] = vfmaq_f16(accd[iy], y1, x_val1);
accd[iy] = vfmaq_f16(accd[iy], y2, x_val2);
}
}
}
}
}
if constexpr (nrc_y == 1) {
auto sum16 = vaddq_f16(accd[0], accd[1]);
auto sum = vaddq_f32(vcvt_f32_f16(vget_low_f16(sum16)), vcvt_f32_f16(vget_high_f16(sum16)));
info.store(ix, 0, d*vaddvq_f32(sum) + dav*row_sum[0]);
} else {
for (int iy = 0; iy < nrc_y; ++iy) {
auto sum = vaddq_f32(vcvt_f32_f16(vget_low_f16(accd[iy])), vcvt_f32_f16(vget_high_f16(accd[iy])));
info.store(ix, iy, d*vaddvq_f32(sum) + dav*row_sum[iy]);
}
}
}
}
template <int nrc_y>
static void mul_mat_iq4_kt_F32_T(int n, const void * vx, size_t bx, const DataInfo& info, int nrc_x) {
assert(n%QK_K == 0);
const int nb = n/QK_K;
constexpr int kNumGroups = 64;
Trellis1 trellis;
float32x4_t accd[nrc_y * 2];
const float * y[nrc_y];
float row_sum[nrc_y];
for (int iy = 0; iy < nrc_y; ++iy) {
y[iy] = (const float *)info.src1_row(iy);
auto sum = vdupq_n_f32(0);
for (int i = 0; i < n/4; ++i) sum = vaddq_f32(sum, vld1q_f32(y[iy] + 4*i));
row_sum[iy] = vaddvq_f32(sum);
}
for (int ix = 0; ix < nrc_x; ++ix) {
const float * dptr = (const float *)((const char*)vx + ix*bx);
const float d = dptr[0] * 31.75f * 1.01f;
const float row_av = dptr[1];
const block_iq4_kt * x = (const block_iq4_kt *)(dptr + 2);
for (int iy = 0; iy < nrc_y * 2; ++iy) accd[iy] = vdupq_n_f32(0.0f);
for (int i = 0; i < nb; ++i) {
const uint32_t * shb = x[i].qs;
const uint8_t * ql = (const uint8_t *)(shb + 8);
const uint8_t * qh = ql + kNumGroups;
for (int ib = 0; ib < 4; ++ib) {
const uint16_t x_scale1 = (int16_t)((shb[ib+0] & 0xff) >> 1) - 64;
const uint16_t x_scale2 = (int16_t)((shb[ib+4] & 0xff) >> 1) - 64;
const float16x8_t scale1 = vcvtq_f16_s16(vdupq_n_s16(x_scale1));
const float16x8_t scale2 = vcvtq_f16_s16(vdupq_n_s16(x_scale2));
const uint32_t offset1 = 4096 + ((shb[ib+0] & 1) << 15);
const uint32_t offset2 = 4096 + ((shb[ib+4] & 1) << 15);
uint32_t sh1 = shb[ib+0] >> 8;
uint32_t sh2 = shb[ib+4] >> 8;
for (int j = 0; j < 4; ++j) {
uint32_t val1 = ql[8*ib+2*j+ 0] + ((qh[8*ib+2*j+0] << 8) & 0xf00) + ((sh1 & 7) << 12) + offset1;
uint32_t val2 = ql[8*ib+2*j+32] + ((qh[8*ib+2*j+0] << 4) & 0xf00) + ((sh2 & 7) << 12) + offset2;
uint32_t val3 = ql[8*ib+2*j+ 1] + ((qh[8*ib+2*j+1] << 8) & 0xf00) + ((sh1 & 56) << 9) + offset1;
uint32_t val4 = ql[8*ib+2*j+33] + ((qh[8*ib+2*j+1] << 4) & 0xf00) + ((sh2 & 56) << 9) + offset2;
sh1 >>= 6;
sh2 >>= 6;
auto x1 = trellis.gen8_f32(val1, val3, scale1);
auto x2 = trellis.gen8_f32(val2, val4, scale2);
for (int iy = 0; iy < nrc_y; ++iy) {
auto y1 = vld1q_f32_x2(y[iy] + i*QK_K + 32*ib + 8*j);
auto y2 = vld1q_f32_x2(y[iy] + i*QK_K + 32*ib + 8*j + 128);
accd[iy*2 + 0] = vfmaq_f32(accd[iy*2 + 0], y1.val[0], x1.val[0]);
accd[iy*2 + 1] = vfmaq_f32(accd[iy*2 + 1], y1.val[1], x1.val[1]);
accd[iy*2 + 0] = vfmaq_f32(accd[iy*2 + 0], y2.val[0], x2.val[0]);
accd[iy*2 + 1] = vfmaq_f32(accd[iy*2 + 1], y2.val[1], x2.val[1]);
}
}
}
}
for (int iy = 0; iy < nrc_y; ++iy) {
float32x4_t sum1 = vaddq_f32(accd[iy*2], accd[iy*2 + 1]);
float result = d*vaddvq_f32(sum1) + row_av*row_sum[iy];
info.store(ix, iy, result);
}
}
}
}
bool iqk_set_kernels_ktquants(int ne00, int typeA, int typeB, std::array<mul_mat_t, IQK_MAX_NY>& kernels, mul_mat_t& func16) {
return false;
if (ne00%QK_K == 0 && ggml_type(typeB) == GGML_TYPE_F32 && ggml_type(typeA) == GGML_TYPE_IQ4_KT) {
IQK_SET_MUL_MAT_FUNCTIONS(mul_mat_iq4_kt_F32_T, kernels);
func16 = nullptr;
return true;
}
if (ne00%QK_K != 0 || ggml_type(typeB) != GGML_TYPE_F16) {
return false;
}
func16 = nullptr;
switch (typeA) {
case GGML_TYPE_IQ2_KT:
IQK_SET_MUL_MAT_FUNCTIONS(mul_mat_iq2_kt_F16_T, kernels);
break;
case GGML_TYPE_IQ3_KT:
IQK_SET_MUL_MAT_FUNCTIONS(mul_mat_iq3_kt_F16_T, kernels);
break;
case GGML_TYPE_IQ4_KT:
IQK_SET_MUL_MAT_FUNCTIONS(mul_mat_iq4_kt_F16_T, kernels);
break;
default:
return false;
}
return true;
}
#endif

6
ggml/src/iqk/iqk_mul_mat.cpp
View File

@@ -651,6 +651,10 @@ bool MulMat::prepare(int typeA, int typeB, int ne00, MulMat& m, int /*Ny*/) {
case GGML_TYPE_IQ1_S_R4:
case GGML_TYPE_IQ1_M_R4:
return iqk_set_kernels_1bit(ne00, typeA, typeB, m.funcs, m.func16);
case GGML_TYPE_IQ2_KT:
case GGML_TYPE_IQ3_KT:
case GGML_TYPE_IQ4_KT:
return iqk_set_kernels_ktquants(ne00, typeA, typeB, m.funcs, m.func16);
default:
return false;
}
@@ -926,4 +930,4 @@ extern "C" IQK_API bool iqk_moe_fused_up_gate(long /*Nx*/, long /*Ny*/, long /*n
return false;
}
#endif
#endif