From 4af24881b3fe923b7bc7d83e07ba171f93a6f14c Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Wed, 3 Jan 2024 10:58:19 +0200 Subject: [PATCH 01/17] iq2_xxs: basics --- ggml-quants.c | 169 ++++++++++++++++++++++++++++++++++++++++++++++++++ ggml-quants.h | 12 ++++ ggml.c | 26 ++++++++ ggml.h | 3 + 4 files changed, 210 insertions(+) diff --git a/ggml-quants.c b/ggml-quants.c index 55a9496d1b37b..4a522e0145b8c 100644 --- a/ggml-quants.c +++ b/ggml-quants.c @@ -2340,6 +2340,138 @@ size_t ggml_quantize_q6_K(const float * src, void * dst, int n, int k, int64_t * return (n/QK_K*sizeof(block_q6_K)); } +// ====================== "True" 2-bit (de)-quantization + +void quantize_row_iq2_xxs_reference(const float * restrict x, block_iq2_xxs * restrict y, int k) { + (void)x; + (void)y; + (void)k; + assert(k % QK_K == 0); + //fprintf(stderr, "=========================== %s: not implemented\n", __func__); +} + +static const uint64_t iq2xxs_grid[256] = { + 0x0808080808080808, 0x080808080808082b, 0x0808080808081919, 0x0808080808082b08, + 0x0808080808082b2b, 0x0808080808190819, 0x0808080808191908, 0x08080808082b0808, + 0x08080808082b082b, 0x08080808082b2b08, 0x08080808082b2b2b, 0x0808080819080819, + 0x0808080819081908, 0x0808080819190808, 0x0808080819192b08, 0x08080808192b0819, + 0x08080808192b1908, 0x080808082b080808, 0x080808082b08082b, 0x080808082b082b2b, + 0x080808082b2b082b, 0x0808081908080819, 0x0808081908081908, 0x0808081908190808, + 0x0808081908191919, 0x0808081919080808, 0x080808192b081908, 0x080808192b192b08, + 0x0808082b08080808, 0x0808082b0808082b, 0x0808082b082b082b, 0x0808082b2b08082b, + 0x0808190808080819, 0x0808190808081908, 0x0808190808190808, 0x08081908082b0819, + 0x08081908082b1908, 0x0808190819080808, 0x080819081908082b, 0x0808190819082b08, + 0x08081908192b0808, 0x080819082b080819, 0x080819082b081908, 0x080819082b190808, + 0x080819082b2b1908, 0x0808191908080808, 0x080819190808082b, 0x0808191908082b08, + 0x08081919082b0808, 0x080819191908192b, 0x08081919192b2b19, 0x080819192b080808, + 0x080819192b190819, 0x0808192b08082b19, 0x0808192b08190808, 0x0808192b19080808, + 0x0808192b2b081908, 0x0808192b2b2b1908, 0x08082b0808080808, 0x08082b0808081919, + 0x08082b0808082b08, 0x08082b0808191908, 0x08082b08082b2b08, 0x08082b0819080819, + 0x08082b0819081908, 0x08082b0819190808, 0x08082b081919082b, 0x08082b082b082b08, + 0x08082b1908081908, 0x08082b1919080808, 0x08082b2b0808082b, 0x08082b2b08191908, + 0x0819080808080819, 0x0819080808081908, 0x0819080808190808, 0x08190808082b0819, + 0x0819080819080808, 0x08190808192b0808, 0x081908082b081908, 0x081908082b190808, + 0x081908082b191919, 0x0819081908080808, 0x0819081908082b08, 0x08190819082b0808, + 0x0819081919190808, 0x0819081919192b2b, 0x081908192b080808, 0x0819082b082b1908, + 0x0819082b19081919, 0x0819190808080808, 0x0819190808082b08, 0x08191908082b0808, + 0x08191908082b1919, 0x0819190819082b19, 0x081919082b080808, 0x0819191908192b08, + 0x08191919192b082b, 0x0819192b08080808, 0x0819192b0819192b, 0x08192b0808080819, + 0x08192b0808081908, 0x08192b0808190808, 0x08192b0819080808, 0x08192b082b080819, + 0x08192b1908080808, 0x08192b1908081919, 0x08192b192b2b0808, 0x08192b2b19190819, + 0x082b080808080808, 0x082b08080808082b, 0x082b080808082b2b, 0x082b080819081908, + 0x082b0808192b0819, 0x082b08082b080808, 0x082b08082b08082b, 0x082b0819082b2b19, + 0x082b081919082b08, 0x082b082b08080808, 0x082b082b0808082b, 0x082b190808080819, + 0x082b190808081908, 0x082b190808190808, 0x082b190819080808, 0x082b19081919192b, + 0x082b191908080808, 0x082b191919080819, 0x082b1919192b1908, 0x082b192b2b190808, + 0x082b2b0808082b08, 0x082b2b08082b0808, 0x082b2b082b191908, 0x082b2b2b19081908, + 0x1908080808080819, 0x1908080808081908, 0x1908080808190808, 0x1908080808192b08, + 0x19080808082b0819, 0x19080808082b1908, 0x1908080819080808, 0x1908080819082b08, + 0x190808081919192b, 0x19080808192b0808, 0x190808082b080819, 0x190808082b081908, + 0x190808082b190808, 0x1908081908080808, 0x19080819082b0808, 0x19080819192b0819, + 0x190808192b080808, 0x190808192b081919, 0x1908082b08080819, 0x1908082b08190808, + 0x1908082b19082b08, 0x1908082b1919192b, 0x1908082b192b2b08, 0x1908190808080808, + 0x1908190808082b08, 0x19081908082b0808, 0x190819082b080808, 0x190819082b192b19, + 0x190819190819082b, 0x19081919082b1908, 0x1908192b08080808, 0x19082b0808080819, + 0x19082b0808081908, 0x19082b0808190808, 0x19082b0819080808, 0x19082b0819081919, + 0x19082b1908080808, 0x19082b1919192b08, 0x19082b19192b0819, 0x19082b192b08082b, + 0x19082b2b19081919, 0x19082b2b2b190808, 0x1919080808080808, 0x1919080808082b08, + 0x1919080808190819, 0x1919080808192b19, 0x19190808082b0808, 0x191908082b080808, + 0x191908082b082b08, 0x1919081908081908, 0x191908191908082b, 0x191908192b2b1908, + 0x1919082b2b190819, 0x191919082b190808, 0x191919082b19082b, 0x1919191908082b2b, + 0x1919192b08080819, 0x1919192b19191908, 0x19192b0808080808, 0x19192b0808190819, + 0x19192b0808192b19, 0x19192b08192b1908, 0x19192b1919080808, 0x19192b2b08082b08, + 0x192b080808081908, 0x192b080808190808, 0x192b080819080808, 0x192b0808192b2b08, + 0x192b081908080808, 0x192b081919191919, 0x192b082b08192b08, 0x192b082b192b0808, + 0x192b190808080808, 0x192b190808081919, 0x192b191908190808, 0x192b19190819082b, + 0x192b19192b081908, 0x192b2b081908082b, 0x2b08080808080808, 0x2b0808080808082b, + 0x2b08080808082b2b, 0x2b08080819080819, 0x2b0808082b08082b, 0x2b08081908081908, + 0x2b08081908192b08, 0x2b08081919080808, 0x2b08082b08190819, 0x2b08190808080819, + 0x2b08190808081908, 0x2b08190808190808, 0x2b08190808191919, 0x2b08190819080808, + 0x2b081908192b0808, 0x2b08191908080808, 0x2b0819191908192b, 0x2b0819192b191908, + 0x2b08192b08082b19, 0x2b08192b19080808, 0x2b08192b192b0808, 0x2b082b080808082b, + 0x2b082b1908081908, 0x2b082b2b08190819, 0x2b19080808081908, 0x2b19080808190808, + 0x2b190808082b1908, 0x2b19080819080808, 0x2b1908082b2b0819, 0x2b1908190819192b, + 0x2b1908192b080808, 0x2b19082b19081919, 0x2b19190808080808, 0x2b191908082b082b, + 0x2b19190819081908, 0x2b19191919190819, 0x2b192b082b080819, 0x2b192b19082b0808, + 0x2b2b08080808082b, 0x2b2b080819190808, 0x2b2b08082b081919, 0x2b2b081908082b19, + 0x2b2b082b08080808, 0x2b2b190808192b08, 0x2b2b2b0819190808, 0x2b2b2b1908081908, +}; + +static const uint8_t ksigns_iq2xs[128] = { + 0, 129, 130, 3, 132, 5, 6, 135, 136, 9, 10, 139, 12, 141, 142, 15, + 144, 17, 18, 147, 20, 149, 150, 23, 24, 153, 154, 27, 156, 29, 30, 159, + 160, 33, 34, 163, 36, 165, 166, 39, 40, 169, 170, 43, 172, 45, 46, 175, + 48, 177, 178, 51, 180, 53, 54, 183, 184, 57, 58, 187, 60, 189, 190, 63, + 192, 65, 66, 195, 68, 197, 198, 71, 72, 201, 202, 75, 204, 77, 78, 207, + 80, 209, 210, 83, 212, 85, 86, 215, 216, 89, 90, 219, 92, 221, 222, 95, + 96, 225, 226, 99, 228, 101, 102, 231, 232, 105, 106, 235, 108, 237, 238, 111, + 240, 113, 114, 243, 116, 245, 246, 119, 120, 249, 250, 123, 252, 125, 126, 255, +}; +static const uint8_t kmask_iq2xs[8] = {1, 2, 4, 8, 16, 32, 64, 128}; + +void dequantize_row_iq2_xxs(const block_iq2_xxs * restrict x, float * restrict y, int k) { + assert(k % QK_K == 0); + const int nb = k / QK_K; + + uint32_t aux32[2]; + const uint8_t * aux8 = (const uint8_t *)aux32; + + for (int i = 0; i < nb; i++) { + + const float d = GGML_FP16_TO_FP32(x[i].d); + + for (int ib32 = 0; ib32 < QK_K/32; ++ib32) { + memcpy(aux32, x[i].qs + 4*ib32, 2*sizeof(uint32_t)); + const float db = d * (0.5f + (aux32[1] >> 28)); + for (int l = 0; l < 4; ++l) { + const uint8_t * grid = (const uint8_t *)(iq2xxs_grid + aux8[l]); + const uint8_t signs = ksigns_iq2xs[(aux32[1] >> 7*l) & 127]; + for (int j = 0; j < 8; ++j) { + y[j] = db * grid[j] * (signs & kmask_iq2xs[j] ? -1.f : 1.f); + } + y += 8; + } + } + } +} + +void quantize_row_iq2_xxs(const float * restrict x, void * restrict vy, int k) { + assert(k % QK_K == 0); + block_iq2_xxs * restrict y = vy; + quantize_row_iq2_xxs_reference(x, y, k); +} + +size_t ggml_quantize_iq2_xxs(const float * src, void * dst, int n, int k, int64_t * hist) { + assert(k % QK_K == 0); + (void)hist; // TODO: collect histograms + + for (int j = 0; j < n; j += k) { + block_iq2_xxs * restrict y = (block_iq2_xxs *)dst + j/QK_K; + quantize_row_iq2_xxs_reference(src + j, y, k); + } + return (n/QK_K*sizeof(block_iq2_xxs)); +} + //===================================== Q8_K ============================================== void quantize_row_q8_K_reference(const float * restrict x, block_q8_K * restrict y, int k) { @@ -7065,3 +7197,40 @@ void ggml_vec_dot_q6_K_q8_K(const int n, float * restrict s, const void * restri } #endif + +void ggml_vec_dot_iq2_xxs_q8_K(const int n, float * restrict s, const void * restrict vx, const void * restrict vy) { + assert(n % QK_K == 0); + + const block_iq2_xxs * restrict x = vx; + const block_q8_K * restrict y = vy; + + const int nb = n / QK_K; + + uint32_t aux32[2]; + const uint8_t * aux8 = (const uint8_t *)aux32; + + float sumf = 0.f; + for (int i = 0; i < nb; ++i) { + const float d = GGML_FP16_TO_FP32(x[i].d) * y[i].d; + const uint16_t * restrict q2 = x[i].qs; + const int8_t * restrict q8 = y[i].qs; + int32_t bsum = 0; + for (int ib32 = 0; ib32 < QK_K/32; ++ib32) { + memcpy(aux32, q2, 2*sizeof(uint32_t)); + q2 += 4; + const uint32_t ls = 2*(aux32[1] >> 28) + 1; + int32_t sumi = 0; + for (int l = 0; l < 4; ++l) { + const uint8_t * grid = (const uint8_t *)(iq2xxs_grid + aux8[l]); + const uint8_t signs = ksigns_iq2xs[(aux32[1] >> 7*l) & 127]; + for (int j = 0; j < 8; ++j) { + sumi += grid[j] * q8[j] * (signs & kmask_iq2xs[j] ? -1 : 1); + } + q8 += 8; + } + bsum += sumi * ls; + } + sumf += d * bsum; + } + *s = 0.125f * sumf; +} diff --git a/ggml-quants.h b/ggml-quants.h index 62c1df6cbd274..8dd911d4182fa 100644 --- a/ggml-quants.h +++ b/ggml-quants.h @@ -165,6 +165,14 @@ typedef struct { } block_q8_K; static_assert(sizeof(block_q8_K) == sizeof(float) + QK_K + QK_K/16*sizeof(int16_t), "wrong q8_K block size/padding"); +// (Almost) "true" 2-bit quantization. +// Due to the need to use blocks as per ggml dsign, it ends up using +// 2.0625 bpw because of the 16-bit scale for each block of 256. +typedef struct { + ggml_fp16_t d; + uint16_t qs[QK_K/8]; +} block_iq2_xxs; +static_assert(sizeof(block_iq2_xxs) == sizeof(ggml_fp16_t) + QK_K/8*sizeof(uint16_t), "wrong iq2_xxs block size/padding"); // Quantization void quantize_row_q4_0_reference(const float * restrict x, block_q4_0 * restrict y, int k); @@ -180,6 +188,7 @@ void quantize_row_q4_K_reference(const float * restrict x, block_q4_K * restrict void quantize_row_q5_K_reference(const float * restrict x, block_q5_K * restrict y, int k); void quantize_row_q6_K_reference(const float * restrict x, block_q6_K * restrict y, int k); void quantize_row_q8_K_reference(const float * restrict x, block_q8_K * restrict y, int k); +void quantize_row_iq2_xxs_reference(const float * restrict x, block_iq2_xxs * restrict y, int k); void quantize_row_q4_0(const float * restrict x, void * restrict y, int k); void quantize_row_q4_1(const float * restrict x, void * restrict y, int k); @@ -194,6 +203,7 @@ void quantize_row_q4_K(const float * restrict x, void * restrict y, int k); void quantize_row_q5_K(const float * restrict x, void * restrict y, int k); void quantize_row_q6_K(const float * restrict x, void * restrict y, int k); void quantize_row_q8_K(const float * restrict x, void * restrict y, int k); +void quantize_row_iq2_xxs(const float * restrict x, void * restrict y, int k); // Dequantization void dequantize_row_q4_0(const block_q4_0 * restrict x, float * restrict y, int k); @@ -209,6 +219,7 @@ void dequantize_row_q4_K(const block_q4_K * restrict x, float * restrict y, int void dequantize_row_q5_K(const block_q5_K * restrict x, float * restrict y, int k); void dequantize_row_q6_K(const block_q6_K * restrict x, float * restrict y, int k); void dequantize_row_q8_K(const block_q8_K * restrict x, float * restrict y, int k); +void dequantize_row_iq2_xxs(const block_iq2_xxs * restrict x, float * restrict y, int k); // Dot product void ggml_vec_dot_q4_0_q8_0(int n, float * restrict s, const void * restrict vx, const void * restrict vy); @@ -222,3 +233,4 @@ void ggml_vec_dot_q3_K_q8_K(int n, float * restrict s, const void * restrict vx, void ggml_vec_dot_q4_K_q8_K(int n, float * restrict s, const void * restrict vx, const void * restrict vy); void ggml_vec_dot_q5_K_q8_K(int n, float * restrict s, const void * restrict vx, const void * restrict vy); void ggml_vec_dot_q6_K_q8_K(int n, float * restrict s, const void * restrict vx, const void * restrict vy); +void ggml_vec_dot_iq2_xxs_q8_K(int n, float * restrict s, const void * restrict vx, const void * restrict vy); diff --git a/ggml.c b/ggml.c index 62f0f18ef3b70..adb387100780e 100644 --- a/ggml.c +++ b/ggml.c @@ -573,6 +573,17 @@ static const ggml_type_traits_t type_traits[GGML_TYPE_COUNT] = { .vec_dot = ggml_vec_dot_q6_K_q8_K, .vec_dot_type = GGML_TYPE_Q8_K, }, + [GGML_TYPE_IQ2_XXS] = { + .type_name = "iq2_xxs", + .blck_size = QK_K, + .type_size = sizeof(block_iq2_xxs), + .is_quantized = true, + .to_float = (ggml_to_float_t) dequantize_row_iq2_xxs, + .from_float = quantize_row_iq2_xxs, + .from_float_reference = (ggml_from_float_t) quantize_row_iq2_xxs_reference, + .vec_dot = ggml_vec_dot_iq2_xxs_q8_K, + .vec_dot_type = GGML_TYPE_Q8_K, + }, [GGML_TYPE_Q8_K] = { .type_name = "q8_K", .blck_size = QK_K, @@ -2111,6 +2122,7 @@ enum ggml_type ggml_ftype_to_ggml_type(enum ggml_ftype ftype) { case GGML_FTYPE_MOSTLY_Q4_K: wtype = GGML_TYPE_Q4_K; break; case GGML_FTYPE_MOSTLY_Q5_K: wtype = GGML_TYPE_Q5_K; break; case GGML_FTYPE_MOSTLY_Q6_K: wtype = GGML_TYPE_Q6_K; break; + case GGML_FTYPE_MOSTLY_IQ2_XXS: wtype = GGML_TYPE_IQ2_XXS; break; case GGML_FTYPE_UNKNOWN: wtype = GGML_TYPE_COUNT; break; case GGML_FTYPE_MOSTLY_Q4_1_SOME_F16: wtype = GGML_TYPE_COUNT; break; } @@ -7436,6 +7448,7 @@ static void ggml_compute_forward_add( case GGML_TYPE_Q4_K: case GGML_TYPE_Q5_K: case GGML_TYPE_Q6_K: + case GGML_TYPE_IQ2_XXS: { ggml_compute_forward_add_q_f32(params, src0, src1, dst); } break; @@ -7700,6 +7713,7 @@ static void ggml_compute_forward_add1( case GGML_TYPE_Q4_K: case GGML_TYPE_Q5_K: case GGML_TYPE_Q6_K: + case GGML_TYPE_IQ2_XXS: { ggml_compute_forward_add1_q_f32(params, src0, src1, dst); } break; @@ -7814,6 +7828,7 @@ static void ggml_compute_forward_acc( case GGML_TYPE_Q4_K: case GGML_TYPE_Q5_K: case GGML_TYPE_Q6_K: + case GGML_TYPE_IQ2_XXS: default: { GGML_ASSERT(false); @@ -10455,6 +10470,7 @@ static void ggml_compute_forward_out_prod( case GGML_TYPE_Q4_K: case GGML_TYPE_Q5_K: case GGML_TYPE_Q6_K: + case GGML_TYPE_IQ2_XXS: { ggml_compute_forward_out_prod_q_f32(params, src0, src1, dst); } break; @@ -10629,6 +10645,7 @@ static void ggml_compute_forward_set( case GGML_TYPE_Q4_K: case GGML_TYPE_Q5_K: case GGML_TYPE_Q6_K: + case GGML_TYPE_IQ2_XXS: default: { GGML_ASSERT(false); @@ -10823,6 +10840,7 @@ static void ggml_compute_forward_get_rows( case GGML_TYPE_Q4_K: case GGML_TYPE_Q5_K: case GGML_TYPE_Q6_K: + case GGML_TYPE_IQ2_XXS: { ggml_compute_forward_get_rows_q(params, src0, src1, dst); } break; @@ -11459,6 +11477,7 @@ static void ggml_compute_forward_alibi( case GGML_TYPE_Q4_K: case GGML_TYPE_Q5_K: case GGML_TYPE_Q6_K: + case GGML_TYPE_IQ2_XXS: case GGML_TYPE_Q8_K: case GGML_TYPE_I8: case GGML_TYPE_I16: @@ -11533,6 +11552,7 @@ static void ggml_compute_forward_clamp( case GGML_TYPE_Q4_K: case GGML_TYPE_Q5_K: case GGML_TYPE_Q6_K: + case GGML_TYPE_IQ2_XXS: case GGML_TYPE_Q8_K: case GGML_TYPE_I8: case GGML_TYPE_I16: @@ -18648,6 +18668,12 @@ size_t ggml_quantize_chunk(enum ggml_type type, const float * src, void * dst, i block_q6_K * block = (block_q6_K*)dst + start / QK_K; result = ggml_quantize_q6_K(src + start, block, n, n, hist); } break; + case GGML_TYPE_IQ2_XXS: + { + GGML_ASSERT(start % QK_K == 0); + block_iq2_xxs * block = (block_iq2_xxs*)dst + start / QK_K; + result = ggml_quantize_iq2_xxs(src + start, block, n, n, hist); + } break; case GGML_TYPE_F16: { int elemsize = sizeof(ggml_fp16_t); diff --git a/ggml.h b/ggml.h index 64f4e45e880fa..c55e598b4fea3 100644 --- a/ggml.h +++ b/ggml.h @@ -339,6 +339,7 @@ extern "C" { GGML_TYPE_Q5_K = 13, GGML_TYPE_Q6_K = 14, GGML_TYPE_Q8_K = 15, + GGML_TYPE_IQ2_XXS = 16, GGML_TYPE_I8, GGML_TYPE_I16, GGML_TYPE_I32, @@ -373,6 +374,7 @@ extern "C" { GGML_FTYPE_MOSTLY_Q4_K = 12, // except 1d tensors GGML_FTYPE_MOSTLY_Q5_K = 13, // except 1d tensors GGML_FTYPE_MOSTLY_Q6_K = 14, // except 1d tensors + GGML_FTYPE_MOSTLY_IQ2_XXS = 15, // except 1d tensors }; // available tensor operations: @@ -2067,6 +2069,7 @@ extern "C" { GGML_API size_t ggml_quantize_q4_K(const float * src, void * dst, int n, int k, int64_t * hist); GGML_API size_t ggml_quantize_q5_K(const float * src, void * dst, int n, int k, int64_t * hist); GGML_API size_t ggml_quantize_q6_K(const float * src, void * dst, int n, int k, int64_t * hist); + GGML_API size_t ggml_quantize_iq2_xxs(const float * src, void * dst, int n, int k, int64_t * hist); GGML_API size_t ggml_quantize_chunk(enum ggml_type type, const float * src, void * dst, int start, int n, int64_t * hist); From 7ef6389694d188e2ec4ef103b69bc07359d718ad Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Wed, 3 Jan 2024 15:46:59 +0200 Subject: [PATCH 02/17] iq2_xxs: scalar and AVX2 dot products Needed to change Q8_K to have quants in the -127...127 range, else the IQ2_XXS AVX implementation becomes very awkward. The alternative would have been to use Q8_0 instead. Perhaps I'll change later, for now this is what we have. --- ggml-quants.c | 84 ++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 83 insertions(+), 1 deletion(-) diff --git a/ggml-quants.c b/ggml-quants.c index 4a522e0145b8c..6e66ed84131c5 100644 --- a/ggml-quants.c +++ b/ggml-quants.c @@ -2494,7 +2494,9 @@ void quantize_row_q8_K_reference(const float * restrict x, block_q8_K * restrict x += QK_K; continue; } - const float iscale = -128.f/max; + //const float iscale = -128.f/max; + // We need this change for IQ2_XXS, else the AVX implementation becomes very awkward + const float iscale = -127.f/max; for (int j = 0; j < QK_K; ++j) { int v = nearest_int(iscale*x[j]); y[i].qs[j] = MIN(127, v); @@ -7198,6 +7200,41 @@ void ggml_vec_dot_q6_K_q8_K(const int n, float * restrict s, const void * restri #endif +static const int8_t keven_signs_q2xs[1024] = { + 1, 1, 1, 1, 1, 1, 1, 1, -1, 1, 1, 1, 1, 1, 1, -1, 1, -1, 1, 1, 1, 1, 1, -1, -1, -1, 1, 1, 1, 1, 1, 1, + 1, 1, -1, 1, 1, 1, 1, -1, -1, 1, -1, 1, 1, 1, 1, 1, 1, -1, -1, 1, 1, 1, 1, 1, -1, -1, -1, 1, 1, 1, 1, -1, + 1, 1, 1, -1, 1, 1, 1, -1, -1, 1, 1, -1, 1, 1, 1, 1, 1, -1, 1, -1, 1, 1, 1, 1, -1, -1, 1, -1, 1, 1, 1, -1, + 1, 1, -1, -1, 1, 1, 1, 1, -1, 1, -1, -1, 1, 1, 1, -1, 1, -1, -1, -1, 1, 1, 1, -1, -1, -1, -1, -1, 1, 1, 1, 1, + 1, 1, 1, 1, -1, 1, 1, -1, -1, 1, 1, 1, -1, 1, 1, 1, 1, -1, 1, 1, -1, 1, 1, 1, -1, -1, 1, 1, -1, 1, 1, -1, + 1, 1, -1, 1, -1, 1, 1, 1, -1, 1, -1, 1, -1, 1, 1, -1, 1, -1, -1, 1, -1, 1, 1, -1, -1, -1, -1, 1, -1, 1, 1, 1, + 1, 1, 1, -1, -1, 1, 1, 1, -1, 1, 1, -1, -1, 1, 1, -1, 1, -1, 1, -1, -1, 1, 1, -1, -1, -1, 1, -1, -1, 1, 1, 1, + 1, 1, -1, -1, -1, 1, 1, -1, -1, 1, -1, -1, -1, 1, 1, 1, 1, -1, -1, -1, -1, 1, 1, 1, -1, -1, -1, -1, -1, 1, 1, -1, + 1, 1, 1, 1, 1, -1, 1, -1, -1, 1, 1, 1, 1, -1, 1, 1, 1, -1, 1, 1, 1, -1, 1, 1, -1, -1, 1, 1, 1, -1, 1, -1, + 1, 1, -1, 1, 1, -1, 1, 1, -1, 1, -1, 1, 1, -1, 1, -1, 1, -1, -1, 1, 1, -1, 1, -1, -1, -1, -1, 1, 1, -1, 1, 1, + 1, 1, 1, -1, 1, -1, 1, 1, -1, 1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, -1, -1, 1, -1, 1, -1, 1, 1, + 1, 1, -1, -1, 1, -1, 1, -1, -1, 1, -1, -1, 1, -1, 1, 1, 1, -1, -1, -1, 1, -1, 1, 1, -1, -1, -1, -1, 1, -1, 1, -1, + 1, 1, 1, 1, -1, -1, 1, 1, -1, 1, 1, 1, -1, -1, 1, -1, 1, -1, 1, 1, -1, -1, 1, -1, -1, -1, 1, 1, -1, -1, 1, 1, + 1, 1, -1, 1, -1, -1, 1, -1, -1, 1, -1, 1, -1, -1, 1, 1, 1, -1, -1, 1, -1, -1, 1, 1, -1, -1, -1, 1, -1, -1, 1, -1, + 1, 1, 1, -1, -1, -1, 1, -1, -1, 1, 1, -1, -1, -1, 1, 1, 1, -1, 1, -1, -1, -1, 1, 1, -1, -1, 1, -1, -1, -1, 1, -1, + 1, 1, -1, -1, -1, -1, 1, 1, -1, 1, -1, -1, -1, -1, 1, -1, 1, -1, -1, -1, -1, -1, 1, -1, -1, -1, -1, -1, -1, -1, 1, 1, + 1, 1, 1, 1, 1, 1, -1, -1, -1, 1, 1, 1, 1, 1, -1, 1, 1, -1, 1, 1, 1, 1, -1, 1, -1, -1, 1, 1, 1, 1, -1, -1, + 1, 1, -1, 1, 1, 1, -1, 1, -1, 1, -1, 1, 1, 1, -1, -1, 1, -1, -1, 1, 1, 1, -1, -1, -1, -1, -1, 1, 1, 1, -1, 1, + 1, 1, 1, -1, 1, 1, -1, 1, -1, 1, 1, -1, 1, 1, -1, -1, 1, -1, 1, -1, 1, 1, -1, -1, -1, -1, 1, -1, 1, 1, -1, 1, + 1, 1, -1, -1, 1, 1, -1, -1, -1, 1, -1, -1, 1, 1, -1, 1, 1, -1, -1, -1, 1, 1, -1, 1, -1, -1, -1, -1, 1, 1, -1, -1, + 1, 1, 1, 1, -1, 1, -1, 1, -1, 1, 1, 1, -1, 1, -1, -1, 1, -1, 1, 1, -1, 1, -1, -1, -1, -1, 1, 1, -1, 1, -1, 1, + 1, 1, -1, 1, -1, 1, -1, -1, -1, 1, -1, 1, -1, 1, -1, 1, 1, -1, -1, 1, -1, 1, -1, 1, -1, -1, -1, 1, -1, 1, -1, -1, + 1, 1, 1, -1, -1, 1, -1, -1, -1, 1, 1, -1, -1, 1, -1, 1, 1, -1, 1, -1, -1, 1, -1, 1, -1, -1, 1, -1, -1, 1, -1, -1, + 1, 1, -1, -1, -1, 1, -1, 1, -1, 1, -1, -1, -1, 1, -1, -1, 1, -1, -1, -1, -1, 1, -1, -1, -1, -1, -1, -1, -1, 1, -1, 1, + 1, 1, 1, 1, 1, -1, -1, 1, -1, 1, 1, 1, 1, -1, -1, -1, 1, -1, 1, 1, 1, -1, -1, -1, -1, -1, 1, 1, 1, -1, -1, 1, + 1, 1, -1, 1, 1, -1, -1, -1, -1, 1, -1, 1, 1, -1, -1, 1, 1, -1, -1, 1, 1, -1, -1, 1, -1, -1, -1, 1, 1, -1, -1, -1, + 1, 1, 1, -1, 1, -1, -1, -1, -1, 1, 1, -1, 1, -1, -1, 1, 1, -1, 1, -1, 1, -1, -1, 1, -1, -1, 1, -1, 1, -1, -1, -1, + 1, 1, -1, -1, 1, -1, -1, 1, -1, 1, -1, -1, 1, -1, -1, -1, 1, -1, -1, -1, 1, -1, -1, -1, -1, -1, -1, -1, 1, -1, -1, 1, + 1, 1, 1, 1, -1, -1, -1, -1, -1, 1, 1, 1, -1, -1, -1, 1, 1, -1, 1, 1, -1, -1, -1, 1, -1, -1, 1, 1, -1, -1, -1, -1, + 1, 1, -1, 1, -1, -1, -1, 1, -1, 1, -1, 1, -1, -1, -1, -1, 1, -1, -1, 1, -1, -1, -1, -1, -1, -1, -1, 1, -1, -1, -1, 1, + 1, 1, 1, -1, -1, -1, -1, 1, -1, 1, 1, -1, -1, -1, -1, -1, 1, -1, 1, -1, -1, -1, -1, -1, -1, -1, 1, -1, -1, -1, -1, 1, + 1, 1, -1, -1, -1, -1, -1, -1, -1, 1, -1, -1, -1, -1, -1, 1, 1, -1, -1, -1, -1, -1, -1, 1, -1, -1, -1, -1, -1, -1, -1, -1, +}; + void ggml_vec_dot_iq2_xxs_q8_K(const int n, float * restrict s, const void * restrict vx, const void * restrict vy) { assert(n % QK_K == 0); @@ -7206,6 +7243,50 @@ void ggml_vec_dot_iq2_xxs_q8_K(const int n, float * restrict s, const void * res const int nb = n / QK_K; +#if defined __AVX2__ + + const uint64_t * signs64 = (const uint64_t *)keven_signs_q2xs; + + uint32_t aux32[4]; + const uint8_t * aux8 = (const uint8_t *)aux32; + + __m256 accumf = _mm256_setzero_ps(); + for (int i = 0; i < nb; ++i) { + const float d = GGML_FP16_TO_FP32(x[i].d) * y[i].d; + const uint16_t * restrict q2 = x[i].qs; + const int8_t * restrict q8 = y[i].qs; + __m256i sumi1 = _mm256_setzero_si256(); + __m256i sumi2 = _mm256_setzero_si256(); + for (int ib32 = 0; ib32 < QK_K/32; ib32 += 2) { + const __m256i q8_1 = _mm256_loadu_si256((const __m256i *)q8); q8 += 32; + const __m256i q8_2 = _mm256_loadu_si256((const __m256i *)q8); q8 += 32; + memcpy(aux32, q2, 4*sizeof(uint32_t)); q2 += 8; + const __m256i q2_1 = _mm256_set_epi64x(iq2xxs_grid[aux8[ 3]], iq2xxs_grid[aux8[ 2]], iq2xxs_grid[aux8[1]], iq2xxs_grid[aux8[0]]); + const __m256i q2_2 = _mm256_set_epi64x(iq2xxs_grid[aux8[11]], iq2xxs_grid[aux8[10]], iq2xxs_grid[aux8[9]], iq2xxs_grid[aux8[8]]); + const __m256i s2_1 = _mm256_set_epi64x(signs64[(aux32[1] >> 21) & 127], signs64[(aux32[1] >> 14) & 127], + signs64[(aux32[1] >> 7) & 127], signs64[(aux32[1] >> 0) & 127]); + const __m256i s2_2 = _mm256_set_epi64x(signs64[(aux32[3] >> 21) & 127], signs64[(aux32[3] >> 14) & 127], + signs64[(aux32[3] >> 7) & 127], signs64[(aux32[3] >> 0) & 127]); + const __m256i q8s_1 = _mm256_sign_epi8(q8_1, s2_1); + const __m256i q8s_2 = _mm256_sign_epi8(q8_2, s2_2); + const __m256i dot1 = _mm256_maddubs_epi16(q2_1, q8s_1); + const __m256i dot2 = _mm256_maddubs_epi16(q2_2, q8s_2); + const uint16_t ls1 = aux32[1] >> 28; + const uint16_t ls2 = aux32[3] >> 28; + const __m256i p1 = _mm256_madd_epi16(dot1, _mm256_set1_epi16(2*ls1+1)); + const __m256i p2 = _mm256_madd_epi16(dot2, _mm256_set1_epi16(2*ls2+1)); + sumi1 = _mm256_add_epi32(sumi1, p1); + sumi2 = _mm256_add_epi32(sumi2, p2); + } + + accumf = _mm256_fmadd_ps(_mm256_set1_ps(d), _mm256_cvtepi32_ps(_mm256_add_epi32(sumi1, sumi2)), accumf); + + } + + *s = 0.125f * hsum_float_8(accumf); + +#else + uint32_t aux32[2]; const uint8_t * aux8 = (const uint8_t *)aux32; @@ -7233,4 +7314,5 @@ void ggml_vec_dot_iq2_xxs_q8_K(const int n, float * restrict s, const void * res sumf += d * bsum; } *s = 0.125f * sumf; +#endif } From 7b72318e6f1733b03c82f7a7c33227b6b40852db Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Wed, 3 Jan 2024 15:35:24 +0100 Subject: [PATCH 03/17] iq2_xxs: ARM_NEON dot product Somehow strangely slow (112 ms/token). --- ggml-quants.c | 43 ++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 42 insertions(+), 1 deletion(-) diff --git a/ggml-quants.c b/ggml-quants.c index 6e66ed84131c5..24e39c73f75ed 100644 --- a/ggml-quants.c +++ b/ggml-quants.c @@ -7243,7 +7243,48 @@ void ggml_vec_dot_iq2_xxs_q8_K(const int n, float * restrict s, const void * res const int nb = n / QK_K; -#if defined __AVX2__ +#if defined(__ARM_NEON) + + const uint64_t * signs64 = (const uint64_t *)keven_signs_q2xs; + + uint32_t aux32[4]; + const uint8_t * aux8 = (const uint8_t *)aux32; + + int8x16x4_t q2u; + int8x16x4_t q2s; + int8x16x4_t q8b; + + float sumf = 0; + for (int i = 0; i < nb; ++i) { + const float d = GGML_FP16_TO_FP32(x[i].d) * y[i].d; + const uint16_t * restrict q2 = x[i].qs; + const int8_t * restrict q8 = y[i].qs; + float sumf1 = 0, sumf2 = 0; + for (int ib32 = 0; ib32 < QK_K/32; ib32 += 2) { + q8b = vld1q_s8_x4(q8); q8 += 64; + memcpy(aux32, q2, 4*sizeof(uint32_t)); q2 += 8; + q2u.val[0] = vcombine_u8(vld1_u8((const void *)(iq2xxs_grid + aux8[ 0])), vld1_u8((const void *)(iq2xxs_grid + aux8[ 1]))); + q2u.val[1] = vcombine_u8(vld1_u8((const void *)(iq2xxs_grid + aux8[ 2])), vld1_u8((const void *)(iq2xxs_grid + aux8[ 3]))); + q2u.val[2] = vcombine_u8(vld1_u8((const void *)(iq2xxs_grid + aux8[ 8])), vld1_u8((const void *)(iq2xxs_grid + aux8[ 9]))); + q2u.val[3] = vcombine_u8(vld1_u8((const void *)(iq2xxs_grid + aux8[10])), vld1_u8((const void *)(iq2xxs_grid + aux8[11]))); + q2s.val[0] = vcombine_u8(vld1_u8((const void *)(signs64 + ((aux32[1] >> 0) & 127))), vld1_u8((const void *)(signs64 + ((aux32[1] >> 7) & 127)))); + q2s.val[1] = vcombine_u8(vld1_u8((const void *)(signs64 + ((aux32[1] >> 14) & 127))), vld1_u8((const void *)(signs64 + ((aux32[1] >> 21) & 127)))); + q2s.val[2] = vcombine_u8(vld1_u8((const void *)(signs64 + ((aux32[3] >> 0) & 127))), vld1_u8((const void *)(signs64 + ((aux32[3] >> 7) & 127)))); + q2s.val[3] = vcombine_u8(vld1_u8((const void *)(signs64 + ((aux32[3] >> 14) & 127))), vld1_u8((const void *)(signs64 + ((aux32[3] >> 21) & 127)))); + q2u.val[0] = vmulq_s8(q2u.val[0], q2s.val[0]); + q2u.val[1] = vmulq_s8(q2u.val[1], q2s.val[1]); + q2u.val[2] = vmulq_s8(q2u.val[2], q2s.val[2]); + q2u.val[3] = vmulq_s8(q2u.val[3], q2s.val[3]); + const int32x4_t p1 = ggml_vdotq_s32(ggml_vdotq_s32(vdupq_n_s32(0), q2u.val[0], q8b.val[0]), q2u.val[1], q8b.val[1]); + const int32x4_t p2 = ggml_vdotq_s32(ggml_vdotq_s32(vdupq_n_s32(0), q2u.val[2], q8b.val[2]), q2u.val[3], q8b.val[3]); + sumf1 += vaddvq_s32(p1) * (0.5f + (aux32[1] >> 28)); + sumf2 += vaddvq_s32(p2) * (0.5f + (aux32[3] >> 28)); + } + sumf += d*(sumf1 + sumf2); + } + *s = 0.25f * sumf; + +#elif defined(__AVX2__) const uint64_t * signs64 = (const uint64_t *)keven_signs_q2xs; From d383f00eeaae5ca1dcadf398ad89b7cd22b085dd Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Wed, 3 Jan 2024 18:25:02 +0100 Subject: [PATCH 04/17] iq2_xxs: WIP Metal Dequantize works, something is still wrong with the dot product. --- ggml-metal.m | 32 +++++ ggml-metal.metal | 301 +++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 333 insertions(+) diff --git a/ggml-metal.m b/ggml-metal.m index fbbdcd8c46726..e422dbb4e484a 100644 --- a/ggml-metal.m +++ b/ggml-metal.m @@ -88,6 +88,7 @@ GGML_METAL_DECL_KERNEL(get_rows_q5_K); GGML_METAL_DECL_KERNEL(get_rows_q6_K); GGML_METAL_DECL_KERNEL(get_rows_i32); + GGML_METAL_DECL_KERNEL(get_rows_iq2_xxs); GGML_METAL_DECL_KERNEL(rms_norm); GGML_METAL_DECL_KERNEL(group_norm); GGML_METAL_DECL_KERNEL(norm); @@ -106,6 +107,7 @@ GGML_METAL_DECL_KERNEL(mul_mv_q4_K_f32); GGML_METAL_DECL_KERNEL(mul_mv_q5_K_f32); GGML_METAL_DECL_KERNEL(mul_mv_q6_K_f32); + GGML_METAL_DECL_KERNEL(mul_mv_iq2_xxs_f32); GGML_METAL_DECL_KERNEL(mul_mv_id_f32_f32); //GGML_METAL_DECL_KERNEL(mul_mv_id_f16_f16); GGML_METAL_DECL_KERNEL(mul_mv_id_f16_f32); @@ -121,6 +123,7 @@ GGML_METAL_DECL_KERNEL(mul_mv_id_q4_K_f32); GGML_METAL_DECL_KERNEL(mul_mv_id_q5_K_f32); GGML_METAL_DECL_KERNEL(mul_mv_id_q6_K_f32); + GGML_METAL_DECL_KERNEL(mul_mv_id_iq2_xxs_f32); GGML_METAL_DECL_KERNEL(mul_mm_f32_f32); GGML_METAL_DECL_KERNEL(mul_mm_f16_f32); GGML_METAL_DECL_KERNEL(mul_mm_q4_0_f32); @@ -133,6 +136,7 @@ GGML_METAL_DECL_KERNEL(mul_mm_q4_K_f32); GGML_METAL_DECL_KERNEL(mul_mm_q5_K_f32); GGML_METAL_DECL_KERNEL(mul_mm_q6_K_f32); + GGML_METAL_DECL_KERNEL(mul_mm_iq2_xxs_f32); GGML_METAL_DECL_KERNEL(mul_mm_id_f32_f32); GGML_METAL_DECL_KERNEL(mul_mm_id_f16_f32); GGML_METAL_DECL_KERNEL(mul_mm_id_q4_0_f32); @@ -145,6 +149,7 @@ GGML_METAL_DECL_KERNEL(mul_mm_id_q4_K_f32); GGML_METAL_DECL_KERNEL(mul_mm_id_q5_K_f32); GGML_METAL_DECL_KERNEL(mul_mm_id_q6_K_f32); + GGML_METAL_DECL_KERNEL(mul_mm_id_iq2_xxs_f32); GGML_METAL_DECL_KERNEL(rope_f32); GGML_METAL_DECL_KERNEL(rope_f16); GGML_METAL_DECL_KERNEL(alibi_f32); @@ -379,6 +384,7 @@ static void ggml_metal_log(enum ggml_log_level level, const char * format, ...){ GGML_METAL_ADD_KERNEL(get_rows_q5_K); GGML_METAL_ADD_KERNEL(get_rows_q6_K); GGML_METAL_ADD_KERNEL(get_rows_i32); + GGML_METAL_ADD_KERNEL(get_rows_iq2_xxs); GGML_METAL_ADD_KERNEL(rms_norm); GGML_METAL_ADD_KERNEL(group_norm); GGML_METAL_ADD_KERNEL(norm); @@ -397,6 +403,7 @@ static void ggml_metal_log(enum ggml_log_level level, const char * format, ...){ GGML_METAL_ADD_KERNEL(mul_mv_q4_K_f32); GGML_METAL_ADD_KERNEL(mul_mv_q5_K_f32); GGML_METAL_ADD_KERNEL(mul_mv_q6_K_f32); + GGML_METAL_ADD_KERNEL(mul_mv_iq2_xxs_f32); GGML_METAL_ADD_KERNEL(mul_mv_id_f32_f32); //GGML_METAL_ADD_KERNEL(mul_mv_id_f16_f16); GGML_METAL_ADD_KERNEL(mul_mv_id_f16_f32); @@ -412,6 +419,7 @@ static void ggml_metal_log(enum ggml_log_level level, const char * format, ...){ GGML_METAL_ADD_KERNEL(mul_mv_id_q4_K_f32); GGML_METAL_ADD_KERNEL(mul_mv_id_q5_K_f32); GGML_METAL_ADD_KERNEL(mul_mv_id_q6_K_f32); + GGML_METAL_ADD_KERNEL(mul_mv_id_iq2_xxs_f32); if ([ctx->device supportsFamily:MTLGPUFamilyApple7]) { GGML_METAL_ADD_KERNEL(mul_mm_f32_f32); GGML_METAL_ADD_KERNEL(mul_mm_f16_f32); @@ -425,6 +433,7 @@ static void ggml_metal_log(enum ggml_log_level level, const char * format, ...){ GGML_METAL_ADD_KERNEL(mul_mm_q4_K_f32); GGML_METAL_ADD_KERNEL(mul_mm_q5_K_f32); GGML_METAL_ADD_KERNEL(mul_mm_q6_K_f32); + GGML_METAL_ADD_KERNEL(mul_mm_iq2_xxs_f32); GGML_METAL_ADD_KERNEL(mul_mm_id_f32_f32); GGML_METAL_ADD_KERNEL(mul_mm_id_f16_f32); GGML_METAL_ADD_KERNEL(mul_mm_id_q4_0_f32); @@ -437,6 +446,7 @@ static void ggml_metal_log(enum ggml_log_level level, const char * format, ...){ GGML_METAL_ADD_KERNEL(mul_mm_id_q4_K_f32); GGML_METAL_ADD_KERNEL(mul_mm_id_q5_K_f32); GGML_METAL_ADD_KERNEL(mul_mm_id_q6_K_f32); + GGML_METAL_ADD_KERNEL(mul_mm_id_iq2_xxs_f32); } GGML_METAL_ADD_KERNEL(rope_f32); GGML_METAL_ADD_KERNEL(rope_f16); @@ -502,6 +512,7 @@ void ggml_metal_free(struct ggml_metal_context * ctx) { GGML_METAL_DEL_KERNEL(get_rows_q5_K); GGML_METAL_DEL_KERNEL(get_rows_q6_K); GGML_METAL_DEL_KERNEL(get_rows_i32); + GGML_METAL_DEL_KERNEL(get_rows_iq2_xxs); GGML_METAL_DEL_KERNEL(rms_norm); GGML_METAL_DEL_KERNEL(group_norm); GGML_METAL_DEL_KERNEL(norm); @@ -520,6 +531,7 @@ void ggml_metal_free(struct ggml_metal_context * ctx) { GGML_METAL_DEL_KERNEL(mul_mv_q4_K_f32); GGML_METAL_DEL_KERNEL(mul_mv_q5_K_f32); GGML_METAL_DEL_KERNEL(mul_mv_q6_K_f32); + GGML_METAL_DEL_KERNEL(mul_mv_iq2_xxs_f32); GGML_METAL_DEL_KERNEL(mul_mv_id_f32_f32); //GGML_METAL_DEL_KERNEL(mul_mv_id_f16_f16); GGML_METAL_DEL_KERNEL(mul_mv_id_f16_f32); @@ -535,6 +547,7 @@ void ggml_metal_free(struct ggml_metal_context * ctx) { GGML_METAL_DEL_KERNEL(mul_mv_id_q4_K_f32); GGML_METAL_DEL_KERNEL(mul_mv_id_q5_K_f32); GGML_METAL_DEL_KERNEL(mul_mv_id_q6_K_f32); + GGML_METAL_DEL_KERNEL(mul_mv_id_iq2_xxs_f32); if ([ctx->device supportsFamily:MTLGPUFamilyApple7]) { GGML_METAL_DEL_KERNEL(mul_mm_f32_f32); GGML_METAL_DEL_KERNEL(mul_mm_f16_f32); @@ -548,6 +561,7 @@ void ggml_metal_free(struct ggml_metal_context * ctx) { GGML_METAL_DEL_KERNEL(mul_mm_q4_K_f32); GGML_METAL_DEL_KERNEL(mul_mm_q5_K_f32); GGML_METAL_DEL_KERNEL(mul_mm_q6_K_f32); + GGML_METAL_DEL_KERNEL(mul_mm_iq2_xxs_f32); GGML_METAL_DEL_KERNEL(mul_mm_id_f32_f32); GGML_METAL_DEL_KERNEL(mul_mm_id_f16_f32); GGML_METAL_DEL_KERNEL(mul_mm_id_q4_0_f32); @@ -560,6 +574,7 @@ void ggml_metal_free(struct ggml_metal_context * ctx) { GGML_METAL_DEL_KERNEL(mul_mm_id_q4_K_f32); GGML_METAL_DEL_KERNEL(mul_mm_id_q5_K_f32); GGML_METAL_DEL_KERNEL(mul_mm_id_q6_K_f32); + GGML_METAL_DEL_KERNEL(mul_mm_id_iq2_xxs_f32); } GGML_METAL_DEL_KERNEL(rope_f32); GGML_METAL_DEL_KERNEL(rope_f16); @@ -1541,6 +1556,7 @@ bool ggml_metal_graph_compute( case GGML_TYPE_Q4_K: [encoder setComputePipelineState:ctx->pipeline_mul_mm_q4_K_f32]; break; case GGML_TYPE_Q5_K: [encoder setComputePipelineState:ctx->pipeline_mul_mm_q5_K_f32]; break; case GGML_TYPE_Q6_K: [encoder setComputePipelineState:ctx->pipeline_mul_mm_q6_K_f32]; break; + case GGML_TYPE_IQ2_XXS: [encoder setComputePipelineState:ctx->pipeline_mul_mm_iq2_xxs_f32]; break; default: GGML_ASSERT(false && "MUL MAT-MAT not implemented"); } [encoder setBuffer:id_src0 offset:offs_src0 atIndex:0]; @@ -1653,6 +1669,12 @@ bool ggml_metal_graph_compute( nth1 = 32; [encoder setComputePipelineState:ctx->pipeline_mul_mv_q6_K_f32]; } break; + case GGML_TYPE_IQ2_XXS: + { + nth0 = 4; + nth1 = 16; + [encoder setComputePipelineState:ctx->pipeline_mul_mv_iq2_xxs_f32]; + } break; default: { GGML_METAL_LOG_ERROR("Asserting on type %d\n", (int)src0t); @@ -1686,6 +1708,7 @@ bool ggml_metal_graph_compute( if (src0t == GGML_TYPE_Q4_0 || src0t == GGML_TYPE_Q4_1 || src0t == GGML_TYPE_Q5_0 || src0t == GGML_TYPE_Q5_1 || src0t == GGML_TYPE_Q8_0 || + src0t == GGML_TYPE_IQ2_XXS || src0t == GGML_TYPE_Q2_K) { // || src0t == GGML_TYPE_Q4_K) { [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 7)/8, ne11, ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; } @@ -1778,6 +1801,7 @@ bool ggml_metal_graph_compute( case GGML_TYPE_Q4_K: [encoder setComputePipelineState:ctx->pipeline_mul_mm_id_q4_K_f32]; break; case GGML_TYPE_Q5_K: [encoder setComputePipelineState:ctx->pipeline_mul_mm_id_q5_K_f32]; break; case GGML_TYPE_Q6_K: [encoder setComputePipelineState:ctx->pipeline_mul_mm_id_q6_K_f32]; break; + case GGML_TYPE_IQ2_XXS: [encoder setComputePipelineState:ctx->pipeline_mul_mm_id_iq2_xxs_f32]; break; default: GGML_ASSERT(false && "MUL_MAT_ID not implemented"); } [encoder setBuffer:id_src0 offset:offs_src0 atIndex:0]; @@ -1893,6 +1917,12 @@ bool ggml_metal_graph_compute( nth1 = 32; [encoder setComputePipelineState:ctx->pipeline_mul_mv_id_q6_K_f32]; } break; + case GGML_TYPE_IQ2_XXS: + { + nth0 = 4; + nth1 = 16; + [encoder setComputePipelineState:ctx->pipeline_mul_mv_id_iq2_xxs_f32]; + } break; default: { GGML_METAL_LOG_ERROR("Asserting on type %d\n", (int)src2t); @@ -1942,6 +1972,7 @@ bool ggml_metal_graph_compute( if (src2t == GGML_TYPE_Q4_0 || src2t == GGML_TYPE_Q4_1 || src2t == GGML_TYPE_Q5_0 || src2t == GGML_TYPE_Q5_1 || src2t == GGML_TYPE_Q8_0 || + src2t == GGML_TYPE_IQ2_XXS || src2t == GGML_TYPE_Q2_K) { // || src2t == GGML_TYPE_Q4_K) { [encoder dispatchThreadgroups:MTLSizeMake((ne21 + 7)/8, _ne1, ne01*ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; } @@ -1982,6 +2013,7 @@ bool ggml_metal_graph_compute( case GGML_TYPE_Q5_K: [encoder setComputePipelineState:ctx->pipeline_get_rows_q5_K]; break; case GGML_TYPE_Q6_K: [encoder setComputePipelineState:ctx->pipeline_get_rows_q6_K]; break; case GGML_TYPE_I32: [encoder setComputePipelineState:ctx->pipeline_get_rows_i32]; break; + case GGML_TYPE_IQ2_XXS: [encoder setComputePipelineState:ctx->pipeline_get_rows_iq2_xxs]; break; default: GGML_ASSERT(false && "not implemented"); } diff --git a/ggml-metal.metal b/ggml-metal.metal index a7d3f9efa5787..33fe32dcb8e38 100644 --- a/ggml-metal.metal +++ b/ggml-metal.metal @@ -2446,6 +2446,12 @@ typedef struct { } block_q6_K; // 210 bytes / block +typedef struct { + half d; + uint16_t qs[QK_K/8]; +} block_iq2_xxs; +// 66 bytes / block for QK_K = 256, so 2.0625 bpw + //====================================== dot products ========================= void kernel_mul_mv_q2_K_f32_impl( @@ -3468,6 +3474,210 @@ kernel void kernel_mul_mv_q6_K_f32( kernel_mul_mv_q6_K_f32_impl(src0, src1, dst, ne00, ne01, ne02, ne10, ne12, ne0, ne1, r2, r3, tgpig, tiisg, sgitg); } +// ======================= "True" 2-bit + +constexpr constant static uint64_t kgrid_iq2xxs[256] = { + 0x0808080808080808, 0x080808080808082b, 0x0808080808081919, 0x0808080808082b08, + 0x0808080808082b2b, 0x0808080808190819, 0x0808080808191908, 0x08080808082b0808, + 0x08080808082b082b, 0x08080808082b2b08, 0x08080808082b2b2b, 0x0808080819080819, + 0x0808080819081908, 0x0808080819190808, 0x0808080819192b08, 0x08080808192b0819, + 0x08080808192b1908, 0x080808082b080808, 0x080808082b08082b, 0x080808082b082b2b, + 0x080808082b2b082b, 0x0808081908080819, 0x0808081908081908, 0x0808081908190808, + 0x0808081908191919, 0x0808081919080808, 0x080808192b081908, 0x080808192b192b08, + 0x0808082b08080808, 0x0808082b0808082b, 0x0808082b082b082b, 0x0808082b2b08082b, + 0x0808190808080819, 0x0808190808081908, 0x0808190808190808, 0x08081908082b0819, + 0x08081908082b1908, 0x0808190819080808, 0x080819081908082b, 0x0808190819082b08, + 0x08081908192b0808, 0x080819082b080819, 0x080819082b081908, 0x080819082b190808, + 0x080819082b2b1908, 0x0808191908080808, 0x080819190808082b, 0x0808191908082b08, + 0x08081919082b0808, 0x080819191908192b, 0x08081919192b2b19, 0x080819192b080808, + 0x080819192b190819, 0x0808192b08082b19, 0x0808192b08190808, 0x0808192b19080808, + 0x0808192b2b081908, 0x0808192b2b2b1908, 0x08082b0808080808, 0x08082b0808081919, + 0x08082b0808082b08, 0x08082b0808191908, 0x08082b08082b2b08, 0x08082b0819080819, + 0x08082b0819081908, 0x08082b0819190808, 0x08082b081919082b, 0x08082b082b082b08, + 0x08082b1908081908, 0x08082b1919080808, 0x08082b2b0808082b, 0x08082b2b08191908, + 0x0819080808080819, 0x0819080808081908, 0x0819080808190808, 0x08190808082b0819, + 0x0819080819080808, 0x08190808192b0808, 0x081908082b081908, 0x081908082b190808, + 0x081908082b191919, 0x0819081908080808, 0x0819081908082b08, 0x08190819082b0808, + 0x0819081919190808, 0x0819081919192b2b, 0x081908192b080808, 0x0819082b082b1908, + 0x0819082b19081919, 0x0819190808080808, 0x0819190808082b08, 0x08191908082b0808, + 0x08191908082b1919, 0x0819190819082b19, 0x081919082b080808, 0x0819191908192b08, + 0x08191919192b082b, 0x0819192b08080808, 0x0819192b0819192b, 0x08192b0808080819, + 0x08192b0808081908, 0x08192b0808190808, 0x08192b0819080808, 0x08192b082b080819, + 0x08192b1908080808, 0x08192b1908081919, 0x08192b192b2b0808, 0x08192b2b19190819, + 0x082b080808080808, 0x082b08080808082b, 0x082b080808082b2b, 0x082b080819081908, + 0x082b0808192b0819, 0x082b08082b080808, 0x082b08082b08082b, 0x082b0819082b2b19, + 0x082b081919082b08, 0x082b082b08080808, 0x082b082b0808082b, 0x082b190808080819, + 0x082b190808081908, 0x082b190808190808, 0x082b190819080808, 0x082b19081919192b, + 0x082b191908080808, 0x082b191919080819, 0x082b1919192b1908, 0x082b192b2b190808, + 0x082b2b0808082b08, 0x082b2b08082b0808, 0x082b2b082b191908, 0x082b2b2b19081908, + 0x1908080808080819, 0x1908080808081908, 0x1908080808190808, 0x1908080808192b08, + 0x19080808082b0819, 0x19080808082b1908, 0x1908080819080808, 0x1908080819082b08, + 0x190808081919192b, 0x19080808192b0808, 0x190808082b080819, 0x190808082b081908, + 0x190808082b190808, 0x1908081908080808, 0x19080819082b0808, 0x19080819192b0819, + 0x190808192b080808, 0x190808192b081919, 0x1908082b08080819, 0x1908082b08190808, + 0x1908082b19082b08, 0x1908082b1919192b, 0x1908082b192b2b08, 0x1908190808080808, + 0x1908190808082b08, 0x19081908082b0808, 0x190819082b080808, 0x190819082b192b19, + 0x190819190819082b, 0x19081919082b1908, 0x1908192b08080808, 0x19082b0808080819, + 0x19082b0808081908, 0x19082b0808190808, 0x19082b0819080808, 0x19082b0819081919, + 0x19082b1908080808, 0x19082b1919192b08, 0x19082b19192b0819, 0x19082b192b08082b, + 0x19082b2b19081919, 0x19082b2b2b190808, 0x1919080808080808, 0x1919080808082b08, + 0x1919080808190819, 0x1919080808192b19, 0x19190808082b0808, 0x191908082b080808, + 0x191908082b082b08, 0x1919081908081908, 0x191908191908082b, 0x191908192b2b1908, + 0x1919082b2b190819, 0x191919082b190808, 0x191919082b19082b, 0x1919191908082b2b, + 0x1919192b08080819, 0x1919192b19191908, 0x19192b0808080808, 0x19192b0808190819, + 0x19192b0808192b19, 0x19192b08192b1908, 0x19192b1919080808, 0x19192b2b08082b08, + 0x192b080808081908, 0x192b080808190808, 0x192b080819080808, 0x192b0808192b2b08, + 0x192b081908080808, 0x192b081919191919, 0x192b082b08192b08, 0x192b082b192b0808, + 0x192b190808080808, 0x192b190808081919, 0x192b191908190808, 0x192b19190819082b, + 0x192b19192b081908, 0x192b2b081908082b, 0x2b08080808080808, 0x2b0808080808082b, + 0x2b08080808082b2b, 0x2b08080819080819, 0x2b0808082b08082b, 0x2b08081908081908, + 0x2b08081908192b08, 0x2b08081919080808, 0x2b08082b08190819, 0x2b08190808080819, + 0x2b08190808081908, 0x2b08190808190808, 0x2b08190808191919, 0x2b08190819080808, + 0x2b081908192b0808, 0x2b08191908080808, 0x2b0819191908192b, 0x2b0819192b191908, + 0x2b08192b08082b19, 0x2b08192b19080808, 0x2b08192b192b0808, 0x2b082b080808082b, + 0x2b082b1908081908, 0x2b082b2b08190819, 0x2b19080808081908, 0x2b19080808190808, + 0x2b190808082b1908, 0x2b19080819080808, 0x2b1908082b2b0819, 0x2b1908190819192b, + 0x2b1908192b080808, 0x2b19082b19081919, 0x2b19190808080808, 0x2b191908082b082b, + 0x2b19190819081908, 0x2b19191919190819, 0x2b192b082b080819, 0x2b192b19082b0808, + 0x2b2b08080808082b, 0x2b2b080819190808, 0x2b2b08082b081919, 0x2b2b081908082b19, + 0x2b2b082b08080808, 0x2b2b190808192b08, 0x2b2b2b0819190808, 0x2b2b2b1908081908, +}; + +constexpr constant static uint8_t ksigns_iq2xs[128] = { + 0, 129, 130, 3, 132, 5, 6, 135, 136, 9, 10, 139, 12, 141, 142, 15, + 144, 17, 18, 147, 20, 149, 150, 23, 24, 153, 154, 27, 156, 29, 30, 159, + 160, 33, 34, 163, 36, 165, 166, 39, 40, 169, 170, 43, 172, 45, 46, 175, + 48, 177, 178, 51, 180, 53, 54, 183, 184, 57, 58, 187, 60, 189, 190, 63, + 192, 65, 66, 195, 68, 197, 198, 71, 72, 201, 202, 75, 204, 77, 78, 207, + 80, 209, 210, 83, 212, 85, 86, 215, 216, 89, 90, 219, 92, 221, 222, 95, + 96, 225, 226, 99, 228, 101, 102, 231, 232, 105, 106, 235, 108, 237, 238, 111, + 240, 113, 114, 243, 116, 245, 246, 119, 120, 249, 250, 123, 252, 125, 126, 255, +}; + +constexpr constant static uint8_t kmask_iq2xs[8] = {1, 2, 4, 8, 16, 32, 64, 128}; + +void kernel_mul_mv_iq2_xxs_f32_impl( + device const void * src0, + device const float * src1, + device float * dst, + constant int64_t & ne00, + constant int64_t & ne01, + constant int64_t & ne02, + constant int64_t & ne10, + constant int64_t & ne12, + constant int64_t & ne0, + constant int64_t & ne1, + constant uint & r2, + constant uint & r3, + uint3 tgpig[[threadgroup_position_in_grid]], + uint tiisg[[thread_index_in_simdgroup]], + uint sgitg[[simdgroup_index_in_threadgroup]]) { + + const int nb = ne00/QK_K; + const int r0 = tgpig.x; + const int r1 = tgpig.y; + const int im = tgpig.z; + + const int first_row = (r0 * N_SIMDGROUP + sgitg) * N_DST; + const int ib_row = first_row * nb; + + const uint i12 = im%ne12; + const uint i13 = im/ne12; + + const uint offset0 = (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02); + + device const block_iq2_xxs * x = (device const block_iq2_xxs *) src0 + ib_row + offset0; + device const float * y = (device const float *) src1 + r1*ne10 + im*ne00*ne1; + + float yl[32]; + float sumf[N_DST]={0.f}, all_sum; + + const int step = sizeof(block_q2_K) * nb; + + const int nb32 = nb * (QK_K / 32); + +#if QK_K == 256 + const int ix = tiisg; + + device const float * y4 = y + 32 * ix; + + uint32_t aux32[2]; + thread const uint8_t * aux8 = (thread const uint8_t *)aux32; + thread uint16_t * aux16 = (thread uint16_t *)aux32; + + for (int ib32 = ix; ib32 < nb32; ib32 += 32) { + + for (int i = 0; i < 32; ++i) { + yl[i] = y4[i]; + } + + const int ibl = ib32 / (QK_K / 32); + const int ib = ib32 % (QK_K / 32); + + device const block_iq2_xxs * xr = x + ibl; + + for (int row = 0; row < N_DST; row++) { + + const float db = xr->d; + device const uint16_t * q2 = xr->qs + 4 * ib; + for (int i = 0; i < 4; ++i) aux16[i] = q2[i]; + const float d = db * (0.5f + (aux32[1] >> 28)); + + float sum = 0; + for (int l = 0; l < 4; ++l) { + constant uint8_t * grid = (constant uint8_t *)(kgrid_iq2xxs + aux8[l]); + const uint8_t signs = ksigns_iq2xs[(aux32[1] >> 7*l) & 127]; + for (int j = 0; j < 8; ++j) { + sum += yl[8*l + j] * grid[j] * (signs & kmask_iq2xs[j] ? -1.f : 1.f); + } + } + sumf[row] += d * sum; + + xr += nb; + } + + y4 += 32 * 32; + } +#else + // TODO +#endif + + for (int row = 0; row < N_DST; ++row) { + all_sum = simd_sum(sumf[row]); + if (tiisg == 0) { + dst[r1*ne0 + im*ne0*ne1 + first_row + row] = all_sum * 0.25f; + } + } +} + +[[host_name("kernel_mul_mv_iq2_xxs_f32")]] +kernel void kernel_mul_mv_iq2_xxs_f32( + device const void * src0, + device const float * src1, + device float * dst, + constant int64_t & ne00, + constant int64_t & ne01, + constant int64_t & ne02, + constant uint64_t & nb00, + constant uint64_t & nb01, + constant uint64_t & nb02, + constant int64_t & ne10, + constant int64_t & ne11, + constant int64_t & ne12, + constant uint64_t & nb10, + constant uint64_t & nb11, + constant uint64_t & nb12, + constant int64_t & ne0, + constant int64_t & ne1, + constant uint & r2, + constant uint & r3, + uint3 tgpig[[threadgroup_position_in_grid]], + uint tiisg[[thread_index_in_simdgroup]], + uint sgitg[[simdgroup_index_in_threadgroup]]) { + + kernel_mul_mv_iq2_xxs_f32_impl(src0, src1, dst, ne00, ne01, ne02, ne10, ne12, ne0, ne1, r2, r3, tgpig, tiisg, sgitg); +} + //============================= templates and their specializations ============================= // NOTE: this is not dequantizing - we are simply fitting the template @@ -3739,6 +3949,31 @@ void dequantize_q6_K(device const block_q6_K *xb, short il, thread type4x4 & reg } } +template +void dequantize_iq2_xxs(device const block_iq2_xxs * xb, short il, thread type4x4 & reg) { + // il is 0...15 for QK_K = 256 => index of block of 32 is il/2 + const float d = xb->d; + const int ib32 = il/2; + il = il%2; + // il = 0 or 1. il = 0 processes the first 16 quants in a block of 32, il = 1 the second 16 + // each block of 32 needs 2 uint32_t's for the quants & scale, so 4 uint16_t's. + device const uint16_t * q2 = xb->qs + 4*ib32; + const uint32_t aux32_g = q2[0] | (q2[1] << 16); + const uint32_t aux32_s = q2[2] | (q2[3] << 16); + thread const uint8_t * aux8 = (thread const uint8_t *)&aux32_g; + const float dl = d * (0.5f + (aux32_s >> 28)) * 0.25f; + constant uint8_t * grid = (constant uint8_t *)(kgrid_iq2xxs + aux8[2*il+0]); + uint8_t signs = ksigns_iq2xs[(aux32_s >> 14*il) & 127]; + for (int i = 0; i < 8; ++i) { + reg[i/4][i%4] = dl * grid[i] * (signs & kmask_iq2xs[i] ? -1.f : 1.f); + } + grid = (constant uint8_t *)(kgrid_iq2xxs + aux8[2*il+1]); + signs = ksigns_iq2xs[(aux32_s >> (14*il+7)) & 127]; + for (int i = 0; i < 8; ++i) { + reg[2+i/4][i%4] = dl * grid[i] * (signs & kmask_iq2xs[i] ? -1.f : 1.f); + } +} + template kernel void kernel_get_rows( device const void * src0, @@ -4278,6 +4513,7 @@ template [[host_name("kernel_get_rows_q3_K")]] kernel get_rows_t kernel_get_rows template [[host_name("kernel_get_rows_q4_K")]] kernel get_rows_t kernel_get_rows; template [[host_name("kernel_get_rows_q5_K")]] kernel get_rows_t kernel_get_rows; template [[host_name("kernel_get_rows_q6_K")]] kernel get_rows_t kernel_get_rows; +template [[host_name("kernel_get_rows_iq2_xxs")]] kernel get_rows_t kernel_get_rows; // // matrix-matrix multiplication @@ -4314,6 +4550,7 @@ template [[host_name("kernel_mul_mm_q3_K_f32")]] kernel mat_mm_t kernel_mul_mm; template [[host_name("kernel_mul_mm_q5_K_f32")]] kernel mat_mm_t kernel_mul_mm; template [[host_name("kernel_mul_mm_q6_K_f32")]] kernel mat_mm_t kernel_mul_mm; +template [[host_name("kernel_mul_mm_iq2_xxs_f32")]] kernel mat_mm_t kernel_mul_mm; // // indirect matrix-matrix multiplication @@ -4362,6 +4599,7 @@ template [[host_name("kernel_mul_mm_id_q3_K_f32")]] kernel mat_mm_id_t kernel_mu template [[host_name("kernel_mul_mm_id_q4_K_f32")]] kernel mat_mm_id_t kernel_mul_mm_id; template [[host_name("kernel_mul_mm_id_q5_K_f32")]] kernel mat_mm_id_t kernel_mul_mm_id; template [[host_name("kernel_mul_mm_id_q6_K_f32")]] kernel mat_mm_id_t kernel_mul_mm_id; +template [[host_name("kernel_mul_mm_id_iq2_xxs_f32")]] kernel mat_mm_id_t kernel_mul_mm_id; // // matrix-vector multiplication @@ -5134,3 +5372,66 @@ kernel void kernel_mul_mv_id_q6_K_f32( tiisg, sgitg); } + +[[host_name("kernel_mul_mv_id_iq2_xxs_f32")]] +kernel void kernel_mul_mv_id_iq2_xxs_f32( + device const char * ids, + device const char * src1, + device float * dst, + constant uint64_t & nbi1, + constant int64_t & ne00, + constant int64_t & ne01, + constant int64_t & ne02, + constant uint64_t & nb00, + constant uint64_t & nb01, + constant uint64_t & nb02, + constant int64_t & ne10, + constant int64_t & ne11, + constant int64_t & ne12, + constant int64_t & ne13, + constant uint64_t & nb10, + constant uint64_t & nb11, + constant uint64_t & nb12, + constant int64_t & ne0, + constant int64_t & ne1, + constant uint64_t & nb1, + constant uint & r2, + constant uint & r3, + constant int & idx, + device const char * src00, + device const char * src01, + device const char * src02, + device const char * src03, + device const char * src04, + device const char * src05, + device const char * src06, + device const char * src07, + uint3 tgpig[[threadgroup_position_in_grid]], + uint tiitg[[thread_index_in_threadgroup]], + uint tiisg[[thread_index_in_simdgroup]], + uint sgitg[[simdgroup_index_in_threadgroup]]) { + device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07}; + + const int64_t bid = tgpig.z/(ne12*ne13); + + tgpig.z = tgpig.z%(ne12*ne13); + + const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx]; + + kernel_mul_mv_iq2_xxs_f32_impl( + src0[id], + (device const float *) (src1 + bid*nb11), + dst + bid*ne0, + ne00, + ne01, + ne02, + ne10, + ne12, + ne0, + ne1, + r2, + r3, + tgpig, + tiisg, + sgitg); +} From dd296101531af342332d00e23c729454121aee14 Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Wed, 3 Jan 2024 18:59:14 +0100 Subject: [PATCH 05/17] iq2_xxs: Metal dot product now works We have PP-512 = 475 t/s TG-128 = 47.3 t/s Not the greatest performance, but not complete garbage either. --- ggml-metal.metal | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-) diff --git a/ggml-metal.metal b/ggml-metal.metal index 33fe32dcb8e38..a6f1b1745caa1 100644 --- a/ggml-metal.metal +++ b/ggml-metal.metal @@ -3601,10 +3601,6 @@ void kernel_mul_mv_iq2_xxs_f32_impl( device const float * y4 = y + 32 * ix; - uint32_t aux32[2]; - thread const uint8_t * aux8 = (thread const uint8_t *)aux32; - thread uint16_t * aux16 = (thread uint16_t *)aux32; - for (int ib32 = ix; ib32 < nb32; ib32 += 32) { for (int i = 0; i < 32; ++i) { @@ -3620,13 +3616,14 @@ void kernel_mul_mv_iq2_xxs_f32_impl( const float db = xr->d; device const uint16_t * q2 = xr->qs + 4 * ib; - for (int i = 0; i < 4; ++i) aux16[i] = q2[i]; - const float d = db * (0.5f + (aux32[1] >> 28)); + device const uint8_t * aux8 = (device const uint8_t *)q2; + const uint32_t aux32 = q2[2] | (q2[3] << 16); + const float d = db * (0.5f + (aux32 >> 28)); float sum = 0; for (int l = 0; l < 4; ++l) { constant uint8_t * grid = (constant uint8_t *)(kgrid_iq2xxs + aux8[l]); - const uint8_t signs = ksigns_iq2xs[(aux32[1] >> 7*l) & 127]; + const uint8_t signs = ksigns_iq2xs[(aux32 >> 7*l) & 127]; for (int j = 0; j < 8; ++j) { sum += yl[8*l + j] * grid[j] * (signs & kmask_iq2xs[j] ? -1.f : 1.f); } From 1c96aa0d7fa25f892ca87d5765c86ea231f393ba Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Wed, 3 Jan 2024 19:32:57 +0100 Subject: [PATCH 06/17] iq2_xxs: slighty faster dot product TG-128 is now 48.4 t/s --- ggml-metal.metal | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/ggml-metal.metal b/ggml-metal.metal index a6f1b1745caa1..94fc712350d9b 100644 --- a/ggml-metal.metal +++ b/ggml-metal.metal @@ -3592,8 +3592,6 @@ void kernel_mul_mv_iq2_xxs_f32_impl( float yl[32]; float sumf[N_DST]={0.f}, all_sum; - const int step = sizeof(block_q2_K) * nb; - const int nb32 = nb * (QK_K / 32); #if QK_K == 256 @@ -3611,11 +3609,12 @@ void kernel_mul_mv_iq2_xxs_f32_impl( const int ib = ib32 % (QK_K / 32); device const block_iq2_xxs * xr = x + ibl; + device const uint16_t * q2 = xr->qs + 4 * ib; + device const half * dh = &xr->d; for (int row = 0; row < N_DST; row++) { - const float db = xr->d; - device const uint16_t * q2 = xr->qs + 4 * ib; + const float db = dh[0]; device const uint8_t * aux8 = (device const uint8_t *)q2; const uint32_t aux32 = q2[2] | (q2[3] << 16); const float d = db * (0.5f + (aux32 >> 28)); @@ -3630,7 +3629,8 @@ void kernel_mul_mv_iq2_xxs_f32_impl( } sumf[row] += d * sum; - xr += nb; + dh += nb*sizeof(block_iq2_xxs)/2; + q2 += nb*sizeof(block_iq2_xxs)/2; } y4 += 32 * 32; From e211fadc8a52af67bad3f543a67000ae471363ec Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Thu, 4 Jan 2024 02:06:23 +0100 Subject: [PATCH 07/17] iq2_xxs: slighty faster dot product TG-128 is now 50.9 t/s --- ggml-metal.m | 12 ++++++++++-- ggml-metal.metal | 16 ++++++++++++++-- 2 files changed, 24 insertions(+), 4 deletions(-) diff --git a/ggml-metal.m b/ggml-metal.m index e422dbb4e484a..47e474d557f5d 100644 --- a/ggml-metal.m +++ b/ggml-metal.m @@ -1708,10 +1708,14 @@ bool ggml_metal_graph_compute( if (src0t == GGML_TYPE_Q4_0 || src0t == GGML_TYPE_Q4_1 || src0t == GGML_TYPE_Q5_0 || src0t == GGML_TYPE_Q5_1 || src0t == GGML_TYPE_Q8_0 || - src0t == GGML_TYPE_IQ2_XXS || + //src0t == GGML_TYPE_IQ2_XXS || src0t == GGML_TYPE_Q2_K) { // || src0t == GGML_TYPE_Q4_K) { [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 7)/8, ne11, ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; } + else if (src0t == GGML_TYPE_IQ2_XXS) { + [encoder setThreadgroupMemoryLength:256*8 atIndex:0]; + [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 7)/8, ne11, ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; + } else if (src0t == GGML_TYPE_Q4_K) { [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 3)/4, ne11, ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; } @@ -1972,10 +1976,14 @@ bool ggml_metal_graph_compute( if (src2t == GGML_TYPE_Q4_0 || src2t == GGML_TYPE_Q4_1 || src2t == GGML_TYPE_Q5_0 || src2t == GGML_TYPE_Q5_1 || src2t == GGML_TYPE_Q8_0 || - src2t == GGML_TYPE_IQ2_XXS || + //src2t == GGML_TYPE_IQ2_XXS || src2t == GGML_TYPE_Q2_K) { // || src2t == GGML_TYPE_Q4_K) { [encoder dispatchThreadgroups:MTLSizeMake((ne21 + 7)/8, _ne1, ne01*ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; } + else if (src0t == GGML_TYPE_IQ2_XXS) { + [encoder setThreadgroupMemoryLength:256*8 atIndex:0]; + [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 7)/8, ne11, ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; + } else if (src2t == GGML_TYPE_Q4_K) { [encoder dispatchThreadgroups:MTLSizeMake((ne21 + 3)/4, _ne1, ne01*ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; } diff --git a/ggml-metal.metal b/ggml-metal.metal index 94fc712350d9b..0ada17bef9789 100644 --- a/ggml-metal.metal +++ b/ggml-metal.metal @@ -3569,6 +3569,7 @@ void kernel_mul_mv_iq2_xxs_f32_impl( constant int64_t & ne1, constant uint & r2, constant uint & r3, + threadgroup int8_t * shared_values [[threadgroup(0)]], uint3 tgpig[[threadgroup_position_in_grid]], uint tiisg[[thread_index_in_simdgroup]], uint sgitg[[simdgroup_index_in_threadgroup]]) { @@ -3594,6 +3595,14 @@ void kernel_mul_mv_iq2_xxs_f32_impl( const int nb32 = nb * (QK_K / 32); + threadgroup uint64_t * values = (threadgroup uint64_t *)shared_values; + { + const int nval = 4; + const int pos = (32*sgitg + tiisg)*nval; + for (int i = 0; i < nval; ++i) values[pos + i] = kgrid_iq2xxs[pos + i]; + threadgroup_barrier(mem_flags::mem_threadgroup); + } + #if QK_K == 256 const int ix = tiisg; @@ -3621,7 +3630,7 @@ void kernel_mul_mv_iq2_xxs_f32_impl( float sum = 0; for (int l = 0; l < 4; ++l) { - constant uint8_t * grid = (constant uint8_t *)(kgrid_iq2xxs + aux8[l]); + const threadgroup uint8_t * grid = (const threadgroup uint8_t *)(values + aux8[l]); const uint8_t signs = ksigns_iq2xs[(aux32 >> 7*l) & 127]; for (int j = 0; j < 8; ++j) { sum += yl[8*l + j] * grid[j] * (signs & kmask_iq2xs[j] ? -1.f : 1.f); @@ -3668,11 +3677,12 @@ kernel void kernel_mul_mv_iq2_xxs_f32( constant int64_t & ne1, constant uint & r2, constant uint & r3, + threadgroup int8_t * shared_values [[threadgroup(0)]], uint3 tgpig[[threadgroup_position_in_grid]], uint tiisg[[thread_index_in_simdgroup]], uint sgitg[[simdgroup_index_in_threadgroup]]) { - kernel_mul_mv_iq2_xxs_f32_impl(src0, src1, dst, ne00, ne01, ne02, ne10, ne12, ne0, ne1, r2, r3, tgpig, tiisg, sgitg); + kernel_mul_mv_iq2_xxs_f32_impl(src0, src1, dst, ne00, ne01, ne02, ne10, ne12, ne0, ne1, r2, r3, shared_values, tgpig, tiisg, sgitg); } //============================= templates and their specializations ============================= @@ -5403,6 +5413,7 @@ kernel void kernel_mul_mv_id_iq2_xxs_f32( device const char * src05, device const char * src06, device const char * src07, + threadgroup int8_t * shared_values [[threadgroup(0)]], uint3 tgpig[[threadgroup_position_in_grid]], uint tiitg[[thread_index_in_threadgroup]], uint tiisg[[thread_index_in_simdgroup]], @@ -5428,6 +5439,7 @@ kernel void kernel_mul_mv_id_iq2_xxs_f32( ne1, r2, r3, + shared_values, tgpig, tiisg, sgitg); From 065cc8cb474be7945d2997047dd926c644899cc5 Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Thu, 4 Jan 2024 02:25:59 +0100 Subject: [PATCH 08/17] iq2_xxs: even faster Metal dot product TG-128 is now 54.1 t/s. Strangely enough, putting the signs lookup table into shared memory has a bigger impact than the grid values being in shared memory. --- ggml-metal.m | 4 ++-- ggml-metal.metal | 10 +++++++--- 2 files changed, 9 insertions(+), 5 deletions(-) diff --git a/ggml-metal.m b/ggml-metal.m index 47e474d557f5d..43536e5f4c861 100644 --- a/ggml-metal.m +++ b/ggml-metal.m @@ -1713,7 +1713,7 @@ bool ggml_metal_graph_compute( [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 7)/8, ne11, ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; } else if (src0t == GGML_TYPE_IQ2_XXS) { - [encoder setThreadgroupMemoryLength:256*8 atIndex:0]; + [encoder setThreadgroupMemoryLength:(256*8+128) atIndex:0]; [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 7)/8, ne11, ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; } else if (src0t == GGML_TYPE_Q4_K) { @@ -1981,7 +1981,7 @@ bool ggml_metal_graph_compute( [encoder dispatchThreadgroups:MTLSizeMake((ne21 + 7)/8, _ne1, ne01*ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; } else if (src0t == GGML_TYPE_IQ2_XXS) { - [encoder setThreadgroupMemoryLength:256*8 atIndex:0]; + [encoder setThreadgroupMemoryLength:(256*8+128) atIndex:0]; [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 7)/8, ne11, ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; } else if (src2t == GGML_TYPE_Q4_K) { diff --git a/ggml-metal.metal b/ggml-metal.metal index 0ada17bef9789..0cc535ac7294d 100644 --- a/ggml-metal.metal +++ b/ggml-metal.metal @@ -3596,10 +3596,14 @@ void kernel_mul_mv_iq2_xxs_f32_impl( const int nb32 = nb * (QK_K / 32); threadgroup uint64_t * values = (threadgroup uint64_t *)shared_values; + threadgroup uint8_t * shared_signs = (threadgroup uint8_t *)(values + 256); { - const int nval = 4; - const int pos = (32*sgitg + tiisg)*nval; + int nval = 4; + int pos = (32*sgitg + tiisg)*nval; for (int i = 0; i < nval; ++i) values[pos + i] = kgrid_iq2xxs[pos + i]; + nval = 2; + pos = (32*sgitg + tiisg)*nval; + for (int i = 0; i < nval; ++i) shared_signs[pos+i] = ksigns_iq2xs[pos+i]; threadgroup_barrier(mem_flags::mem_threadgroup); } @@ -3631,7 +3635,7 @@ void kernel_mul_mv_iq2_xxs_f32_impl( float sum = 0; for (int l = 0; l < 4; ++l) { const threadgroup uint8_t * grid = (const threadgroup uint8_t *)(values + aux8[l]); - const uint8_t signs = ksigns_iq2xs[(aux32 >> 7*l) & 127]; + const uint8_t signs = shared_signs[(aux32 >> 7*l) & 127]; for (int j = 0; j < 8; ++j) { sum += yl[8*l + j] * grid[j] * (signs & kmask_iq2xs[j] ? -1.f : 1.f); } From 06e6908a6b0272884e047782b6f52abb3abc0c7a Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Thu, 4 Jan 2024 06:25:06 +0200 Subject: [PATCH 09/17] iq2_xxs: dequantize CUDA kernel - fix conflict with master --- ggml-cuda.cu | 144 +++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 144 insertions(+) diff --git a/ggml-cuda.cu b/ggml-cuda.cu index 2df64b11156c0..6c8a44bc2f63e 100644 --- a/ggml-cuda.cu +++ b/ggml-cuda.cu @@ -477,6 +477,14 @@ typedef struct { } block_q6_K; static_assert(sizeof(block_q6_K) == sizeof(ggml_fp16_t) + 13*QK_K/16, "wrong q6_K block size/padding"); +#define QR2_XXS 4 +#define QI2_XXS (QK_K / (4*QR2_XXS)) +typedef struct { + half d; + uint16_t qs[QK_K/8]; +} block_iq2_xxs; +static_assert(sizeof(block_iq2_xxs) == sizeof(ggml_fp16_t) + QK_K/8*sizeof(uint16_t), "wrong iq2_xxs block size/padding"); + #define WARP_SIZE 32 #define MATRIX_ROW_PADDING 512 // last row of quant. matrices is a multiple of this to avoid out-of-bounds memory accesses @@ -1292,6 +1300,128 @@ static __global__ void dequantize_block_q6_K(const void * __restrict__ vx, dst_t #endif } +static const __device__ uint64_t kgrid_iq2xxs[256] = { + 0x0808080808080808, 0x080808080808082b, 0x0808080808081919, 0x0808080808082b08, + 0x0808080808082b2b, 0x0808080808190819, 0x0808080808191908, 0x08080808082b0808, + 0x08080808082b082b, 0x08080808082b2b08, 0x08080808082b2b2b, 0x0808080819080819, + 0x0808080819081908, 0x0808080819190808, 0x0808080819192b08, 0x08080808192b0819, + 0x08080808192b1908, 0x080808082b080808, 0x080808082b08082b, 0x080808082b082b2b, + 0x080808082b2b082b, 0x0808081908080819, 0x0808081908081908, 0x0808081908190808, + 0x0808081908191919, 0x0808081919080808, 0x080808192b081908, 0x080808192b192b08, + 0x0808082b08080808, 0x0808082b0808082b, 0x0808082b082b082b, 0x0808082b2b08082b, + 0x0808190808080819, 0x0808190808081908, 0x0808190808190808, 0x08081908082b0819, + 0x08081908082b1908, 0x0808190819080808, 0x080819081908082b, 0x0808190819082b08, + 0x08081908192b0808, 0x080819082b080819, 0x080819082b081908, 0x080819082b190808, + 0x080819082b2b1908, 0x0808191908080808, 0x080819190808082b, 0x0808191908082b08, + 0x08081919082b0808, 0x080819191908192b, 0x08081919192b2b19, 0x080819192b080808, + 0x080819192b190819, 0x0808192b08082b19, 0x0808192b08190808, 0x0808192b19080808, + 0x0808192b2b081908, 0x0808192b2b2b1908, 0x08082b0808080808, 0x08082b0808081919, + 0x08082b0808082b08, 0x08082b0808191908, 0x08082b08082b2b08, 0x08082b0819080819, + 0x08082b0819081908, 0x08082b0819190808, 0x08082b081919082b, 0x08082b082b082b08, + 0x08082b1908081908, 0x08082b1919080808, 0x08082b2b0808082b, 0x08082b2b08191908, + 0x0819080808080819, 0x0819080808081908, 0x0819080808190808, 0x08190808082b0819, + 0x0819080819080808, 0x08190808192b0808, 0x081908082b081908, 0x081908082b190808, + 0x081908082b191919, 0x0819081908080808, 0x0819081908082b08, 0x08190819082b0808, + 0x0819081919190808, 0x0819081919192b2b, 0x081908192b080808, 0x0819082b082b1908, + 0x0819082b19081919, 0x0819190808080808, 0x0819190808082b08, 0x08191908082b0808, + 0x08191908082b1919, 0x0819190819082b19, 0x081919082b080808, 0x0819191908192b08, + 0x08191919192b082b, 0x0819192b08080808, 0x0819192b0819192b, 0x08192b0808080819, + 0x08192b0808081908, 0x08192b0808190808, 0x08192b0819080808, 0x08192b082b080819, + 0x08192b1908080808, 0x08192b1908081919, 0x08192b192b2b0808, 0x08192b2b19190819, + 0x082b080808080808, 0x082b08080808082b, 0x082b080808082b2b, 0x082b080819081908, + 0x082b0808192b0819, 0x082b08082b080808, 0x082b08082b08082b, 0x082b0819082b2b19, + 0x082b081919082b08, 0x082b082b08080808, 0x082b082b0808082b, 0x082b190808080819, + 0x082b190808081908, 0x082b190808190808, 0x082b190819080808, 0x082b19081919192b, + 0x082b191908080808, 0x082b191919080819, 0x082b1919192b1908, 0x082b192b2b190808, + 0x082b2b0808082b08, 0x082b2b08082b0808, 0x082b2b082b191908, 0x082b2b2b19081908, + 0x1908080808080819, 0x1908080808081908, 0x1908080808190808, 0x1908080808192b08, + 0x19080808082b0819, 0x19080808082b1908, 0x1908080819080808, 0x1908080819082b08, + 0x190808081919192b, 0x19080808192b0808, 0x190808082b080819, 0x190808082b081908, + 0x190808082b190808, 0x1908081908080808, 0x19080819082b0808, 0x19080819192b0819, + 0x190808192b080808, 0x190808192b081919, 0x1908082b08080819, 0x1908082b08190808, + 0x1908082b19082b08, 0x1908082b1919192b, 0x1908082b192b2b08, 0x1908190808080808, + 0x1908190808082b08, 0x19081908082b0808, 0x190819082b080808, 0x190819082b192b19, + 0x190819190819082b, 0x19081919082b1908, 0x1908192b08080808, 0x19082b0808080819, + 0x19082b0808081908, 0x19082b0808190808, 0x19082b0819080808, 0x19082b0819081919, + 0x19082b1908080808, 0x19082b1919192b08, 0x19082b19192b0819, 0x19082b192b08082b, + 0x19082b2b19081919, 0x19082b2b2b190808, 0x1919080808080808, 0x1919080808082b08, + 0x1919080808190819, 0x1919080808192b19, 0x19190808082b0808, 0x191908082b080808, + 0x191908082b082b08, 0x1919081908081908, 0x191908191908082b, 0x191908192b2b1908, + 0x1919082b2b190819, 0x191919082b190808, 0x191919082b19082b, 0x1919191908082b2b, + 0x1919192b08080819, 0x1919192b19191908, 0x19192b0808080808, 0x19192b0808190819, + 0x19192b0808192b19, 0x19192b08192b1908, 0x19192b1919080808, 0x19192b2b08082b08, + 0x192b080808081908, 0x192b080808190808, 0x192b080819080808, 0x192b0808192b2b08, + 0x192b081908080808, 0x192b081919191919, 0x192b082b08192b08, 0x192b082b192b0808, + 0x192b190808080808, 0x192b190808081919, 0x192b191908190808, 0x192b19190819082b, + 0x192b19192b081908, 0x192b2b081908082b, 0x2b08080808080808, 0x2b0808080808082b, + 0x2b08080808082b2b, 0x2b08080819080819, 0x2b0808082b08082b, 0x2b08081908081908, + 0x2b08081908192b08, 0x2b08081919080808, 0x2b08082b08190819, 0x2b08190808080819, + 0x2b08190808081908, 0x2b08190808190808, 0x2b08190808191919, 0x2b08190819080808, + 0x2b081908192b0808, 0x2b08191908080808, 0x2b0819191908192b, 0x2b0819192b191908, + 0x2b08192b08082b19, 0x2b08192b19080808, 0x2b08192b192b0808, 0x2b082b080808082b, + 0x2b082b1908081908, 0x2b082b2b08190819, 0x2b19080808081908, 0x2b19080808190808, + 0x2b190808082b1908, 0x2b19080819080808, 0x2b1908082b2b0819, 0x2b1908190819192b, + 0x2b1908192b080808, 0x2b19082b19081919, 0x2b19190808080808, 0x2b191908082b082b, + 0x2b19190819081908, 0x2b19191919190819, 0x2b192b082b080819, 0x2b192b19082b0808, + 0x2b2b08080808082b, 0x2b2b080819190808, 0x2b2b08082b081919, 0x2b2b081908082b19, + 0x2b2b082b08080808, 0x2b2b190808192b08, 0x2b2b2b0819190808, 0x2b2b2b1908081908, +}; + +static const __device__ uint8_t ksigns_iq2xs[128] = { + 0, 129, 130, 3, 132, 5, 6, 135, 136, 9, 10, 139, 12, 141, 142, 15, + 144, 17, 18, 147, 20, 149, 150, 23, 24, 153, 154, 27, 156, 29, 30, 159, + 160, 33, 34, 163, 36, 165, 166, 39, 40, 169, 170, 43, 172, 45, 46, 175, + 48, 177, 178, 51, 180, 53, 54, 183, 184, 57, 58, 187, 60, 189, 190, 63, + 192, 65, 66, 195, 68, 197, 198, 71, 72, 201, 202, 75, 204, 77, 78, 207, + 80, 209, 210, 83, 212, 85, 86, 215, 216, 89, 90, 219, 92, 221, 222, 95, + 96, 225, 226, 99, 228, 101, 102, 231, 232, 105, 106, 235, 108, 237, 238, 111, + 240, 113, 114, 243, 116, 245, 246, 119, 120, 249, 250, 123, 252, 125, 126, 255, +}; + +static const __device__ uint8_t kmask_iq2xs[8] = {1, 2, 4, 8, 16, 32, 64, 128}; + +inline bool ggml_cuda_supports_mmq(enum ggml_type type) { + switch (type) { + case GGML_TYPE_Q4_0: + case GGML_TYPE_Q4_1: + case GGML_TYPE_Q5_0: + case GGML_TYPE_Q5_1: + case GGML_TYPE_Q8_0: + case GGML_TYPE_Q2_K: + case GGML_TYPE_Q3_K: + case GGML_TYPE_Q4_K: + case GGML_TYPE_Q5_K: + case GGML_TYPE_Q6_K: + return true; + default: + return false; + } +} + +template +static __global__ void dequantize_block_iq2_xxs(const void * __restrict__ vx, dst_t * __restrict__ yy) { + + const int i = blockIdx.x; + const block_iq2_xxs * x = (const block_iq2_xxs *) vx; + + const int tid = threadIdx.x; +#if QK_K == 256 + const int il = tid/8; // 0...3 + const int ib = tid%8; // 0...7 + dst_t * y = yy + i*QK_K + 32*ib + 8*il; + const uint16_t * q2 = x[i].qs + 4*ib; + const uint8_t * aux8 = (const uint8_t *)q2; + const uint8_t * grid = (const uint8_t *)(kgrid_iq2xxs + aux8[il]); + const uint32_t aux32 = q2[2] | (q2[3] << 16); + const float d = (float)x[i].d * (0.5f + (aux32 >> 28)) * 0.25f; + const uint8_t signs = ksigns_iq2xs[(aux32 >> 7*il) & 127]; + for (int j = 0; j < 8; ++j) y[j] = d * grid[j] * (signs & kmask_iq2xs[j] ? -1.f : 1.f); +#else + assert(false); +#endif + +} + static __global__ void dequantize_mul_mat_vec_q2_k(const void * __restrict__ vx, const float * __restrict__ yy, float * __restrict__ dst, const int ncols, int nrows) { static_assert(16%K_QUANTS_PER_ITERATION == 0, "16 must be divisible by K_QUANTS_PER_ITERATION"); @@ -5670,6 +5800,12 @@ static void convert_unary_cuda(const void * __restrict__ vx, dst_t * __restrict_ convert_unary<<>>(vx, y, k); } +template +static void dequantize_row_iq2_xxs_cuda(const void * vx, dst_t * y, const int k, cudaStream_t stream) { + const int nb = k / QK_K; + dequantize_block_iq2_xxs<<>>(vx, y); +} + static to_fp16_cuda_t ggml_get_to_fp16_cuda(ggml_type type) { switch (type) { case GGML_TYPE_Q4_0: @@ -5692,6 +5828,8 @@ static to_fp16_cuda_t ggml_get_to_fp16_cuda(ggml_type type) { return dequantize_row_q5_K_cuda; case GGML_TYPE_Q6_K: return dequantize_row_q6_K_cuda; + case GGML_TYPE_IQ2_XXS: + return dequantize_row_iq2_xxs_cuda; case GGML_TYPE_F32: return convert_unary_cuda; default: @@ -5721,6 +5859,8 @@ static to_fp32_cuda_t ggml_get_to_fp32_cuda(ggml_type type) { return dequantize_row_q5_K_cuda; case GGML_TYPE_Q6_K: return dequantize_row_q6_K_cuda; + case GGML_TYPE_IQ2_XXS: + return dequantize_row_iq2_xxs_cuda; case GGML_TYPE_F16: return convert_unary_cuda; default: @@ -7407,6 +7547,7 @@ static int64_t get_row_rounding(ggml_type type) { case GGML_TYPE_Q4_K: case GGML_TYPE_Q5_K: case GGML_TYPE_Q6_K: + case GGML_TYPE_IQ2_XXS: return max_compute_capability >= CC_RDNA2 ? 128 : 64; default: GGML_ASSERT(false); @@ -7427,6 +7568,7 @@ static int64_t get_row_rounding(ggml_type type) { case GGML_TYPE_Q3_K: case GGML_TYPE_Q4_K: case GGML_TYPE_Q5_K: + case GGML_TYPE_IQ2_XXS: return max_compute_capability >= CC_VOLTA ? 128 : 64; case GGML_TYPE_Q6_K: return 64; @@ -8691,6 +8833,8 @@ static void ggml_cuda_mul_mat(const ggml_tensor * src0, const ggml_tensor * src1 use_mul_mat_q = use_mul_mat_q && !(fp16_performance_good && src1->ne[1] > MMQ_MAX_BATCH_SIZE); #endif // CUDA_USE_TENSOR_CORES + use_mul_mat_q = use_mul_mat_q && ggml_cuda_supports_mmq(src0->type); + #endif // defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__) // debug helpers From 8240521901f2a0dcf31e2ca54034bbbaa4ea8d1f Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Thu, 4 Jan 2024 07:32:39 +0200 Subject: [PATCH 10/17] iq2_xxs: quantized CUDA dot product (MMVQ) We get TG-128 = 153.1 t/s --- ggml-cuda.cu | 41 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 41 insertions(+) diff --git a/ggml-cuda.cu b/ggml-cuda.cu index 6c8a44bc2f63e..1cc2f7ca6c9b3 100644 --- a/ggml-cuda.cu +++ b/ggml-cuda.cu @@ -3955,6 +3955,35 @@ static __device__ __forceinline__ float vec_dot_q6_K_q8_1_mul_mat( return vec_dot_q6_K_q8_1_impl_mmq(&x_ql[index_x], &y_qs[index_y], sc, x_dmf[i * (WARP_SIZE/QI6_K) + i/QI6_K], &y_df[index_y/QI8_1]); } +static __device__ __forceinline__ float vec_dot_iq2_xxs_q8_1( + const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & iqs) { +#if QK_K == 256 + const block_iq2_xxs * bq2 = (const block_iq2_xxs *) vbq; + + // iqs is 0...15 + const int ib32 = iqs/2; + const int il = iqs%2; + const uint16_t * q2 = bq2->qs + 4*ib32; + const uint8_t * aux8 = (const uint8_t *)q2; + const uint8_t * grid1 = (const uint8_t *)(kgrid_iq2xxs + aux8[2*il+0]); + const uint8_t * grid2 = (const uint8_t *)(kgrid_iq2xxs + aux8[2*il+1]); + const uint32_t aux32 = q2[2] | (q2[3] << 16); + const float d = (float)bq2->d * (0.5f + (aux32 >> 28)) * (float)bq8_1[ib32].ds.x * 0.25f; + const uint8_t signs1 = ksigns_iq2xs[(aux32 >> 14*il) & 127]; + const uint8_t signs2 = ksigns_iq2xs[(aux32 >> (14*il + 7)) & 127]; + const int8_t * q8 = bq8_1[ib32].qs + 16*il; + int sumi1 = 0, sumi2 = 0; + for (int j = 0; j < 8; ++j) { + sumi1 += q8[j+0] * grid1[j] * (signs1 & kmask_iq2xs[j] ? -1 : 1); + sumi2 += q8[j+8] * grid2[j] * (signs2 & kmask_iq2xs[j] ? -1 : 1); + } + return d * (sumi1 + sumi2); +#else + assert(false); + return 0.f; +#endif +} + template static __device__ __forceinline__ void mul_mat_q( @@ -6055,6 +6084,15 @@ static void mul_mat_vec_q6_K_q8_1_cuda(const void * vx, const void * vy, float * <<>>(vx, vy, dst, ncols, nrows); } +static void mul_mat_vec_iq2_xxs_q8_1_cuda(const void * vx, const void * vy, float * dst, const int ncols, const int nrows, cudaStream_t stream) { + GGML_ASSERT(ncols % QK_K == 0); + const int block_num_y = (nrows + GGML_CUDA_MMV_Y - 1) / GGML_CUDA_MMV_Y; + const dim3 block_nums(block_num_y, 1, 1); + const dim3 block_dims(WARP_SIZE, GGML_CUDA_MMV_Y, 1); + mul_mat_vec_q + <<>>(vx, vy, dst, ncols, nrows); +} + static void ggml_mul_mat_q4_0_q8_1_cuda( const void * vx, const void * vy, float * dst, const int ncols_x, const int nrows_x, const int ncols_y, const int nrows_y, const int nrows_dst, cudaStream_t stream) { @@ -7619,6 +7657,9 @@ static void ggml_cuda_op_mul_mat_vec_q( case GGML_TYPE_Q6_K: mul_mat_vec_q6_K_q8_1_cuda(src0_dd_i, src1_ddq_i, dst_dd_i, ne00, row_diff, stream); break; + case GGML_TYPE_IQ2_XXS: + mul_mat_vec_iq2_xxs_q8_1_cuda(src0_dd_i, src1_ddq_i, dst_dd_i, ne00, row_diff, stream); + break; default: GGML_ASSERT(false); break; From c19d0d09ba369100deb356da297a6e5194bd303a Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Thu, 4 Jan 2024 08:56:17 +0200 Subject: [PATCH 11/17] iq2_xxs: slightly faster CUDA dot product TG-128 is now at 155.1 t/s. --- ggml-cuda.cu | 22 +++++++++++++++++++++- 1 file changed, 21 insertions(+), 1 deletion(-) diff --git a/ggml-cuda.cu b/ggml-cuda.cu index 1cc2f7ca6c9b3..e833679cad4fd 100644 --- a/ggml-cuda.cu +++ b/ggml-cuda.cu @@ -477,7 +477,7 @@ typedef struct { } block_q6_K; static_assert(sizeof(block_q6_K) == sizeof(ggml_fp16_t) + 13*QK_K/16, "wrong q6_K block size/padding"); -#define QR2_XXS 4 +#define QR2_XXS 8 #define QI2_XXS (QK_K / (4*QR2_XXS)) typedef struct { half d; @@ -3960,6 +3960,25 @@ static __device__ __forceinline__ float vec_dot_iq2_xxs_q8_1( #if QK_K == 256 const block_iq2_xxs * bq2 = (const block_iq2_xxs *) vbq; +#if QR2_XXS == 8 + const int ib32 = iqs; + const uint16_t * q2 = bq2->qs + 4*ib32; + const uint8_t * aux8 = (const uint8_t *)q2; + const int8_t * q8 = bq8_1[ib32].qs; + uint32_t aux32 = q2[2] | (q2[3] << 16); + int sumi = 0; + for (int l = 0; l < 4; ++l) { + const uint8_t * grid = (const uint8_t *)(kgrid_iq2xxs + aux8[l]); + const uint8_t signs = ksigns_iq2xs[aux32 & 127]; + for (int j = 0; j < 8; ++j) { + sumi += q8[j] * grid[j] * (signs & kmask_iq2xs[j] ? -1 : 1); + } + q8 += 8; + aux32 >>= 7; + } + const float d = (float)bq2->d * (0.5f + aux32) * (float)bq8_1[ib32].ds.x * 0.25f; + return d * sumi; +#else // iqs is 0...15 const int ib32 = iqs/2; const int il = iqs%2; @@ -3978,6 +3997,7 @@ static __device__ __forceinline__ float vec_dot_iq2_xxs_q8_1( sumi2 += q8[j+8] * grid2[j] * (signs2 & kmask_iq2xs[j] ? -1 : 1); } return d * (sumi1 + sumi2); +#endif #else assert(false); return 0.f; From fd42737c09df113e22c740995f57fb151fe70d80 Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Thu, 4 Jan 2024 10:13:00 +0100 Subject: [PATCH 12/17] iq2_xxs: add to llama ftype enum --- llama.cpp | 3 +++ llama.h | 1 + 2 files changed, 4 insertions(+) diff --git a/llama.cpp b/llama.cpp index 63853d1c3cdae..8e0717db92702 100644 --- a/llama.cpp +++ b/llama.cpp @@ -2222,6 +2222,7 @@ struct llama_model_loader { case GGML_TYPE_Q4_K: ftype = LLAMA_FTYPE_MOSTLY_Q4_K_M; break; case GGML_TYPE_Q5_K: ftype = LLAMA_FTYPE_MOSTLY_Q5_K_M; break; case GGML_TYPE_Q6_K: ftype = LLAMA_FTYPE_MOSTLY_Q6_K; break; + case GGML_TYPE_IQ2_XXS: ftype = LLAMA_FTYPE_MOSTLY_IQ2_XXS; break; default: { LLAMA_LOG_WARN("%s: unknown type %s\n", __func__, ggml_type_name(type_max)); @@ -2593,6 +2594,7 @@ static std::string llama_model_ftype_name(llama_ftype ftype) { case LLAMA_FTYPE_MOSTLY_Q5_K_S: return "Q5_K - Small"; case LLAMA_FTYPE_MOSTLY_Q5_K_M: return "Q5_K - Medium"; case LLAMA_FTYPE_MOSTLY_Q6_K: return "Q6_K"; + case LLAMA_FTYPE_MOSTLY_IQ2_XXS:return "IQ2_XSS - 2.0625 bpw"; default: return "unknown, may not work"; } @@ -9038,6 +9040,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s case LLAMA_FTYPE_MOSTLY_Q5_K_S: case LLAMA_FTYPE_MOSTLY_Q5_K_M: quantized_type = GGML_TYPE_Q5_K; break; case LLAMA_FTYPE_MOSTLY_Q6_K: quantized_type = GGML_TYPE_Q6_K; break; + case LLAMA_FTYPE_MOSTLY_IQ2_XXS:quantized_type = GGML_TYPE_IQ2_XXS; break; default: throw std::runtime_error(format("invalid output file type %d\n", ftype)); } diff --git a/llama.h b/llama.h index 869ff0acf525a..c11075bbcd693 100644 --- a/llama.h +++ b/llama.h @@ -103,6 +103,7 @@ extern "C" { LLAMA_FTYPE_MOSTLY_Q5_K_S = 16, // except 1d tensors LLAMA_FTYPE_MOSTLY_Q5_K_M = 17, // except 1d tensors LLAMA_FTYPE_MOSTLY_Q6_K = 18, // except 1d tensors + LLAMA_FTYPE_MOSTLY_IQ2_XXS = 19, // except 1d tensors LLAMA_FTYPE_GUESSED = 1024, // not specified in the model file }; From 47ae9b8f34ac6ac655ea2b9807f4f7ce6fc63abd Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Sat, 6 Jan 2024 06:32:37 +0100 Subject: [PATCH 13/17] iq2_xxs: fix MoE on Metal --- ggml-metal.m | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/ggml-metal.m b/ggml-metal.m index 43536e5f4c861..6c2a8d04e5292 100644 --- a/ggml-metal.m +++ b/ggml-metal.m @@ -1980,9 +1980,9 @@ bool ggml_metal_graph_compute( src2t == GGML_TYPE_Q2_K) { // || src2t == GGML_TYPE_Q4_K) { [encoder dispatchThreadgroups:MTLSizeMake((ne21 + 7)/8, _ne1, ne01*ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; } - else if (src0t == GGML_TYPE_IQ2_XXS) { + else if (src2t == GGML_TYPE_IQ2_XXS) { [encoder setThreadgroupMemoryLength:(256*8+128) atIndex:0]; - [encoder dispatchThreadgroups:MTLSizeMake((ne01 + 7)/8, ne11, ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; + [encoder dispatchThreadgroups:MTLSizeMake((ne21 + 7)/8, _ne1, ne01*ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; } else if (src2t == GGML_TYPE_Q4_K) { [encoder dispatchThreadgroups:MTLSizeMake((ne21 + 3)/4, _ne1, ne01*ne12*ne13) threadsPerThreadgroup:MTLSizeMake(nth0, nth1, 1)]; From 61c04053a48d153a23724c1acbbdfc084730fde2 Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Mon, 8 Jan 2024 10:09:50 +0100 Subject: [PATCH 14/17] Fix missing MMQ ops when on hipBLAS I had put the ggml_supports_mmq call at the wrong place. --- ggml-cuda.cu | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/ggml-cuda.cu b/ggml-cuda.cu index e833679cad4fd..e77fa06e7368d 100644 --- a/ggml-cuda.cu +++ b/ggml-cuda.cu @@ -8894,10 +8894,10 @@ static void ggml_cuda_mul_mat(const ggml_tensor * src0, const ggml_tensor * src1 use_mul_mat_q = use_mul_mat_q && !(fp16_performance_good && src1->ne[1] > MMQ_MAX_BATCH_SIZE); #endif // CUDA_USE_TENSOR_CORES - use_mul_mat_q = use_mul_mat_q && ggml_cuda_supports_mmq(src0->type); - #endif // defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__) + use_mul_mat_q = use_mul_mat_q && ggml_cuda_supports_mmq(src0->type); + // debug helpers //printf("src0: %8d %8d %8d %8d\n", src0->ne[0], src0->ne[1], src0->ne[2], src0->ne[3]); //printf(" %8d %8d %8d %8d\n", src0->nb[0], src0->nb[1], src0->nb[2], src0->nb[3]); From 7db967e811024e75b4ce75fc50976944500d2d27 Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Mon, 8 Jan 2024 12:12:00 +0100 Subject: [PATCH 15/17] Fix bug in qequantize_row_iq2_xxs The 0.25f factor was missing. Great detective work by @ggerganov! --- ggml-quants.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/ggml-quants.c b/ggml-quants.c index 24e39c73f75ed..80c339b1f7ba7 100644 --- a/ggml-quants.c +++ b/ggml-quants.c @@ -2442,7 +2442,7 @@ void dequantize_row_iq2_xxs(const block_iq2_xxs * restrict x, float * restrict y for (int ib32 = 0; ib32 < QK_K/32; ++ib32) { memcpy(aux32, x[i].qs + 4*ib32, 2*sizeof(uint32_t)); - const float db = d * (0.5f + (aux32[1] >> 28)); + const float db = d * (0.5f + (aux32[1] >> 28)) * 0.25f; for (int l = 0; l < 4; ++l) { const uint8_t * grid = (const uint8_t *)(iq2xxs_grid + aux8[l]); const uint8_t signs = ksigns_iq2xs[(aux32[1] >> 7*l) & 127]; From 5684d79079560c8529b727db9ddaa44b72ca14c2 Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Mon, 8 Jan 2024 13:39:36 +0200 Subject: [PATCH 16/17] Fixing tests --- ggml-quants.c | 16 ++++++++-------- tests/test-quantize-fns.cpp | 5 +++++ 2 files changed, 13 insertions(+), 8 deletions(-) diff --git a/ggml-quants.c b/ggml-quants.c index 80c339b1f7ba7..fd127f2d1558a 100644 --- a/ggml-quants.c +++ b/ggml-quants.c @@ -7263,14 +7263,14 @@ void ggml_vec_dot_iq2_xxs_q8_K(const int n, float * restrict s, const void * res for (int ib32 = 0; ib32 < QK_K/32; ib32 += 2) { q8b = vld1q_s8_x4(q8); q8 += 64; memcpy(aux32, q2, 4*sizeof(uint32_t)); q2 += 8; - q2u.val[0] = vcombine_u8(vld1_u8((const void *)(iq2xxs_grid + aux8[ 0])), vld1_u8((const void *)(iq2xxs_grid + aux8[ 1]))); - q2u.val[1] = vcombine_u8(vld1_u8((const void *)(iq2xxs_grid + aux8[ 2])), vld1_u8((const void *)(iq2xxs_grid + aux8[ 3]))); - q2u.val[2] = vcombine_u8(vld1_u8((const void *)(iq2xxs_grid + aux8[ 8])), vld1_u8((const void *)(iq2xxs_grid + aux8[ 9]))); - q2u.val[3] = vcombine_u8(vld1_u8((const void *)(iq2xxs_grid + aux8[10])), vld1_u8((const void *)(iq2xxs_grid + aux8[11]))); - q2s.val[0] = vcombine_u8(vld1_u8((const void *)(signs64 + ((aux32[1] >> 0) & 127))), vld1_u8((const void *)(signs64 + ((aux32[1] >> 7) & 127)))); - q2s.val[1] = vcombine_u8(vld1_u8((const void *)(signs64 + ((aux32[1] >> 14) & 127))), vld1_u8((const void *)(signs64 + ((aux32[1] >> 21) & 127)))); - q2s.val[2] = vcombine_u8(vld1_u8((const void *)(signs64 + ((aux32[3] >> 0) & 127))), vld1_u8((const void *)(signs64 + ((aux32[3] >> 7) & 127)))); - q2s.val[3] = vcombine_u8(vld1_u8((const void *)(signs64 + ((aux32[3] >> 14) & 127))), vld1_u8((const void *)(signs64 + ((aux32[3] >> 21) & 127)))); + q2u.val[0] = vcombine_s8(vld1_s8((const void *)(iq2xxs_grid + aux8[ 0])), vld1_s8((const void *)(iq2xxs_grid + aux8[ 1]))); + q2u.val[1] = vcombine_s8(vld1_s8((const void *)(iq2xxs_grid + aux8[ 2])), vld1_s8((const void *)(iq2xxs_grid + aux8[ 3]))); + q2u.val[2] = vcombine_s8(vld1_s8((const void *)(iq2xxs_grid + aux8[ 8])), vld1_s8((const void *)(iq2xxs_grid + aux8[ 9]))); + q2u.val[3] = vcombine_s8(vld1_s8((const void *)(iq2xxs_grid + aux8[10])), vld1_s8((const void *)(iq2xxs_grid + aux8[11]))); + q2s.val[0] = vcombine_s8(vld1_s8((const void *)(signs64 + ((aux32[1] >> 0) & 127))), vld1_s8((const void *)(signs64 + ((aux32[1] >> 7) & 127)))); + q2s.val[1] = vcombine_s8(vld1_s8((const void *)(signs64 + ((aux32[1] >> 14) & 127))), vld1_s8((const void *)(signs64 + ((aux32[1] >> 21) & 127)))); + q2s.val[2] = vcombine_s8(vld1_s8((const void *)(signs64 + ((aux32[3] >> 0) & 127))), vld1_s8((const void *)(signs64 + ((aux32[3] >> 7) & 127)))); + q2s.val[3] = vcombine_s8(vld1_s8((const void *)(signs64 + ((aux32[3] >> 14) & 127))), vld1_s8((const void *)(signs64 + ((aux32[3] >> 21) & 127)))); q2u.val[0] = vmulq_s8(q2u.val[0], q2s.val[0]); q2u.val[1] = vmulq_s8(q2u.val[1], q2s.val[1]); q2u.val[2] = vmulq_s8(q2u.val[2], q2s.val[2]); diff --git a/tests/test-quantize-fns.cpp b/tests/test-quantize-fns.cpp index a2459a2867c5c..cee712618be3d 100644 --- a/tests/test-quantize-fns.cpp +++ b/tests/test-quantize-fns.cpp @@ -134,6 +134,11 @@ int main(int argc, char * argv[]) { continue; } + if ((ggml_type)i == GGML_TYPE_IQ2_XXS) { + printf("Skip %s due to missing quantization functionality\n", ggml_type_name((ggml_type) i)); + continue; + } + printf("Testing %s\n", ggml_type_name((ggml_type) i)); if (qfns.from_float && qfns.to_float) { From bad5f7f33d9e141397f574337315eaf3f5dbe98f Mon Sep 17 00:00:00 2001 From: Iwan Kawrakow Date: Mon, 8 Jan 2024 16:58:37 +0200 Subject: [PATCH 17/17] PR suggestion --- ggml-cuda.cu | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/ggml-cuda.cu b/ggml-cuda.cu index e77fa06e7368d..e0ea890b1afd8 100644 --- a/ggml-cuda.cu +++ b/ggml-cuda.cu @@ -5843,18 +5843,18 @@ static void dequantize_row_q6_K_cuda(const void * vx, dst_t * y, const int k, cu #endif } -template -static void convert_unary_cuda(const void * __restrict__ vx, dst_t * __restrict__ y, const int k, cudaStream_t stream) { - const int num_blocks = (k + CUDA_DEQUANTIZE_BLOCK_SIZE - 1) / CUDA_DEQUANTIZE_BLOCK_SIZE; - convert_unary<<>>(vx, y, k); -} - template static void dequantize_row_iq2_xxs_cuda(const void * vx, dst_t * y, const int k, cudaStream_t stream) { const int nb = k / QK_K; dequantize_block_iq2_xxs<<>>(vx, y); } +template +static void convert_unary_cuda(const void * __restrict__ vx, dst_t * __restrict__ y, const int k, cudaStream_t stream) { + const int num_blocks = (k + CUDA_DEQUANTIZE_BLOCK_SIZE - 1) / CUDA_DEQUANTIZE_BLOCK_SIZE; + convert_unary<<>>(vx, y, k); +} + static to_fp16_cuda_t ggml_get_to_fp16_cuda(ggml_type type) { switch (type) { case GGML_TYPE_Q4_0: