Skip to content

hrx: reuse Q6 row64 tile coordinates in tiled prefill #189

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
AaronStGeorge wants to merge 1 commit into from
Closed

hrx: reuse Q6 row64 tile coordinates in tiled prefill #189

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 13 additions & 7 deletions ggml/src/ggml-hrx/kernel-corpus/kernels/loom-libs/motifs/dequant.loom
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2707,11 +2707,9 @@ func.def inline @ggml_q6k_f32_vector4(%weight: buffer, %weight_row_byte_base: of
func.return %values : vector<4xf32>
}

func.def inline @ggml_q6k_native_row64_f32_vector4(%weight: buffer, %row: index, %input_size: index, %q6_block: index, %q6_group: index, %packet: index) -> (vector<4xf32>) {
func.def inline @ggml_q6k_row64_tile_lane_f32_vector4(%weight: buffer, %tile: index, %row_lane0: index, %q6_group: index, %packet: index) -> (vector<4xf32>) {
%c2 = index.constant 2 : index
%c4 = index.constant 4 : index
%c64 = index.constant 64 : index
%c256 = index.constant 256 : index
%tile_bytes = index.constant 13440 : offset
%scale_byte_add = index.constant 128 : offset
%raw_byte_add = index.constant 1152 : offset
Expand All @@ -2721,10 +2719,7 @@ func.def inline @ggml_q6k_native_row64_f32_vector4(%weight: buffer, %row: index,
%c32_f32v = vector.constant 32.0 : vector<4xf32>
%bounded_group = index.assume %q6_group [range(%q6_group, 0, 7)] : index
%bounded_packet = index.assume %packet [range(%packet, 0, 7)] : index
%blocks = index.div %input_size, %c256 : index
%row_group = index.div %row, %c64 : index
%row_lane = index.rem %row, %c64 : index
%tile = index.madd %row_group, %blocks, %q6_block : index
%row_lane = index.assume %row_lane0 [range(%row_lane0, 0, 63)] : index
%tile_base = index.scale %tile, %tile_bytes : index, offset -> offset
%scale_byte_base = index.add %tile_base, %scale_byte_add : offset
%raw_byte_base = index.add %tile_base, %raw_byte_add : offset
Expand Down Expand Up @@ -2763,6 +2758,17 @@ func.def inline @ggml_q6k_native_row64_f32_vector4(%weight: buffer, %row: index,
func.return %values : vector<4xf32>
}

func.def inline @ggml_q6k_native_row64_f32_vector4(%weight: buffer, %row: index, %input_size: index, %q6_block: index, %q6_group: index, %packet: index) -> (vector<4xf32>) {
%c64 = index.constant 64 : index
%c256 = index.constant 256 : index
%blocks = index.div %input_size, %c256 : index
%row_group = index.div %row, %c64 : index
%row_lane = index.rem %row, %c64 : index
%tile = index.madd %row_group, %blocks, %q6_block : index
%values = func.call @ggml_q6k_row64_tile_lane_f32_vector4(%weight, %tile, %row_lane, %q6_group, %packet) : (buffer, index, index, index, index) -> (vector<4xf32>)
func.return %values : vector<4xf32>
}

func.def inline @ggml_q8_0_f32_vector4(%weight: buffer, %row_byte_base: offset, %q8_block: index, %packet: index) -> (vector<4xf32>) {
%c0 = index.constant 0 : index
%c4 = index.constant 4 : index
Expand Down
7 changes: 5 additions & 2 deletions ...kernel-corpus/kernels/loom-libs/motifs/matmul/tiled_input_f32_dequant_accumulate_f32.loom
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,7 @@ func.decl @ggml_dot_u8_s8_vector8_f32(%weight: vector<2xi32>, %activation: vecto
config.def @ggml.mul_mat.activation_format = 0 : index

func.decl @ggml_q4k_native_row64_f32_vector4(%weight: buffer, %row: index, %input_size: index, %q4_block: index, %q4_group: index, %packet: index) -> (vector<4xf32>)
func.decl @ggml_q6k_native_row64_f32_vector4(%weight: buffer, %row: index, %input_size: index, %q6_block: index, %q6_group: index, %packet: index) -> (vector<4xf32>)
func.decl @ggml_q6k_row64_tile_lane_f32_vector4(%weight: buffer, %tile: index, %row_lane: index, %q6_group: index, %packet: index) -> (vector<4xf32>)

func.decl @ggml_dequant_f32_vector4(%iq4nl_table: vector<16xi8>, %iq3s_grid_low0: vector<32xf32>, %iq3s_grid_low1: vector<32xf32>, %iq3s_grid_low2: vector<32xf32>, %iq3s_grid_low3: vector<32xf32>, %iq3s_grid_low4: vector<32xf32>, %iq3s_grid_low5: vector<32xf32>, %iq3s_grid_low6: vector<32xf32>, %iq3s_grid_low7: vector<32xf32>, %iq3s_grid_low8: vector<32xf32>, %iq3s_grid_low9: vector<32xf32>, %iq3s_grid_low10: vector<32xf32>, %iq3s_grid_low11: vector<32xf32>, %iq3s_grid_low12: vector<32xf32>, %iq3s_grid_low13: vector<32xf32>, %iq3s_grid_low14: vector<32xf32>, %iq3s_grid_low15: vector<32xf32>, %iq3s_grid_high0: vector<32xf32>, %iq3s_grid_high1: vector<32xf32>, %iq3s_grid_high2: vector<32xf32>, %iq3s_grid_high3: vector<32xf32>, %iq3s_grid_high4: vector<32xf32>, %iq3s_grid_high5: vector<32xf32>, %iq3s_grid_high6: vector<32xf32>, %iq3s_grid_high7: vector<32xf32>, %iq3s_grid_high8: vector<32xf32>, %iq3s_grid_high9: vector<32xf32>, %iq3s_grid_high10: vector<32xf32>, %iq3s_grid_high11: vector<32xf32>, %iq3s_grid_high12: vector<32xf32>, %iq3s_grid_high13: vector<32xf32>, %iq3s_grid_high14: vector<32xf32>, %iq3s_grid_high15: vector<32xf32>, %weight_format: index, %weight: buffer, %row_byte_base: offset, %row: index, %input_size: index, %quant_block: index, %quant_group: index, %packet: index, %k: index) -> (vector<4xf32>)

Expand Down Expand Up @@ -102,6 +102,7 @@ func.def inline @ggml_matmul_tiled_input_f32_dequant_accumulate_f32(%full_token_
%result_fragment_view = buffer.view %result_stage[%wave_result_stage_offset] : buffer -> view<16x16xf16, %result_fragment_layout>
%result_physical_view = buffer.view %result_stage[%wave_result_stage_offset] : buffer -> view<16x16xf16>
%channel_tile_base = index.mul %channel_tile, %c64 : index
%q6_blocks_per_row_group = index.div %bounded_input_size, %c256 : index
%token_tile_base = index.mul %token_tile, %c32 : index
%load_packet = index.rem %workitem, %c8 : index
%load_k = index.mul %load_packet, %c4 : index
Expand All @@ -125,6 +126,8 @@ func.def inline @ggml_matmul_tiled_input_f32_dequant_accumulate_f32(%full_token_
%weight_reuses_header = scalar.ori %weight_is_q5, %weight_is_q4_row64 : i1
%weight_is_iq4 = index.cmp eq, %weight_format, %reuse_iq4_xs : index
%result00, %result01, %result10, %result11 = scf.for %quant_block = [%quant_block_begin to %quant_block_end step %c1](%block_acc00 = %init00 : vector<8xf16>, %block_acc01 = %init01 : vector<8xf16>, %block_acc10 = %init10 : vector<8xf16>, %block_acc11 = %init11 : vector<8xf16>) -> (vector<8xf16>, vector<8xf16>, vector<8xf16>, vector<8xf16>) {
// All rows in this output tile share the same row64 weight tile.
%q6_tile = index.madd %channel_tile, %q6_blocks_per_row_group, %quant_block : index
// Retain Q4 row64 or Q5_K headers and Q5 packet high bits for all eight K32 groups; payloads are still loaded per group.
%weight_dm, %weight_scales, %weight_qh = scf.if %weight_reuses_header -> (vector<4x2xf16>, vector<4x3xi32>, vector<4x4xi8>) {
%zero_dm = vector.constant 0.0 : vector<4x2xf16>
Expand Down Expand Up @@ -229,7 +232,7 @@ func.def inline @ggml_matmul_tiled_input_f32_dequant_accumulate_f32(%full_token_
scf.yield %half : vector<4xf16>
} else {
%q6_or_generic = scf.if %weight_is_q6_row64 -> (vector<4xf16>) {
%wide = func.call @ggml_q6k_native_row64_f32_vector4(%weight_noalias, %channel, %bounded_input_size, %quant_block, %quant_group, %load_packet) : (buffer, index, index, index, index, index) -> (vector<4xf32>)
%wide = func.call @ggml_q6k_row64_tile_lane_f32_vector4(%weight_noalias, %q6_tile, %local_row, %quant_group, %load_packet) : (buffer, index, index, index, index) -> (vector<4xf32>)
%half = vector.fptrunc %wide : vector<4xf32> to vector<4xf16>
scf.yield %half : vector<4xf16>
} else {
Expand Down