3 people 971facc38e opencl: add optimized q8_0 mm kernel for adreno (#18871) ...

* Add Q8_0 OpenCL kernel

Co-authored-by: yunjie <yunjie@qti.qualcomm.com>

* opencl: fix build for non-adreno

* opencl: refactor q8_0

* opencl: enforce subgroup size of 64 for adreno for q8_0

* For A750 and older generations, subgroup size can be 64 or 128.
  This kernel assumes subgroup size 64.

* opencl: suppress warning when adreno kernels are disabled

---------

Co-authored-by: yunjie <yunjie@qti.qualcomm.com>
Co-authored-by: Li He <lih@qti.qualcomm.com>

2026-01-30 10:19:27 -08:00

..

add_id.cl

opencl: add swiglu_oai and add_id (#15121)

2025-08-06 12:12:17 -07:00

add.cl

opencl: allow mixed f16/f32 add (#15140)

2025-08-12 02:42:41 -07:00

argsort.cl

opencl: add new ops - argsort, div, sub, addrows, sigmoid, group_norm (#13787)

2025-05-27 12:56:08 -07:00

clamp.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

concat.cl

OpenCL: Add concat, tsembd, upscale, tanh, pad and repeat (#13840)

2025-06-02 16:53:36 -07:00

conv2d_f16_f32.cl

opencl: add conv2d kernel (#14403)

2025-07-21 10:03:19 -07:00

conv2d.cl

opencl: add conv2d kernel (#14403)

2025-07-21 10:03:19 -07:00

cpy.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

cvt.cl

opencl: add optimized q8_0 mm kernel for adreno (#18871)

2026-01-30 10:19:27 -08:00

diag_mask_inf.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

div.cl

opencl: add f16 for add, sub, mul, div (#14984)

2025-08-01 13:15:44 +02:00

embed_kernel.py

Introducing experimental OpenCL backend with support for Qualcomm Adreno GPUs (#10693)

2024-12-13 12:23:52 -08:00

expm1.cl

opencl: add EXPM1 op (#18704)

2026-01-09 10:13:13 -08:00

fill.cl

opencl: add FILL op support (#18682)

2026-01-07 22:04:50 -08:00

flash_attn_f16.cl

opencl: add attn sinks support for FA kernels (#15706)

2025-09-01 23:26:53 -07:00

flash_attn_f32_f16.cl

opencl: add attn sinks support for FA kernels (#15706)

2025-09-01 23:26:53 -07:00

flash_attn_f32.cl

opencl: fix FA for f32 (#16584)

2025-10-15 10:48:28 -07:00

gelu.cl

opencl: add GELU_ERF (#14476)

2025-07-04 23:24:56 -07:00

gemm_moe_mxfp4_f32.cl

opencl: transposed gemm/gemv moe kernel with mxfp4,f32 (#16602)

2025-10-17 17:55:32 -07:00

gemv_moe_mxfp4_f32.cl

opencl: transposed gemm/gemv moe kernel with mxfp4,f32 (#16602)

2025-10-17 17:55:32 -07:00

gemv_noshuffle_general_q8_0_f32.cl

opencl: add optimized q8_0 mm kernel for adreno (#18871)

2026-01-30 10:19:27 -08:00

gemv_noshuffle_general.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

gemv_noshuffle.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

get_rows.cl

opencl: support ne3 in get_rows (#15866)

2025-09-30 09:55:13 -07:00

glu.cl

opencl: add swiglu_oai and add_id (#15121)

2025-08-06 12:12:17 -07:00

group_norm.cl

OpenCL: add fused group_norm/norm, mul, add (#15314)

2025-08-26 23:36:05 -07:00

im2col_f16.cl

opencl: fix im2col when KW!=KH (#14803)

2025-07-21 13:55:10 -07:00

im2col_f32.cl

opencl: fix im2col when KW!=KH (#14803)

2025-07-21 13:55:10 -07:00

mean.cl

opencl: add sqr, sqrt, mean and ssm_conv (#17476)

2025-11-26 13:29:58 -08:00

mul_mat_Ab_Bi_8x4.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

mul_mat_f16_f32.cl

opencl: add tiled mul_mat_f16_f32 (#14535)

2025-07-10 14:58:12 -07:00

mul_mm_f16_f32_kq_kqv.cl

opencl: add kernel to handle mat mul in attention to improve encoding speed (#17181)

2025-11-15 17:33:10 -08:00

mul_mm_f16_f32_l4_lm.cl

opencl: fix boundary handling for mul_mm (#16875)

2025-10-30 16:00:20 -07:00

mul_mm_f32_f32_l4_lm.cl

opencl: fix boundary handling for mul_mm (#16875)

2025-10-30 16:00:20 -07:00

mul_mm_q8_0_f32_8x4.cl

opencl: add optimized q8_0 mm kernel for adreno (#18871)

2026-01-30 10:19:27 -08:00

mul_mm_q8_0_f32_l4_lm.cl

opencl: fix boundary handling for mul_mm (#16875)

2025-10-30 16:00:20 -07:00

mul_mv_f16_f16.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

mul_mv_f16_f32_1row.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

mul_mv_f16_f32_l4.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

mul_mv_f16_f32.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

mul_mv_f32_f32.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

mul_mv_id_mxfp4_f32_flat.cl

opencl: optimize mxfp4 kernels (#16037)

2025-09-18 12:03:34 -07:00

mul_mv_id_mxfp4_f32.cl

opencl: add initial mxfp4 support via mv (#15270)

2025-08-15 09:52:14 -07:00

mul_mv_id_q4_0_f32_8x_flat.cl

opencl: add mul_mv_id_q4_0_f32_8x_flat (#14003)

2025-06-10 16:55:58 -07:00

mul_mv_id_q8_0_f32_flat.cl

opencl: initial q8_0 mv support (#15732)

2025-09-21 14:48:44 -07:00

mul_mv_id_q8_0_f32.cl

opencl: initial q8_0 mv support (#15732)

2025-09-21 14:48:44 -07:00

mul_mv_mxfp4_f32_flat.cl

opencl: optimize mxfp4 kernels (#16037)

2025-09-18 12:03:34 -07:00

mul_mv_mxfp4_f32.cl

opencl: add initial mxfp4 support via mv (#15270)

2025-08-15 09:52:14 -07:00

mul_mv_q4_0_f32_1d_8x_flat.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

mul_mv_q4_0_f32_1d_16x_flat.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

mul_mv_q4_0_f32_8x_flat.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

mul_mv_q4_0_f32_v.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

mul_mv_q4_0_f32.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

mul_mv_q6_k_f32_flat.cl

opencl: add flattened q6_K mv (#19054)

2026-01-26 19:36:24 -08:00

mul_mv_q6_k_f32.cl

opencl: add flattened q6_K mv (#19054)

2026-01-26 19:36:24 -08:00

mul_mv_q8_0_f32_flat.cl

opencl: initial q8_0 mv support (#15732)

2025-09-21 14:48:44 -07:00

mul_mv_q8_0_f32.cl

opencl: initial q8_0 mv support (#15732)

2025-09-21 14:48:44 -07:00

mul.cl

opencl: add f16 for add, sub, mul, div (#14984)

2025-08-01 13:15:44 +02:00

norm.cl

OpenCL: add fused group_norm/norm, mul, add (#15314)

2025-08-26 23:36:05 -07:00

pad.cl

opencl: support pad_ext (#15888)

2025-09-30 10:45:45 -07:00

relu.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

repeat.cl

OpenCL: Add concat, tsembd, upscale, tanh, pad and repeat (#13840)

2025-06-02 16:53:36 -07:00

rms_norm.cl

opencl: fix rms_norm_mul (#17250)

2025-11-15 17:40:14 -08:00

rope.cl

opencl: support imrope (#16914)

2025-11-03 11:47:57 -08:00

scale.cl

ggml : add ggml_scale_bias (#14417)

2025-07-09 18:16:12 +02:00

set_rows.cl

opencl: add fastdiv and use it in set_rows, ported from cuda (#17090)

2025-11-10 15:00:13 -08:00

sigmoid.cl

opencl: add new ops - argsort, div, sub, addrows, sigmoid, group_norm (#13787)

2025-05-27 12:56:08 -07:00

silu.cl

opencl: split ggml-opencl.cl into multiple files and cleanup (#12886)

2025-04-15 12:26:00 -07:00

softmax_4_f16.cl

opencl: support sink in soft_max (attn sinks) (#15152)

2025-08-07 21:47:03 -07:00

softmax_4_f32.cl

opencl: support sink in soft_max (attn sinks) (#15152)

2025-08-07 21:47:03 -07:00

softmax_f16.cl

opencl: support sink in soft_max (attn sinks) (#15152)

2025-08-07 21:47:03 -07:00

softmax_f32.cl

opencl: support sink in soft_max (attn sinks) (#15152)

2025-08-07 21:47:03 -07:00

softplus.cl

opencl: add SOFTPLUS op support (#18726)

2026-01-10 21:57:44 -08:00

solve_tri.cl

OpenCL: add SOLVE_TRI op support (#18846)

2026-01-15 11:17:17 -08:00

sqr.cl

opencl: add sqr, sqrt, mean and ssm_conv (#17476)

2025-11-26 13:29:58 -08:00

sqrt.cl

opencl: add sqr, sqrt, mean and ssm_conv (#17476)

2025-11-26 13:29:58 -08:00

ssm_conv.cl

opencl: add sqr, sqrt, mean and ssm_conv (#17476)

2025-11-26 13:29:58 -08:00

sub.cl

opencl: add f16 for add, sub, mul, div (#14984)

2025-08-01 13:15:44 +02:00

sum_rows.cl

opencl: add new ops - argsort, div, sub, addrows, sigmoid, group_norm (#13787)

2025-05-27 12:56:08 -07:00

tanh.cl

OpenCL: Add concat, tsembd, upscale, tanh, pad and repeat (#13840)

2025-06-02 16:53:36 -07:00

transpose.cl

opencl: unpack q4_0 for adreno in get_tensor (#18278)

2025-12-22 10:19:01 -08:00

tri.cl

opencl: add TRI op support (#18979)

2026-01-21 22:05:54 -08:00

tsembd.cl

ggml : fix padding in timestep embedding kernels (#15932)

2025-09-16 15:25:57 +02:00

upscale.cl

opencl : update upscale to support align corners (#14488)

2025-07-02 09:07:42 +02:00