nihui fcde729c7b riscv: add int8 convolution winograd (#7093)

2026-10-14 14:58:05 +08:00

..

perf

x86: optimize permute with SIMD (#6762)

2026-10-07 23:21:38 +08:00

CMakeLists.txt

fix crash reading past end of memory-mapped model file (#7021)

2026-10-12 00:01:46 +08:00

prng.h

format code style and setup restyled.io (#1840)

2020-06-13 22:17:51 +08:00

test_absval.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_batchnorm.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_bias.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_binaryop_1.cpp

fix(vulkan): include element packing in BinaryOp shape hints (#7040)

2026-10-12 14:07:04 +08:00

test_binaryop_2.cpp

fix(vulkan): include element packing in BinaryOp shape hints (#7040)

2026-10-12 14:07:04 +08:00

test_binaryop_3.cpp

fix(vulkan): include element packing in BinaryOp shape hints (#7040)

2026-10-12 14:07:04 +08:00

test_binaryop_4.cpp

fix(vulkan): include element packing in BinaryOp shape hints (#7040)

2026-10-12 14:07:04 +08:00

test_binaryop_5.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_binaryop.cpp

fix(vulkan): include element packing in BinaryOp shape hints (#7040)

2026-10-12 14:07:04 +08:00

test_bnll.cpp

support 4d mat in more operators (#6737)

2026-05-22 15:31:01 +08:00

test_c_api.cpp

kvcache allocator (#6901)

2026-08-17 15:44:30 +08:00

test_cast.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_celu.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_clip.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_command.cpp

adapt to vkwebgpu and vkopencl (#6876)

2026-08-04 14:37:26 +08:00

test_concat_oom.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_concat.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_convolution1d.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_convolution3d.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_convolution_1.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_convolution_2.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_convolution_3a.cpp

riscv: add int8 convolution winograd (#7093)

2026-10-14 14:58:05 +08:00

test_convolution_3b.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_convolution_3c.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_convolution_4.cpp

unroll 16 for convolution x86 bf16s avx512 (#6680)

2026-05-10 16:06:50 +08:00

test_convolution_oom.cpp

riscv: add int8 convolution winograd (#7093)

2026-10-14 14:58:05 +08:00

test_convolution.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_convolutiondepthwise1d.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_convolutiondepthwise3d.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_convolutiondepthwise_1.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_convolutiondepthwise_oom.cpp

add more oom test (#6759)

2026-05-30 14:59:33 +08:00

test_convolutiondepthwise.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_copyto_1.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_copyto.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_cpu.cpp

riscv: add int8 convolution winograd (#7093)

2026-10-14 14:58:05 +08:00

test_crop_1.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_crop_2.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_crop_3.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_crop_oom.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_crop.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_cumulativesum.cpp

x86: optimize CumulativeSum with SIMD Kogge-Stone prefix scan (#6693)

2026-09-18 17:24:37 +08:00

test_datareader.cpp

fix crash reading past end of memory-mapped model file (#7021)

2026-10-12 00:01:46 +08:00

test_deconvolution1d.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_deconvolution3d.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_deconvolution_oom.cpp

add more oom test (#6759)

2026-05-30 14:59:33 +08:00

test_deconvolution.cpp

x86: reduce Col2im memory access for non-overlapping deconvolution (#7015)

2026-10-08 19:44:27 +08:00

test_deconvolutiondepthwise1d.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_deconvolutiondepthwise3d.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_deconvolutiondepthwise_1.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_deconvolutiondepthwise_oom.cpp

add more oom test (#6759)

2026-05-30 14:59:33 +08:00

test_deconvolutiondepthwise.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_deepcopy.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_deformableconv2d_1.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_deformableconv2d_2.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_deformableconv2d_3.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_deformableconv2d_4.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_deformableconv2d.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_dequantize.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_diag.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_dropout.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_einsum.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_eltwise.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_elu.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_embed.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_erf.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_exp.cpp

support 4d mat in more operators (#6737)

2026-05-22 15:31:01 +08:00

test_expanddims.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_expression.cpp

pnnx: preserve batch semantics across reshape and flatten (#6934)

2026-10-08 19:08:41 +08:00

test_flatten_oom.cpp

add more oom test (#6759)

2026-05-30 14:59:33 +08:00

test_flatten.cpp

gemm x86 tile-N threading, drop testutil layer hook function (#6428)

2025-11-28 15:48:48 +08:00

test_flip.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_fold.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_gelu.cpp

arm/riscv: avoid fp16 GELU overflow (#7060)

2026-10-12 14:07:42 +08:00

test_gemm_0.h

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_0a.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_0b.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_0c.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_0d.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_0e.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_0f.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_1.h

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_1a.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_gemm_1b.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_gemm_2.h

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_2a.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_2b.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_2c.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_2d.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_gemm_2e.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_gemm_3.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_gemm_4.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_gemm_5.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_block_quant.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_gemm_nt_int8.cpp

mips/loongarch: remove dead code and optimize transpose pack (#6749)

2026-05-28 23:02:28 +08:00

test_gemm_nt.cpp

unroll n-tile x16 for gemm x86 avx512 (#6434)

2025-12-02 15:54:12 +08:00

test_gemm_oom.cpp

avoid test failure on fp16 denormal flush to zero (#6891)

2026-08-05 20:05:07 +08:00

test_gemm.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_glu.cpp

support 4d mat in more operators (#6737)

2026-05-22 15:31:01 +08:00

test_gpu.cpp

fix spirv compile failure diagnostics read past shader buffer (#6882)

2026-09-15 21:49:34 +08:00

test_gridsample.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_groupnorm.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_gru.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_hardsigmoid.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_hardswish.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_innerproduct_oom.cpp

add more oom test (#6759)

2026-05-30 14:59:33 +08:00

test_innerproduct.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_input.cpp

support 4d mat in more operators (#6737)

2026-05-22 15:31:01 +08:00

test_instancenorm.cpp

support 4d mat in more operators (#6737)

2026-05-22 15:31:01 +08:00

test_interp_1.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_interp_scale.cpp

fix interp scale factor sampling and pnnx recomputed sizes (#7075)

2026-10-10 00:00:45 +08:00

test_interp.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_inversespectrogram.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_layernorm.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_log.cpp

support 4d mat in more operators (#6737)

2026-05-22 15:31:01 +08:00

test_lrn.cpp

fix lrn even across-channel window size (#7070)

2026-10-12 14:08:14 +08:00

test_lstm.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_mat_batch.cpp

fix test mat batch fp16/bf16 cast (#7044)

2026-10-05 15:22:37 +08:00

test_mat_pixel_affine.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_mat_pixel_drawing.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_mat_pixel_resize.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_mat_pixel_rotate.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_mat_pixel.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_matmul.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_memorydata.cpp

pnnx: store bool MemoryData attributes as raw int8 (#7006)

2026-09-17 10:09:00 +08:00

test_mish.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_modelbin.cpp

fix crash reading past end of memory-mapped model file (#7021)

2026-10-12 00:01:46 +08:00

test_multiheadattention_1.cpp

Fix x86 bf16 GEMM packing on i386 (#6708)

2026-05-09 13:22:23 +08:00

test_multiheadattention_block_quant.cpp

fix MHA singleton mask broadcast across backends (#6869)

2026-10-10 19:47:13 +08:00

test_multiheadattention_kvcache_allocator.cpp

sdpa flash attention x86 optimization, llm models enable gqa, dispatch for avx2 bfloat2float (#6923)

2026-09-01 17:35:42 +08:00

test_multiheadattention_kvcache.cpp

fix MHA singleton mask broadcast across backends (#6869)

2026-10-10 19:47:13 +08:00

test_multiheadattention_oom.cpp

x86/arm/mips/loongarch/riscv gemm block quantization optimization (#6831)

2026-08-04 14:34:22 +08:00

test_multiheadattention.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_mvn.cpp

support 4d mat in more operators (#6737)

2026-05-22 15:31:01 +08:00

test_net.cpp

fix crash reading past end of memory-mapped model file (#7021)

2026-10-12 00:01:46 +08:00

test_noop.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_normalize.cpp

x86: optimize normalize with SIMD (#6761)

2026-09-20 10:49:20 +08:00

test_packing.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_padding_oom.cpp

add more oom test (#6759)

2026-05-30 14:59:33 +08:00

test_padding.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_paramdict.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_permute_packing.cpp

x86: optimize permute with SIMD (#6762)

2026-10-07 23:21:38 +08:00

test_permute.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_pipeline_cache.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_pixelshuffle.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_pooling1d.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_pooling3d.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_pooling.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_power.cpp

support 4d mat in more operators (#6737)

2026-05-22 15:31:01 +08:00

test_prelu.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_priorbox.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_quantize_oom.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_quantize.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_reduction.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_relu.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_reorg.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_requantize_oom.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_requantize.cpp

RVV1.0 Packed INT8 Convolution (#6763)

2026-10-05 23:03:55 +08:00

test_reshape_1.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_reshape_batch.cpp

pnnx: preserve batch semantics across reshape and flatten (#6934)

2026-10-08 19:08:41 +08:00

test_reshape_oom.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_reshape.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_rmsnorm.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_rnn.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_roialign.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_roipooling.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_rotaryembed_oom.cpp

rotaryembed layer (#6407)

2025-11-19 17:28:08 +08:00

test_rotaryembed.cpp

RotaryEmbed: support full-width cos/sin cache for 2D / vision RoPE (#6834)

2026-10-08 18:56:24 +08:00

test_scale.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_sdpa_kvcache_session.cpp

sdpa flash attention x86 optimization, llm models enable gqa, dispatch for avx2 bfloat2float (#6923)

2026-09-01 17:35:42 +08:00

test_sdpa_kvcache.cpp

fix MHA singleton mask broadcast across backends (#6869)

2026-10-10 19:47:13 +08:00

test_sdpa_oom.cpp

sdpa flash attention x86 optimization, llm models enable gqa, dispatch for avx2 bfloat2float (#6923)

2026-09-01 17:35:42 +08:00

test_sdpa.cpp

sdpa flash attention x86 optimization, llm models enable gqa, dispatch for avx2 bfloat2float (#6923)

2026-09-01 17:35:42 +08:00

test_selu.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_shrink.cpp

pnnx convert shrink family to ncnn, fix shrink test (#6329)

2025-09-24 15:22:02 +08:00

test_shufflechannel.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_sigmoid.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_slice_oom.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_slice.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_softmax_oom.cpp

use spdx license header, copyright Tencent (#6152)

2025-07-08 11:01:29 +08:00

test_softmax.cpp

reduce redundant tests that do not contribute to coverage (#7042)

2026-10-05 08:57:20 +08:00

test_softplus.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_spectrogram.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_split.cpp

support 4d mat in more operators (#6737)

2026-05-22 15:31:01 +08:00

test_spp.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_squeeze.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_squeezenet.cpp

adapt to vkwebgpu and vkopencl (#6876)

2026-08-04 14:37:26 +08:00

test_statisticspooling.cpp

support 4d mat in more operators (#6737)

2026-05-22 15:31:01 +08:00

test_swish.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_tanh.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_threshold.cpp

add fp16s test, skip cpu test for fp16p (#6724)

2026-05-15 10:14:31 +08:00

test_tile_oom.cpp

pnnx ncnn convert tensor expand to tile, fix missing tile repeats (#6399)

2025-11-12 10:56:57 +08:00

test_tile.cpp

vulkan: implement tile (#7027)

2026-10-06 21:27:08 +08:00

test_unaryop.cpp

reduce gpu test cases (#7017)

2026-10-01 23:33:05 +08:00

test_unfold.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_yolodetectionoutput.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

test_yolov3detectionoutput.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

testutil.cpp

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00

testutil.h

layer param hardening (#7003)

2026-09-15 19:18:46 +08:00