171b9d1bba
use spdx license header, copyright Tencent ( #6152 )
2025-07-08 11:01:29 +08:00
9f832c19c1
vulkan int8 packing quantize dequantize requantize ( #3731 )
...
* add int8 definitions
* packing vulkan int8/int32, quantize vulkan
* vulkan dequantize
* requantize vulkan
2025-06-26 10:47:25 +08:00
24a3b99f1f
drop layer support_image_storage and option use_image_storage ( #6126 )
...
* fix pyncnn build
2025-06-16 11:23:41 +08:00
211e238639
drop layer forward vkimagemat ( #6124 )
...
vkimagemat was originally used as a mat storage in the hope of improving performance on old adreno gpus, but in fact it is slower than the cpu in most cases and is no longer suitable for the latest adreno architecture and large shapes
2025-06-14 10:36:47 +08:00
eed257df1f
ci update llvmpipe ( #5954 )
...
* check image fp16
2025-03-28 19:05:40 +08:00
19caca3140
port rvv intrinsic 1.0+ ( #5642 )
...
* zfh zvfh xtheadvector infra
* dispatch for rvv and xtheadvector
* dispatch for non-vector zfh
* port xtheadvector recp rsqrt trunc
* general rvv gemm
* c906 and c910 ci
* old tuple code clean
* update riscv64 ci
* update build doc
* drop old th1520 toolchain
2024-12-04 16:16:17 +08:00
8fe62812c9
arm neon optimization for layernorm fp32/bf16s/fp16s ( #5746 )
2024-10-18 21:32:18 +08:00
3752d71200
fix potential fp16s bf16s conflicts on arm vfpv4 ( #5578 )
...
* fix potential fp16s bf16s conflicts on armv7 vfpv4
* but prefer fp16 on armv8.2
2024-07-12 10:25:37 +08:00
da7d1a10f7
test x86 arm convolution oom ( #5492 )
...
* skip mips loongarch riscv oom test atm
* test softmax oom
2024-06-11 11:06:09 +08:00
08b7d99a75
rnn/lstm/gru dynamic quantization ( #5435 )
2024-05-08 19:25:51 +08:00
984d6dd844
promote vfpv4 for auto fp16 storage conversion ( #5325 )
...
* promote vfpv4 for auto fp16 storage conversion
* always report neon and vfpv4 for arm64
2024-02-02 18:50:15 +08:00
5329d32e74
check vulkan fp16 uniform support and implement lfp conversion without fp16u ( #5287 )
2024-01-17 15:53:34 +08:00
556b79ce4d
create layer decoupled ( #5258 )
...
* create layer decoupled
* no more virtual public
* allow build test with shared library
* decouple cpu vulkan
* drop old scripts
2024-01-05 14:08:15 +08:00
nihui