Commit Graph

Select branches

Hide Pull Requests

Conv_NHWC

NNET_231111

NNET_231111_from_master

NNET_GAN

NNET_OpSearch

NNET_anyOp

NNET_bias

NNET_bias_0630

NNET_e2e

NNET_e2e_fix

NNET_e2e_for_merge

NNET_eliminateOP

NNET_gcn

NNET_gcn_fuse

NNET_op_test

NNET_transpose

TC_revision

activation

add_paddle_model

allocator_memPool

ascend

bang-softmax

benchmark

benchmark_conv

benchmark_softmax

broadcast_wanghailu_0916

case-fsrcnn

change_path

constroy/doc_on_ares

conv_half

cpu_backend2

cuda-attention

cuda-transpose

dcj/for_multiple_datatype

ddp

dev-dynamic-graph

dev-dynamic-graph-allocator

dev-leakyrelu

dev-memory

dist/graph

dist_bench

dropout

dump/init

fsrcnn-conv-bias-act-fuse

gencode

gpt

graph-onnx

graphFactory

instance_norm

kunlun_temp

kvcache_attention_fp16

kvcache_backup

master

model_test

nnet_e2e_for_merge

op_timer

optimization-pass

point2point

power-fusion

pure_engine

search_engine

support_fp16

testAccuracy

test_codegen

test_onnx

train_wanghailu_1010

update_doc

update_pybind11

v0630

xpu_allreduce

test-models

Mono Color

• cc62a3216d feat: 整理子项目结构，实现一部分导入 YdrMaster 2023-04-20 14:47:05 +0800
• 9cfe223953 refactor: 重命名 optimization，以后可以分离为 submodule YdrMaster 2023-04-19 11:15:31 +0800
• 34f7d7e9ed feat: 图变换相关类分到单独文件中，并模板化。 YdrMaster 2023-04-14 10:57:52 +0800
• 59a46f3ff9 feat: Rating 评分对分归一化 YdrMaster 2023-04-11 10:02:00 +0800
• 72788e8e0a feat: 实现子图合并 YdrMaster 2023-04-11 09:42:10 +0800
• bd61cf4533 docs: 补充文档 YdrMaster 2023-04-10 10:29:35 +0800
• f0f8915433 docs: 补充文档 YdrMaster 2023-04-07 17:30:25 +0800
• bb5bfb0be8 docs YdrMaster 2023-04-07 16:01:45 +0800
• a56e86dfa9 feat: 提供子图划分、突变生成和评分的类型 YdrMaster 2023-04-07 15:04:24 +0800
• 5129d312d2 feat: 重写 YdrMaster 2023-04-07 11:33:01 +0800
• 45f7e891f1 feat: 增加 io_id 以支持子图划分 YdrMaster 2023-04-06 09:38:53 +0800
• b4b5157bd4 feat: 增加更多示例 YdrMaster 2023-04-04 17:17:20 +0800
• 63cc93aadc feat: 将数据和张量分离，单独保存 YdrMaster 2023-04-04 16:45:16 +0800
• ddaf6685b3 fix: compile YdrMaster 2023-04-04 15:32:41 +0800
• 385586d57b docs: 补充文档 YdrMaster 2023-04-04 13:49:14 +0800
• 320468b627 try: 提出另一种图表示 YdrMaster 2023-04-04 10:51:01 +0800
• 8f38a41fb6 fix: compile YdrMaster 2023-04-03 14:11:01 +0800
• a6a0141234 feat: 整理每个子图的突变和评价算法 YdrMaster 2023-03-30 14:52:23 +0800
• 36b0c5855c feat: 添加 PassManager 的运行 YdrMaster 2023-03-30 11:02:42 +0800
• 0ad0150b87 feat: 添加 Pass 的构造 YdrMaster 2023-03-30 10:50:40 +0800
• 334e0cccbc feat: 添加互换图排序依据的实现 YdrMaster 2023-03-30 10:07:00 +0800
• cc6c18b00f feat: 调整优化接口 YdrMaster 2023-03-29 16:14:29 +0800
• d9da06eb67 init optimization-pass whjthu 2023-03-28 21:32:04 +0800
• 55c69978ef fix bug of unary_list_kernel dev-memory wanghailu 2023-08-03 15:38:21 +0800
• e3da455aba Merge branch 'dev-memory' of github.com:InfiniTensor/InfiniTensor into dev-memory wanghailu 2023-08-03 15:36:17 +0800
• 9b10a74788
  支持fp16 dtype (#96) zhangyunze 2023-08-02 16:38:16 +0800
• 0f63e1bb0f update cuda_conv_fp16 conv_half zhangyue207 2023-07-27 09:28:27 +0800
• 1b51bc54bb Merge branch 'dtype_assign' into conv_half zhangyue207 2023-07-27 09:21:12 +0800
• e936ee2850 fix: 按照review意见修改 zhangyunze 2023-07-24 16:58:38 +0800
• 9f21ed3ea1 fix: format probs zhangyue207 2023-07-24 15:20:52 +0800
• 90752cdf0c add: conv_fp16 zhangyue207 2023-07-24 14:33:24 +0800
• 23e16fbf7d add: conv_fp16 zhangyue207 2023-07-24 14:30:58 +0800
• c4aec6c38b fix: set default clang-format to 14 version zhangyunze 2023-07-20 13:52:16 +0800
• 8675db68ac feat: support Float16 dtype zhangyunze 2023-07-20 11:13:09 +0800
• 1d6b6157ce dcj: replace "DataType::Float32" with "op->getDType()" to support more DataType kilinchange 2023-07-17 09:52:52 +0800
• ec22ed80ab Conv Kernel FP16 zhangyue207 2023-07-19 10:06:47 +0800
• 83d378d1c5 add conv_half kernel zhangyue207 2023-07-18 16:55:00 +0800
• 951704d262 fmt YdrMaster 2023-07-21 13:04:54 +0800
• 429b04a744 fix bang_unary_fusion_kernl wanghailu 2023-07-21 10:54:33 +0800
• 9f6343de7a fmt YdrMaster 2023-07-20 16:02:12 +0800
• 2781939955 add bang unary cache kernel wanghailu 2023-07-20 15:25:48 +0800
• 47b2c32bfe feat: 添加内存拷贝的属性定义 YdrMaster 2023-07-20 15:29:06 +0800
• 2d66d5941d feat: 为 MemoryGraph 实现 Operator YdrMaster 2023-07-20 14:14:29 +0800
• c0d6ea5440 feat: 添加构图用的方法 YdrMaster 2023-07-20 13:34:29 +0800
• b8dabe71d3 feat: 添加向访存图增加输入输出的功能 YdrMaster 2023-07-20 11:25:48 +0800
• ac9f2a0b15 feat: 设计访存图定义 YdrMaster 2023-07-20 10:20:50 +0800
• f85bbf7625 build: 实现格式化 git added c/c++ 源码的脚本 YdrMaster 2023-07-21 10:27:52 +0800
• 1dc65e2788
  build: 实现格式化 git added c/c++ 源码的脚本 (#98) Derui Yang 2023-07-21 12:29:50 +0800
• 0fa4e8efe1 fix bang_unary_fusion_kernl wanghailu 2023-07-21 10:54:33 +0800
• 608f997042 Add comments in conv benchmark benchmark_conv bolun 2023-07-20 16:57:27 +0800
• 97970c5d94 Add benchmark for conv bolun 2023-07-20 16:49:14 +0800
• 93778f212e fmt YdrMaster 2023-07-20 16:02:12 +0800
• a3d8ce36a0 add bang unary cache kernel wanghailu 2023-07-20 15:25:48 +0800
• a997d0c693 feat: 添加内存拷贝的属性定义 YdrMaster 2023-07-20 15:29:06 +0800
• a84fb0463b feat: 为 MemoryGraph 实现 Operator YdrMaster 2023-07-20 14:14:29 +0800
• 4b762cc8d9 Merge branch 'benchmark_softmax' of github.com:InfiniTensor/InfiniTensor into benchmark_conv bolun 2023-07-20 14:15:37 +0800
• 1c55c74151 Add comments benchmark_softmax bolun 2023-07-20 14:14:06 +0800
• ddaddf375e Init benchmark for conv bolun 2023-07-20 14:03:05 +0800
• 440df742f3 feat: 添加构图用的方法 YdrMaster 2023-07-20 13:34:29 +0800
• 9117a0fcc8 feat: 添加向访存图增加输入输出的功能 YdrMaster 2023-07-20 11:25:48 +0800
• 7945693131 Add: benchmark for cudnn softmax bolun 2023-07-20 11:03:07 +0800
• f9ca0273a3 feat: 设计访存图定义 YdrMaster 2023-07-20 10:20:50 +0800
• fdb2d30868 fix: Makefile benchmark bolun 2023-07-19 11:29:15 +0800
• f532784d4f update CMakeList.txt Makefile for micro benchmark zhangyue207 2023-07-19 11:10:18 +0800
• 454b7651a8 update CMakeList.txt Makefile for micro benchmark zhangyue207 2023-07-19 10:55:51 +0800
• 48322dbf27 Conv Kernel FP16 zhangyue207 2023-07-19 10:06:47 +0800
• 523946cb8b add conv_half kernel zhangyue207 2023-07-18 16:55:00 +0800
• 898c7b3a0a fix formatting dcj/for_multiple_datatype kilinchange 2023-07-17 14:16:53 +0800
• 5c2a32dbbc dcj: replace "DataType::Float32" with "op->getDType()" to support more DataType kilinchange 2023-07-17 09:52:52 +0800
• f7de8113e0
  fix: 修正 README.md (#93) YdrMaster 2023-07-11 10:03:38 +0800
• 7023454e32
  Update docs (#92) YdrMaster 2023-07-10 02:31:45 +0800
• ab74b6a321
  Update doc 0627 (#89) Hardy 2023-07-06 16:57:10 +0800
• 10f2dd2f28 add how to use docker Conv_NHWC wanghailu 2023-07-04 16:29:46 +0800
• 27053db478 update test report wanghailu 2023-06-30 16:30:45 +0800
• c8f6290c03 fix code wanghailu 2023-06-30 14:56:18 +0800
• 1a925ba5c7 fix: 修改 clip kernel YdrMaster 2023-06-30 14:26:56 +0800
• d6338feb1b fix: 修改 clip kernel YdrMaster 2023-06-30 14:19:28 +0800
• 1ec2fd09e5 fix: 修改 clip 接口 YdrMaster 2023-06-30 14:16:09 +0800
• 3141676b18 add test report markdown wanghailu 2023-06-26 14:32:35 +0800
• b59484562a clang format wanghailu 2023-06-27 13:32:47 +0800
• 26f418713d fix some code of cast and elementwise wanghailu 2023-06-27 11:18:36 +0800
• 0185b54600 fix some code of mlu operator wanghailu 2023-06-27 11:08:15 +0800
• f41e9c9ed2 fix some operator code wanghailu 2023-06-27 10:26:08 +0800
• 3c1359b166 fix code wanghailu 2023-06-27 15:57:33 +0800
• 3caeeabacc remove the build Dependencies of test-cpp, avoid twice build wanghailu 2023-06-27 15:43:27 +0800
• d423480c8f delete README_CN.md. because the file has been split into INSTALL_GUIDE_CN.md and USER_GUIDE_CN.md at 2023.06.23 wanghailu 2023-06-27 15:35:32 +0800
• 2539c3efd9 update doc of mlu wanghailu 2023-06-27 15:29:28 +0800
• 75a9605fa9
  Merge branch 'pure_engine' into Conv_NHWC Hardy 2023-06-30 13:34:19 +0800
• c54499456d add reduce operator on mlu wanghailu 2023-06-30 11:21:37 +0800
• 9ffa70e5a0 style: 整理 reduce 变换代码 YdrMaster 2023-06-30 09:56:11 +0800
• 6af8fc573b clang format wanghailu 2023-06-29 17:34:56 +0800
• 5fe48f49ea fix reducemean on mlu wanghailu 2023-06-29 17:19:43 +0800
• b8d8241e36 feat: 支持编译时转换任何张量 YdrMaster 2023-06-29 16:00:08 +0800
• fab08fcfa1
  Constroy/fix reducemean (#91) Haojie Wang 2023-06-29 08:27:00 +0800
• 579cdbbb81
  fix ReduceMean and element_wise (#90) constroy 2023-06-29 07:15:07 +0800
• 4340522720 fix batchnorm bug wanghailu 2023-06-28 17:25:26 +0800
• 829c4bfe96 fix conv operator wanghailu 2023-06-28 13:24:47 +0800
• 0b1c4e4681 feat: 添加不变换一些张量 YdrMaster 2023-06-28 11:14:26 +0800
• 1ee4a60af0 Add: convert expression to operator NNET_bias_0630 NNET_bias Liyan Zheng 2023-06-28 11:06:17 +0800
• ac624c2279 fix conv bug wanghailu 2023-06-28 10:27:04 +0800