tanghuikang
|
dbcdcc2daa
|
Clean code
|
4 years ago |
i-robot
|
0376e934a5
|
!30943 PyNative RunOp without Actor
Merge pull request !30943 from caifubi/master-pynative-no-actor
|
4 years ago |
caifubi
|
d26d7b21ff
|
PyNative RunOp without Actor
|
4 years ago |
Parastoo Ashtari
|
c61e56cbe2
|
fix kernel by kernel and mindRT parameter bug
|
4 years ago |
xiao_yao1994
|
13b400499f
|
fix unify backend binary bug
|
4 years ago |
lby
|
2b1b539b36
|
parse core type from pre-build for tbe kernel compile
|
4 years ago |
ZPaC
|
04ee8f4dd9
|
Support reference for gpu and cpu for parameter update
|
4 years ago |
huoxinyou
|
6d974ca877
|
修复bngrad动态shape输出infer不正确问题 修复多个连续NopNode在动态shape场景下,infer问题
|
4 years ago |
zhoufeng
|
f49b195c39
|
extract common as an independent shared library
Signed-off-by: zhoufeng <zhoufeng54@huawei.com>
|
4 years ago |
i-robot
|
6c8f942737
|
!30544 fix codecheck
Merge pull request !30544 from xulei/fix_codecheck
|
4 years ago |
i-robot
|
5227cb1dff
|
!30527 【PS】fix process can not exit normaly when exception has been thrown
Merge pull request !30527 from zyli2020/master
|
4 years ago |
xulei
|
4cf320cdbd
|
fix codecheck
|
4 years ago |
lizhenyu
|
f378fd3302
|
fix process can not exit normaly when exception has been thrown
|
4 years ago |
kswang
|
f26870d437
|
clean code
|
4 years ago |
ttudu
|
451ebd1bd1
|
dynamic_shape
|
4 years ago |
i-robot
|
71f7173f9b
|
!30119 Memory alignment for GPU AllReduce
Merge pull request !30119 from caifubi/master-allreduce-memory-align
|
4 years ago |
limingqi107
|
6f1790ef77
|
optimize the host and device copy in the control flow
|
4 years ago |
marui
|
d47e79b04c
|
Refactor ccsrc directories and CMakeLists files
|
4 years ago |
caifubi
|
1f3fe4cd2b
|
Fix AllReduce memory align size
|
4 years ago |
LaiYongqiang
|
531fa9ca42
|
use device HBM free memory to calculate MindSpore manage memory size
|
4 years ago |
ttudu
|
e2d1fb515a
|
fix event error in mindrt
|
4 years ago |
i-robot
|
a8f2dba3fa
|
!29536 Support cell dump for Ascend kernel-by-kernel dump
Merge pull request !29536 from parastooashtari/cell-kernel-by-kernel-dump
|
4 years ago |
i-robot
|
48f4fd2f52
|
!29639 build core.so
Merge pull request !29639 from zhoufeng/dynamic-core
|
4 years ago |
i-robot
|
9161257f9b
|
!29164 Dynamic Shape for BatchNorm and BatchNormGrad
Merge pull request !29164 from zhengzuohe/BatchNormRebase
|
4 years ago |
i-robot
|
776d937266
|
!29625 Add GPU matrix_diag_part ops, optimize related testcase
Merge pull request !29625 from wuwenbing/dev
|
4 years ago |
zhoufeng
|
8e97ce5cd2
|
build core.so
Signed-off-by: zhoufeng <zhoufeng54@huawei.com>
|
4 years ago |
wenbean
|
b9872dd961
|
Add GPU matrix_diag_part ops, optimize related testcase
|
4 years ago |
i-robot
|
92602761e8
|
!29604 unified runtime support sub graph parameter output in the control flow
Merge pull request !29604 from limingqi107/new_actor_runtime
|
4 years ago |
huoxinyou
|
87fcafcf40
|
Fix the datatype bug of tiling_key in ai_core_dynamic_kernel
BatchNorm (BNTrainingReduce, BNTrainingUpdate, bn_split)
BatchNormGrad (BNTrainingUpdateGrad, BNTrainingReduceGrad, bn_grad_split)
|
4 years ago |
limingqi107
|
e4c0519263
|
unified runtime support sub graph parameter output in the control flow
|
4 years ago |
tanghuikang
|
e939ae611a
|
Swap temporary when mem on device not enough
|
4 years ago |
Parastoo Ashtari
|
bc228d4d58
|
support cell dump for ascend kernel-by-kernel mode
|
4 years ago |
i-robot
|
f2466fbff2
|
!29443 update ascend stream assign && add PROF log info
Merge pull request !29443 from lyqlola/master
|
4 years ago |
liyiqi
|
adb33a15b5
|
update ascend stream assign && add PROF log info
|
4 years ago |
i-robot
|
ace1e89caf
|
!29510 Fix AllReduce bug
Merge pull request !29510 from caifubi/master-fix-pre-assign-cnode-memory
|
4 years ago |
i-robot
|
e3a5051e94
|
!29166 Dynamic shape for SplitD
Merge pull request !29166 from zhengzuohe/SplitD
|
4 years ago |
wYann
|
39e89f73ac
|
support dynamic datasink on GPU
|
4 years ago |
i-robot
|
1192616ac2
|
!29250 custom op support julia
Merge pull request !29250 from r1chardf1d0/master
|
4 years ago |
caifubi
|
b616b28e50
|
Launch AllReduce failed
|
4 years ago |
i-robot
|
c3139a605b
|
!29424 unfied runtime support sub graph share root graph parameters in contorl flow
Merge pull request !29424 from limingqi107/new_actor_runtime
|
4 years ago |
r1chardf1d0
|
044e110f7a
|
custom op support julia
|
4 years ago |
limingqi107
|
567c667517
|
unfied runtime support sub graph share root graph parameters in contorl flow
|
4 years ago |
i-robot
|
3ed062fcfa
|
!28983 ms_function needs to be compiled and executed in graph
Merge pull request !28983 from caifubi/master-pynative-add-bprop-flag
|
4 years ago |
caifubi
|
1e48cf1742
|
Pynative ms_function mix execution
1. Add bprop flag to bp-graph in PyNative mode.
2. Run ms_function in actor-DAG.
3. Use aclMemcopyAsync in DtoD copy.
4. Insert event before RunTask.
|
4 years ago |
TronZhang
|
deac002bed
|
refactor kernel mod class and subclass
|
4 years ago |
caifubi
|
9bfdcedaef
|
Wait device task finish when python exit
|
4 years ago |
Parastoo Ashtari
|
c6f5fb06f2
|
add comments for dump and debugger code.
|
4 years ago |
zhengzuohe
|
b9dc3cad2b
|
Dynamic shape for SplitD
|
4 years ago |
i-robot
|
5e7a38d1e2
|
!29083 fix bug of data-parallel mix-precision in PyNative
Merge pull request !29083 from caifubi/master-pynative-data-parallel
|
4 years ago |
i-robot
|
68189870e4
|
!29109 Fix Core Dump after RuntimeError
Merge pull request !29109 from hwjiaorui/segmentation-fault
|
4 years ago |