laiyongqiang
|
33d1427a14
|
optimize is all nop node detect in mem reuse
|
5 years ago |
laiyongqiang
|
eb37669e3b
|
fix bug to remove reshape when reshape is depend's input
|
5 years ago |
mindspore-ci-bot
|
a663f2066c
|
!2285 [Code Review] code review fix
Merge pull request !2285 from jjfeing/master
|
5 years ago |
jjfeing
|
c26274f324
|
fix code review bug
|
5 years ago |
limingqi107
|
0f4397cece
|
fix all nop node graph execute
|
5 years ago |
mindspore-ci-bot
|
4642df207a
|
!2210 gpu optimize the max device memory config
Merge pull request !2210 from limingqi107/max_device_memory_optimize
|
5 years ago |
limingqi107
|
55b3557c0d
|
gpu optimize the max device memory config
|
5 years ago |
limingqi107
|
20083679a0
|
gpu memreuse supports summary node
|
5 years ago |
limingqi107
|
b83f90a8d8
|
gpu optimize Nop node
|
5 years ago |
mindspore-ci-bot
|
5da171a735
|
!2076 fix summary nodes memory reuse refcount
Merge pull request !2076 from laiyongqiang/summary_refcnt
|
5 years ago |
laiyongqiang
|
09d5a7227e
|
fix summary nodes memory reuse refcount
|
5 years ago |
lizhenyu
|
2d16316c55
|
fix code review warnings
|
5 years ago |
lizhenyu
|
0a55ebf6e9
|
fix code review defects
|
5 years ago |
laiyongqiang
|
79843c3528
|
use VisitKernelWithReturnType instead of VisitKernel to get node's input
|
5 years ago |
mindspore-ci-bot
|
3d28ba10ef
|
!1631 fix reviewbot of mem_reuse
Merge pull request !1631 from yangjie159/fix_reviewbot_of_mem_reuse
|
5 years ago |
lizhenyu
|
9df3a8613c
|
fix get nullptr when use graph manager
|
5 years ago |
yangjie159
|
37904cea01
|
fix reviewBot of mem_reuse
|
5 years ago |
yangjie159
|
cbf5390b34
|
refactor memreuse allocator
remove stream id from membuf
add kernel dependence to determine whether is usable
|
5 years ago |
lizhenyu
|
23a57476da
|
change gpu kernel runtime to support memory swap
|
5 years ago |
lizhenyu
|
d97f849484
|
add memory swap manager module
|
5 years ago |
lizhenyu
|
c5ac2cc38c
|
add memory copy module
|
5 years ago |
lizhenyu
|
af3307ec6e
|
add mem swap module header file
|
5 years ago |
yangjie159
|
72f9fe22d4
|
modify ascend device max memoryGB to 26
fix mem_reuse_allocator ReleaseNodeUnusedOutput
|
6 years ago |
limingqi107
|
d9197b591a
|
gpu optimize the use of reference count
|
6 years ago |
limingqi107
|
63f3a2caac
|
gpu optimize some return values of dynamic memory pool
|
6 years ago |
kswang
|
1e8997f4cb
|
optimize sort for mem reuse and fix memreuse bug
|
6 years ago |
limingqi107
|
2891f0d20d
|
gpu dynamic memory pool supports multi-allReduce
|
6 years ago |
zhoufeng
|
c2b3360d69
|
update clang format rule
|
6 years ago |
gukecai
|
f8208c7c52
|
Support GetNext Parallel
|
6 years ago |
limingqi107
|
99f12f9105
|
gpu uses dynamic memory pool by default
|
6 years ago |
kswang
|
04be6a37f0
|
add getptr for memreuse
|
6 years ago |
zhunaipan
|
930a1fb0a8
|
initial version
Signed-off-by: leonwanghui <leon.wanghui@huawei.com>
|
6 years ago |