Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
320 commits
Select commit Hold shift + click to select a range
d1a1105
Update test_cross_entropy_loss.py
HydrogenSulfate Aug 15, 2021
23cc214
Update test_cross_entropy_loss.py
HydrogenSulfate Aug 15, 2021
0bf3248
Update test_cross_entropy_loss.py
HydrogenSulfate Aug 15, 2021
f2df33e
Update loss.py
HydrogenSulfate Aug 15, 2021
f6dc4b6
Update test_cross_entropy_loss.py
HydrogenSulfate Aug 15, 2021
7afd7f3
Update test_cross_entropy_loss.py
HydrogenSulfate Aug 15, 2021
ee070fb
Update test_cross_entropy_loss.py
HydrogenSulfate Aug 16, 2021
0046768
Update loss.py
HydrogenSulfate Aug 16, 2021
52804cd
Update loss.py
HydrogenSulfate Aug 16, 2021
0c2d6bc
Update loss.py
HydrogenSulfate Aug 16, 2021
11e9d4e
Update loss.py
HydrogenSulfate Aug 16, 2021
cf6e543
Update loss.py
HydrogenSulfate Aug 18, 2021
e838cac
Update test_cross_entropy_loss.py
HydrogenSulfate Aug 26, 2021
ad52248
fix the crash when input variable is bool type, test=develop (#35176)
winter-wang Aug 27, 2021
2006fbc
gelu/logsigmoid add AsExtra (#35198)
zhupengyang Aug 27, 2021
7272526
fix count_api_without_core_ops (#35170)
wanghuancoder Aug 27, 2021
48bf7cb
Polish DeviceEvent interface and Remove #ifdef in InterpreterCore (#3…
Aurelius84 Aug 27, 2021
7debae3
Add fusion_gru and multi_gru to PTQ (Post-Training Quantization) (#33…
wozna Aug 27, 2021
1533d7e
[hybrid] Fix row parallel linear bias (#35186)
wangxicoding Aug 27, 2021
234ce93
sparse_momentum_op is used to save w@GRAD memory for gather_op (#34942)
GuoxiaWang Aug 27, 2021
ceee71a
Add unpool2d op & Expose max_unpool2d API (#35056)
tink2123 Aug 27, 2021
5a72cf4
add more models for model_benchmark_ci,test=document_fix (#35178)
xiegegege Aug 27, 2021
be29b8e
add uniform_ op and UT (#33934)
zoooo0820 Aug 27, 2021
5dcff7c
test=document_fix (#35222)
GuoxiaWang Aug 27, 2021
31cd106
test=document_fix (#35221)
GuoxiaWang Aug 29, 2021
20cfa8b
Abstract GenerateDeviceEventFlag to shield platforms (#35219)
Aurelius84 Aug 30, 2021
8f94d34
Add cpu/gpu for PR-CI-CPU-Py2 (#35174)
tianshuo78520a Aug 30, 2021
37d281c
Set value (#34886)
2742195759 Aug 30, 2021
c0bdef5
del message;test=document_fix (#35248)
tianshuo78520a Aug 30, 2021
0043fa8
[paddle-TRT]support matmul set to int8 in multihead (#34917)
ceci3 Aug 30, 2021
6ff179a
[Op Def] Add extra def of linear_interp & linear_interp_v2 & addmm (#…
Ray2020BD Aug 30, 2021
ca4d2fc
- candidate fix (#35231)
jczaja Aug 30, 2021
e864667
fix using boost::none as the init value when using paddle::optional (…
MingMingShangTian Aug 30, 2021
b94d7ff
[NPU] Add log_loss op (#35010)
Ray2020BD Aug 30, 2021
3956514
[Paddle Inference-TRT]Adding six op unittest codes of TRT INT8 (#35130)
xiaoxiaohehe001 Aug 30, 2021
531a890
change exit code and polish infer_ut summary style (#35254)
OliverLPH Aug 31, 2021
aaaa996
NPU add fill_zeros_like kernel (#35246)
Aganlengzi Aug 31, 2021
561841d
NPU add elementwise_mod (#35245)
Aganlengzi Aug 31, 2021
ef53625
support fuse layers for ptq (#35015)
XGZhang11 Aug 31, 2021
54f0701
fix the pass compat check position error, test=develop (#35272)
winter-wang Aug 31, 2021
2c0d667
fix windows batch file error:The system cannot find the batch label s…
zhwesky2010 Aug 31, 2021
2100816
Add AsExtra() for conditional_block_op.h (#35268)
zhhsplendid Aug 31, 2021
3d76d00
fix CI skip cc test error (#35264)
Avin0323 Aug 31, 2021
a2afcac
add trt error information. (#35277)
b3602sss Aug 31, 2021
96e7d90
update infer trt ut. (#35261)
jiweibo Aug 31, 2021
d9f59fd
Put code style check on gpu_ci (#35309)
tianshuo78520a Aug 31, 2021
2f3b393
New whl release strategy with pruned nv_fatbin (#35239)
jim19930609 Aug 31, 2021
00c9aeb
fix bug that cmake find python (#35304)
zhwesky2010 Aug 31, 2021
6116f9a
Revert "Revert "Add copy from tensor (#34406)" (#35173)" (#35256)
shangzhizhou Aug 31, 2021
f6004ab
[NPU] fix cmake for ascend ci, test=develop (#35255)
qili93 Aug 31, 2021
079c585
[Dy2Stat]Add model ResNet50 for Dy2stat AMP training (#35276)
Aurelius84 Aug 31, 2021
e299155
transformer opt python files (#35206)
xingfeng01 Aug 31, 2021
572bad8
Support CostInfo and MemProfiler in InterpreterCore (#34981)
Aurelius84 Aug 31, 2021
884011a
reverse xpu adamw to the combination of ops version. (#35286)
zhaoyinglia Sep 1, 2021
d387820
Support settiem by Bool index (#35133)
zyfncg Sep 1, 2021
3b52f68
[Dy2Stat]Support append method and initialized value for List in Cont…
Aurelius84 Sep 1, 2021
d9afa83
label prelu op (#35315)
juncaipeng Sep 1, 2021
12df57f
add ElementwiseTernary, Reduce, ReadDataStride (#35075)
AnnaTrainingG Sep 1, 2021
5c27c2c
[NPU]shard index op for npu (#35281)
sljlp Sep 1, 2021
5baccfd
add support ops for quantization (#35312)
juncaipeng Sep 1, 2021
7ca28bb
support KL label smooth (#35177)
QingshuChen Sep 1, 2021
3c21f26
Stablize depthwise conv (#35161)
jerrywgz Sep 1, 2021
b53887f
fix bug:When axes in paddle.slice is a tuple, an error occurs. (#35267)
hbwx24 Sep 1, 2021
b24f84c
[Dy2stat]modify dy2stat error message in compile time (#35320)
0x45f Sep 1, 2021
7f17f9a
bugfix for mp accuracy (#35326)
JZ-LIANG Sep 1, 2021
5eefc8c
[NPU] skip NPU UT if no npu files changed, test=develop (#35338)
qili93 Sep 1, 2021
4f54891
add input and output description docs for vision transform (#34926)
LielinJiang Sep 1, 2021
5fa7d9c
support setting linewidth when printing tensor (#35175)
zhiqiu Sep 1, 2021
7743cdf
add strided_slice_grad op for npu (#35204)
baoachun Sep 1, 2021
c56d697
modify fetch logic, use D2H Stream (#35191)
wanghuancoder Sep 1, 2021
bee511d
[NPU] set constant before copy data (#35335)
FeixLiu Sep 1, 2021
264ff9e
[HybridParallel]Support finetinue model for PipelineParallel (#35287)
ForFishes Sep 1, 2021
a647b80
[HeterPs] merge dense && data norm && g2sum (#35029)
Thunderbrook Sep 1, 2021
070cab1
Added slice BF16/FP32 FWD/BWD kernels (#34332)
jakpiase Sep 1, 2021
df57df9
add AsExtra for grid_sampler_op (#35339)
LielinJiang Sep 1, 2021
280d742
[npu] add update_loss_scaling npu min value (#35270)
Baibaifan Sep 2, 2021
a622b70
[Auto Parallel] Logical Partition & Dist Op (#35117)
JZ-LIANG Sep 2, 2021
25871e0
add axis check for elementwise op while the dimension of x is equal t…
wangxinxin08 Sep 2, 2021
b28cc73
fix static error in summary (#35303)
wangna11BD Sep 2, 2021
8525dd1
add npu code not exec linux/windows cases (#35363)
lelelelelez Sep 2, 2021
6e638d7
Refactor transpose cuda kernel impl. (#35308)
limin2021 Sep 2, 2021
67ed7e1
[hybrid] [npu] fit npu nan/inf check (#35171)
FeixLiu Sep 2, 2021
e57a88b
[NPU] Add label_smooth_op (#34828)
Ray2020BD Sep 2, 2021
7e5fb46
Add SVD Op and it's GPU and CPU kernel (#34953)
2742195759 Sep 2, 2021
bb63396
[NPU] Support npu kernel for gather_nd op (#34800)
JZZ-NOTE Sep 2, 2021
ba6a312
add log_softmax_op_npu (#35006)
juneweng Sep 3, 2021
648e377
add AsExtra() mark for layer_norm (#35415)
zoooo0820 Sep 3, 2021
e913796
[NPU] Add elementwise_pow_grad npu op (#35278)
wjj19950828 Sep 3, 2021
668bfb3
[NPU] add 32 extra bytes for npu memory slot (#35347)
zhiqiu Sep 3, 2021
4fe1bb4
fix mean/variance when is_test=True (#35328)
ceci3 Sep 3, 2021
fc8d46c
add lookup_table_v2_op AsExtra (#35265)
juneweng Sep 3, 2021
ef7bc36
Gather op (#35353)
juneweng Sep 3, 2021
a6cc567
[Dy2Stat]Modify dy2stat error message in runtime and format error mes…
0x45f Sep 3, 2021
ccd42db
fix flatten infershape (#35321)
danleifeng Sep 3, 2021
e77fd2e
tensor formatter (#35399)
Thunderbrook Sep 3, 2021
cec1abc
[NPU] update npu ut skil scirpts, test=develop (#35360)
qili93 Sep 3, 2021
0712611
fix a quantization bug (#35407)
XGZhang11 Sep 3, 2021
6cdc1dc
add AsExtra to the reshape op (#35358)
zmxdream Sep 3, 2021
42d3650
add AsExtra to concat op (#35380)
zmxdream Sep 3, 2021
8ba58eb
add AsExtra to partial_sum op (#35381)
zmxdream Sep 3, 2021
8305ba3
fix bn_infer and optimize momentum for kunlun (#35250)
tangzhiyi11 Sep 3, 2021
a9dfebb
[NPU]add conv2d_transpose npu op (#35232)
Sep 3, 2021
c171eca
Unify the implementation of AlignedVector and simplify the codes of d…
Xreki Sep 3, 2021
b333dac
[iscan] bugfix: DLTP-33615 / DLTP-33953 / DLTP-33968 / DLTP-34166 (#3…
WorgenZhang Sep 3, 2021
f13dcfb
Add AsExtra for transpose, lstm, gru (#35317)
joey12300 Sep 3, 2021
f014e30
[NPU] add int64_t kernels for YoloV3, test=develop (#35045)
qili93 Sep 3, 2021
4e67cd1
[NPU] Add huber_loss op (#34826)
Ray2020BD Sep 3, 2021
b6adfd9
support lodtensorarray for send/recv (#35279)
Sep 3, 2021
80c0cc9
modify gc logic, use new device_event (#35208)
wanghuancoder Sep 3, 2021
e8a8816
disable test_standalone_executor temporarily (#35436)
Aurelius84 Sep 3, 2021
e877248
update inference trt ut framework (#35418)
jiweibo Sep 4, 2021
ffc3d36
[WIP] paddle.where api add broadcast, when x_shape == y_shape, and x_…
windstamp Sep 5, 2021
18934c5
update trt ut. (#35458)
jiweibo Sep 6, 2021
c3ad777
Revert hccl check nan (#35438)
FeixLiu Sep 6, 2021
5675042
replase pass with error exception (#35367)
xingfeng01 Sep 6, 2021
60c5ada
support numpy dtype and polish code of list index. (#35404)
hbwx24 Sep 6, 2021
4bea0ff
[NPU]add depthwise_conv_npu_grad op (#35374)
Sep 6, 2021
c2f76b0
add AsExtra tag for conv transpose op (#35354)
wangxinxin08 Sep 6, 2021
8c73c1b
Support Reset for DeviceEvent (#35443)
Aurelius84 Sep 6, 2021
13bbb6b
add kernel, stride check (#35106)
LDOUBLEV Sep 6, 2021
295253a
add pool2d pool3d extra() (#35393)
LDOUBLEV Sep 6, 2021
70a9b65
transpose/slice/stride_slice/squeeze/unsqueeze op_def_enhance-1.0 (#3…
Sep 6, 2021
97798f9
Add grad grad for AvgPool2D (#35388)
weisy11 Sep 6, 2021
7ef04da
Add fusion_lstm INT8 PTQ (#35334)
wozna Sep 6, 2021
49797d8
Add the extra flag for the some ops (#35442)
wawltor Sep 6, 2021
266fcbe
support double in deformable conv (#35330)
jerrywgz Sep 6, 2021
28b6407
[HIP] fix op not support AMD GPU bug, the flag PADDLE_WITH_ROCM is in…
xymyeah Sep 7, 2021
2b1efc3
transfer the static.accurcay to v2 op (#35494)
wawltor Sep 7, 2021
3c8eeb5
[Dy2Stat]Open test_resnet_amp on Windows (#35323)
Aurelius84 Sep 7, 2021
5bb1285
support test different infer_ut suite type (#35435)
OliverLPH Sep 7, 2021
1445103
merge from latest develop branch, test=document_fix (#34995)
betterpig Sep 7, 2021
071e815
Fix DryRun unittest failed from test_standalon_executor.py (#35433)
Aurelius84 Sep 7, 2021
7907e24
add AsExtra in data_norm op (#35420)
jeff41404 Sep 7, 2021
0c71edc
operators/flatten_op.cc add AsExtra (#35471)
dyning Sep 7, 2021
ed97be0
fix int8 (#35504)
ceci3 Sep 7, 2021
36cdb6e
Fix for reshape2 oneDNN op (#35455)
jakpiase Sep 7, 2021
e9ae8dd
[oneDNN] Disable cache matmul v1 & refactoring (#35331)
jczaja Sep 7, 2021
e928274
[NPU] log_softmax_grad, test=develop (#35484)
qili93 Sep 7, 2021
cf40894
[NPU] Add norm_grad kernel (#35237)
windstamp Sep 7, 2021
f57a240
disable added ut check,test=document_fix (#35535)
XieYunshen Sep 7, 2021
cec36ea
Add DPADDLE_WITH_CUDA for GCC (#35448)
Aurelius84 Sep 7, 2021
d47a97d
fix trace op stack overflow (#35419)
jeff41404 Sep 7, 2021
cc6d2b0
[NPU] update batch norm op, test=develop (#35223)
qili93 Sep 7, 2021
b211f02
add as-extra for softplus/leaky_relu/softmax (#35493)
Sep 7, 2021
eae4bf5
Modify the elementwise op according to the kernel primitive API (#34…
AnnaTrainingG Sep 7, 2021
8307b0c
add conv op check for illegal input or attributes (#35337)
wangxinxin08 Sep 7, 2021
c6e0ced
support multi-node (#35396)
yaoxuefeng6 Sep 7, 2021
2b10521
Add depth while cloning benchmark code (#35548)
xiegegege Sep 7, 2021
1635c02
Fix scatter_nd_add doc (#35542)
sneaxiy Sep 7, 2021
c727ec4
[NPU] add get_float_status op and refine NPU check_nan_inf (#35274)
wangxicoding Sep 7, 2021
b3787d1
add the matmul v2 grad kernel
wawltor Sep 8, 2021
a53460a
Work queue group (#35470)
liutiexing Sep 8, 2021
3dab2e2
Add op define extra for norm and frobenius norm op. (#35329)
ZHUI Sep 8, 2021
db6242e
[NPU] release gil before op run (#35370)
zhiqiu Sep 8, 2021
18a963a
add some file for jetson-op-test-CE (#35431)
Sep 8, 2021
5199c74
support weight sharing for pipeline (#35351)
Sep 8, 2021
c4a3e8b
merge CMakeList.txt manual (#35378)
Sep 8, 2021
39540b0
add checkers for auto parallel apis (#35486)
Sep 8, 2021
afd1b37
hidden the auto parallel apis (#35385)
Sep 8, 2021
4e62af8
Add FP16 PRelu (#35532)
juncaipeng Sep 8, 2021
ad5f749
fix the bug of layer_norm when batch_size=1 (#35480)
zkh2016 Sep 8, 2021
5f36988
Enable program passes on Fleet APIs (#34955)
sneaxiy Sep 8, 2021
3c457a3
Fix scatter_nd_add and gather bug (#35544)
sneaxiy Sep 8, 2021
e133d8e
fix bug (#35482)
GuoxiaWang Sep 8, 2021
692ac3e
Add AsExtra in relu6 Op Maker (#35472)
hong19860320 Sep 8, 2021
51cc73f
Intergrate GLOOParallelContext to support Multi-CPU Core for Dygraph …
2742195759 Sep 8, 2021
28abd5d
Slice bug (#35357)
shangzhizhou Sep 8, 2021
7aa4d87
add clip_by_norm fp16 kernel (#35446)
zhiqiu Sep 8, 2021
82b33be
Modify the reduce op according to the kernel primitive api (#35282)
AnnaTrainingG Sep 8, 2021
ce7c18f
mark WhileOp AsExtra attribute (#35499)
CtfGo Sep 8, 2021
0eb7c94
refactor new executor (#35537)
wanghuancoder Sep 8, 2021
b95c5ae
modified pool_op for higher performance (#33144)
AnnaTrainingG Sep 8, 2021
a2dbb0c
[hybrid] check pipeline persist var which changed in forward and used…
wangxicoding Sep 8, 2021
db5fd2a
multiply supports bool
will-jl944 Sep 8, 2021
1215535
[Auto Parallel] Integrate all modules (#35483)
aoyulong Sep 8, 2021
2133f3d
add API Tensor.T for reverse dim of Tensor (#35379)
zhwesky2010 Sep 8, 2021
1159f75
modify unittest parallel rule to avoid UT failure (#35567)
zhwesky2010 Sep 8, 2021
ef61da8
Refactor softmax_cudnn kernel impl for code reuse. (#35350)
limin2021 Sep 8, 2021
abe70d3
Change depth as 1 when cloning benchmark code,test=document_fix (#35590)
xiegegege Sep 8, 2021
0cb413d
add backward inplace for dygraph (#35412)
zhiqiu Sep 8, 2021
eb1fbf1
Add matrix_rank Op and it's GPU and CPU kernel (#34823)
0x45f Sep 9, 2021
cf8bf03
add a fusion op: fused_residual_dropout_bias (#34963)
zkh2016 Sep 9, 2021
16a2fda
test=document_fix (#35592)
tianshuo78520a Sep 9, 2021
92810e6
test=document_fix (#35606)
lelelelelez Sep 9, 2021
7fcb9e3
Add extra flags for attr of affine_grid_op (#35581)
wanghaoshuang Sep 9, 2021
2d6871d
Update quant_layers.py (#35392)
XGZhang11 Sep 9, 2021
81e702a
quant: fix a export bug (#35410)
XGZhang11 Sep 9, 2021
f05e444
optimization of index_select forward op (#32863)
Zjq9409 Sep 9, 2021
4beaa75
mark extra attr for unsqueeze2 (#35528)
weisy11 Sep 9, 2021
512329b
add asExtra for nce op (#35474)
pangyoki Sep 9, 2021
42847d2
conv3d (#35507)
b3602sss Sep 10, 2021
6f7aca9
Fix scatter and gather bug (#35595)
sneaxiy Sep 10, 2021
3896bdb
add opdef extra (#35514)
shangzhizhou Sep 10, 2021
d798505
fix bn/in/squeeze/syncbn extra (#35502)
ceci3 Sep 10, 2021
9c9eba1
fix extra op for expand, expand_as, tile, unstack (#35598)
bjjwwang Sep 10, 2021
652da1f
Add As_extra to dropout op and lrn op (#35349)
huangjun12 Sep 10, 2021
d53e567
fix bug of recompute in hybridparallel (#35588)
ForFishes Sep 10, 2021
0b6623d
[NPU] support gradient_accumulator (#35044)
ronny1996 Sep 10, 2021
29cacee
add elementwise trt converter test cases (#35552)
baoachun Sep 10, 2021
922e23b
change trt_tile_op half diff and add some func for CE (#35597)
Sep 10, 2021
749945b
add prelu trt converter test case (#35512)
baoachun Sep 10, 2021
3d9603d
import ska flat_hash_map (#34464)
MingMingShangTian Sep 10, 2021
11965bc
add llvm::SmallVector to paddle (#34832)
MingMingShangTian Sep 10, 2021
5bdca05
Support float16 when using ClipGradByGlobalNorm. (#33565)
huangxu96 Sep 10, 2021
4e509f4
add cumprod op (#35185)
liyagit21 Sep 10, 2021
deb40f0
fix api doc of paddle.any' (#35631)
shangzhizhou Sep 10, 2021
98d047d
add api_op fill_diagonal_tensor (#34515)
zhiboniu Sep 10, 2021
47d15a3
set gradient_merge_cond persistable to false (#35578)
ZHUI Sep 10, 2021
d8bfe83
add the extra for op rnn/sequence_conv/sequence_pool/sequence_softmax…
zhwesky2010 Sep 10, 2021
966f042
Fix warning (#34875)
sunzhongkai588 Sep 10, 2021
a4b67f7
re-submit softmax_with_cross_entropy hard label (#35283)
xingfeng01 Sep 10, 2021
49e243c
test=document_fix (#35655)
xingfeng01 Sep 10, 2021
2c922d6
[Dygraph 4D Parallel] Sharding Support MP-PP-DP Parallelism (#35580)
JZ-LIANG Sep 10, 2021
d856f87
Set attribute "with_quant_attr" into quantized operators (#35583)
wanghaoshuang Sep 10, 2021
fe4deac
test=document_fix (#35654)
lelelelelez Sep 10, 2021
523f46f
change metaclass of Layer from pybind11_builtins.pybind11_type to typ…
zhiqiu Sep 10, 2021
246a9b6
fix prelu float16 bug (#35584)
GuoxiaWang Sep 10, 2021
cd5115f
Clear VLOG in DeviceEvent (#35633)
Aurelius84 Sep 11, 2021
4f4962c
re-submit softmax_with_cross_entropy hard label (#35283) (#35660)
xingfeng01 Sep 11, 2021
ec25291
Add cpu npu cembedding (#35467)
Baibaifan Sep 11, 2021
8412d6c
register the with_quant_attr attribute for all operattor. test=develo…
winter-wang Sep 11, 2021
adaa207
refactor gc (#35525)
wanghuancoder Sep 11, 2021
8342403
fix some op extra error, test=develop (#35667)
winter-wang Sep 12, 2021
b480664
Update scales when var is unsigned (#35599)
wozna Sep 12, 2021
4e23371
Added clip BF16/FP32 FWD/BWD kernels (#35601)
jakpiase Sep 12, 2021
1ee237c
add lstm qat models scales (#35382)
lidanqing-vv Sep 12, 2021
b8c6e18
[ROCM] fix top_k_v2 with large shape (#33783)
Ray2020BD Sep 13, 2021
dc3c845
support hybrid parallel inference helper class (#35576)
GuoxiaWang Sep 13, 2021
7b743ba
catch dimentions error when input is empty in static.nn.group_norm (#…
zoooo0820 Sep 13, 2021
03026ce
Revert "change '/' method from scale Op to elementwise_div Op (#33279…
MingMingShangTian Sep 13, 2021
787209f
add group_norm trt converter test case (#35524)
baoachun Sep 13, 2021
48ec02f
Add public api for dlpack. (#35620)
DesmonDay Sep 13, 2021
2bb4431
[HybridParallel]Fix scaler bug in pipeline_parallel/model_parallel (#…
ForFishes Sep 13, 2021
6622304
Add searchsorted op (#35159)
Yanxing-Shi Sep 13, 2021
666da14
[NPU] add npu unit test if title has NPU key word, test=develop (#35566)
qili93 Sep 13, 2021
42559f7
add gather_nd trt converter test cases (#35464)
baoachun Sep 13, 2021
75d5e3b
add gather trt converter test case (#35523)
baoachun Sep 13, 2021
4b6f809
Support int16_t in fill_constant_op (#35619)
ZzSean Sep 13, 2021
ecfe837
upload global scatter and global gather operators related files (#35546)
youth123 Sep 13, 2021
cabc5f3
[Bugfix] reshape with zero input tensor (#35642)
JZ-LIANG Sep 13, 2021
fb65268
add flatten/flatten2 converter test cases (#35462)
baoachun Sep 13, 2021
53e294c
[RC22] Fix linear with matmul_op replace (#35445)
Ray2020BD Sep 13, 2021
1a7b3ff
fix cumprod docs (#35647)
liyagit21 Sep 13, 2021
86a6be1
add xpu_wait & new implementation replace memcpy in adam, adamw (#35437)
taixiurong Sep 13, 2021
f521a30
refine svd; unexpose tensor.svd; fix english document; set timeout=40…
2742195759 Sep 13, 2021
d4f84d4
Implement FunctionTraits to support two kinds of elementwise functor …
Xreki Sep 13, 2021
0460608
Fix CPU CI build-time count (#35677)
tianshuo78520a Sep 13, 2021
e641c63
fix instance norm index error (#35341)
ceci3 Sep 13, 2021
5f31737
fix interpolate launch error (#35577)
tink2123 Sep 13, 2021
40d4a29
Revert "Implement FunctionTraits to support two kinds of elementwise …
Xreki Sep 13, 2021
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
3 changes: 3 additions & 0 deletions CMakeLists.txt
Original file line number Diff line number Diff line change
Expand Up @@ -222,6 +222,9 @@ option(WITH_MIPS "Compile PaddlePaddle with mips support" OFF)
option(WITH_MUSL "Compile with musl libc instead of gblic" OFF)
option(WITH_UNITY_BUILD "Compile with UnityBuild mode" OFF)
option(WITH_STRIP "Strip so files of Whl packages" OFF)
option(NEW_RELEASE_CUBIN "PaddlePaddle next-level release strategy for pypi cubin package" OFF)
option(NEW_RELEASE_JIT "PaddlePaddle next-level release strategy for backup jit package" OFF)
option(WITH_ASCEND_INT64 "Compile with int64 kernel for ascend NPU" OFF)

# PY_VERSION
if(NOT PY_VERSION)
Expand Down
8 changes: 8 additions & 0 deletions cmake/configure.cmake
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,10 @@ if(WITH_TESTING)
add_definitions(-DPADDLE_WITH_TESTING)
endif(WITH_TESTING)

if(WITH_INFERENCE_API_TEST)
add_definitions(-DPADDLE_WITH_INFERENCE_API_TEST)
endif(WITH_INFERENCE_API_TEST)

if(NOT WITH_PROFILER)
add_definitions(-DPADDLE_DISABLE_PROFILER)
endif(NOT WITH_PROFILER)
Expand Down Expand Up @@ -86,6 +90,10 @@ if(WITH_ASCEND_CL)
add_definitions(-DPADDLE_WITH_ASCEND_CL)
endif()

if(WITH_ASCEND_INT64)
add_definitions(-DPADDLE_WITH_ASCEND_INT64)
endif()

if(WITH_XPU)
message(STATUS "Compile with XPU!")
add_definitions(-DPADDLE_WITH_XPU)
Expand Down
20 changes: 19 additions & 1 deletion cmake/cuda.cmake
Original file line number Diff line number Diff line change
Expand Up @@ -3,10 +3,22 @@ if(NOT WITH_GPU)
endif()


if (WITH_NV_JETSON)
if(WITH_NV_JETSON)
add_definitions(-DWITH_NV_JETSON)
set(paddle_known_gpu_archs "53 62 72")
set(paddle_known_gpu_archs10 "53 62 72")
elseif(NEW_RELEASE_CUBIN)
message("Using New Release Strategy - Cubin Packge")
add_definitions(-DNEW_RELEASE_CUBIN)
set(paddle_known_gpu_archs "35 37 50 52 60 61 70 75 80 86")
set(paddle_known_gpu_archs10 "50 60 70 75")
set(paddle_known_gpu_archs11 "60 70 75 80")
elseif(NEW_RELEASE_JIT)
message("Using New Release Strategy - JIT Packge")
add_definitions(-DNEW_RELEASE_JIT)
set(paddle_known_gpu_archs "35 37 50 52 60 61 70 75 80 86")
set(paddle_known_gpu_archs10 "35 50 60 70 75")
set(paddle_known_gpu_archs11 "35 50 60 70 75 80")
else()
set(paddle_known_gpu_archs "35 50 52 60 61 70 75 80")
set(paddle_known_gpu_archs10 "35 50 52 60 61 70 75")
Expand Down Expand Up @@ -130,11 +142,17 @@ function(select_nvcc_arch_flags out_variable)
set(cuda_arch_bin ${CUDA_ARCH_BIN})
endif()

if(NEW_RELEASE_JIT)
set(cuda_arch_ptx "${cuda_arch_ptx}${cuda_arch_bin}")
set(cuda_arch_bin "")
endif()

# remove dots and convert to lists
string(REGEX REPLACE "\\." "" cuda_arch_bin "${cuda_arch_bin}")
string(REGEX REPLACE "\\." "" cuda_arch_ptx "${CUDA_ARCH_PTX}")
string(REGEX MATCHALL "[0-9()]+" cuda_arch_bin "${cuda_arch_bin}")
string(REGEX MATCHALL "[0-9]+" cuda_arch_ptx "${cuda_arch_ptx}")

list(REMOVE_DUPLICATES cuda_arch_bin)
list(REMOVE_DUPLICATES cuda_arch_ptx)

Expand Down
2 changes: 1 addition & 1 deletion cmake/external/python.cmake
Original file line number Diff line number Diff line change
Expand Up @@ -27,7 +27,7 @@ print(s.get_config_var('LDVERSION') or s.get_config_var('VERSION'));
OUTPUT_VARIABLE _PYTHON_VALUES
ERROR_VARIABLE _PYTHON_ERROR_VALUE)

if(NOT _PYTHON_SUCCESS MATCHES 0)
if(NOT _PYTHON_SUCCESS EQUAL 0)
set(PYTHONLIBS_FOUND FALSE)
return()
endif()
Expand Down
2 changes: 1 addition & 1 deletion cmake/external/xpu.cmake
Original file line number Diff line number Diff line change
Expand Up @@ -35,7 +35,7 @@ ELSE ()
ENDIF()

SET(XPU_BASE_URL_WITHOUT_DATE "https://baidu-kunlun-product.cdn.bcebos.com/KL-SDK/klsdk-dev")
SET(XPU_BASE_URL "${XPU_BASE_URL_WITHOUT_DATE}/20210804")
SET(XPU_BASE_URL "${XPU_BASE_URL_WITHOUT_DATE}/20210909")
SET(XPU_XRE_URL "${XPU_BASE_URL}/${XPU_XRE_DIR_NAME}.tar.gz" CACHE STRING "" FORCE)
SET(XPU_XDNN_URL "${XPU_BASE_URL}/${XPU_XDNN_DIR_NAME}.tar.gz" CACHE STRING "" FORCE)
SET(XPU_XCCL_URL "${XPU_BASE_URL_WITHOUT_DATE}/20210623/${XPU_XCCL_DIR_NAME}.tar.gz" CACHE STRING "" FORCE)
Expand Down
4 changes: 4 additions & 0 deletions cmake/generic.cmake
Original file line number Diff line number Diff line change
Expand Up @@ -431,6 +431,8 @@ function(cc_test_run TARGET_NAME)
if (APPLE)
set_tests_properties(${TARGET_NAME} PROPERTIES TIMEOUT 20)
endif()
elseif(WITH_TESTING AND NOT TEST ${TARGET_NAME})
add_test(NAME ${TARGET_NAME} COMMAND ${CMAKE_COMMAND} -E echo CI skip ${TARGET_NAME}.)
endif()
endfunction()

Expand Down Expand Up @@ -459,6 +461,8 @@ function(cc_test TARGET_NAME)
COMMAND ${TARGET_NAME}
ARGS ${cc_test_ARGS})
endif()
elseif(WITH_TESTING AND NOT TEST ${TARGET_NAME})
add_test(NAME ${TARGET_NAME} COMMAND ${CMAKE_COMMAND} -E echo CI skip ${TARGET_NAME}.)
endif()
endfunction(cc_test)

Expand Down
26 changes: 24 additions & 2 deletions cmake/inference_lib.cmake
Original file line number Diff line number Diff line change
Expand Up @@ -72,6 +72,12 @@ function(copy_part_of_thrid_party TARGET DST)
copy(${TARGET}
SRCS ${MKLML_LIB} ${MKLML_IOMP_LIB} ${MKLML_INC_DIR}
DSTS ${dst_dir}/lib ${dst_dir}/lib ${dst_dir})
if(WITH_STRIP)
add_custom_command(TARGET ${TARGET} POST_BUILD
COMMAND strip -s ${dst_dir}/lib/libiomp5.so
COMMAND strip -s ${dst_dir}/lib/libmklml_intel.so
COMMENT "striping libiomp5.so\nstriping libmklml_intel.so")
endif()
endif()
elseif(${CBLAS_PROVIDER} STREQUAL EXTERN_OPENBLAS)
set(dst_dir "${DST}/third_party/install/openblas")
Expand All @@ -94,8 +100,17 @@ function(copy_part_of_thrid_party TARGET DST)
DSTS ${dst_dir} ${dst_dir}/lib ${dst_dir}/lib)
else()
copy(${TARGET}
SRCS ${MKLDNN_INC_DIR} ${MKLDNN_SHARED_LIB} ${MKLDNN_SHARED_LIB_1} ${MKLDNN_SHARED_LIB_2}
DSTS ${dst_dir} ${dst_dir}/lib ${dst_dir}/lib ${dst_dir}/lib)
SRCS ${MKLDNN_INC_DIR} ${MKLDNN_SHARED_LIB}
DSTS ${dst_dir} ${dst_dir}/lib)
if(WITH_STRIP)
add_custom_command(TARGET ${TARGET} POST_BUILD
COMMAND strip -s ${dst_dir}/lib/libmkldnn.so.0
COMMENT "striping libmkldnn.so.0")
endif()
add_custom_command(TARGET ${TARGET} POST_BUILD
COMMAND ${CMAKE_COMMAND} -E create_symlink libmkldnn.so.0 ${dst_dir}/lib/libdnnl.so.1
COMMAND ${CMAKE_COMMAND} -E create_symlink libmkldnn.so.0 ${dst_dir}/lib/libdnnl.so.2
COMMENT "Make a symbol link of libmkldnn.so.0")
endif()
endif()

Expand Down Expand Up @@ -225,6 +240,13 @@ copy(inference_lib_dist
SRCS ${src_dir}/inference/capi_exp/pd_*.h ${paddle_inference_c_lib}
DSTS ${PADDLE_INFERENCE_C_INSTALL_DIR}/paddle/include ${PADDLE_INFERENCE_C_INSTALL_DIR}/paddle/lib)

if(WITH_STRIP AND NOT WIN32)
add_custom_command(TARGET inference_lib_dist POST_BUILD
COMMAND strip -s ${PADDLE_INFERENCE_C_INSTALL_DIR}/paddle/lib/libpaddle_inference_c.so
COMMAND strip -s ${PADDLE_INFERENCE_INSTALL_DIR}/paddle/lib/libpaddle_inference.so
COMMENT "striping libpaddle_inference_c.so\nstriping libpaddle_inference.so")
endif()

# fluid library for both train and inference
set(fluid_lib_deps inference_lib_dist)
add_custom_target(fluid_lib_dist ALL DEPENDS ${fluid_lib_deps})
Expand Down
2 changes: 2 additions & 0 deletions cmake/operators.cmake
Original file line number Diff line number Diff line change
Expand Up @@ -183,6 +183,8 @@ function(op_library TARGET)
list(REMOVE_ITEM miopen_cu_cc_srcs "affine_grid_cudnn_op.cu.cc")
list(REMOVE_ITEM miopen_cu_cc_srcs "grid_sampler_cudnn_op.cu.cc")
list(REMOVE_ITEM hip_srcs "cholesky_op.cu")
list(REMOVE_ITEM hip_srcs "matrix_rank_op.cu")
list(REMOVE_ITEM hip_srcs "svd_op.cu")
list(REMOVE_ITEM hip_srcs "multinomial_op.cu")
list(REMOVE_ITEM hip_srcs "decode_jpeg_op.cu")
hip_library(${TARGET} SRCS ${cc_srcs} ${hip_cc_srcs} ${miopen_cu_cc_srcs} ${miopen_cu_srcs} ${mkldnn_cc_srcs} ${hip_srcs} DEPS ${op_library_DEPS}
Expand Down
6 changes: 4 additions & 2 deletions paddle/fluid/framework/CMakeLists.txt
Original file line number Diff line number Diff line change
Expand Up @@ -354,10 +354,10 @@ cc_library(executor_cache SRCS executor_cache.cc DEPS parallel_executor)
if(WITH_PSCORE)
get_property(RPC_DEPS GLOBAL PROPERTY RPC_DEPS)
cc_test(dist_multi_trainer_test SRCS dist_multi_trainer_test.cc DEPS
conditional_block_op executor ${RPC_DEPS})
conditional_block_op executor gloo_wrapper ${RPC_DEPS})
else()
cc_test(dist_multi_trainer_test SRCS dist_multi_trainer_test.cc DEPS
conditional_block_op executor)
conditional_block_op executor gloo_wrapper)
endif()
cc_library(prune SRCS prune.cc DEPS framework_proto boost)
cc_test(prune_test SRCS prune_test.cc DEPS op_info prune recurrent_op device_context)
Expand Down Expand Up @@ -447,3 +447,5 @@ cc_library(paddle_framework DEPS ${FLUID_FRAMEWORK_MODULES})
if(WITH_TESTING AND TEST selected_rows_test)
set_tests_properties(selected_rows_test PROPERTIES TIMEOUT 120)
endif()

cc_test(scope_guard_test SRCS scope_guard_test.cc)
21 changes: 17 additions & 4 deletions paddle/fluid/framework/attribute.h
Original file line number Diff line number Diff line change
Expand Up @@ -322,8 +322,19 @@ class TypedAttrChecker {
typedef std::function<void(const T&)> ValueChecker;

public:
explicit TypedAttrChecker(const std::string& attr_name)
: attr_name_(attr_name) {}
explicit TypedAttrChecker(const std::string& attr_name,
proto::OpProto_Attr* attr)
: attr_name_(attr_name), attr_(attr) {}

TypedAttrChecker& AsExtra() {
attr_->set_extra(true);
return *this;
}

TypedAttrChecker& AsQuant() {
attr_->set_quant(true);
return *this;
}

TypedAttrChecker& InEnum(const std::unordered_set<T>& range) {
value_checkers_.push_back(EnumInContainer<T>(range));
Expand Down Expand Up @@ -398,6 +409,7 @@ class TypedAttrChecker {

private:
std::string attr_name_;
proto::OpProto_Attr* attr_;
std::vector<ValueChecker> value_checkers_;
std::vector<DefaultValueChecker> default_value_setter_;
};
Expand All @@ -408,8 +420,9 @@ class OpAttrChecker {

public:
template <typename T>
TypedAttrChecker<T>& AddAttrChecker(const std::string& attr_name) {
attr_checkers_.push_back(TypedAttrChecker<T>(attr_name));
TypedAttrChecker<T>& AddAttrChecker(const std::string& attr_name,
proto::OpProto_Attr* attr) {
attr_checkers_.push_back(TypedAttrChecker<T>(attr_name, attr));
AttrChecker& checker = attr_checkers_.back();
return *(checker.target<TypedAttrChecker<T>>());
}
Expand Down
22 changes: 21 additions & 1 deletion paddle/fluid/framework/block_desc.cc
Original file line number Diff line number Diff line change
Expand Up @@ -263,7 +263,27 @@ void BlockDesc::MoveFrom(BlockDesc *block) {
}
ops_.clear();
for (const auto &src_op : block->ops_) {
AppendOp()->CopyFrom(*src_op);
auto *dst_op = AppendOp();
dst_op->CopyFrom(*src_op);
for (const auto &pair : src_op->GetAttrMap()) {
const auto &attr_name = pair.first;
const auto &attr_value = pair.second;
auto attr_type = static_cast<proto::AttrType>(attr_value.which() - 1);
if (attr_type == proto::AttrType::BLOCK) {
auto block_id = BOOST_GET_CONST(BlockDesc *, attr_value)->ID();
dst_op->SetBlockAttr(attr_name, prog_->MutableBlock(block_id));
VLOG(10) << "Set block attr " << attr_name << " id " << block_id;
} else if (attr_type == proto::AttrType::BLOCKS) {
auto old_blocks = BOOST_GET_CONST(std::vector<BlockDesc *>, attr_value);
std::vector<BlockDesc *> new_blocks;
new_blocks.reserve(old_blocks.size());
for (auto *b : old_blocks) {
VLOG(10) << "Set block attr " << attr_name << " id " << b->ID();
new_blocks.push_back(prog_->MutableBlock(b->ID()));
}
dst_op->SetBlocksAttr(attr_name, new_blocks);
}
}
}
need_update_ = true;
Flush();
Expand Down
Loading