issues
search
microsoft
/
nnfusion
A flexible and efficient deep neural network (DNN) compiler that generates high-performance executable from a DNN model description.
MIT License
959
stars
163
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Add backward implementation for Abs op
#333
xysmlx
closed
3 years ago
0
[BUG] Unit test segfault
#332
xysmlx
opened
3 years ago
0
add conv grad
#331
xiayuqing0622
closed
3 years ago
0
Update install dependencies
#330
siahuat0727
closed
3 years ago
3
Yuqxia/merge
#329
xiayuqing0622
closed
1 year ago
0
[BUG] ir_based_fusion generates wrong Antares IR
#328
xysmlx
closed
3 years ago
1
No performance improvement with NNFusion on ResNet50
#327
jsfs2019
opened
3 years ago
1
[ENHANCEMENT] Policy of setting tuning step for Antares IR in kernel tuning
#326
xysmlx
opened
3 years ago
0
[BUG] Support ONNX TopK operator
#325
void-main
opened
3 years ago
4
[BUG]
#324
thetffs
opened
3 years ago
0
cuda codegen supports fhost_entry
#323
xiayuqing0622
closed
3 years ago
0
Avoid unnecessary rebuilds for `test.sh`
#322
colinyoyo26
closed
2 years ago
4
Fix resize operator for align corner mode
#321
wenxcs
closed
3 years ago
0
Correct `get_ready_bes` for `RangeBlockKernelScheduler`
#320
colinyoyo26
closed
3 years ago
4
[BUG] KernelRegisteration may fetch Antares kernel of wrong backend
#319
xysmlx
opened
3 years ago
0
[BUG] nnfusion CLI fails with "error while loading shared libraries: libcontrib_custom_operators.so: cannot open shared object file: No such file or directory"
#318
harveylihr
opened
3 years ago
0
Dump antares performance
#317
mzmssg
closed
3 years ago
0
Dump antares performance
#316
mzmssg
closed
3 years ago
2
Fix graph order
#315
mzmssg
closed
3 years ago
0
test nnfbot
#314
xiayuqing0622
closed
3 years ago
0
support specify per-kernel tuning step with fkernel_tuning_config
#313
jlxue
closed
3 years ago
0
Having trouble using tools/nnfusion/kernel_db
#312
kk2049
closed
3 years ago
2
[ENHANCEMENT] Support running on cuda11.0
#311
jsfs2019
closed
3 years ago
2
add -fir_based_fusion; support 1 op : multi antares kernels for cuda backend
#310
xiayuqing0622
closed
3 years ago
0
Fix reshape op IR
#309
jlxue
closed
3 years ago
0
Support SR model
#308
wenxcs
closed
3 years ago
0
Add ScatterND ONNX import and IR
#307
jlxue
closed
3 years ago
0
add onnx compatible depth2space
#306
xiayuqing0622
closed
3 years ago
2
spargen open source?
#305
leiwen83
opened
3 years ago
1
Attempt to fix graph visit order
#304
mzmssg
closed
3 years ago
0
Disable dedicated pass for hlsl and enhance antares wrapper
#303
mzmssg
closed
3 years ago
0
support fuse and tune element-wise kernels in fused IR format
#302
jlxue
closed
3 years ago
1
[BUG] cuda + antares v0.2 workflow broken
#301
mzmssg
closed
3 years ago
2
Automatically optimize ONNX Graph with external tool
#300
xysmlx
closed
3 years ago
2
add subgraph match and subgraph fusion
#299
xiayuqing0622
closed
3 years ago
3
Support fold onnx symbolic dimension
#298
mzmssg
closed
3 years ago
2
Disable dedicated pass for hlsl and fix bugs
#297
mzmssg
closed
3 years ago
0
[BUG] Out-of-order inputs for runtime constant folding
#296
mzmssg
closed
3 years ago
1
[BUG] Antares IR translate functions affect building graph with generic_operators when kernel_tuning disabled
#295
xysmlx
opened
3 years ago
2
Check antares mode in inplace analysis
#294
mzmssg
closed
3 years ago
2
[Hold for pipeline test] Support control-flow
#293
xysmlx
closed
2 years ago
0
Sort graph outputs in the order of ONNX output nodes
#292
heheda12345
closed
3 years ago
7
Fix bug and improve performance for column reduction cuda_gpu kernel
#291
xysmlx
closed
3 years ago
0
[BUG] Python bert training example broken
#290
xysmlx
closed
3 years ago
1
Script for printing each operator's output in ONNX model
#289
xysmlx
closed
3 years ago
0
Parse tunning response of antares v0.2
#288
mzmssg
closed
3 years ago
2
NNFusion v0.4 (JUN~) Release Plan
#287
wenxcs
opened
3 years ago
0
Support dump optimized ONNX model in ort_run_frozen script
#286
xysmlx
closed
3 years ago
0
[ENHANCEMENT] Confusing semantics of the Convolution op
#285
xysmlx
opened
3 years ago
1
[BUG] Convolution op config format does not consistent with frontend converter
#284
xysmlx
closed
3 years ago
1
Previous
Next