b3579380dc
improve compatibility with c++ compilers, prepare for CMake
2019-02-15 17:27:12 +01:00
3d9c8530a0
Use Tensor Cores only when (channels % 8 == 0) and (filters % 8 == 0)
2019-02-12 23:13:25 +03:00
28106c0fd8
Optimized memory allocation for XNOR on CPU
2019-02-12 22:16:11 +03:00
449fcfed75
Fix for GCC on ARM 32/64-bit
2019-02-12 22:15:35 +03:00
00e992a600
Compile fix
2019-02-12 02:12:46 +03:00
5448e07445
Try to fuse conv_xnor+shortcut -> conv_xnor
2019-02-12 02:05:15 +03:00
9e138adf09
more accurate time measurements in Demo
2019-02-12 02:01:10 +03:00
7dff7365cb
Minor demo fix
2019-02-11 23:16:30 +03:00
f154d2070a
Fixed RNN (LSTM, RNN, CRNN, GRU) for CUDNN_HALF=1
2019-02-08 00:51:41 +03:00
9e07605bc5
get_connected_workspace_size() and get_convolutional_workspace_size()
2019-02-08 00:51:20 +03:00
6832290eee
Fixed set_batch_network(), when workspace larger for smaller batch
2019-02-08 00:49:51 +03:00
58de6b2d3d
Minor fix for CHECK_CUDA()
2019-02-08 00:48:27 +03:00
98103552fb
Minor fix
2019-02-07 18:19:34 +03:00
6c28da5def
Draw top5 accuracy on the Loss-chart for training Classifier
2019-02-07 18:05:58 +03:00
fc663f6efe
Another minor fix
2019-02-07 15:02:36 +03:00
9bb7455a0e
Minor fix
2019-02-07 14:47:43 +03:00
c999f53e9d
Merge pull request #2359 from aughey/master
...
Change for py code
2019-02-07 14:43:01 +03:00
7587d47c46
Partial fixed
2019-02-06 18:54:38 -06:00
0543278a5b
Partial fixed
2019-02-07 00:15:31 +03:00
022ce74fe9
Rewriting darknet_video.py to reuse darknet.py as a lib
2019-02-06 14:59:23 -06:00
64b217aa86
Update Readme.md
2019-02-06 14:51:02 +03:00
c50b0e0c8a
Minor Python and C API improvement
2019-02-06 14:38:12 +03:00
b76f1c0006
Merge pull request #2352 from aughey/master
...
Changes to better support python bindings
2019-02-06 14:23:26 +03:00
285088adc4
Fixed checking CC for enabling Tensor Cores
2019-02-06 01:55:42 +03:00
e1bbeb8367
CUDNN_HALF and CC 7.5 by default in darknet.sln
2019-02-05 11:59:58 -06:00
fa1415e3c2
CUDNN_HALF and CC 7.5 by default in darknet.sln
2019-02-05 20:43:07 +03:00
c00d3c92db
Making a fast API compatible way to copy image data.
...
This can improve on the python array_to_image function
where we already have an allocated image struct and simply
need to copy the data into the correct format/shape without
reallocating.
Aside: I think array_to_image does two frees because the pointer
data_as returns doesn't own the memory that is ultimately freed
in free_image.
2019-02-05 11:40:21 -06:00
7e9416aa80
Making a pointer version of network_predict for python.
...
The python binding requires the network struct to be passed
as a pointer to it rather than a struct copy.
2019-02-05 11:35:45 -06:00
8726d7b0db
Optimizing network_predict_image to resize only if necessary.
...
This speeds up the processing considerably if the user is so
kind to resize the image prior to doing detection.
2019-02-05 11:30:38 -06:00
edfdf2c20e
Fixed bug in Tensor Cores training
2019-02-05 19:33:10 +03:00
12b6e93893
CHECK_CUDA is used everywhere
2019-02-05 16:18:36 +03:00
ce2e0eff00
DEBUG=1 fixed
2019-02-05 00:36:17 +03:00
d767e8ca38
Minor fixes
2019-02-04 23:29:06 +03:00
5446d19576
Checks Compute Capability and forcibly disables Tensor Cores for CC < 7.0
2019-02-04 23:28:40 +03:00
f7cb538b32
Compile fix
2019-02-03 00:37:00 +03:00
584f840b40
CUDA_CHECK definition for debug
2019-02-03 00:19:04 +03:00
61156239e0
Minor performance improvement
2019-02-03 00:18:30 +03:00
dc7e7f035d
improve XNOR Tensor Cores GEMM - N 2x unrolled - minor performance improvement
2019-02-02 17:57:30 +03:00
41814fc4b3
Minor fixes
2019-02-02 15:16:57 +03:00
ff0733ed40
Speedup repack_input_kernel_bin()
2019-02-02 15:16:25 +03:00
2d747cab2b
Minor fixes
2019-02-02 03:16:30 +03:00
f91d5a5e09
Fixed __shfl() and __ballot() warnings
2019-02-02 03:16:05 +03:00
e1ec8a8b07
Update Readme.md
2019-02-02 00:58:09 +03:00
f09a9c3315
XNOR uses Tensor Cores on Turing GPU CC>=7.3 (not Volta)
2019-02-02 00:24:34 +03:00
e17bd9ba8f
Minor fix
2019-02-01 01:32:26 +03:00
a607784626
Added crnn.train.cfg just for test
2019-02-01 01:32:03 +03:00
c7309c1fdb
Fixed CRNN (RNN based on Convolution) layer
2019-02-01 01:30:02 +03:00
bd91d0a908
Add try-catch to the http_stream.cpp
2019-01-31 14:33:05 +03:00
c71354ab2e
Added cudaGetLastError() for cudaHostAlloc() to reset last cuda error
2019-01-31 14:22:07 +03:00
381f90ebb8
Fixed CUDA error checking
2019-01-29 13:46:30 +03:00
