Commit Graph

  • 26b863d58d Only run github actions on master branch. update_cuda Joe Evans 2023-09-02 20:23:53 -07:00
  • 547d557a88 Remove jetson image Joe Evans 2023-09-01 17:03:45 -07:00
  • efd203b1bb Add config file for cuda 11.8 Joe Evans 2023-09-01 17:00:54 -07:00
  • eb6159f7ee Remove edge build, update cuda version to build and test with. Joe Evans 2023-09-01 14:19:57 -07:00
  • 2daabcd0dc Disable jetson builds. Joe Evans 2023-09-01 14:16:46 -07:00
  • cf7d6408da Additional updates Joe Evans 2023-09-01 13:14:14 -07:00
  • 9abe5a9859 Update cuda versions and remove deprecated versions. Joe Evans 2023-09-01 12:30:49 -07:00
  • 961cc87689 package lib arm compute in wheel acl-build Ubuntu 2023-06-23 19:06:18 +00:00
  • e9f92b57b1 copy openblas binary Ubuntu 2023-06-23 00:37:07 +00:00
  • 4adef653ef add acl support Ubuntu 2023-06-22 23:33:42 +00:00
  • 76d73dbd77 Don't explicitly release CUDA resources at main Python process exit (#21182) v1.9.x Dick Carter 2023-02-28 10:34:39 -08:00
  • 4dc889802a update protobuf tag (#21186) Manu Seth 2023-02-24 17:31:05 -08:00
  • b84609d3fc Bump tzinfo from 1.2.6 to 1.2.10 in /docs/static_site/src (#21139) master dependabot[bot] 2023-01-26 13:28:45 -08:00
  • 1e4c3c27e3 Bump pyyaml from 5.1 to 5.4 in /cd/utils dependabot/pip/cd/utils/pyyaml-5.4 dependabot[bot] 2023-01-12 15:32:34 +00:00
  • 699d1ce01a Upgrade boto3/awscli in docker containers and build environments. (#21168) v1.x Joe Evans 2023-01-12 07:31:46 -08:00
  • 9f013ddad0 Upgrade boto3 to latest version. (#21167) Joe Evans 2023-01-12 07:31:22 -08:00
  • 7adf66e230 Bump addressable from 2.7.0 to 2.8.1 in /docs/static_site/src dependabot/bundler/docs/static_site/src/addressable-2.8.1 dependabot[bot] 2023-01-04 12:10:42 +00:00
  • 48d7f4af70 Port top-level-project updates from v1.x branch (#21162) Joe Evans 2023-01-04 04:09:23 -08:00
  • 5cdf130f72 [v1.9.x] TLP Updates (#21148) (#21149) Joe Evans 2022-11-28 13:51:10 -08:00
  • 26a5ad1f39 [v1.9.x] TLP Updates (#21148) Joe Evans 2022-11-21 09:02:56 -08:00
  • 7acfb3a50e [BUGFIX] Fix nms kernel's out of range access issue (#21018) Triston 2022-10-26 15:09:13 -07:00
  • 7d602e3b23 [DOC] Fix the table in Improving accuracy with INC (#21140) Andrzej Kotłowski 2022-09-26 14:15:06 +02:00
  • c8922fedff Python string formatting (#21136) hankaj 2022-09-16 13:01:27 +02:00
  • bd6405b787 Add quantized batch norm operator fused with ReLU (#21137) hankaj 2022-09-15 18:13:20 +02:00
  • f803641b53 [DOC] Add custom strategy script to quantization with INC example (#21134) Andrzej Kotłowski 2022-09-08 10:04:27 +02:00
  • 3a19f0e50d [FEATURE] Dnnl sum primitive path (#21132) Kacper Pietkun 2022-08-31 09:01:31 +02:00
  • 8d933fdcdb Add proper link to scripts in quantization with INC example (#21133) Andrzej Kotłowski 2022-08-29 11:32:07 +02:00
  • 2d72ce465a [DOC] Add tutotrial about improving accuracy of quantization with oneDNN (#21127) Andrzej Kotłowski 2022-08-26 13:10:50 +02:00
  • e2ed553f89 [v1.x] Port CICD changes (#21123, #21126 and #21128) from v1.9.x (#21129) Joe Evans 2022-08-25 22:03:15 -07:00
  • 878c3c9e5d [v1.9.x] Restore Cuda 10.x CD builds (#21128) Joe Evans 2022-08-24 08:37:14 -07:00
  • 1a418e4e1c [FEATURE] Add query_keys transformer version without split (#21115) AdamGrabowski 2022-08-23 15:33:02 +02:00
  • 67abb85131 [v1.9.x] Refactor dockerfiles in CI, migrate some ubuntu docker containers to use docker-compose. Update CI to use Cuda 11.7 (#21126) Joe Evans 2022-08-22 13:58:58 -07:00
  • daac02c785 Fix fused resnet low accuracy (#21122) hankaj 2022-08-22 13:53:44 +02:00
  • 71dab74cf5 Refactor CD to support newer cuda versions (11.0-11.7) (#21123) Joe Evans 2022-08-18 14:53:07 -07:00
  • 7748ae7edf docs: Fix a few typos (#21094) Tim Gates 2022-08-17 04:43:51 +10:00
  • 1058369f8a [BUGFIX] _npi_repeats with swap (#21112) Kacper Pietkun 2022-08-12 09:31:54 +02:00
  • 736313f4e7 Add support for bool data type for condition in where operator (#21103) bgawrych 2022-08-12 09:25:31 +02:00
  • 6d1fbe35d2 Add size threshold for few oneDNN operators (#21106) bgawrych 2022-08-04 17:08:23 +02:00
  • 9975ab41a6 [BUGFIX] Reenable fwd conv engine 5 on test_group_conv2d_16c (#21104) Dick Carter 2022-08-04 01:41:47 -07:00
  • 97e25cfc7a [submodule] Upgrade oneDNN to v2.6.1 (#21108) bartekkuncer 2022-08-04 08:46:28 +02:00
  • 0b4ecdbc4a [BUGFIX] Fix threadsafety and shutdown issues with threaded_engine_perdevice (#21110) Dick Carter 2022-08-02 02:59:57 -07:00
  • dedb8c97af [WIP] [BUGFIX] Fix flakey TemporaryDirectory() cleanup on Windows (#21107) Dick Carter 2022-08-01 10:21:45 -07:00
  • db39bb1126 Fix multivariate normal bug (#21105) hankaj 2022-07-29 13:06:00 +02:00
  • 5e5e0e3fc1 [BUGFIX] Fix SupportDNNL for multiple inputs (#21102) AdamGrabowski 2022-07-22 14:43:37 +02:00
  • ecb5026116 Comments formatting fix (#21101) DominikaJedynak 2022-07-21 09:43:23 +02:00
  • 702e47594b [v1.x] Fix for fc with sum when types are incompatible (#21042) DominikaJedynak 2022-07-20 11:25:32 +02:00
  • 183e012f01 Get rid of warnings (#21099) hankaj 2022-07-19 17:00:19 +02:00
  • 7b1daf9bc3 Requantize scale fix (#21100) DominikaJedynak 2022-07-19 16:44:52 +02:00
  • cca8f4e8c6 Reduce overhead in sg_onednn_fully_connected for floats (#21092) bgawrych 2022-07-19 09:51:42 +02:00
  • ef0415d645 [BUGFIX] Fix floor divide (#21096) Kacper Pietkun 2022-07-18 15:50:14 +02:00
  • cf15e0a478 [BUGFIX] Fix remove Cast fuse (#21086) bartekkuncer 2022-07-15 13:47:18 +02:00
  • f6d1ed1872 Improve bf16 support (#21002) Paweł Głomski 2022-07-15 10:56:38 +02:00
  • ded6096126 [FEATURE] Add pytest with benchmarking operator (#21088) AdamGrabowski 2022-07-15 09:57:57 +02:00
  • e522bea513 [BUGFIX] Fix Gluon2.0 guide (#21090) bartekkuncer 2022-07-15 09:50:44 +02:00
  • e36c9f075a Refactor fc_sum_fuse (#21077) bartekkuncer 2022-07-07 16:35:29 +02:00
  • 26243eea86 Fix broadcast ops descriptions (#21087) bartekkuncer 2022-07-07 14:49:46 +02:00
  • 84b1626b66 oneDNN FullyConnected weight caching & refactor (#21047) bgawrych 2022-07-06 12:02:58 +02:00
  • b713dc5aa3 [BUGFIX] Fix DNNL requantize operator overflow error (#21079) Andrzej Kotłowski 2022-07-05 09:42:02 +02:00
  • 5abdc77f3c [FEATURE] Add _npi_power_scalar and _npi_multiply_scalar fuse (#20976) bartekkuncer 2022-07-05 09:39:50 +02:00
  • a0def52a3e Fix oneDNN RNN weights reorder (#21065) bgawrych 2022-06-30 10:38:45 +02:00
  • 9745d36ff4 Improve masked_softmax performance with temperature parameter (#21082) AdamGrabowski 2022-06-29 10:34:31 +02:00
  • d8872c876b Fix building of master website due to removed blog page. (#21083) Joe Evans 2022-06-28 08:53:03 -07:00
  • cdffaf0994 [FEATURE] Add tanh approximation for GeLU activation (#21034) bgawrych 2022-06-28 14:53:16 +02:00
  • c486a0e304 [master] Node elimination graph pass (#21046) PiotrWolinski - Intel 2022-06-23 16:58:47 +02:00
  • afbef154ed Type fix for FullyConnected with sum (#21043) DominikaJedynak 2022-06-23 15:29:23 +02:00
  • b4aca83e31 Use requested mem in dot op to reduce memory usage (#21067) bartekkuncer 2022-06-23 13:37:47 +02:00
  • ef2be51265 Refactor SupportDNNL functions (#21032) AdamGrabowski 2022-06-23 09:41:21 +02:00
  • 1eeda3357b [BUGFIX] Fix mkldnn segfault in reshape operator (#21056) RafLit 2022-06-23 09:38:47 +02:00
  • 63aea9e031 [FEATURE] Add quantization for npi_add with oneDNN (#21041) Andrzej Kotłowski 2022-06-22 08:27:26 +02:00
  • 1ad198d639 [FEATURE] Refactor SwapAxis operator. (#21024) AdamGrabowski 2022-06-21 17:46:47 +02:00
  • 1dba76998d Diversify default RNG seed (#21058) bartekkuncer 2022-06-21 10:23:18 +02:00
  • 08f578b946 Fix test_bf16_binary_broadcast_elemwise_mixed_input (#20986) AdamGrabowski 2022-06-21 08:53:09 +02:00
  • b322bee0e7 [FEATURE] Add property removing duplicate Cast operations (#21020) bartekkuncer 2022-06-20 13:06:55 +02:00
  • 7265ac50ab Bump kramdown from 2.1.0 to 2.4.0 in /docs/static_site/src dependabot/bundler/docs/static_site/src/kramdown-2.4.0 dependabot[bot] 2022-06-20 06:42:24 +00:00
  • 36f4f58887 Fix pip package description (#21064) bartekkuncer 2022-06-20 08:41:48 +02:00
  • 6b2d4d41f5 Merge #21055 from v1.9.x - website changes. (#21068) Joe Evans 2022-06-17 11:02:20 -07:00
  • 9491442d14 Update oneDNN quantization tutorial (#21060) Andrzej Kotłowski 2022-06-14 15:27:26 +02:00
  • 3263aeeebb fix linkcheck (#21057) bgawrych 2022-06-14 15:25:47 +02:00
  • 4e161d02d4 update readme for mx1.9.1 (#21050) waytrue17 2022-06-10 14:51:54 -07:00
  • 79fe506207 [v1.x] Port #21055 from v1.9.x branch and fix CD docker python pipeline (#21062) Joe Evans 2022-06-09 14:48:07 -07:00
  • 744751539a [v1.9.x] Website updates to address ASF website check tool results (#21055) Joe Evans 2022-06-08 16:13:07 -07:00
  • 15cacdbdbf broadcast_like CPU optimization (#21004) bgawrych 2022-06-07 09:15:21 +02:00
  • 102388a055 [master] Remove dnnl_ops-inl.h file (#20997) PiotrWolinski - Intel 2022-06-02 08:40:48 +02:00
  • ac61ad2b7f Fix next_impl in deconvolution (#20750) (#21051) waytrue17 2022-06-01 13:31:30 -07:00
  • 9fe13b555c [v1.9.x] Fix CD pipeline for python docker (#21049) waytrue17 2022-05-31 19:55:14 -07:00
  • 9152b9078f Fix linkcheck (#21045) AdamGrabowski 2022-05-31 22:48:30 +02:00
  • ab30c56bce [master] Fix issue with fc_eltwise fusing (#20958) PiotrWolinski - Intel 2022-05-30 13:36:45 +02:00
  • 358ef6fa37 [website][v1.9.x] Update website for 1.9.1 release (#21033) (#21039) waytrue17 2022-05-25 13:56:35 -07:00
  • 1687b442d9 [website] Update website for mxnet 1.9.1 (#21038) waytrue17 2022-05-25 11:29:08 -07:00
  • 8843ca37bd [website][v1.9.x] Update website for 1.9.1 release (#21033) waytrue17 2022-05-23 15:15:27 -07:00
  • 9ca3b27635 [master] Enabled tests using the whole batch for calibration (#21008) DominikaJedynak 2022-05-23 07:54:38 +02:00
  • c0fee5843d change cuda version arg name v1.9.1-test Wei Chu 2022-05-20 14:02:19 -07:00
  • 0ff39ddbbb remove collect_test_results_unix and collect_test_results_windows in ci Wei Chu 2022-05-20 11:12:29 -07:00
  • 36cb197a79 add debug lines Wei Chu 2022-05-19 15:18:48 -07:00
  • 2a381e83de update year in notice (#21021) waytrue17 2022-05-19 11:54:51 -07:00
  • b1b4c25c17 update year in notice (#21023) waytrue17 2022-05-19 11:54:01 -07:00
  • 65a01503ad fix LD path for cuda files Wei Chu 2022-05-19 11:47:33 -07:00
  • 0497efd4e4 fix transformer optimization for gpt-2 (#21007) bgawrych 2022-05-19 08:59:07 +02:00
  • a9574d2b39 Fix quantized_elemwise_add (#21031) Andrzej Kotłowski 2022-05-18 07:35:38 +02:00
  • 3098af4a80 AMP improvements + enable bf16 input for quantize_v2 (#20983) Paweł Głomski 2022-05-17 14:34:09 +02:00