aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Merge branch 'hotfix/v01.05.01'HEADv01.05.01masterYuan Zhao2020-03-094-4/+19
|\
| * Add TIDL_SUBGRAPH_NUM_EVES env varYuan Zhao2020-03-063-3/+18
| | | | | | | | | | | | | | | | | | | | - Current subgraph implementation will initialize and use all available EVEs and DSPs, with streaming/batch inputs in mind. There are cases we only need 1 EVE and 1 DSP, for example, demonstrating subgraph offloading on a single input. This commit adds an environment variable, TIDL_SUBGRAPH_NUM_EVES, to specify number of EVEs used for subgraph inferencing. - MCT-1243
| * Start hotfix v01.05.01Yuan Zhao2020-03-061-1/+1
|/
* Merge branch 'release/v01.05.00'v01.05.00Yuan Zhao2020-02-2812-50/+132
|\
| * Update version to 1.5.0 in manifest and docsYuan Zhao2020-02-282-6/+6
| |
| * Clean up required subgraph cfg file entriesYuan Zhao2020-02-187-35/+109
| | | | | | | | | | | | | | - Added environment variable TIDL_SUBGRAPH_DIR for locating the subgraph config files. - Updated documentation for subgraph runtime. - MCT-1227
| * replace 2 dsp + 2 group layer use cases with 1 dspCaleb Robey2020-01-071-5/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | reference to PLSDK-3189. The BBAI only has enough CMEM for 4 EVEs, 1 DSP, and 2 group layers. In the case of all of our networks, the difference between 1 and 2 dsps is essentially nonexistent. The following is the benchmarks run side by side: CMDLINE: ./mcbench -g 1 -d 2 -e 4 -c ../test/testvecs/config/ CMDLINE: ./mcbench -g 1 -d 2 -e 4 -c ../test/testvecs/config/ Input: ../test/testvecs/input/preproc_0_224x224_multi.y frame Input: ../test/testvecs/input/preproc_0_224x224_multi.y frame Loop total time: 1189ms Loop total time: 1189ms FPS:42.06 FPS:42.06 mcbench PASSED mcbench PASSED CMDLINE: ./mcbench -g 1 -d 2 -e 4 -c ../test/testvecs/config/ CMDLINE: ./mcbench -g 1 -d 2 -e 4 -c ../test/testvecs/config/ Input: ../test/testvecs/input/preproc_0_224x224_multi.y frame Input: ../test/testvecs/input/preproc_0_224x224_multi.y frame Loop total time: 3066ms Loop total time: 3066ms FPS:16.31 FPS:16.31 mcbench PASSED mcbench PASSED CMDLINE: ./mcbench -g 2 -d 1 -e 4 -c ../test/testvecs/config/ | CMDLINE: ./mcbench -g 2 -d 2 -e 4 -c ../test/testvecs/config/ Input: ../test/testvecs/input/preproc_2_224x224_multi.y frame Input: ../test/testvecs/input/preproc_2_224x224_multi.y frame Loop total time: 1822ms | Loop total time: 1835ms FPS:27.44 | FPS:27.24 mcbench PASSED mcbench PASSED CMDLINE: ./mcbench -g 2 -d 1 -e 4 -c ../test/testvecs/config/ | CMDLINE: ./mcbench -g 2 -d 2 -e 4 -c ../test/testvecs/config/ Input: ../test/testvecs/input/preproc_2_224x224_multi.y frame Input: ../test/testvecs/input/preproc_2_224x224_multi.y frame Loop total time: 1823ms | Loop total time: 1841ms FPS:27.42 | FPS:27.16 mcbench PASSED mcbench PASSED CMDLINE: ./mcbench -g 2 -d 1 -e 4 -c ../test/testvecs/config/ | CMDLINE: ./mcbench -g 2 -d 2 -e 4 -c ../test/testvecs/config/ Input: ../test/testvecs/input/preproc_2_224x224_multi.y frame Input: ../test/testvecs/input/preproc_2_224x224_multi.y frame Loop total time: 1793ms | Loop total time: 1817ms FPS:27.89 | FPS:27.52 mcbench PASSED mcbench PASSED CMDLINE: ./mcbench -g 2 -d 1 -e 4 -c ../test/testvecs/config/ | CMDLINE: ./mcbench -g 2 -d 2 -e 4 -c ../test/testvecs/config/ Input: ../test/testvecs/input/preproc_0_224x224_multi.y frame Input: ../test/testvecs/input/preproc_0_224x224_multi.y frame Loop total time: 4269ms | Loop total time: 4285ms FPS:11.71 | FPS:11.67 mcbench PASSED mcbench PASSED CMDLINE: ./mcbench -g 2 -d 1 -e 4 -c ../test/testvecs/config/ | CMDLINE: ./mcbench -g 2 -d 2 -e 4 -c ../test/testvecs/config/ Input: ../test/testvecs/input/preproc_0_224x224_multi.y frame Input: ../test/testvecs/input/preproc_0_224x224_multi.y frame Loop total time: 892.9ms | Loop total time: 915ms FPS:55.99 | FPS:54.64 mcbench PASSED mcbench PASSED CMDLINE: ./mcbench -g 2 -d 1 -e 4 -c ../test/testvecs/config/ | CMDLINE: ./mcbench -g 2 -d 2 -e 4 -c ../test/testvecs/config/ Input: ../test/testvecs/input/preproc_0_224x224_multi.y frame Input: ../test/testvecs/input/preproc_0_224x224_multi.y frame Loop total time: 2008ms | Loop total time: 2014ms FPS:24.9 | FPS:24.82 mcbench PASSED mcbench PASSED
| * Minor version update in docsYuan Zhao2019-12-301-3/+3
| |
| * Fix versioning for shared libaryYuan Zhao2019-12-121-2/+11
| | | | | | | | | | - by putting proper SONAME in built shared libraries. - MCT-1231
| * Merge tag 'v01.04.00' into developYuan Zhao2019-12-075-0/+32
| |\ | |/ |/| | | TIDL-API v01.04.00 for Processor SDK 6.2
* | Merge branch 'release/v01.04.00'v01.04.00Yuan Zhao2019-12-0720-5/+3107
|\ \
| * | Update changelog for v01.04.00 releaseYuan Zhao2019-12-071-0/+9
| | |
| * | mcbench: Adjust network heap sizes, so that all test cases can fit into CMEM ↵Djordje Senicic2019-11-271-19/+7
| | | | | | | | | | | | | | | | | | of 384MB Signed-off-by: Djordje Senicic <x0157990@ti.com>
| * | examples:classification: Detect number of EVEs, DSPs and CMEM size on SoCDjordje Senicic2019-11-271-0/+7
| | | | | | | | | | | | Signed-off-by: Djordje Senicic <x0157990@ti.com>
| * | mcbench: Add test cases for AM5729Djordje Senicic2019-11-273-0/+28
| | | | | | | | | | | | | | | | | | | | | | | | - Add one line of comment in each script, to indicate SoC used with specific the script - Add all_5729.sh, script with benchmarking test cases for AM5729 device, 2xDSP+4xEVE - PLSDK-3140 Signed-off-by: Djordje Senicic <x0157990@ti.com>
| | * Bump up develop branch version to 1.5.0Yuan Zhao2019-11-251-1/+1
| |/
| * Subgraph: use Layer2Group map in config fileYuan Zhao2019-11-244-62/+106
| | | | | | | | | | | | | | | | - If Layer2Group map exists in subgraph config file, use it. Otherwise, try derive the map from network layer types. - Added TidlFreeSubgraph() for subgraph resource de-allocation - Code changes based on review comments. - MCT-1223
| * Subgraph example: multi-threaded batch processingYuan Zhao2019-11-207-42/+472
| | | | | | | | | | | | | | | | | | | | | | - Compared different batch size in subgraph execution example - Compared async/future implementation vs thread pool implementation, async/future has slightly worse (~1%) performance, but it is much easier to program - Recommended inference is multi-threaded batch processing, where batch_size can be obtained from TidlGetPreferredBatchSize(), number of threads can be set to 2. - MCT-1223
| * Parse data conversion info from subgraph configYuan Zhao2019-11-209-59/+349
| | | | | | | | - MCT-1224
| * Subgraph: support batch processingYuan Zhao2019-11-057-114/+286
| | | | | | | | - MCT-1223
| * Subgraph: add a mobilenet v1 exampleYuan Zhao2019-10-304-0/+1364
| | | | | | | | - Using the TidlRunSubgraph() interface
| * Subgraph: add top level API TidlRunSubgraphYuan Zhao2019-10-303-72/+80
| | | | | | | | | | | | - TidlRunSubgraph() should be the interface function that TVM/TFLite calls to offload subgraph to TIDL - MCT-1222
| * Subgraph data conversion at boundariesYuan Zhao2019-10-305-10/+182
| | | | | | | | | | | | - Data layout: NCHW <-> NHWC - Data type: 8-bit quantized <-> float - MCT-1222
| * Subgraph offloading to TIDL: first commitYuan Zhao2019-10-284-1/+591
| | | | | | | | | | | | | | | | - ResM class provides top level encapsulation - All allocation of core resources and buffers, and all creation of Executor, ExecutionObject, ExecutionObjectPipeline are encapsulated. - Auto-partition last few layers to DSP if profitable, also encapsulated. - MCT-1223, MCT-1224
| * Merge tag 'v01.03.03' into developYuan Zhao2019-09-3012-34/+18
| |\ | |/ |/| | | TIDL-API 1.3.3 for PSDK 6.1
* | Merge branch 'hotfix/v01.03.03'v01.03.03Yuan Zhao2019-09-3013-35/+19
|\ \
| * | Update the changelogYuan Zhao2019-09-271-1/+3
| | |
| * | Revert "Adjust example heap sizes with new TIDL library"Yuan Zhao2019-09-274-7/+3
| | | | | | | | | | | | This reverts commit c3786ddb01e187a983811d1cd3e08f6dfa20dd2e.
| * | Revert "More example sizes adjustment with new TIDL lib"Yuan Zhao2019-09-278-22/+4
| | | | | | | | | | | | This reverts commit 268aecd993dec4faec1a414d6aac4c43b0c059ed.
| * | mcbench: Add MobileNetV2 test casesDjordje Senicic2019-09-272-4/+8
| | | | | | | | | | | | | | | | | | - PLSDK-3078 Signed-off-by: Djordje Senicic <x0157990@ti.com>
| * | Start hotfix v01.03.03Yuan Zhao2019-09-271-1/+1
|/ /
| * Merge tag 'v01.03.02' into developYuan Zhao2019-09-1629-35/+58
| |\ | |/ |/| | | TIDL-API 1.3.2 for Processor SDK 6.1
* | Merge branch 'hotfix/v01.03.02'v01.03.02Yuan Zhao2019-09-1630-36/+59
|\ \ | | | | | | | | | TIDL-API 1.3.2 for Processor SDK 6.1
| * | Fix min OpenCL version to 1.1.19.00Yuan Zhao2019-09-161-1/+1
| | | | | | | | | | | | - MCT-1221
| * | Added double quotes "" to input data fileJianzhong Xu2019-09-1621-21/+21
| | |
| * | PLSDK-2956: add MobileNetV2 model (.bin files) and inference config filesJianzhong Xu2019-09-164-0/+19
| | |
| * | Fix classification example for tensorflow modelsYuan Zhao2019-09-136-15/+19
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | - Copy original image to show image before pre-processing, because pre-processing will change BGR to RGB for tensorflow models - Subtract 1 from output object class index, because tensorflow outputs 1001 bytes and uses index-0 for background. Regular imagenet labels only have 1000 entries. - Fix path to inceptionnet net and params binaries in the config file. - MCT-1221
| * | PLSDK-2986: Calibrate TIDL models for TF with correct raw image (preproc ↵Jianzhong Xu2019-09-122-0/+0
| | | | | | | | | | | | type 2).
| * | Start hotfix v01.03.02Yuan Zhao2019-09-091-1/+1
|/ /
| * Merge tag 'v01.03.01' into developYuan Zhao2019-09-0619-52/+88
| |\ | |/ |/| | | TIDL-API 1.3.1 for Processor SDK 6.1
* | Merge branch 'release/v01.03.01'v01.03.01Yuan Zhao2019-09-0651-390/+599
|\ \ | | | | | | | | | TIDL-API 1.3.1 for Processor SDK 6.1
| * | More example sizes adjustment with new TIDL libYuan Zhao2019-09-058-4/+22
| | | | | | | | | | | | - MCT-1217
| * | Update changelog for version 1.3.1Yuan Zhao2019-09-031-0/+14
| | |
| * | Update TIDL-API manifest for version 1.3.1Yuan Zhao2019-09-031-3/+3
| | |
| * | Update squeeze net reference outputYuan Zhao2019-09-031-0/+0
| | | | | | | | | | | | - 6 out of 1000 outputs changed with the newly imported network
| * | PLSDK-2986: update TIDL models for mobilenetV1, inceptionNetV1, squeezeNetV1.Jianzhong Xu2019-08-306-39/+39
| | |
| * | Adjust example heap sizes with new TIDL libraryYuan Zhao2019-08-305-6/+10
| | | | | | | | | | | | | | | | | | - The latest TIDL library increased memory requirement slightly. Adjust heap sizes in the examples accordingly. - MCT-1217
| | * Bump develop branch to version 1.4.0Yuan Zhao2019-08-301-2/+2
| |/
| * Print out imagenet object index in imagenet exampleYuan Zhao2019-08-281-2/+7
| | | | | | | | | | | | | | - The object index is helpful information, in addition to text label that has already been printed out. - Offset tensorflow model output by 1 to remove backgroup index - MCT-1216
| * Control heap size and alloc opt using env varsYuan Zhao2019-08-275-4/+45
| | | | | | | | | | | | | | | | | | - TIDL_PARAM_HEAP_SIZE_EVE, TIDL_PARAM_HEAP_SIZE_DSP, TIDL_NETWORK_HEAP_SIZE_EVE, TIDL_NETWORK_HEAP_SIZE_DSP, TIDL_EXTMEM_ALLOC_OPT_EVE, TIDL_EXTMEM_ALLOC_OPT_DSP are provided to overwrite the heap sizes and heap allocation optimization level (1 or 2) that are specified by default or by application. - MCT-1215