Hi,
I have copied object detection trained models from caffe-jacinto-models github repo ( github.com/.../ssd512x512_ds_PSP_dsFac_32_fc_0_hdDS8_1_kerMbox_3_1stHdSameOpCh_1)
And i am trying to compare the execustion time for Dense Model and Sparse Model.
As per the datasheet, Sparse model execution must be faster than dense model. But, I am getting different result.
Setup :
1. TDA2x EVM Board is connected to CCS using XDS560 JTAG.
2. Build and generated eve_test_dl_algo.out (release mode) for test app (in tidl folder)
3. Using CCS Load Program eve_test_dl_algo.out executed on EVE
Output:
1. Sparse Model:
TEST_REPORT_INIT_PROFILE_DATA : TSC cycles = 2117750600, SCTM VCOP BUSY cycles = 0, SCTM VCOP Overhead = 0
I/O Bytes : 0 ( 0 + 0 ) : 0 ( 0 + 0 )
Processing Frame Number : 0
TEST_REPORT_PROCESS_PROFILE_DATA : TSC cycles = 69244619476, SCTM VCOP BUSY cycles = 108141852, SCTM VCOP Overhead = 0
I/O Bytes : 0 ( 0 + 0 ) : 0 ( 0 + 0 )
- Execution Time = TSC cycles / EVE Frequncy
- Considering only PROCESS PROFILE DATA
- Execution Time: 129.4291953 seconds (EVE Frequncy: 535 MHz)
2. Dense Model:
TEST_REPORT_INIT_PROFILE_DATA : TSC cycles = 3165035504, SCTM VCOP BUSY cycles = 0, SCTM VCOP Overhead = 0
I/O Bytes : 0 ( 0 + 0 ) : 0 ( 0 + 0 )
Processing Frame Number : 0
TEST_REPORT_PROCESS_PROFILE_DATA : TSC cycles = 62898624182, SCTM VCOP BUSY cycles = 275218342, SCTM VCOP Overhead = 0
I/O Bytes : 0 ( 0 + 0 ) : 0 ( 0 + 0 )
- Execution Time = TSC cycles / EVE Frequncy
- Considering only PROCESS PROFILE DATA
- Execution Time: 117.5675218 seconds (EVE Frequncy: 535 MHz)
Whether this setup is correct ?
Sparse Model execution time is more that Dense Model ?
Regards,
Sagar
Attached Log:
