SPRADG0A April   2024  – August 2024 AM62P , AM62P-Q1

 

  1.   Abstract
  2.   2
  3.   Trademarks
  4. 1Introduction
    1. 1.1 Change Cortex-A53 Clock Frequency
  5. 2Processor Core and Compute Benchmarks
    1. 2.1 Dhrystone
    2. 2.2 CoreMark-Pro
    3. 2.3 Fast Fourier Transform
    4. 2.4 Cryptographic Benchmarks
    5. 2.5 IPC Mailbox Latency
  6. 3Memory System Benchmarks
    1. 3.1 Memory Bandwidth and Latency
      1. 3.1.1 LMBench
      2. 3.1.2 STREAM
    2. 3.2 Critical Memory Access Latency
    3. 3.3 UDMA: DDR to DDR Data Copy
  7. 4Graphics Processing Unit Benchmarks
    1. 4.1 Glmark2
    2. 4.2 GFXBench5
  8. 5Video Codec
  9. 6References
  10. 7Revision History

2.1 Dhrystone

Dhrystone benchmark focuses on the processor core performance. The benchmark runs from warm L1 caches in all modern processors. The benchmark scales linearly with clock speed. Even though the benchmark was introduced in 1984 by Reinhold P. Weicker, Dhrystone still gets used in embedded processing. The industry has adopted the VAX 11/780 as the reference 1 MIPS machine. The VAX 11/780 achieves 1757 Dhrystones per second. The score is calculated by normalizing the time the benchmark loop runs by the reference 1 MIPS machine score of 1757. A common issue is to further normalize to DMIPS/MHz/core as the score scales linearly with clock speed. For standard Arm cores, the DMIPS/MHz is identical to the same compiler and flags. Dhrystone is a single core benchmark, a simple sum of multiple cores running the benchmark in parallel is sometimes used.

The Dhrystone (Version 2.1, C Language ) benchmark is included in the SDK and can be performed by simply running the command dhrystone. Due to the short execution time, TI recommends to run the test for high number of iterations to measure accurate results. More than 100 million iterations are used in the tests implemented for Arm-Cortex-A53. The code block below shows a short version of the terminal printout for Dhrystone benchmark execution.

root@am62pxx-evm:~# dhrystone
	
Dhrystone Benchmark, Version 2.1 (Language: C)
	
Program compiled without 'register' attribute
	
Please give the number of runs through the benchmark: 100000000
	
Execution starts, 100000000 runs through Dhrystone
Execution ends
	
Final values of the variables used in the benchmark:
.
.
.	
	
Microseconds for one run through Dhrystone:    0.1
Dhrystones per Second:                      7142857.0

Table 2-1 shows the results for this benchmark with the compiler and operating system details. The aggregate scores for AM62Ax with four A53 cores running at 1.25GHz and 1.4GHz are 14,229DMIPS and 16,261DMIPS, respectively.

Table 2-1 Dhrystone Benchmarks
Arm-Cortex-A53(1.25GHz) Arm-Cortex-A53 (1.4GHz)
Dhrystones/s 6,250,000 7,142,857
Normalized dhrystones (divide by 1757 reference for 1MIPS) 3,557 4,065
DMIPS/MHz each core ~3 ~3
Compiler and flags GCC 11.4 -march=ARMv8 -O3
Operating system Linux 6.1.80 (2023)