Evaluation of soft-core processors on a Xilinx Virtex-5 field programmable gate array. Page: 23 of 37
This report is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided to Digital Library by the UNT Libraries Government Documents Department.
The following text was automatically extracted from the image on this page using optical character recognition software:
As discussed in the previous sections, compiler optimization levels 1 through 3 were used to test
performance of each processor. Figure 6 shows the speed-up seen when utilizing a certain
V5 Dhrystone Comparison (100MHz): Cache Enabled, FPU Enabled
1= 70.61 DMIPS (SRAM)
- 1 20
-0 U MB: SRAM
-01 -02 -03
Figure 6. Processor compiler optimizations.
Figure 6 only shows the processors running code out of SRAM with both the cache and FPU
enabled. The intent of the figure is to demonstrate how much of a speed-up the compiler
optimizations contributed to each processor. As seen in Figure 6, the uB sees a larger speed-up
than the Leon3 when utilizing optimization level 3. The Leon, on the other hand, only sees a
slight speed-up with each optimization level when compared to the slowest runtime of the uB.
One of the reasons that there is a difference between the processors in terms of compiler
optimizations is because of the compiler used for each processor.
Since compiler optimization level 2 is the standard optimization level used for program
deployment, the following results shown are for this level only.
Though the FT memory controller used for SRAM on the Leon3FT was enabled in the hardware
configuration, the EDACing on the memory controller could not be enabled in software due to
the SRAM used on the ML507 board, as mentioned in Section 184.108.40.206.
4.1 Virtex-5 Dhrystone
The performance of each processor running the Dhrystone application is shown in Figure 7. The
Dhrystone application measures fixed-point operations and is mainly used to test the
performance of either using cache or no cache. The chart represents the performance for each
processor at 100 MHz. The chart is normalized to the slowest performance measurement, which
was the DDR2 memory on the uB. Part of the reason that the DDR2 may have performed slower
Here’s what’s next.
This report can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Report.
Learn, Mark Walter. Evaluation of soft-core processors on a Xilinx Virtex-5 field programmable gate array., report, April 1, 2011; United States. (https://digital.library.unt.edu/ark:/67531/metadc836670/m1/23/: accessed May 27, 2019), University of North Texas Libraries, Digital Library, https://digital.library.unt.edu; crediting UNT Libraries Government Documents Department.