- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
May I know first which device you are referring to actually? Cyclone V SoC, Arria 10 SoC etc.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello!
I am sorry. I have forgotten to specify the device, it is Cyclone V SOC, part number is 5CSEBA4U19I7N.
Now I am using the following configuration:
1. AXI FPGA-to-HPS bridge clock (h2f_axi_clk) is 200MHz, the data width is 128 bits.
2. There are baremetal, Intes`s Hwlib and FreeRTOS.
The device is working properly. The maximum transfer rates (between CPU and FPGA on-chip memory) that I have achieved are:
When access through a range of virtual addresses with attribute ALT_MMU_ATTR_NC:
1. CPU write data to FPGA is ~1216 MByte/sec.
2. CPU read data from FPGA is ~450 MByte/sec.
When access through a range of virtual addresses with attribute ALT_MMU_ATTR_DEVICE:
1. CPU write data to FPGA is ~571 MByte/sec.
2. CPU read data from FPGA is ~406 MByte/sec.
Data sizes range is 32768 bytes.
These rates are enough for our application, but I would like to try to increase their without
using DMA, simply increasing the frequency of h2f_axi_clk.
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page