LWH2F Throughput

Hi,

For completeness I'm adding some additional measurements (I've already sent them to Altera via E-Mail).

The following tests were made using the LWH2F or H2F bridge with a clock rate of 200 MHz. The width of the H2F bridge is 128 bit. The tests are run from CPU 2 (A76). I've measured the write throughput.

lwh2f kernel driver ioremap              *ptr=i            18.5 MB/s
lwh2f kernel driver ioremap              memcpy      69.9 MB/s
lwh2f kernel driver ioremap_wc       memcpy    205.1 MB/s
lwh2f kernel driver ioremap_cache memcpy    [kernel panic]
lwh2f userspace (O_SYNC)               memcpy     68.2 MB/s
lwh2f userspace (no O_SYNC)         memcpy     68.3 MB/s

h2f kernel driver ioremap                  memcpy     72.2 MB/s
h2f kernel driver ioremap_wc           memcpy   258.0 MB/s
h2f kernel driver ioremap_cache     memcpy   [kernel panic]
h2f userspace (no O_SYNC)             memcpy     73.6 MB/s

The ioremap_* variants should result in the following flags:
                               cached   arm64 flags
ioremap                 no           device memory, nGnRE
ioremap_wc          no           normal memory
ioremap_cache    yes          normal memory

- ioremap_wc is much faster than ioremap(). The performance improvements is due to longer bursts (4 cycle bursts, see below). Still, there is a long latency between requests (around 250ns), the requests are not pipelined, and the bursts aren't as long as they could be (According to the TRM max burst length is 16 for LWH2F and 8 for H2F).
- The theoretical throughput is 800 MB/s for LWH2f and 3200 MB/s for H2F. Above measurements show significantly lower throughput.

Kind Regards,
Eric Opitz

----

Write CPU2 (A76) H2F ioremap_wc