New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Gigabit Ethernet performance issue #4
Comments
I would suggest switching to
Eases testing since then you simply do a
I think applying all the A64 BSP kernel patches first makes sense since there GbE performance also improved greatly over time (and I would suspect that's not magic that happened but code changes ;) ) |
Just for the record, I tried to test through all possible TX/RX delay settings on OPi PC 2 already but it seems the values from DT are ignored. As a comparison how it looks like when doing the same with Pine64:
Here increased RX values immediately lead to low througput or no networking at all. Will let the test continue today but right now it's obvious that I remember that longsleep mentioned that he removed every contents from On a related note: Can't do any more tests from now on since I send my OPi PC 2 today to jemk (one of the linux-sunxi guys doing mainlining work. He'll look into DRAM and video issues the next time). |
Please see this possible explanation for And please also feel free to join linux-sunxi community. Some people seem to be rude from time to time (me included) but that can be dealt with by ignoring :) |
@ThomasKaiser |
Most probably the most simple way is to join IRC. I lurked around some time there quietly before to get the idea how the 'big guys' behave. Many people read the IRC log so simply ask questions and don't be surprised if they will be answered 12h later ;) |
I ended up now with this script to test through TX/RX delay (on Pine64). As soon as the values in
|
Hi all, maybe it is a bit off-topic but I hit a super low network speed with h5 chips. In details, I connected a orange pc 2 board with 1000M switch but ethtool shows it's only 100M-FD link. and the performance is much slower than both of two, maximum speed is around 70Kb/s. sysbench result on cpu, memory, and fileio seems to be good and iperf3 on lo device is around 2G/s which is acceptable. I am not sure what's going wrong with the network. lspci doesn't work so I can't tell if it is a A-chip or so. Please let me know if anything can help with it. thanks in advance. |
Currently the script only demonstrates that changes to device tree have no effect. Maybe since u-boot takes settings from sys_config.fex and overrides DT prior to kernel loading. This has to be resolved first. |
@ThomasKaiser thanks for the response. So the performance issue won't be resolved until u-boot is fixed, right? |
I was referring to this pretty limited version of the Without at least the version string u-boot complains but such a limited version is enough to make it run. And my hope is that by stripping down I can't do any testing since I sent my dev sample to another dev (busy with DRAM initialization and video stuff). |
@ThomasKaiser Okay, I'll give it a try. |
Another thought: https://irclog.whitequark.org/linux-sunxi/2016-11-20#18225898; |
@Chilledheart
Good Luck |
BTW: armbian/build#546 (comment) @BuddyZhang1 it would be really cool if you would start to address the various issues (network speed, non working dvfs and especially the inability to tweak any settings due to BSP weirdness) since currently users might be happy to get something that works but as soon as they realize what's wrong or first reviews/tests come up the usual 'Orange Pi are crap' shitstorm will start (we had that 2015, no need for that a second time) |
@apritzel's |
@ThomasKaiser
Hi, friends:
Today, I have test Gigabit Ethernet performanc on different mode, and report as follow:
According test, Gigabit Ethernet performance is not good, I think there is issue on ethernet driver. Do you know any patch to fix it.
The text was updated successfully, but these errors were encountered: