|
From: | Pomeroy, Marty |
Subject: | Re: [lwip-users] lwip with fatfs on a STM32 speed problem |
Date: | Wed, 29 Jan 2014 11:14:36 -0500 |
Yeah,
the problem is not lwIP! I'm using an M3 at 120MHz, and have megabit
performance. So optimizing how you fill buffers is a waste of time.
As Krzysztof pointed out, you need to find the real cause of the slow behavior
and fix that. Try sending a pre-packaged chunk from flash or
sram.
Have
you verified that Nagle's Algorithm is not hindering? If Nagle is on (at
the non-lwIP end), it throttles back to like five packets per
second. Put Wireshark on the wire somewhere and see the spacing of the
packets.
I'm
pretty sure Nagle's Algorithm is off by default in lwIP. I don't recall
actively turning it off, but it is not happening in our lwIP
stack.
Marty
From: address@hidden [mailto:address@hidden On Behalf Of Dave Sandl Sent: Tuesday, January 28, 2014 10:37 AM To: Mailing list for lwIP users Subject: Re: [lwip-users] lwip with fatfs on a STM32 speed problem I do not currently have my code / project in front of me, but I was
reading some of the lwip docs and I had a question on the tcp_write
usage.
Is it OK to loop a function with tcp_write and tcp_sndbuf (to check when full) in order to fill the write buffer with data from the SD card? Could I run a loop like this every time the tcp_sent callback is called? Would this increase speed? Thanks Dave From: address@hidden Date: Tue, 28 Jan 2014 09:41:34 +0100 To: address@hidden Subject: Re: [lwip-users] lwip with fatfs on a STM32 speed problem What performance do you achieve when sending random data from flash
instead of reading it from FAT FS? Have you measured which module is responsible
for this speed?
Regards, Krzysztof Wesołowski On Tue, Jan 28, 2014 at 8:23 AM, Noam weissman <address@hidden> wrote: Hi Dave, _______________________________________________ lwip-users mailing list address@hidden https://lists.nongnu.org/mailman/listinfo/lwip-users |
[Prev in Thread] | Current Thread | [Next in Thread] |