On Fri, 2005-04-01 at 18:07, Libor Michalek wrote:
> You need to increase the amount of memory that the user is allowed
> to lock. The following command in each shell from which you are running
> ttcp:
>
> limit memorylocked unlimited
>
> In 2.6 mlock() looks at the rlimits for the process
On Fri, Apr 01, 2005 at 05:56:00PM -0500, Hal Rosenstock wrote:
> On Thu, 2005-03-31 at 16:41, Libor Michalek wrote:
> >
> > ./ttcp.aio.x -r -l 98304 -a 20 -f M
> > ./ttcp.aio.x -t -l 98304 -n 20 -a 20 -f M 192.168.0.100
> >
> > > > For the FMR pool cache miss async results I used
On Thu, 2005-03-31 at 16:41, Libor Michalek wrote:
> On Thu, Mar 31, 2005 at 04:25:28PM -0500, Hal Rosenstock wrote:
> > On Wed, 2005-03-30 at 19:43, Libor Michalek wrote:
> > > The program has a decent help for available parameters, but here are
> > > some reasonable defaults:
> > >
> > > serv
On Thu, Mar 31, 2005 at 04:25:28PM -0500, Hal Rosenstock wrote:
> On Wed, 2005-03-30 at 19:43, Libor Michalek wrote:
> > The program has a decent help for available parameters, but here are
> > some reasonable defaults:
> >
> > server:
> >
> > ./ttcp.aio.x -r -l 65536 -a 20
> >
> > clien
On Wed, 2005-03-30 at 19:43, Libor Michalek wrote:
> The program has a decent help for available parameters, but here are
> some reasonable defaults:
>
> server:
>
> ./ttcp.aio.x -r -l 65536 -a 20
>
> client:
>
> ./ttcp.aio.x -t -l 65536 -n 10 -a 20 192.168.0.100
Are these the
On Wed, Mar 30, 2005 at 05:40:33PM -0800, Peter Buckingham wrote:
> Libor Michalek wrote:
> > The recompile modifications to use SDP protocol familt for regular
> > ttcp are pretty straight forward:
> >
> > #include
> > #undef AF_INET
> > #define AF_INET AF_INET_SDP
>
> if i want to do
Libor Michalek wrote:
The recompile modifications to use SDP protocol familt for regular
ttcp are pretty straight forward:
#include
#undef AF_INET
#define AF_INET AF_INET_SDP
if i want to do in-kernel SDP can i just change my in-kernel sockets
address family too, or is it more complicat
On Tue, Mar 29, 2005 at 07:29:04PM -0800, Grant Grundler wrote:
> On Tue, Mar 29, 2005 at 06:12:28PM -0800, Libor Michalek wrote:
>
> > I could checkin the source for the modified ttcp's but I'm not sure
> > exactly where... (gen2/users/libor ???)
>
> I wouldn't bother for the "#define AF_INET
On Wed, Mar 30, 2005 at 10:09:24AM -0500, Hal Rosenstock wrote:
> On Mon, 2005-03-28 at 20:03, Libor Michalek wrote:
> > I haven't looked closely at the code yet, but I did try it out
> > with SDP/AIO on a pair of x86 systems with Tavors and a pair of
> > x86_64 systems with Arbels. With a small
Quoting r. Hal Rosenstock <[EMAIL PROTECTED]>:
> Subject: Re: [openib-general] [PATCH] FMR support in mthca
>
> On Mon, 2005-03-28 at 20:03, Libor Michalek wrote:
> > I haven't looked closely at the code yet, but I did try it out
> > with SDP/AIO on a pair of x8
On Mon, 2005-03-28 at 20:03, Libor Michalek wrote:
> I haven't looked closely at the code yet, but I did try it out
> with SDP/AIO on a pair of x86 systems with Tavors and a pair of
> x86_64 systems with Arbels. With a small change to core/fmr_pool.c
> and enabling pool creation in SDP it worke
On Wed, Mar 30, 2005 at 07:44:08AM +0200, Michael S. Tsirkin wrote:
> > As Libor reported, SDP is working and getting the expected good
> > AIO performance with this code.
> What benchmark do you use to measure AIO performance?
I just asked libor as well. It was a hacked version of ttcp:
Quoting r. Roland Dreier <[EMAIL PROTECTED]>:
> Subject: Re: [openib-general] [PATCH] FMR support in mthca
>
> Thanks, I applied this in a few pieces with some minor cleanups.
>
> As Libor reported, SDP is working and getting the expected good
> AIO performance wit
Thanks, I applied this in a few pieces with some minor cleanups.
As Libor reported, SDP is working and getting the expected good
AIO performance with this code.
Thanks,
Roland
___
openib-general mailing list
openib-general@openib.org
http://openib.org
On Tue, Mar 29, 2005 at 06:12:28PM -0800, Libor Michalek wrote:
> The recompile modifications to use SDP protocol familt for regular
> ttcp are pretty straight forward:
>
> #include
> #undef AF_INET
> #define AF_INET AF_INET_SDP
Yes, that's pretty easy.
> I could checkin the source f
On Tue, Mar 29, 2005 at 05:08:14PM -0800, Grant Grundler wrote:
> On Mon, Mar 28, 2005 at 05:03:51PM -0800, Libor Michalek wrote:
> > I haven't looked closely at the code yet, but I did try it out
> > with SDP/AIO on a pair of x86 systems with Tavors and a pair of
> > x86_64 systems with Arbels.
On Mon, Mar 28, 2005 at 05:03:51PM -0800, Libor Michalek wrote:
> I haven't looked closely at the code yet, but I did try it out
> with SDP/AIO on a pair of x86 systems with Tavors and a pair of
> x86_64 systems with Arbels. With a small change to core/fmr_pool.c
> and enabling pool creation in
Libor> Here is the change I made to core/fmr_pool.c:
BTW I already committed this fix.
- R.
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/
On Sun, Mar 27, 2005 at 05:31:13PM +0200, Michael S. Tsirkin wrote:
> OK, here's an updated version of the patch. This passed basic
> tests: allocate/free, map/remap/unmap.
>
> For Tavor, MTTs for FMR are separate from regular MTTs, and are reserved
> at driver initialization. This is done to limi
On Sun, Mar 27, 2005 at 05:31:13PM +0200, Michael S. Tsirkin wrote:
> For Tavor, MTTs for FMR are separate from regular MTTs, and are reserved
> at driver initialization. This is done to limit the amount of
> virtual memory needed to map the MTTs.
> For Arbel, there's no such limitation, and all MT
Excellent! I'll review this ASAP, and see if Libor and I can get
SDP/AIO going with it.
- R.
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mai
OK, here's an updated version of the patch. This passed basic
tests: allocate/free, map/remap/unmap.
For Tavor, MTTs for FMR are separate from regular MTTs, and are reserved
at driver initialization. This is done to limit the amount of
virtual memory needed to map the MTTs.
For Arbel, there's no s
Michael> Good, glad to help. I will try to address your comments
Michael> next week (its already weekend here).
No problem. Libor won't be back until Monday so I won't even try SDP
until then.
Roland> What if we just reserve something like 64K MPTs and MTTs
Roland> for FMRs and i
Quoting r. Roland Dreier <[EMAIL PROTECTED]>:
> Subject: Re: [openib-general] [PATCH] fmr support in mthca
>
> Thanks for implementing this. You saved me a lot of work and Libor
> will be very happy when he gets back next week.
Good, glad to help. I will try to address your
Thanks for implementing this. You saved me a lot of work and Libor
will be very happy when he gets back next week.
Some comments from my first read through:
> This patch implements FMR support. I also rolled into it two fixes
> for regular mrs that I posed previously, let me know if its a prob
This patch implements FMR support. I also rolled into it two fixes
for regular mrs that I posed previously, let me know if its a problem.
This seems to be working fine for me, although I only did relatively basic
tests. Both Tavor and Arbel Native modes are supported. I made some tradeoffs
for sim
26 matches
Mail list logo