From: "Chris Leech" <[EMAIL PROTECTED]>
Date: Thu, 16 Mar 2006 11:24:05 -0800
> > Thanks, that clarifies things. So, if I've understood correctly, the
> > benefit kicks in when:
> >
> > 1) I/OAT is enabled :)
> > 2) The user posts a recv() (or the like) of >= 2K
> > 3) There is >= 2K of data avai
On 3/16/06, Scott Feldman <[EMAIL PROTECTED]> wrote:
> Do you have any data to share on header split? Also, can other non-
> Intel nics use I/OAT copy, and if so, is header-split a requirement
> for the copy?
I don't have any header-split data. The I/OAT copy offload will work
for any TCP traffi
On Mar 16, 2006, at 11:20 AM, Chris Leech wrote:
On 3/16/06, Leonid Grossman <[EMAIL PROTECTED]> wrote:
Hi Chris,
Do you know what part of the performance delta is contributed by the
offload for copy operations, and what part comes from other I/OAT
features like header separation, etc. ?
Thi
> Thanks, that clarifies things. So, if I've understood correctly, the
> benefit kicks in when:
>
> 1) I/OAT is enabled :)
> 2) The user posts a recv() (or the like) of >= 2K
> 3) There is >= 2K of data available to give them
>
> yes?
Yes
-
To unsubscribe from this list: send the line "unsubscrib
Chris Leech wrote:
I must be missing something - if the MTU was 1500 bytes, how did the
receiver's offloaded copies get to the 2k level? Were several arriving
TCP segments aggregated?
Most of the overhead (get_user_pages) is per recv, not on a per packet
basis. Regardless of packet size, we
should have kept this on list
-- Forwarded message --
From: Chris Leech <[EMAIL PROTECTED]>
Date: Mar 16, 2006 11:13 AM
Subject: Re: I/OAT performance data
To: Rick Jones <[EMAIL PROTECTED]>
> I must be missing something - if the MTU was 1500 bytes, how did t
oops, should have kept this on list
-- Forwarded message --
From: Chris Leech <[EMAIL PROTECTED]>
Date: Mar 16, 2006 10:56 AM
Subject: Re: I/OAT performance data
To: Rick Jones <[EMAIL PROTECTED]>
> When it says "buffer size" for the Chariot stuff, is
On 3/16/06, Leonid Grossman <[EMAIL PROTECTED]> wrote:
> Hi Chris,
> Do you know what part of the performance delta is contributed by the
> offload for copy operations, and what part comes from other I/OAT
> features like header separation, etc. ?
This is showing the offloaded copy as the only dif
Behalf Of Chris Leech
> Sent: Thursday, March 16, 2006 10:33 AM
> To: netdev
> Subject: I/OAT performance data
>
> Sorry this took so long. The attached PDF show the benefit
> of I/OAT for bulk data receives on 1, 2, 4, 6, and 8 gigabit
> Ethernet ports.
> The baseline i
Chris Leech wrote:
When it says "buffer size" for the Chariot stuff, is that the socket
buffer size, or the size of the buffer(s) being passed to the transport?
That's the I/O size for the application, being passed to the transport.
Was the MTU 1500 or 9000 bytes?
1500 byte MTU
Can the Ch
Chris Leech wrote:
Sorry this took so long. The attached PDF show the benefit of I/OAT
for bulk data receives on 1, 2, 4, 6, and 8 gigabit Ethernet ports.
The baseline is a 2.6.15 kernel with an updated e1000 driver.
When it says "buffer size" for the Chariot stuff, is that the socket
buffer
11 matches
Mail list logo
