> -----Original Message----- > From: David S. Miller [mailto:[EMAIL PROTECTED] > Sent: Friday, August 05, 2005 1:19 AM > To: Leonid Grossman > Cc: netdev@vger.kernel.org > Subject: Re: Super TSO performance drop
> I'm not talking about the application, I'm talking about the > new kernel segmenting engine. You will need to actively > investigate the kernel issues, is what I'm trying to say. > Hi David, Basically, it looks like with SuperTSO most of the traffic in our tests comes down to the driver with mss 0 (TSO is mostly "off"). With the original TSO, is was always "on" (see below). Could you describe how SuperTSO algorithm for switching TSO "on" and "off" works, it seems to be fairly complex for us to mess with? For a fast network with extremely low error rate, I'd expect TSO to be on. Leonid -------------------------------------------------------------- Hi Leonid, I found that original TSO will always fill in 64K IP payload no matter what block size application uses. With SuperTSO situation is different, especially with bigger block sizes. With 1M block sizes, TSO special function is enabled only 20% of the time... This is obviously causing performance degradation. Bellow is a typical printout: original TSO, 1024k payload, 2 threads (most of the time): --------------------------------------------------------- [ 2890.043854] mss 9560 frg_cnt 14 frg0 1600 frg1 4096 frg2 4096 frg3 4096 frg4 4096 frg5 4096 frg6 4096 frg7 4096 frg8 4096 frg9 4096 frg10 4096 frg11 4096 frg12 4096 frg13 3040 total 57414 Super TSO, 1024k payload, 2 threads (most of the time): -------------------------------------------------------- [ 288.594989] mss 0 frg_cnt 3 frg0 2624 frg1 4096 frg1 2840 total 9614 Dima - To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
