Date: Tue, 10 Nov 1998 14:19:15 -0600 From: "SnowFox" <snowfox@newtoy.com> To: <questions@FreeBSD.ORG> Subject: Slow ed revisited - D-Link DE-220P & if_ed.c multibuffering Message-ID: <001a01be0ce7$66047760$7c000080@newtoy.com>
next in thread | raw e-mail | index | archive | help
I wrote about 120-130k/s transfer rates with a D-Link DE-220P NE2000 compatible card - cable wiggling and all didn't seem to help, so I went spelunking in the code today. (*Huge hugs for open source - I learned how network cards work today... this if fun!*) =) Okay, if_ed.c detects the card as an NE2000 compatible and assigns it a 16k buffer. This much is nice. Unfortunately, ed_probe_Novell_generic defaults to creating two transmission buffers, and this is what's slowing me down. Hard-coding "sc->txb_cnt = 1" gets me upper 900k/s rates. (Bonus points for anyone who can tell me why? I'd gone into the code assuming I was on an 8k buffer witch-hunt, but the specs on D-Link's page say 16k.) I'd like to get if_ed.c patched to fix this condition for future releases, however I'm not sure how to best go about detecting this specific card. I know the plug and play string ends in "DLK2201" if the card hasn't been flashed with specific port and interrupt settings and "DLK2202" if it has, but I don't see anything that looks like an ISA PnP hunt section in FreeBSD for a reference on how to cleanly use this information. Anyone know better than me on how to handle this, or has anyone seen a good ISA PnP primer? To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?001a01be0ce7$66047760$7c000080>