Date: Thu, 18 Jan 1996 13:49:22 -0800 From: "Mike O'Brien" <obrien@aero.org> To: freebsd-scsi@freebsd.org Subject: Messing with the TUNE code Message-ID: <9601182149.AA16938@antares.aero.org>
next in thread | raw e-mail | index | archive | help
I'm an old-line UNIX hacker. I messed a lot with the Rand kernel (we never went to V7, we already had all that stuff), 32/V, 3BSD, 4BSD, 4.1, 4.2, you get the idea. Curious as am about anything that slows down the boot process, I went grotting around the code this morning looking for the little wonder that was sitting for a long time, then printing "100 nSec ok, using 150 nSec." Ok, cool. Timing the DMA and using the fastest stable setting, sounds good. Hmmm. 100 nSec is the lowest one in the table. Since the code always backs off, we'll never use it, no matter what. Hey! Wait a minute! That's 50% of the available DMA bandwidth we're throwing away! 100 vs. 150 nSec, hmmm. Ok, here's the question: what are the codes for extending the table downward? Can it be extended downward? Not far, given memory bandwidth, I'd say. Even one slot would be enough, though. I mean, I could hotwire the code to use 100 nSec and see if my disk goes kaflooey and takes my file system with it, but... And shame, plus week-old bananas, on whoever hardwired those "7"s into the code. Mike O'Brien
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9601182149.AA16938>