karl@ddsw1.MCS.COM (Karl Denninger) (02/01/90)
Hello! We're looking for specific programming information/instructions for the AHA1542 board, and in particular a way to do a few things which should boost performance. We've noticed that ISC 2.0.2 seems to be issuing a hard reset to the card, and since the adapter's BIOS never gets control after that, some performance-limiting factors come into play. In particular, we've found: o The "transfer rate" jumpers on the controller appear to do nothing. o The "synchronous negotiation" jumper appears to do nothing. Both of these have major impacts on performance under MSDOS, where the BIOS on the card gets to read the jumpers and "do things". We would like to be able to: o Set the transfer rate of the board. Our motherboards can handle up to 8Mb without any problems, being limited to 5 is bogus. o Enable synch transfers. They work under MSDOS, and result in a huge speed increase! Our devices can certainly support this option; again verified with "SCSICNTL" from Roy Neese (thanks Roy!) We've seen as high as 1,500KB/sec (that's 1.5MB/sec folks!) with some "real" tests under MSDOS, but only realize about 700KB/sec under ISC. This is with a Maxtor 8760S. Howver, the 4380, a SMALLER drive with a lower data rate, gives over 1.1MB/sec of transfer rate under ISC! I have to believe that someone tuned that driver for the lower transfer rate, and some juggling would greatly improve things. Has anyone managed to play around with the board under ISC 2.0.2, and get any or all of this to work? There are a number of interesting #defines in /usr/include/sys/aha1540.h, but many of the options don't even appear to be used in the driver itself! Any assistance appreciated; if someone has working code that's even better. We're not afraid to go poking around in the driver or even writing something to poke the board, driver level or not! The Chantel Driver suite is not a realistic option; besides the cost problems ($500 per system is a little steep) people I have talked to who have it say it's not that fast; the HPDD with some tweaking seems to be the way to go. Thanks in advance -- email replies, I will summarize! -- Karl Denninger (karl@ddsw1.MCS.COM, <well-connected>!ddsw1!karl) Public Access Data Line: [+1 708 566-8911], Voice: [+1 708 566-8910] Macro Computer Solutions, Inc. "Quality Solutions at a Fair Price"
jr@frog.UUCP (John Richardson) (02/06/90)
In article <1990Jan31.234823.228@ddsw1.MCS.COM>, karl@ddsw1.MCS.COM (Karl Denninger) writes: > > The Chantel Driver suite is not a realistic option; besides the cost > problems ($500 per system is a little steep) people I have talked to > who have it say it's not that fast; the HPDD with some tweaking seems to > be the way to go. Is there interest in a host and target independant SCSI driver for a cheaper price? I develope SCSI drivers to do striping, shadowing, etc, which are host and target independant. I am interested to know what demand there could be for a source level SCSI driver product. JR
keithe@tekgvs.LABS.TEK.COM (Keith Ericson) (02/07/90)
In article <11827@frog.UUCP> jr@frog.UUCP (John Richardson) writes: > > I develope SCSI drivers to do striping, shadowing, etc, which are host >and target independant. I am interested to know what demand there could be >for a source level SCSI driver product. > Well I, for one, have been waiting what seems like _for(steekeng)ever_ to get a decent UNIX with a decent SCSI driver/system in it. (ESIX seems to have a decent SCSI driver but its accompanying file system is, uh, er, "slow" (can you spell "interminable?"); Intel has been looking for/promising a SCSI system for their/Bell-Tech UNIX for as long as we've been working with them; ISC is (so far) the best, but it's as expensive as ESIX is slow. So, if you can come up with a working add-on SCSI driver - and figure out which UNIXes are robust enough to actually _use_ the available speed - I think you'd have yourself a little niche. At least, until the much-promised SysVR4 with it's promised SCSI support gets here ("real soon now," right?). kEITHe
karl@ddsw1.MCS.COM (Karl Denninger) (02/08/90)
In article <11827@frog.UUCP> jr@frog.UUCP (John Richardson) writes: >In article <1990Jan31.234823.228@ddsw1.MCS.COM>, karl@ddsw1.MCS.COM (Karl Denninger) writes: >> >> The Chantel Driver suite is not a realistic option; besides the cost >> problems ($500 per system is a little steep) people I have talked to >> who have it say it's not that fast; the HPDD with some tweaking seems to >> be the way to go. > > Is there interest in a host and target independant SCSI driver for >a cheaper price? > > I develope SCSI drivers to do striping, shadowing, etc, which are host >and target independant. I am interested to know what demand there could be >for a source level SCSI driver product. > > JR I would think that there would be lots of interest IF: 1) You can boot from it. 2) You integrate it with (or supplant) the ISC HPDD functionality. 3) We can get the speed of the HPDD or better. We'd buy it if it met these conditions and was reasonably-priced. -- Karl Denninger (karl@ddsw1.MCS.COM, <well-connected>!ddsw1!karl) Public Access Data Line: [+1 708 566-8911], Voice: [+1 708 566-8910] Macro Computer Solutions, Inc. "Quality Solutions at a Fair Price"