Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!usc!zaphod.mps.ohio-state.edu!mips!pacbell.com!decwrl!sgi!shinobu!odin!anchor!olson From: olson@anchor.esd.sgi.com (Dave Olson) Newsgroups: comp.sys.sgi Subject: Re: Hardware question on 4D/20 Message-ID: <9914@odin.corp.sgi.com> Date: 3 Jul 90 17:27:33 GMT References: <1990Jul3.063306.28718@ariel.unm.edu> Sender: news@odin.corp.sgi.com Distribution: usa Organization: Silicon Graphics, Inc. Mountain View, CA Lines: 39 jcmiller@hydra.unm.edu (Jeff Miller) writes: | Hello fellow SGI users!! | | I have a couple of hardware questions for all of you GURU's out there: | | (1) What is involved in upgrading a 4D/20 to a 4D/25? I am willing to perform | the surgery myself. I believe that it involves replacing the R3000 and R3010 | CPU and FPU with faster components and, of course, faster RAM. In addition, | there seems to be room to install 64Kb of cache RAM in place of the existing | 8Kb cache. Is there any special hardware swith to tell the CPU that it has | more cache? | ( Yes, I know the proceding proposal definitely voids my warranty-- if I | still had one :-) I won't speak to the rest, but some other things that you get with a 4D25 are a faster SCSI chip, and a new I/O controller chip. The combination boosts disk throughput considerably. Since the I/O chip is an SGI chip, you can't just buy one. The SCSI chip is the WD 33C93A, instead of the 33C93. Just changing the SCSI chip won't help a whole lot without the new I/O chip (I seem to recall about 10% improvement with just the 93A). The types of the I/O and SCSI chips are determined independently of each other. I also seem to recall that there was some minor board layout required to accomodate the higher clock rate, and some PAL changes. The cache is sized at boot time; I don't know if the 4D20 cpu board really supports more cache than is present; I am not a hardware person. And, as you note, you have completely violated your warranty, and if you blow it, you lose... -- Dave Olson Life would be so much easier if we could just look at the source code.