On Tue, May 20, 2008 at 08:37:24PM -0400, Smith, Todd wrote: > Hello, > > Sorry about the cryptic subject, but I was trying to provide enough tags to > archive against later. I was doing some cleaning up the other day and I > rediscovered my old VX40 Multia UDB that had been collecting dust for quite > some time. I had purchased it to run Linux on in the late 90's and it was my > first non Intel Linux machine running RedHat 4.0. I had some fun with it and > had put it away because I had nothing better to do with it. Fast forward, > almost a decade and I find it again and I remembered the good times that I > had with it and I decided that I would try to rebuild and run it again. > > My goal is to rebuild the machine as a distributed computing node running > SETI and probably a couple of other DC projects that I am involved in that > have Alpha ports. My first question is about memory, I know that I need 72 > pin FPM True Parity SIMMs, but will 128MB SIMMs work? I tried some HP memory > that I had laying around that was x36 FPM Parity memory, but it wouldn't > POST. Obviously, 64MB SIMMs work just fine, but I would like to boost RAM to > the maximum possible. The VX40 only has 256KB of L2 Cache, and I seem to > remember from that time period in the Intel space there was a correlation > between L2 Cache and maximum memory size? Can anyone shed some light on > maximum memory size possible on a UDB?
No idea about the ram for the alpha, but I think what you are remembering about caching and ram was the old intel 430VX(and TX) chipset, which had too short a cache tag which meant any ram above 64MB was uncached. It didn't matter if you had 256K or 512k cache, it still wouldn't cache above 64MB. I think one of the other 430 chipsets did have a difference in cache range depending on the amount of cache in the system. It was entirely a case of crapy design by intel to make it cheaper. -- Len Sorensen -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
