[Captain Obvious]

I was wondering if I could get a little help from someone on here...

I have a 8500 that I want to put a G3 card (400MHz/1Mb) into and upgrade the OS. I had RAM from this computer and an older 7600 that I decommissioned a while back. So I took it in to get the RAM tested and labeled. I also looked at the labeled info on the modules. The speed the technician labeled the RAM as does not match what I read on the chip (The speed is printed on a separate line, a dash usually precedes the speed number. With the trailing zero left off, so you may see "-6", "-7", "-8", "-10", or "-12", which represents 60ns, 70ns, 80ns, 100ns, and 120ns)

So I want to know who is right: What is on the chip or his tested speed. Below is what each of the chips are and in parenthesis is the tested speed I was told.

LG Semicon 64Mb ~ 60ns (100ns)
NEC 32Mb ~ 60 NS (100ns)
Mitsubishi 16Mb ~ 60ns (125ns)
Texas Inst 16Mb ~ 60ns (125ns)
IBM 8Mb ~ 70ns (100ns)
IBM 8Mb ~ 70ns (100ns)


ALSO i was wondering what configuration I should put the chips into? Should I leave some of them out? I have eight slots open so I want to be sure to optimize the RAM i have and not put ones in that would slow the computer down.

[zac4mac]

I'd trust the original label.

If you use that RAM, it doesn't really matter where you put them.
You may even have problems using such a mismatched hodge-podge.

The 8500 has the capability to interleave memory. This gives a noticeable increase in performance.
The idea there is to put memory in pairs starting with a4-b4 and working up to a1-b1.

I had an HP 16MB stick that appeared to work fine, but killed my ethernet(only symptom)

PM8500/G3-400@454MHz/56.8MHz bus
240MB interleaved EDO 60ns
OS 9.1

It's much slower than my G4, but fast enough to still be very useful. I use it at work for Office98 and grafix.

[drewman]

Memory is much like CPUs - some modules can run faster than what they are rated.

Memory manufacturers try to make every memory module run as fast as possible. If they start getting good yields of 60ns memory, yet had lots of orders for 100ns memory - guess what happens - that's right the 60ns modules get labeled 100ns and sold that way.

So, if testing reveals that a 100ns module runs at 60ns, it wouldn't surprise me in the least.

However, how long did the tech test it for? I would run your machine with some sort of memory testing program for a whole day and see if you get any errors - if not, you are golden.

drewman