Jaimie said:
*Frequently*? What environment are you working in?
Since losing yet another ZX Spectrum to dodgy 4-bit RAM chips back in
1983, I didn't have another memory failure until 2009 (DDR2).
Cheers - Jaimie
I found the FPM/EDO era to be very reliable. And
I had a couple machines here with eight sticks of
RAM in them, of that type.
SDRAM, DDR, and onwards, not so much.
I bought eight 512MB SDRAM sticks, and three of them
died. And those died, while sitting in a motherboard
and not being used, in a (dry) house. Not in the
garage or anything. The remaining five sticks, still
stored in their antistatic tray, all tested good.
(Sticks in antistatic tray, stored in same environment.)
I've also lost two batches of "generic" RAM which
was on sale at local computer stores. Needless to
say, there have been no local purchases of that type
since. The RAM in that case, died at the 1.5 year mark.
One of the computer stores went bankrupt, so no opportunity
to discuss it with them.
I've had one stick of a four stick set of Ballistix die on
me. That was DDR400 CAS2. Other Crucial (non-Ballistix)
purchases were OK.
None of the systems had overclocked RAM. The Ballistix, with
nominal 2.5V chips, was running 2.65 or 2.7V or so (as instructed
on the tin). Not really an excessive voltage. Those particular
DIMMs were known not to tolerate higher voltages, and I didn't
try. The other generic RAMs, were all running nominal.
All in all, I'm not impressed.
The FPM/EDO, I just can't figure out why that stuff was
so much better. One difference in that era, was extra
care and attention to signal undershoot. It was after
that era, that inputs were changed on RAMs, to allow them
to tolerate rather large undershoot (up to -2V ?). Hard
to believe that would have anything to do with it, but
I can't see what other factor might come into play. The
geometry is getting smaller, but yet, we have processors
with 22nm features, that are as reliable as can be. Same
with video card GPUs, with large silicon dies, small
features, temperature stress, the whole bit, and they
aren't dropping like flies. GPUs still die, but generally
as a side effect of a dead and melted cooling fan, and the
GPU gets damaged through no fault of its own.
The RAMs aren't cool to the touch, so probably not
a condensation issue of some sort.
Paul