Quote:
|
Originally Posted by unregistered
oh I get answers, always the same - no benefit
if water cannot be justified, how so a TEC ?
|
We've tested PC-4000 memory in a case that has 0 airflow with all major components water-cooled (so there were no fans in the case) and running MemTest resulted in the memory getting hot enough to start producing errors. What happened is the memory would pass ~3-15 times successfully than start failing slowly with an error or two every odd pass after that till it gets progressively worse and started throwing errors up each pass. Putting a fan in front of the RAM resolved the errors progressively (but very quickly, within a pass or two) till no more errors were generated. This was reproduceable on 3 systems.
The systems were running the PC-4000 (DDR-500) memory at 3-3-3-8 200MHz, so it was
underclocked to a certain degree.
Anecdoctal, but with that information in hand it appears cooling the memory more efficiently can result in better overclockability or just less opportunity for errors...?
Naturally, this isn't the same experience we've had with GPU memory which appears to run fine without any airflow.