OCZ PC2700 DDR rev. 3.2

DDR to use at 190 FSB – Joe

SUMMARY: Stabile at 170 MHz with aggressive timings, 190 MHz with more conservative settings.

John Apps at the Ramstore.ca was nice enough to send a sample of OCZ’s PC2700, rev 3.2 (512 M). OCZ claims “Now running up to CAS 2 2-5-2 1T OCZ PC-2700 Rev 3.2.”

After I did the testing, I carefully removed the aluminum heatspreader to see what’s underneath:

These are Winbond W942508BH-6 chips. Winbond states “The -6 is compliant to the 166 MHz/CL2.5 specification.” Anything more aggressive runs the chip out of spec. (Note that the chip is spec’d for CAS 2 @ 133 MHz and CAS 2.5 @ 166 MHz.)

A side note: I’ve taken off aluminum heatspreaders on a few chips and, IMHO, I think they are close to worthless. I’ve found that at best, the sticky tape used contacts something like 30% of the chip area. This is like using thermal grease on less than half the CPU.

Further, I seriously doubt that the heat transfer from the tape to the aluminum heatspreader is all that great. Removing the heatspreader on any RAM voids warranty and possibly could damage the RAM, so I’m not advocating anyone do this at all. However, if you can get RAM without the heatspreaders, I think an 80 mm fan aimed at the DIMM slots will do a LOT more good.

Interesting that Crucial does not offer their RAM with heatspreaders (I spoke to them about it – “not necessary”) – is there a message here?

I tested the OCZ PC2700 using an ABIT BD7II RAID motherboard (250 FSB max) with an P4 1700, running at an 8x multiplier (this is an Intel Engineering Sample); this avoids any issues due to the CPU running out of spec. PCI settings were fixed at 33 MHz. I boosted RAM voltage to 2.7 volts.

To test the RAM, I used Ultra X’s RAM Stress Test Professional (R.S.T. Pro). This is a stand-alone PCI card – it does not need hard drives or an OS to run. The motherboard boots up and then the R.S.T. Pro takes over. The testing is quite rigorous and, after running at least 5 loops, will identify problems down to the chip and sectors within the chip.

I ran two tests at different RAM timings with the following results:

OCZ’s PC2700 rev 3.2 runs a little better than their spec with aggressive memory timings (consistent with OCZ’s claim) and about 14.5% better at more conservative timings.

Thanks again to Ramstore.ca for sending this our way.

Email Joe