PNY sent us a 1GB kit of PC6400 memory. This caused us a few problems as we can't find PNY memory on sale at any of the usual outlets so we had to resort to some detective work. In addition to the PNY sticker on each module there's also a sticker that says â€˜Elixir M2Y51264TU8880B-25C', which we found on sale at Scan for Â£22 inc. VAT.
That's very cheap but 1GB is barely enough memory for a Windows XP PC and is hopelessly inadequate for Vista so we're going to use the PNY/Elixir as a baseline to gauge the performance of the Gigabyte motherboard and E6750 processor.
The PNY modules are utterly basic and are nothing more than a piece of PCB with eight BGA (Ball Grad Array) memory chips mounted on one side only.
On the standard 333MHz FSB with the memory at its rated effective speed of 800MHz, we got a score of 7015 in PCMark05 with 9212 in the CPU element of the test and 7315 in the memory part.
Skip ahead to our DDR3 testing and you'll see that the Kingston and OCZ memory scored the same overall mark on the standard FSB. In a sense this is obvious, but it's worth spelling out the point that any PC2-800 memory provides sufficient bandwidth for any Core 2 Duo processor running on the standard front-side bus. You only need fast memory if you're overclocking and while you can usually eke out one or two percent extra by running the memory faster than the FSB it's hardly worth the effort.
Overclocking is a completely different proposition as we raised the FSB from 333MHz to 450MHz to give a CPU speed of 3.60GHz. The Gigabyte GA-P35C-DS3R offers a reasonable selection of memory multipliers (although it doesn't allow any adjustment to latency timings) and using the lowest 2.0x multiplier increased the memory speed to 900MHz. This speed increase lengthened the memory latencies slightly as they are set by SPD (Serial Presence Detect) and the increase in system performance was astounding. Overall PCMark05 leapt from 7015 to 8427, which is a boost of 20 percent, while the CPU and memory elements of the test climbed even higher.
We didn't make any adjustments to voltage settings as we prefer to avoid cremating our hardware and we used a quiet Zalman CNPS8000 cooler throughout so there may well be some untapped overclocking potential, but even as it stands we were happy.
The PNY enabled us to run 10MHz more FSB than with any other memory in this round-up, no doubt because it's a 2 x 512MB kit rather than 2 x 1GB. The result is that the PNY memory - the cheapest on test here - gave the highest performance of the entire round-up, and it's all thanks to the 450MHz FSB. Having said that we wouldn't run 1GB of RAM so the PNY is not for us.