|
Advice Beginners BIOS Guide CPUs Links Mainboards Memory Network Storage Video/Sound Cards Contact Forum SiteMap Sponsors WebNews Home
|
. | . |
|
CPU Intel P4 840 D P4 820 D P4 630 P4 640 P4 650 P4 660 P4 670 AMD Athlon64 3500+ 3700+ 3800+ 4000+ X2-3800+ X2-4200+ X2-4400+ X2-4600+ X2-4800+ 1-Way Opteron Opteron 144 Opteron 146 Opteron 148 Opteron 150 Opteron 152 2-Way Opteron Opteron 240 Opteron 242 Opteron 244 Opteron 246 Opteron 248 Opteron 250 Opteron 252 2-Way Dual Core Opteron Opteron 270 Opteron 275 nVidia GF 7800GT GF 6800GT GF 6600GT ATI R X850 XT PE R X850 XT R X800 XT PE R X800 XT R X800 XL Memory Corsair Crucial Kingston Mushkin OCZ |
LOSTCIRCUITS |
|
| DDR400 The New Platform | |
| (Review by MS, May 6, 2003) |
Summary
About two years have passed since our last memory "test" or review and quite frankly I never thought I would do one of those again. Yet, times have changed, it is no longer DDR266 or PC2100 that we are overclocking but official DDR400 or PC3200, and still, here we are, overclocking. So what els is new? Dual channel platforms like Intel's Canterwood and nVidia's nForce2 are pushing the system memory further than any of the older single channel controllers ever did. Especially the Canterwood with its Performance Acceleration Technology proves to stress any modules beyond the normal, causing them to bog down at lower frequencies than any other platform.
Low latency modules are pushing into the market, even against the resistance of mainboard manufacturers who still don't know what to do with an SPD and, thus, are causing all kinds of grotesque scenarios. At the same time, even memory vendors have learned how to pronounce CAS and start throwing it into their marchitecture like their life depended on the number of CAS latencies they can fit into their modules ....
.... only, they really mean totally different timing parameters.
There is only so much of Babylonian CASfusion anybody can or should take and the new breed of module provides an excellent excuse to, once again, remember what memory is all about.
We got Corsair, KingMax, Mushkin, TwinMOS and Samsung and a few surprises..
About 2 ½ years ago, the introduction of DDR saw what appeared almost insurmountable handicaps with respect of getting anything running faster than maybe 266. Iwill's Ali M1647 chipset based KA266 was about the fastest thing on earth running up to 142 MHz bus speed, everybody else tried and tried and tried again until finally ASUS came through with the A7M266 to open up the ceiling beyond 150 MHz bus speed. Until then, there was hardly any benefit of DDR over SDRAM, especially with high-speed solutions from Corsair and EMS pushing the 166 MHz envelope at CAS latencies of 2. Keep in mind also that the most widely used platforms at the time were still running on PIII power and, therefore, not suited to take any advantage of DDR anyway.
Today's candidates, clockwise from top: KingMax 256MB DDR-400 (MPXB62D-68KX3), Samsung Original (PC3200U-30440-A1), Mushkin PC3500 (990976), Corsair (XMS3202v1.1), Corsair TWINX Platinum Series (XMS3206v1.1), TwinMOS (PC3200 CL2.5), Corsair (XMS3200v1.1). Center: Mushkin PC3200 (990902). All modules shown are 256 MB density and use eight 256Mbit components except for the KingMax using sixteen 128 Mbit discretes. The Samsung Original and the KingMax are rated CL3, the Corsairs are CL2 and the TwinMOS and Mushkin are CL2.5 modules. Click for larger image.
That was a bit over two years ago, in the following months we saw Micron, Hynix and Samsung approach 155, 160 MHz and finally, with a little help from ramping up the DDR voltages, there were isolated reports of 180 MHz clock rate. Samsung were the first with their, by now legendary C-revision to reliably run faster than DDR400 (albeit at high latencies) but as so often, since there was not even official support for DDR333 at the time, the cost effectiveness of a die shrink and the consequently increased number of dies per waver got the upper hand. The side effect were scores of frustrated overclockers in the wake of the shrink since the darn or D-revision simply did not live up to the expectations put into it.
next page: => Realigning the Performance Focus =>
If you enjoyed reading this article and found it useful, please consider making a small donation to LostCircuits.