Navigate: 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 What are you shopping for?

256 MByte DDR DIMM performance

Increasing the density of a module naturally causes a slight hit in terms of overclocking capability. Reasons are the increased capacitance of the memory cells and / or the higher number of chips which can influence the CMD rate latency. If the scheme holds, one could hypothesize that eight chips with 256 Mbit will run at a lower command latency but many not be able to reach the extreme overclocking potential of sixteen 128 Mbit chips (since only 8 chips will be fully active at the time). Granted, this is a mere hypothetical consideration, but let's see what happens:

Infineon 256 MB vs Corsair 256 MB. "1" shows the results at 2:2:2 1T, in "2" the command latency is changed to 2 cycles. "3" shows the performance at the highest latencies.

At the 1T command latency, the Corsair XMS 2400 are limited to 148 MHz whereas the Infineon DIMM runs up to 154 MB. Please note that the limitation of the Corsair module has nothing to do with CAS, tRCD or tRP but merely depends on the command latency. The VIA KT266 and Apollo Pro266 are the only chipsets that let the user define the command latency directly, there may be a related setting in the ALi MAGiK1 chipset hidden in the Failsafe to Ultra setting, though.

If the command latency is changed, the Corsair XMS 2400 DIMMs run up to an astonishing 161 MHz at 2:2:2 latency settings and with the latencies set to max values, the DIMMs are pretty much open ended. The Infineon DIMM, on the other hand, which is capable of running at the lowest latency settings up to 154 MHz does not profit from either increasing command rate or latencies. This may be related to the increased chip density or else a deliberate tweak done by Infineon who emphasized latency instead of bandwidth.

For the average user, the path taken by Infineon is certainly the more viable option. 154 MHz FSB is in most boards the limit at which PCI bus related errors start occurring. Related to this issue is the weakness of several AMD 761-based boards to run more than 128 MB DDR under overclocked conditions. Looking back at some of these reports, a command rate issue could very well account for the problems, however, there is not enough documentation available on the chipset registers to make a conclusive statement. Intriguingly, the Infineon DIMM was shown to run on the EPoX 8K7A up to 154 MHz at low latency settings

Conclusion

The one important finding of this roundup is that, regardless of which DIMMs were used, all of them met their specs or actually exceeded them by a wide margin. Keep in mind that the PC 2100 specifications call for 133 MHz clock rate at 2.5:3:3 latencies and there was no single DIMM that had a problem at those settings.

Interesting further are the differential characteristics of the DIMMs regarding the access latencies. As a rule of thumb, the time required for opening a row (bank activate) depends on the internal resistance of the RAS lines vs the amount of row addresses that need to be served by one line as well as the overall energy required to transfer the data from the DRAM cells via the bitlines into the sense amplifiers. Because this RAS-to-CAS-Delay is relatively secondary for performance, the main focus has been on speeding up the actual read process, that is the CAS latency.

Therefore, it is not too surprising that most DIMMs hold up extremely well if only tRCD is increased while everything else is running at the shortest latencies available. In the extreme case, that is, the Corsair DIMMs, the delta between 2:2:2 and 2:3:2 turned out a whopping 13 MHz, that is, the DIMMs that maxxed out at 147 MHz 2:2:2 would still run 160 MHz at CAS-2, as long as the Row-to-Column delay was set to 3 cycles.

The sample tested in this review is too small to make an absolute statement about which DIMM runs up to what speed. Keep in mind that there are differences even between chips from the same wafer. This means that a cautious estimate will include a margin of error in the order of 4-5% for each DIMM module tested here.

DIMM performance vs. chipset limitation

If one looks over the data, the one emerging picture is that only at 2:2:2 timing were there any performance differences. At 2:3:2, most DIMMs ended up at the same frequency, that is 160 MHz (with a few exceptions). Likewise, at 2.5:3:3, most DIMMs were running 166 MHz stable but going higher, ended up in a grey-zone where errors can no longer be traced back to any definite system component.

Bottomline of this review is that any of the brand name DIMMs have the potential to run in PC2700 mode without further modification. It is up to the chipset and board manufacturers now to come up with solutions to utilize this potential.

next page:    => more =>