Corsair and OCZ: New Standards in Fast DDR2
by Wesley Fink on August 4, 2006 5:50 AM EST- Posted in
- Memory
Corsair CM2X1024-6400C3: Memory Performance
Most top Intel motherboards provide a wide range of memory ratios that match available DDR2 memory. Most end-users select the memory ratio that matches their DDR2 memory speed. For those reasons, we first test all of the stock ratios at the fastest stable timings we can achieve at the given ratio. With ratios, CPU speed remains the same at 2.93GHz in our memory test bed, and memory speed is then changed by selecting different ratios.
There are some downsides to this approach. With the memory controller in the chipset, instead of part of the processor as in AM2 systems, there is a small performance penalty for speeds other than a 1:1 ratio (DDR2-533 in this case). However the penalty is in reality very small and memory scales nicely through the various speed options.
The performance of Micron DDR2 memory chips was already impressive in the high-performance DDR2 in the Conroe Buyers Guide, but Corsair takes the DDR2 performance standard even higher with their 6400C3. This memory is clearly the best of the best, allowing 3-2-2 timings to DDR2-533, 3-2-3 at DDR2-667, 3-3-3 at DDR2-800, and an astounding 4-3-4-11 at DDR2-1067. The base chips are from the same manufacturing lines, but Corsair binning and custom SPD programming have squeezed even more from these excellent memory chips. The only real concern, of course, is the very high cost, but that will not likely matter much if the absolute top performance is your system goal.
The fastest possible timings with today's DDR2 are 3-2-2. If you look closely you will see the Corsair 6400C3 works fine at the fastest timings that can be selected at both DDR2-400 and the 1:1 ratio of DDR2-533. Even at DDR2-667 only minor compromises are required, since the Corsair is completely stable at 3-2-3 timings. It is important to keep in mind that the first 5 table rows, from DDR2-400 to DDR2-1067 represent a constant CPU speed of 2.93GHz with memory running at faster speeds through ratios. Any performance increases over this DDR2-400 to DDR2-1067 range is therefore a result of increased memory speed. 1:1 should always have the least overhead in the current Intel universe, but this DDR2 memory still scales well by speed over the full range. Memory bandwidth improves with memory speed, as does raw calculation performance. However, this does not always translate into linear gaming performance. You can see this more clearly in the new performance scaling charts on page 7.
Corsair 6400C3 topped out at DDR2-1112, which is the second highest stable speed ever achieved with our DDR2 memory samples. This is particularly interesting when you consider that this Corsair memory, rated at DDR2-800, actually reaches a slightly higher top speed than we could achieve with Corsair DDR2-1067. Corsair clearly binned this memory for tight timings across the band and at a lower than average voltage in this class range. The 6400C3 achieves that design goal, and always does so with tighter timings than we would expect.
Most top Intel motherboards provide a wide range of memory ratios that match available DDR2 memory. Most end-users select the memory ratio that matches their DDR2 memory speed. For those reasons, we first test all of the stock ratios at the fastest stable timings we can achieve at the given ratio. With ratios, CPU speed remains the same at 2.93GHz in our memory test bed, and memory speed is then changed by selecting different ratios.
There are some downsides to this approach. With the memory controller in the chipset, instead of part of the processor as in AM2 systems, there is a small performance penalty for speeds other than a 1:1 ratio (DDR2-533 in this case). However the penalty is in reality very small and memory scales nicely through the various speed options.
 |
| Click to enlarge |
The performance of Micron DDR2 memory chips was already impressive in the high-performance DDR2 in the Conroe Buyers Guide, but Corsair takes the DDR2 performance standard even higher with their 6400C3. This memory is clearly the best of the best, allowing 3-2-2 timings to DDR2-533, 3-2-3 at DDR2-667, 3-3-3 at DDR2-800, and an astounding 4-3-4-11 at DDR2-1067. The base chips are from the same manufacturing lines, but Corsair binning and custom SPD programming have squeezed even more from these excellent memory chips. The only real concern, of course, is the very high cost, but that will not likely matter much if the absolute top performance is your system goal.
The fastest possible timings with today's DDR2 are 3-2-2. If you look closely you will see the Corsair 6400C3 works fine at the fastest timings that can be selected at both DDR2-400 and the 1:1 ratio of DDR2-533. Even at DDR2-667 only minor compromises are required, since the Corsair is completely stable at 3-2-3 timings. It is important to keep in mind that the first 5 table rows, from DDR2-400 to DDR2-1067 represent a constant CPU speed of 2.93GHz with memory running at faster speeds through ratios. Any performance increases over this DDR2-400 to DDR2-1067 range is therefore a result of increased memory speed. 1:1 should always have the least overhead in the current Intel universe, but this DDR2 memory still scales well by speed over the full range. Memory bandwidth improves with memory speed, as does raw calculation performance. However, this does not always translate into linear gaming performance. You can see this more clearly in the new performance scaling charts on page 7.
Corsair 6400C3 topped out at DDR2-1112, which is the second highest stable speed ever achieved with our DDR2 memory samples. This is particularly interesting when you consider that this Corsair memory, rated at DDR2-800, actually reaches a slightly higher top speed than we could achieve with Corsair DDR2-1067. Corsair clearly binned this memory for tight timings across the band and at a lower than average voltage in this class range. The 6400C3 achieves that design goal, and always does so with tighter timings than we would expect.
Facebook
Twitter
RSS
18 Comments
View All Comments
epsilonparadox - Friday, August 4, 2006 - link
That also is your opinion. To me an elitist is someone who demands a higher standard that the common folk. An elitist to me in the computer industry is someone who buys items that he/she knows will raise the level of standards among his peers. The items this person purchases doesn't necessarily have to be the most expensive but usually demands a higher premium because of the proven value.yacoub - Friday, August 4, 2006 - link
Alas, labeling one an "elitist" has a distinctly negative connotation to it, and that is not an opinion, that is simply how it is (fact, if you will).So by discussing the term elitist, you further the point that those who seek the best bang for the buck are certainly not elitists in the commonly accepted sense. Not to mention your comment regarding a price premium also goes against the core ideology of the overclocker - ie, those who look for what is most cost-effective and NOT premiumly priced. Generally speaking they are items that (at least initially) are lesser-known and often rather cheap because their real potential is commonly unknown. If anything the nature of the overclocker is the opposite of your elitist, in that they are interested in taking common things and making them into superstars, instead of just buying premium-priced items that supposedly have better performance according to the manufacturer's marketing department.
rjm55 - Friday, August 4, 2006 - link
I want, therefore I have.araczynski - Friday, August 4, 2006 - link
thanks for a good read, i agree that the overclocking scale is looks like its designed by some marketing group, stop using that kind of junk.when shown in real scale those small differences are nothing.
jmke - Friday, August 4, 2006 - link
Would be cool to see the charts start at scale 0, which would show the rather small impact of overclocking/getting higher rated memory modules. With a scale starting close to the min. score, the increase shown in the chart is not quite realistically represented ;)Did you run into weird results with: DDR2 PC4200 low timings vs DDR2 PC5300 medium/high timings ?
Spacecomber - Friday, August 4, 2006 - link
Your comment beat me too it. I also was going to chide Anandtech for producing charts that dont have a scale starting from 0. You should stick to tables, if all you are trying to do is show the number of points difference between one speed to the next.Wesley Fink - Friday, August 4, 2006 - link
Tables are already included as a summary of results for each tested memory. The Corsir table is on page 4 and the OCZ on page 5. The charts are there to show the impact of memory speed scaling on performance, which is not well-illustrated in the table or in our common bar charts. We added a paragraph explaining that scales are reduced to better show the small differences in performance between these two memories.You are both right that charts using zero-based scales will make the differences appear smaller than the reduced scale charts, but we were trying to compare the small differences in the Corsair and OCZ. We do prefer zero-based charts and an explanation was definitely needed.
Googer - Saturday, August 5, 2006 - link
With the performance of these DIMMs being about the same it really comes down to price, warranty, and reliablity as the deciding factors when you are shopping for RAM.