ASRock 775Dual-VSTA: Does DDR2 matter?
by Gary Key on August 8, 2006 6:35 AM EST- Posted in
- Motherboards
Memory Performance
Our VIA based ASRock motherboard provides two DDR2 and three DDR memory ratios. The majority of end-users will select the memory ratio that matches their memory speed. We are testing four ratios at the fastest stable timings we can achieve and still pass our benchmark test suite. With these set ratios, CPU speed remains the same at 1.86GHz in our test platform with memory speed being varied by selecting the different ratios.
There are some downsides to this approach. With the memory controller in the chipset, instead of part of the processor as in AMD Athlon 64 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, though the performance between various chipset designs can vary a great deal as we will see in a future article.
Due to performance reasons we did not test the DDR266 setting as we believe most people will not need this setting. Although the BIOS offers a 1T Command Rate, we never could get this setting absolutely stable without drastically raising the memory latency settings. We even tried our more expensive memory modules with the same results. Our memory settings were derived from extensive stress testing with a variety of applications. While certain settings that allowed lower latencies worked well with certain applications, the final settings we arrived at had to work with all applications.
The memory performance result at DDR2-533 (1:1 ratio) leads the other memory speeds in all cases. This particular setting delivered the best raw performance although we will see in our application and game benchmarks that this advantage is negated by other platform components.
Normally memory bandwidth improves with increases in memory speed and reductions in memory timings. To evaluate memory bandwidth SiSoft Sandra 2007 Professional was used to provide a closer look at scaling. As we have been saying for years, however, the Buffered benchmark usually does not correlate well with real performance in applications on the same computer. For that reason, our memory bandwidth tests have always included an UnBuffered Sandra memory score. The UnBuffered result turns off the buffering schemes, and we have found the results correlate better with real-world performance.
In this case, we find the combination of our memory settings at DDR2-667 and the VIA memory controller generates Sandra results that are up 32% lower than the DDR2-533 settings. The DDR2-533 results are up to 48% better than the DDR-333 scores with the DDR2-667 and DDR-400 scores being comparable. The Sandra memory score is really made up of both read and write operations. It is also a synthetic benchmark that does not always reflect real world performance.
To provide more detail on the impact of memory performance, we also compared pure number crunching with Version 1.5 of Super Pi, using the time to calculate 2 million places of Pi at the different memory speeds. Our other memory test is the latency measurement from the latest version of Everest. The results are interesting as the DDR2-667 ties DDR2-533 in Super Pi while DDR-400 trails slightly and DDR-333 brings up the rear. Looking at the latency figures, DDR2-667 is almost equal to DDR2-533 and DDR-400 trails slightly, while the results for DDR-333 are terrible and follow the Sandra bandwidth figures when compared to our DDR2-533 scores.
Our VIA based ASRock motherboard provides two DDR2 and three DDR memory ratios. The majority of end-users will select the memory ratio that matches their memory speed. We are testing four ratios at the fastest stable timings we can achieve and still pass our benchmark test suite. With these set ratios, CPU speed remains the same at 1.86GHz in our test platform with memory speed being varied by selecting the different ratios.
There are some downsides to this approach. With the memory controller in the chipset, instead of part of the processor as in AMD Athlon 64 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, though the performance between various chipset designs can vary a great deal as we will see in a future article.
Due to performance reasons we did not test the DDR266 setting as we believe most people will not need this setting. Although the BIOS offers a 1T Command Rate, we never could get this setting absolutely stable without drastically raising the memory latency settings. We even tried our more expensive memory modules with the same results. Our memory settings were derived from extensive stress testing with a variety of applications. While certain settings that allowed lower latencies worked well with certain applications, the final settings we arrived at had to work with all applications.
 |
| Click to enlarge |
The memory performance result at DDR2-533 (1:1 ratio) leads the other memory speeds in all cases. This particular setting delivered the best raw performance although we will see in our application and game benchmarks that this advantage is negated by other platform components.
Normally memory bandwidth improves with increases in memory speed and reductions in memory timings. To evaluate memory bandwidth SiSoft Sandra 2007 Professional was used to provide a closer look at scaling. As we have been saying for years, however, the Buffered benchmark usually does not correlate well with real performance in applications on the same computer. For that reason, our memory bandwidth tests have always included an UnBuffered Sandra memory score. The UnBuffered result turns off the buffering schemes, and we have found the results correlate better with real-world performance.
In this case, we find the combination of our memory settings at DDR2-667 and the VIA memory controller generates Sandra results that are up 32% lower than the DDR2-533 settings. The DDR2-533 results are up to 48% better than the DDR-333 scores with the DDR2-667 and DDR-400 scores being comparable. The Sandra memory score is really made up of both read and write operations. It is also a synthetic benchmark that does not always reflect real world performance.
To provide more detail on the impact of memory performance, we also compared pure number crunching with Version 1.5 of Super Pi, using the time to calculate 2 million places of Pi at the different memory speeds. Our other memory test is the latency measurement from the latest version of Everest. The results are interesting as the DDR2-667 ties DDR2-533 in Super Pi while DDR-400 trails slightly and DDR-333 brings up the rear. Looking at the latency figures, DDR2-667 is almost equal to DDR2-533 and DDR-400 trails slightly, while the results for DDR-333 are terrible and follow the Sandra bandwidth figures when compared to our DDR2-533 scores.
Facebook
Twitter
RSS
55 Comments
View All Comments
Sunrise089 - Wednesday, August 9, 2006 - link
From the link you provided:"The 4MB L2 cache can increase performance by as much as 10% in some situations. Such a performance improvement is definitely tangible, and as applications grow larger in their working data sets then the advantage of a larger cache will only become more visible. "
Also keep in mind a 3.5% average is much like claiming a new GPU is unnecessary because it only effect's your computing experience 10% of the time. If it's effects are highly pronoumced during those times it could still be worthwhile. The 4meg parts have a 10% inprovement in some apps and games, and that is a pretty big deal.
Paladin165 - Wednesday, August 9, 2006 - link
cmon man, get real. I would never buy a new GPU if it only offered a 10% increase. Also, the big increases are in DIVX compression, itunes, content creation, etc. Most of the game benchmarks are around 1%. In fact the 5% oblivion dungeon benchmark is a little hard for me to beleive, as oblivion shows virtually no gain from extra cache in other tests (http://www.anandtech.com/cpuchipsets/showdoc.aspx?...">http://www.anandtech.com/cpuchipsets/showdoc.aspx?.... The only game that shows a substaintial increase is Quake at 6.7%.The situation is in fact the opposite of a GPU. The GPU increases your performance during the "10% of the time" when it counts, during games. The extra cache on the Conroe seems to increase performance when it doesn't count.
For the record, I agree with you about AMD carrying the better budget offerings at present. Intel has only the ASrock board to offer the budget gamer, which comes with some big sacrafices (lack of overclocking, 4XPCI-E). However, I am buying this board so that I can afford a better core 2 processor (6400 or 6600), which I will put in a better budget motherboard in a few months. I figure I'm basically "renting" this motherboard for 6 months or so. So I think going with budget conroe offers a much better upgrade path. Once we have cheap boards that can overclock the 6300/6400 30-40% (which should happen in only a month or two) your argument will be obselete because the overclocked Allendales absolutely crush the X2 chips.
Sunrise089 - Wednesday, August 9, 2006 - link
"Once we have cheap boards that can overclock the 6300/6400 30-40% (which should happen in only a month or two) your argument will be obselete because the overclocked Allendales absolutely crush the X2 chips."That's true, and I look forward to it. I just think AMD has spoiled me with the idea that there is very little reason to spend extra $$$ so long as I'm willing to work a bit and accept some risk. If Intel prices move a bit to make the same situation true, I will be happily enjoying my fast Conroe machine.
araczynski - Tuesday, August 8, 2006 - link
wow, thanks for a great article, something actually applicable to the real world ;)looking forward to the next parts. although I'm still going for an e6600 with a highly overclockable mobo (probably the asus pdxyzdeluxesomethingorother), its good to see things in perspective.
johnsonx - Tuesday, August 8, 2006 - link
I don't normally join the ranks of the review bashers (who of course mostly come out of the woodwork when Intel is reviewed favorably against AMD), but this review seems to have a fatal flaw.The stated purpose is to show what you do or don't gain by trading in your DDR memory when moving to a Core 2 Duo platform. The review compares DDR to DDR2 performance on the same platform, which at first glance seems reasonable. However, in the prior review of another mainboard based on the same PT880Pro chipset (the review linked at the bottom of page 3), your conclusion about that board states:
The charts in that review show that BOTH DDR and DDR2 performance was equally dismal.
Since the primary reason for purchasing this board is for DDR support, presumably no one who wants to upgrade to DDR2 would bother with it. Therefore any performance comparison of DDR to DDR2 on this board is pointless for the stated purpose of the article. It artificially cripples DDR2 performance, making DDR look quite good and even superior.
For a proper comparison, you need a reasonably priced competitor that doesn't suffer such low DDR2 performance; perhaps a P965Express board, or even a 945 board if any of those have Conroe support.
Gary Key - Wednesday, August 9, 2006 - link
Our comments were based on the performance at the time of review with an early but shipping bios revision that as it turns out had some memory timing issues. The latest bios (1.5) has altered the performance of this board to some degree although DDR2-667 performance is not where it should be. In some cases the performance differences are less than 5% now when using optimized DDR2-533 settings. While this board does not have the overclocking capability of the P965 boards we have tested to date, it still performs within reason at stock speeds.
You are correct. However, we did state (first page) that a full comparsion would be provided in our final article. While I wanted to post a P965 result in this article we had not received our "budget" P965 boards from Foxconn, Epox, ECS, or MSI yet (two are here now, other two next week). I felt like a comparison to a 965 board that will be in the $100 price range would be better than showing results against a $150~$225 P965 board. Also, ASRock had just provided a new bios update (memory compatibility) for the 945P and 945PL boards that we are still testing as of today. It appears at first glance our original memory performance issues with these boards have been solved also. It appears now that the VIA memory performance at DDR2-533 is within 5% of our better performing P965 boards. We will have a full comparison up next week and hopefully we can figure out a way to get DDR2-667 working at CAS3 in a stable manner (it works but we still have not completed a full round of testing without a failure, BF2 and Q4 seem the most sensitive to lockup but SuperPI passes, LOL).
I appreciate your comments and look forward to your thoughts on the final article.
NeverGuy - Thursday, August 10, 2006 - link
IS this board basically the same as the ASUS P5VDC-X that is selling for about the same price? Would the ASUS board have CPU voltage adjustment?Gary Key - Thursday, August 10, 2006 - link
Close to the same, the main difference is the Asus version does not support Core 2 Duo but does offer slightly better overclocking options.
johnsonx - Wednesday, August 9, 2006 - link
Yep, I didn't see that line. You have to admit though it's buried pretty deep in that paragraph. I still think THIS article would have been far more informative with 1 other value DDR2 comparison point. But the whole picture will be clear with the final article, so I'll just look forward to that.
On a side note, I'm quite interested in this series of articles, yet I have almost zero interest in actually buying any Core 2 products for myself or for customers. Aside from the occaisional high power 3D CAD box (for running ProE usually), I travel exclusively in the budget and low-midrange market. Intel hasn't had anything compelling for me there in years, and still doesn't. I guess Core 2 info is more interesting than AM2 info, even if I'm not going to buy it.
Calin - Wednesday, August 9, 2006 - link
In reviews here, I saw a tendency to use grand words for small feats. I won't consider the 10% this board is behind others "dismal performance" - so for very little money, you could get good performance - that is the thing every budget buyer wants.