First Look: AM2 DDR2 vs. 939 DDR Performance
by Wesley Fink on April 17, 2006 12:05 AM EST- Posted in
- CPUs
Socket 939 Fast DDR-400 vs. AM2 Fast DDR-2
DDR has a somewhat narrow operating range. Mainstream DDR might extend from DDR400 to perhaps DDR433. The very best, and most expensive, DDR might extend from DDR400 to a bit over DDR600. The useful range of fast DDR is even less because as the timings increase for higher speed the bandwidth starts to fall. Therefore top bandwidth with many of the top DDR products peaks somewhere in the DDR500 to DDR550 range and bandwidth normally drops as we move higher in frequency at poorer timings.
While ratios can be used with lower performing memory, the main purpose of ratios is to offset the performance limitations of cheaper memory so the user can enjoy the increased performance of overclocked processor without having to similarly invest in expensive memory capable of 1:1 memory overclocking. The point is that we will not be investigating memory ratios on DDR memory, since they are not generally used to increase memory bandwidth and performance.
DDR2 is a different animal when it comes to ratios. As DDR2 matured, memory makers and processor manufacturers have sanctioned higher and higher speed grades of DDR2 that are achieved with memory ratios. Users select and use higher rated DDR2 products to increase memory performance. On the Intel platform there was theoretically a price to pay since the memory controller was on the chipset.. This meant 1:1 DDR2 memory is theoretically top performance without compromise, while other speeds had some degrading from maximum performance due to the overhead of the FSB ratios.
AMD uses HyperTransport, which means there is no FSB. All memory speeds should theoretically perform as if they were 1:1 on HT. This simply means additional DDR2 speeds should carry no penalty at all. A user can choose DDR2-400, DDR2-533, DDR2-667, DDR2-800, DDR2-1066, and other DDR2 speeds based on performance needs and their budget, as there is no one speed which is best in a HyperTransport design.
For all of these reasons, performance of AM2 was benchmarked at DDR2-400, DDR2-533, DDR2-667, and DDR2-800. Due to limitations of current AM2 motherboards in that no timings faster than 3 are available for memory, all 4 speeds were evaluated at 3-3-3 memory timings. This is a fast memory timing for DDR2 at any of those speeds, as the fastest DDR timings we have tested at lower DDR2 speeds is 3-2-2. At DDR2-800, 3-3-3 are the fastest timings ever tested at this speed.
In addition, performance at DDR2-800 3-3-3, the fastest stock speed we could set on current AM2 boards with this memory, was compared to DDR performance at DDR400 2-2-2, which is generally the fastest stock speed on current Socket 939 motherboards. The performance difference is reported in %.
Benchmark results were very interesting. From a broad perspective, memory bandwidth and latency of DDR400 was matched by DDR2 at a speed of just over DDR2-533. Put another way, DDR2-533 provides equivalent bandwidth and latency to DDR400 on this 4th spin of AM2. DDR2-667 provided a bit better performance, and DDR2-800 3-3-3 provided a 12% improvement (lower number) in Latency over DDR400 and a 13.3% to 28.6% improvement in bandwidth depending on the memory benchmark used. It is interesting that memory writes enjoyed the biggest bandwidth boost.
Unfortunately, the added memory bandwidth did not translate into the kinds of increases in real-world gaming performance that many might expect. Gaming Performance at DDR2-800 3-3-3 at the same CPU speed increased 1% to 6.7% in our tests - with all except Call of Duty 2 in the 1% to 4% range. Call of Duty 1.2 was used in these tests, which supports dual processors, and test results did not follow the same expected scaling of other benchmarks. For this reason, the Call of Duty 2 results are considered suspect until we have more experience with Revision 1.2.
Since increased memory bandwidth did not translate into similar increases in gaming performance we can only conclude that this iteration of AM2 is not particularly memory starved - a result that was really expected. This does open the door for future upgrades and revisions of AM2 that might make better use of the available memory bandwidth with fast DDR2 memory.
DDR has a somewhat narrow operating range. Mainstream DDR might extend from DDR400 to perhaps DDR433. The very best, and most expensive, DDR might extend from DDR400 to a bit over DDR600. The useful range of fast DDR is even less because as the timings increase for higher speed the bandwidth starts to fall. Therefore top bandwidth with many of the top DDR products peaks somewhere in the DDR500 to DDR550 range and bandwidth normally drops as we move higher in frequency at poorer timings.
While ratios can be used with lower performing memory, the main purpose of ratios is to offset the performance limitations of cheaper memory so the user can enjoy the increased performance of overclocked processor without having to similarly invest in expensive memory capable of 1:1 memory overclocking. The point is that we will not be investigating memory ratios on DDR memory, since they are not generally used to increase memory bandwidth and performance.
DDR2 is a different animal when it comes to ratios. As DDR2 matured, memory makers and processor manufacturers have sanctioned higher and higher speed grades of DDR2 that are achieved with memory ratios. Users select and use higher rated DDR2 products to increase memory performance. On the Intel platform there was theoretically a price to pay since the memory controller was on the chipset.. This meant 1:1 DDR2 memory is theoretically top performance without compromise, while other speeds had some degrading from maximum performance due to the overhead of the FSB ratios.
AMD uses HyperTransport, which means there is no FSB. All memory speeds should theoretically perform as if they were 1:1 on HT. This simply means additional DDR2 speeds should carry no penalty at all. A user can choose DDR2-400, DDR2-533, DDR2-667, DDR2-800, DDR2-1066, and other DDR2 speeds based on performance needs and their budget, as there is no one speed which is best in a HyperTransport design.
For all of these reasons, performance of AM2 was benchmarked at DDR2-400, DDR2-533, DDR2-667, and DDR2-800. Due to limitations of current AM2 motherboards in that no timings faster than 3 are available for memory, all 4 speeds were evaluated at 3-3-3 memory timings. This is a fast memory timing for DDR2 at any of those speeds, as the fastest DDR timings we have tested at lower DDR2 speeds is 3-2-2. At DDR2-800, 3-3-3 are the fastest timings ever tested at this speed.
In addition, performance at DDR2-800 3-3-3, the fastest stock speed we could set on current AM2 boards with this memory, was compared to DDR performance at DDR400 2-2-2, which is generally the fastest stock speed on current Socket 939 motherboards. The performance difference is reported in %.
Benchmark results were very interesting. From a broad perspective, memory bandwidth and latency of DDR400 was matched by DDR2 at a speed of just over DDR2-533. Put another way, DDR2-533 provides equivalent bandwidth and latency to DDR400 on this 4th spin of AM2. DDR2-667 provided a bit better performance, and DDR2-800 3-3-3 provided a 12% improvement (lower number) in Latency over DDR400 and a 13.3% to 28.6% improvement in bandwidth depending on the memory benchmark used. It is interesting that memory writes enjoyed the biggest bandwidth boost.
Unfortunately, the added memory bandwidth did not translate into the kinds of increases in real-world gaming performance that many might expect. Gaming Performance at DDR2-800 3-3-3 at the same CPU speed increased 1% to 6.7% in our tests - with all except Call of Duty 2 in the 1% to 4% range. Call of Duty 1.2 was used in these tests, which supports dual processors, and test results did not follow the same expected scaling of other benchmarks. For this reason, the Call of Duty 2 results are considered suspect until we have more experience with Revision 1.2.
Since increased memory bandwidth did not translate into similar increases in gaming performance we can only conclude that this iteration of AM2 is not particularly memory starved - a result that was really expected. This does open the door for future upgrades and revisions of AM2 that might make better use of the available memory bandwidth with fast DDR2 memory.
Facebook
Twitter
RSS
37 Comments
View All Comments
Ecmaster76 - Saturday, April 15, 2006 - link
Wouldn't it be the best way to suck up bandwidth? We already knew a single core had enough bandwidth from the 754-939 transition.I guess its not that big a deal now because the parts aren't even for sale. But as long as you do tsome multitasking articles on the final hardware I'll be happy.
IntelUser2000 - Saturday, April 15, 2006 - link
So... the conclusion is that DDR2-800 is needed to outperform DDR400. DDR2-667 is slower than DDR400. The Inquirer is correct, contrary to some other opinions.Wesley Fink - Saturday, April 15, 2006 - link
Latency and Bandwidth of DDR2-533 are about the same as DDR400, and DDR2-667 and DDR2-800 are both faster. Games and Applications SHOULD be faster on both DDR2-667 and DDR2-800. However, given current aopplications and the AM2 memory controller, the applications and games are about the same at DDR2-667 - and DDR2-800 is faster as you state.This is likely the result of the late pre-release AM2 memory controller and applications/games themselves not being optimized for DDR2 on AM2. Both will likely be fixed very quickly, and actual performance of DDR2-533 should then be roughly on par with DDR400 - with 667 and 800 both faster.
psychobriggsy - Saturday, April 15, 2006 - link
If 2.5GHz was a 40% overclock, then the AM2 processor tested must have been running at 1.8GHz (9x200, i.e., it was also unlocked for you to get 10x250).It's not surprising that a 1.8GHz K8 processor wouldn't benefit from DDR2 much - indeed I expect that most of the application improvements were just from the slightly reduced latency at DDR2-800 rather than actually having more bandwidth available.
I wonder what a stock 2.8GHz AM2 X2 with DDR2-800 would get against a stock 2.8GHz 939 X2 with DDR-400? No, I don't think it will be a miracle, but just possibly it will start actually needing the extra bandwidth available, which could lead to a greater gap between the two platforms.
I guess we'll find out in under 2 months.
Wesley Fink - Saturday, April 15, 2006 - link
The AM2 processor was NOT a 1.8GHz x2. I stated the CPU could overclock 40% at stock speeds, but the 10x250 is not that 40% overclcok. We chose that ratio because it is a clock speed our DDR2 memory could handle and it was also a speed doable on 939 for a reasonable comparison. We actually had two AM2 processors this round, a top-line AM2 and a more mainstream processor. We looked at performance from both, and it was the same at the same processor speed. The DDR2 memory controller reports as Rev. F.As for the question about manufacture date, we are providing as little information about these pre-release processors and motherboards as possible to protect our sources. We have several sources who work with us to bring you the latest news before anyone else, and we don't want to compromise those relationships. Therefore we are not providing any information that might make it easier for AMD and others to trace our sources.
We can assure you these are the latest Rev. AM2 shipped to AMD partners in early April as we have evaluated 4 versions since mid-January. This is also the first rev. to fully support DDR2-800.
Viditor - Saturday, April 15, 2006 - link
Fair enough...had to ask.
Viditor - Saturday, April 15, 2006 - link
My concern is that while Anand is just receiving the part, if it's off of a recent production run then AMD will have some problems with the launch. Usually you need a full turn's worth of product in inventory for a launch...if they are just turning out final product now, then my guess is there will be shortages come July. Of course since they have doubled their capacity recently it will be much less, but still...it doesn't bode well.Anand, could you confirm the manufacture date of the chip you tested for us please?
Viditor - Saturday, April 15, 2006 - link
Apologies...I should have addressed the previous question to Wesley (sorry mate).I know that APM allows AMD to change anything (even down to individual dice on the wafer) at any point in the process, but my concern is inventory levels here. If you could please confirm the production date on the chip you used for testing, it would help me significantly with my analysis of the upcoming launch.
Cheers!
Jynx980 - Saturday, April 15, 2006 - link
Typo on page 3, paragraph 5:Wesley Fink - Saturday, April 15, 2006 - link
The extra "we" has been removed.