Memory Bandwidth and Scaling
Everyone should already know that memory bandwidth improves with increases in memory speed and reductions in memory timings. To better understand the behavior of AM2 and Core 2 Duo memory bandwidth we used SiSoft Sandra 2007 Professional to provide a closer look at memory bandwidth scaling.
The most widely reported Sandra score is the Standard or Buffered memory score. This benchmark takes into account the buffering schemes like MMX, SSE, SSE2, SSE3, and other buffering tools that are used to improve memory performance. As you can clearly see in the Buffered result the AM2 on-chip memory controller holds a huge lead in bandwidth over Core 2 Duo. At DDR2-800 the AM2 lead in memory bandwidth is over 40%.
As we have been saying for years, however, the Buffered benchmark does not correlate well with real performance in games 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 well with real-world performance.
The Intel Core 2 Duo and AMD AM2 behave quite differently in UNBuferred tests. In these results AM2 and Core2 Duo are very close in memory bandwidth - much closer than in Standard tests. Core 2 Duo shows wider bandwidth below DDR2-800, but this will likely change when the AM2 controller matures and supports values below 3 in memory timings as the Core 2 Duo currently supports.
The Sandra memory score is really made up of both read and write operations. By taking a closer look at the Read and Write components we can get a clearer picture of how the two memory controllers operate. Everest from Lavalys provides benchmarking tools that can individually measure Read and Write operations.
The READ results are particularly interesting, since you can see that the READ component of Core 2 Duo performance is much larger than the WRITE results on Core 2 Duo. This is the result of the intelligent read-aheads in memory which Intel has used to lower the apparent latency of memory on the Core 2 Duo platform. Actual READ performance on Core 2 Duo now looks almost the same as AM2 to DDR2-533. AM2 starts pulling away in READ at DDR2-677 and has a slightly steeper increase slope as memory speed increases. The increases in READ speed in Core 2 Duo are a result of the intelligent read-aheads in memory. Performance without this feature would show Core 2 Duo much slower in READ operations than AM2.
This is most clearly illustrated by looking at Everest Write scores. Memory read-ahead does not help when you are writing memory, so core 2 Duo exhibits much lower WRITE performance than AM2 as we would expect. This means if all else were equal (and it isn't) the AM2 would perform much better in Memory Write tasks. Surprisingly the WRITE component of Core 2 Duo appears a straight line just below 5000 MB/s. AM2 starts at 5900 at DDR2-400 and WRITE rises to around 8000MB/s at DDR2-667. Write then appears to level off, with higher memory speeds having little to no impact on AM2 WRITE performance.
Everyone should already know that memory bandwidth improves with increases in memory speed and reductions in memory timings. To better understand the behavior of AM2 and Core 2 Duo memory bandwidth we used SiSoft Sandra 2007 Professional to provide a closer look at memory bandwidth scaling.
The most widely reported Sandra score is the Standard or Buffered memory score. This benchmark takes into account the buffering schemes like MMX, SSE, SSE2, SSE3, and other buffering tools that are used to improve memory performance. As you can clearly see in the Buffered result the AM2 on-chip memory controller holds a huge lead in bandwidth over Core 2 Duo. At DDR2-800 the AM2 lead in memory bandwidth is over 40%.
As we have been saying for years, however, the Buffered benchmark does not correlate well with real performance in games 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 well with real-world performance.
The Intel Core 2 Duo and AMD AM2 behave quite differently in UNBuferred tests. In these results AM2 and Core2 Duo are very close in memory bandwidth - much closer than in Standard tests. Core 2 Duo shows wider bandwidth below DDR2-800, but this will likely change when the AM2 controller matures and supports values below 3 in memory timings as the Core 2 Duo currently supports.
The Sandra memory score is really made up of both read and write operations. By taking a closer look at the Read and Write components we can get a clearer picture of how the two memory controllers operate. Everest from Lavalys provides benchmarking tools that can individually measure Read and Write operations.
The READ results are particularly interesting, since you can see that the READ component of Core 2 Duo performance is much larger than the WRITE results on Core 2 Duo. This is the result of the intelligent read-aheads in memory which Intel has used to lower the apparent latency of memory on the Core 2 Duo platform. Actual READ performance on Core 2 Duo now looks almost the same as AM2 to DDR2-533. AM2 starts pulling away in READ at DDR2-677 and has a slightly steeper increase slope as memory speed increases. The increases in READ speed in Core 2 Duo are a result of the intelligent read-aheads in memory. Performance without this feature would show Core 2 Duo much slower in READ operations than AM2.
This is most clearly illustrated by looking at Everest Write scores. Memory read-ahead does not help when you are writing memory, so core 2 Duo exhibits much lower WRITE performance than AM2 as we would expect. This means if all else were equal (and it isn't) the AM2 would perform much better in Memory Write tasks. Surprisingly the WRITE component of Core 2 Duo appears a straight line just below 5000 MB/s. AM2 starts at 5900 at DDR2-400 and WRITE rises to around 8000MB/s at DDR2-667. Write then appears to level off, with higher memory speeds having little to no impact on AM2 WRITE performance.
Facebook
Twitter
RSS
118 Comments
View All Comments
Calin - Wednesday, July 26, 2006 - link
I wonder if Conroe is losing a part of its bandwidth to the prefetcher. It might be so (the prefetcher is busy bringing things that will be or won't be needed, and discards some of them. Meanwhile, the memory bandwidth usable decreases).defter - Tuesday, July 25, 2006 - link
I don't recall that other benchmarks contained DDR2-400, DDR2-533, DDR2-667, DDR2-800 comparison.Basically, this review shows that if you don't want to pay a big premium for DDR2-800, Conroe is even more attractive. This isn't very suprising, because Conroe's 1066MHz (8.4GB/s) FSB can be saturated by dual-channel DDR2-533 (2x4.2GB/s = 8.4GB/s).
classy - Tuesday, July 25, 2006 - link
The Conroe buying guide was a great article, much like the Anandtech of old. But this review is too much like the recent ones absolutely of little or no value. What was the purpose of this read? We know already know core is faster.....we know already know as it is has been shown numerous times around the net memory scaling. Where is the originality at anymore? Want an example? There is a clear price difference between the two now. If a person goes AMD, they will be able to afford a better graphics card then say someone who goes Core. So why not a comparison along those lines. There is so much more to the puzzle then the same old lets run some benchmarks.bob661 - Tuesday, July 25, 2006 - link
I would like to see more articles on memory and motherboards myself for the Conroe.Wesley Fink - Tuesday, July 25, 2006 - link
From the conclusion : "The deep price cuts announced by AMD yesterday will definitely help. The new numbers indicate AM2 will be very competitive at the low end to low-mid of the processor food chain - a spot they have held in the past and where they have still managed to survive. The low end looks very competitive, and AMD is positioned close enough to mid-range in performance to keep Intel honest. There is no mistaking, however, that Intel Core 2 Duo owns the mid to high-end of the current processor market."OcHungry - Wednesday, July 26, 2006 - link
Now wait a minute- The E6800 costs ~ $1200 (if you find one), the FX62 is $800, which is 50% less. According to your review/benchmarks you propose E6800 is about 20% better performer than FX62. Considering price/performance ratio, FX62 should be the ideal choice for high end users/enthusiasts as far as price/performance is concerned.Am I wrong? This is without considering motherboard and video card limitation(s) of Intel platform. The high end enthusiasts would not run single graphic card or IGP system. Have you figured into price/performance of CPU, Motherboard and video card? Or SLI and Crossfire?
What about 4x4?
AMD is releasing 4x4's on several x2 CPU's (I think other than FX62) that will cost less than $1000. Are you saying a $1200 E6800 is a better performer than a less than $1000 4X4?
wouldn’t a 4x4 make the top end of the enthusiast’s market?
In my country a 4x4 make a stronger candidate for "the new king to be born".
DigitalFreak - Tuesday, July 25, 2006 - link
Give up, Wesley. :-)The AMD fanboys will keep trying to destroy your conclusions, since it doesn't benefit them.
duploxxx - Tuesday, July 25, 2006 - link
why would they, you might call me a fanboy but for sure the statement of wesley is true.the high performanmce market is for now (ehh from the moment you can buy them) back at the intel site... (the low-mid market is for both altough amd has an advantage here on lower price in combination with lower price on mobo's and global availability) the perfromance crown for the e6700 and xe6800and will remain there until amd pulls out something new, even if it is by 4x4... perfromance crown is performance crown... some people just do and buy anything as long as it is the fastest... same with gpu cards.. they'll buy them every month.
drebo - Tuesday, July 25, 2006 - link
If you consider an Athlon64 X2 5000+ or FX-62 "low-mid" tier performance, I suppose you could come to this conclusion.
Fact is, on a dollar-for-dollar basis, AMD still leads in, or atleast matches, performance. If I look at my main processor vendor's future pricing on Core 2 Duo processors(they're not scheduled to get any in until August 18th, by the way) and compare them with Athlon64 X2 processors under the new pricing scheme, this is what it looks like:
Athlon64 X2 3800+ @ $149
Athlon64 X2 4200+ @ $183
Core 2 Duo E6300 @ $199.58
Athlon64 X2 4400+ @ 219.79
Athlon64 X2 4600+ @ 235.00
Core 2 Duo E6400 @ 239.58
Athlon64 X2 4800+ @ 279.74
Athlon64 X2 5000+ @ 294.74
Core 2 Duo E6600 @ 334.32
Core 2 Duo E6700 @ 553.26
Athlon64 FX-60/62 @ 811.00
Core 2 Duo X6800 @ 1021.68
Now, that gives us a few pairing(on which Athlon64 X2 processors always come out below) by which to compare: 4200+ vs E6300, 4600+ vs E6400, 5000+ vs E6600. Now, according to Anandtech's own Conroe vs Athlon64 benchmarks, the Athlon64 meets or exceeds the corresponding Conroe processor in probably 90% of the non-synthetic benchmarks. THIS is the important thing. I can take an FX-60 and just as easily compare it to a Pentium D 930 and say that the FX-60 is better, discounting the fact that the FX-60 is nearly 4x the cost.
The same is true here, and I've seen countless review sites doing it. Yes, the E6700 and X6800 processors outperform anything AMD has to offer, but they also cost more than anything AMD has to offer. AMD no longer has a processor at the $1000 price point, and down the line, AMD processors consistently cost less than Core 2 Duo processors they perform just as well against. It is this that is the important thing and it is this than none of the review sites or press has addressed. AMD is still quite competative dollar-for-dollar, particularly when you factor in motherboard cost.
coldpower27 - Tuesday, July 25, 2006 - link
Try meets or is slightly behind. AMD processors rarely "exceeds" in real world applications in the slightest.an E6600 is faster then the 5000+
an E6400 is faster then the 4600+
and E6300 is faster then the 4200+
as has been shown by Anandtech.