Intel's Core 2 Extreme & Core 2 Duo: The Empire Strikes Back
by Anand Lal Shimpi on July 14, 2006 12:00 AM EST- Posted in
- CPUs
Memory Latency: No Integrated Memory Controller Necessary
One thing we pointed out in our earlier preview coverage of Intel's Core 2 Extreme is that the new processors have extremely low latency memory access, despite relying on the same memory controller as the previous generation of Intel CPUs.
Without an on-die memory controller Intel's Core 2 processor must use the memory controller in its chipset, which currently means the 128-bit DDR2 memory controller in either Intel's 965 or 975X chipset. The confusing thing is that although the Core 2 processors use the same memory controller as the old NetBurst processors, memory latency has been improved tremendously:
Intel's Core 2 processors now offer even quicker memory access than AMD's Athlon 64 X2, without resorting to an on-die memory controller. While Intel will eventually add one, the fact of the matter is that it's simply not necessary for competitive memory performance today thanks to Intel's revamped architecture. Update:As many astute readers have pointed out, Core 2's prefetchers are able to work their magic with ScienceMark 2.0, which results in the significant memory latency advantage over AMD's Athlon 64 FX-62. This advantage will not always exist; where it doesn't, AMD will continue to have lower latency memory access and where it does, Intel can gain performance advantages similar to what ScienceMark 2.0 shows.
| CPU | Everest |
CPU-Z 1.35 (8192KB, 128-byte stride) | CPU-Z 1.35 (8192KB, 64-byte stride) |
| AMD Athlon 64 FX-62 (2.8GHz) | 45.9 ns | 43.2 ns | 19.3 ns |
| Intel Core 2 Extreme X6800 (2.93GHz) | 59.8 ns | 52.8 ns | 10.9 ns |
Because of the Core 2's intelligent prefetchers doing too good of a job with ScienceMark memory latency test, we wanted to also showcase situations where Core 2 would feature slower memory accesses than the AMD platform with its integrated memory controller. Everest's results are more in line with what we'd expect to see, with the FX-62 offering over 23% faster memory accesses than the X6800. CPU-Z's latency tool also reported somewhat similar findings, with an 18% performance advantage due to AMD's integrated memory controller. CPU-Z also provided us with numbers that showcase how well Core 2 can perform if its prefetchers are able to "guess" correctly; at lower strides the Core 2 Extreme manages faster memory access than the FX-62. The 128-byte stride numbers are indicative of what will happen if the pre-fetchers are not able to get the Core 2 the data it needs, when it needs it, while the 64-byte numbers show you what can happen when things go well.
Facebook
Twitter
RSS
202 Comments
View All Comments
coldpower27 - Friday, July 14, 2006 - link
Are there supposed to be there as they aren't functioning in Firefox 1.5.0.4coldpower27 - Friday, July 14, 2006 - link
You guys fixed it awesome.Orbs - Friday, July 14, 2006 - link
On "The Test" page (I think page 2), you write:please look back at the following articles:
But then there are no links to the articles.
Anyway, Anand, great report! Very detailed with tons of benchmarks using a very interesting gaming configuration, and this review was the second one I read (so it was up pretty quickly). Thanks for not saccrificing quality just to get it online first, and again, great article.
Makes me want a Conroe!
Calin - Friday, July 14, 2006 - link
Great article, and thanks for a well done job. Conroe is everything Intel marketing machine shown it to be.stepz - Friday, July 14, 2006 - link
The Core 2 doesn't have smaller emoty latency than K8. You're seeing the new advanced prefetcher in action. But don't just believe me, check with the SM2.0 author.Anand Lal Shimpi - Friday, July 14, 2006 - link
That's what Intel's explanation implied as well, when they are working well the prefetchers remove the need for an on-die memory controller so long as you have an unsaturated FSB. Inevitably there will be cases where AMD is still faster (from a pure latency perspective), but it's tough to say how frequently that will happen.Take care,
Anand
stepz - Friday, July 14, 2006 - link
Excuse me. You state "Intel's Core 2 processors now offer even quicker memory access than AMD's Athlon 64 X2, without resorting to an on-die memory controller.". That is COMPLETELY wrong and misleading. (see: http://www.aceshardware.com/forums/read_post.jsp?i...">http://www.aceshardware.com/forums/read_post.jsp?i... )It would be really nice from journalistic integrity point of view and all that, if you posted a correction or atleast silently changed the article to not be spreading incorrect information.
Oh... and you really should have smelt something fishy when a memory controller suddenly halves its latency by changing the requestor.
stepz - Friday, July 14, 2006 - link
To clarify. Yes the prefetching and espescially the speculative memory op reordering does wonders for realworld performance. But then let the real-world performance results speak for themselves. But please don't use broken synthetic tests. The advancements help to hide latency from applications that do real work. They don't reduce the actual latency of memory ops that that test was supposed to test. Given that the prefetcher figures out the access pattern of the latency test, the test is utterly meaningless in any context. The test doesn't do anything close to realworld, so if its main purpose is broken, it is utterly useless.JarredWalton - Friday, July 14, 2006 - link
Modified comments from a similar thread further down:Given that the prefetcher figures out the access pattern of the latency test, the test is utterly meaningless in any context."
That's only true if the prefetcher can't figure out access patterns for all other applications as well, and from the results I'm pretty sure it can. You have to remember, even with the memory latency of approximately 35 ns, that delay means the CPU now has about 100 cycles to go and find other stuff to do. At an instruction fetch rate of 4 instructions per cycle, that's a lot of untapped power. So, while it waits on main memory access one, it can be scanning the next accesses that are likely to take place and start queuing them up and priming the RAM. The net result is that you may never actually be able to measure latency higher than 35-40 ns or whatever.
The way I think of it is this: pipeline issues aside, a large portion of what allowed Athlon 64 to outperform NetBurst was reduced memory latency. Remember, Pentium 4 was easily able to outperform Athlon XP in the majority of benchmarks -- it just did so at higher clock speeds. (Don't *even* try to tell me that the Athlon XP 3200+ was as fast as a Pentium 4 3.2 GHz! LOL. The Athlon 64 3200+ on the other hand....) AMD boosted performance by about 25% by adding an integrated memory controller. Now Intel is faster at similar clock speeds, and although the 4-wide architectural design helps, not to mention 4MB shared L2, they almost certainly wouldn't be able to improve performance without improving memory latency -- not just in theory, but in actual practice. Looking at the benchmarks, I have to think that our memory latency scores are generally representative of what applications see.
If you have to engineer a synthetic application specifically to fool the advanced prefetcher and op reordering, what's the point? To demonstrate a "worst case" scenario that doesn't actually occur in practical use? In the end, memory latency is only one part of CPU/platform design. The Athlon FX-62 is 61.6% faster than the Pentium XE 965 in terms of latency, but that doesn't translate into a real world performance difference of anywhere near 60%. The X6800 is 19.3% faster in memory latency tests, and it comes out 10-35% faster in real world benchmarks, so again there's not an exact correlation. Latency is important to look at, but so is memory bandwidth and the rest of the architecture.
The proof is in the pudding, and right now the Core 2 pudding tastes very good. Nice design, Intel.
coldpower27 - Friday, July 14, 2006 - link
But why are you posting the Manchester core's die size?What about the Socket AM2 Windsor 2x512KB model which has a die size of 183mm2?