Intel Core 2 Extreme X6800 Preview from Taiwan
by Anand Lal Shimpi & Gary Key on June 6, 2006 7:35 PM EST- Posted in
- CPUs
Memory Latency and Bandwidth
We've never been able to look at some of the low level characteristics of Intel's Core architecture, and although we didn't have enough time to do a thorough run of low level benchmarks we were able to run ScienceMark 2.0 in order to get an idea of how the Core 2 Extreme stacked up against the FX-62 in terms of memory latency and bandwidth.
We had seen Conroe performance results that showed the new architecture being able to offer fairly competitive memory access latencies to AMD's architecture, without the need of an on-die memory controller. Our ScienceMark 2.0 results confirm just that:
While AMD still offers lower memory latency, the Core 2 Extreme X6800 is very close in comparison - especially considering that it has no on-die memory controller. With lower clock speeds than its Pentium D siblings and a faster FSB, memory access latency is reduced tremendously with Conroe. On a larger scale, through a very effective cache subsystem as well as memory disambiguation, Conroe can offer significantly improved memory performance compared to its predecessors, including the Athlon 64 X2/FX.
ScienceMark's memory bandwidth results offer a very telling story, showing us the bandwidth limitations of Intel's FSB architecture. While the FX-62's peak theoretical bandwidth is not achieved in real world, you can see how AMD's Direct Connect architecture offers higher limits for chip-to-chip communication.
Facebook
Twitter
RSS
134 Comments
View All Comments
bob661 - Tuesday, June 6, 2006 - link
To be honest, I thought it would be a bit more than shown here but it's still pretty good. If I was an OCer I'd jump on this. Since I'm not, I won't be buying one for myself. My wife will get one since I'm sure some of my customers will be interested in this platform and I want to be familiar with it.thestain - Tuesday, June 6, 2006 - link
You need higher resolutions and everything turned on to highest imnage quality settings to test this and.. the fsb limit will be more than likely exposed by this sort of test. Test it! Yes, Battlefield 2, etc.. could bring the system to a crawl, so.. ?? does this mean it is not important?I have a Dell 2405.. set to 1920X1200... I don't change this for games.. and most still play... some crawl.. a little, but.. I am a prospective buyer for the top cpus from Intel and AMD.. and.. while Conroe looks awfully impressive, it worries me that real world enthusiast class gaming test are not being done.
No excuses.. they need to be done!!
Try 1920X1200 resolution.. if you need to put the cards in SLI on both boards.. and this is important.. since Intel chipsets limit SLI.. but it still needs to be tested.
Turn on everything to the highest setting.. everything.. and lets see how Intel's cpu handles this kind of stress on system memory.
and.. if Anandtech and Intel are afraid to do this.. it is very telling indeed.
Mike
JarredWalton - Tuesday, June 6, 2006 - link
First, limited testing time and hardware limits what benches can be run. Second, SLI doesn't work on anything but NVIDIA chipsets, which don't exist for Core 2 Duo yet, so single GPU results are the only option for apples-to-apples. (CrossFire should work on 975X Core 2 Duo, but wasn't available at the time.) Third, I have a 2405FPW as well, and I can run everything from 2.0 GHz X2 3800+ to 2.6 GHz FX-60 with a single 7900 GTX GPU and get essentially the same performance at 1920x1200.If you're only going to play games at WUXGA, massively upgrading processor performance won't help. End of story - at least until GPUs that are twice as fast become available. This is not an excuse; it is reality. Benchmarking CPUs under GPU-constrained situations doesn't make sense. You can see the results of CPU scaling with a single 7800 GTX in http://www.anandtech.com/cpuchipsets/showdoc.aspx?...">this X2 overclocking article; 7900 GTX would scale better, but eventually you still run into GPU limitations.
None of the above changes the fact that Core 2 Extreme is substantially faster than FX-62 in current applications. SLI/CF support in the future should actually benefit more, as there's CPU overhead associated with such configurations. Video encoding will also benefit greatly from Conroe's improved architecture.
In regards to your above comment about 2004 benchmarks, 2005/2006 versions of SysMark and Winstones do not exist, which is why we continue to use older versions. We would love to run PCMark06 and a bunch of other benchmarks (we had some of them available), but time constraints did not allow for comprehensive testing. Rest assured that we have seen nothing from Conroe that indicates serious problems in any applications; at worst it is slightly outpacing FX-62 performance, and at best it can be 50% faster. In general, a 20-25% performance advantage over FX-62 is common.
DigitalFreak - Wednesday, June 7, 2006 - link
Yeah, what he said. :-)neweggster - Tuesday, June 6, 2006 - link
No offense Mike but did you say enthusiast? DELL is no where part of that word. LOL DELL, just the sound of them makes me gasp. In all good fun and humor, no offense. =)Calin - Wednesday, June 7, 2006 - link
That Dell is the monitor. I think even real enthusiasts have nothing against Dell monitorspeternelson - Tuesday, June 6, 2006 - link
In the table of processors, 805 and 820 are shown as "2Mx2". This is incorrect.
Both of these processors only have "1Mx2". Please amend.
Also weren't there some other new low end procs like 925 on the Intel roadmap, due to launch around the same time as Conroe? These may be more attractive than 805 or 820, neither of which can do virtualisation.
eRacer - Tuesday, June 6, 2006 - link
The Pentium D 945 (3.4GHz, 2x2MB, no VT) should be on that list as well. It is on the July 23rd price list for only $163.http://www.hkepc.com/bbs/attachments_dir/ext_jpg/p...">Intel price list
thestain - Tuesday, June 6, 2006 - link
2004 benchmarks.. Don't you have 2006"Poeple who pay the top price to buy these cpus will not be playing games or running applications at 1024 x768,
At least test these two systems at 1600X1200 or higher.
Do you have any multi-threaded test? Something newer?
Anantech is testing the Core 2 Duo at where it will perform best and this is somewhat suspicious. Does anyone who spends the kind of money people will dish out for the x6800 really use an old, rusty monitor with 1024 X768 resolution?
Conroe might kick butt, but at least give a platform where it goes beyond being tested for single core performance at high resolution, ok??
Some of us would like to see how it can perform when breaking a sweat in a multi-tasking environment..
and.. those early memory bandwidth results.. sure are not consistent with the rest of the results are they, any ideas as to why?
rqle - Tuesday, June 6, 2006 - link
i can benchmark a 1ghz cpu with SLI/Crossfire at 2500x2500 and compare it to an AMD64/conroe and claim the AMD64/Conroe is only 3FPS faster OMG WTH.isolating variable, isnt that what you learn in 4th grade science project.