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
Anand Lal Shimpi - Wednesday, June 7, 2006 - link
You can't really compare our test results to those found on other sites simply because I'm not sure what testing methodology was employed, as well as due to stated differences in hardware (e.g. HDD differences can impact Office Productivity tests considerably). You'll notice that our numbers are in line with what we've published previously for SYSMark 2004.The one thing I can guarantee is that the AMD and Intel systems we tested were as close to the same spec as possible so the numbers are directly comparable to one another, which is what matters the most.
Take care,
Anand
fikimiki - Wednesday, June 7, 2006 - link
I understand 3% difference, but not 20% because of WD Raptor and SLI configuration. RAM is the answer so I wait for benchmark with CL4 and CL3.It is also important to see 64-bit performance for both competitors.
Gary Key - Thursday, June 8, 2006 - link
It is also the answer for Intel to some degree. I think you will understand this once the NVIDIA 590 Intel memory benchmarks are available along with production level 965/975 boards capable of running the Core 2 Duo at 3-2-2-8 at 800MHz, something AMD is not capable of yet. ;-) Our initial memory tests were set to each board's default settings, the way a typical system will delivered to the customer, this ensured there were not any "unbiased" tweaks used for either system.
Accord99 - Thursday, June 8, 2006 - link
Why do you expect it to be more, not when CAS4 vs CAS5 has barely shown more than a few percent benefit. Here a couple of reviews using CAS4 DDR2-800. The FX-62 scores 220 and 219 respectively:http://www.hwupgrade.it/articoli/cpu/1494/socket-a...">http://www.hwupgrade.it/articoli/cpu/14...2-le-nuo... (with 36GB Raptor)
http://www.hardwarezone.com/articles/print.php?cid...">http://www.hardwarezone.com/articles/print.php?cid... (with Seagate Barracuda 7200.7 80GB)
neweggster - Tuesday, June 6, 2006 - link
Just read Toms Hardware review and from that comparison to this in FEAR min FPS it looks like the 2.66 vs the FX-62 is only a small difference of around 5fps with conroe wining. Based on that and Anands article I think its clear to say that a 2.93 conroe only going around 270mhz more over its 2.66 offers more performance in terms of clock cycle performance then what the FX-62 clock for clock can do. If you see the results you can see that the 2.66 conroe is 58fps on min FEAR bench and Fx-62 is 53, and they also have the FX-62 there OC to 3.00 GHz and its result was 55fps, notice the OC FX-62 gains only 2 FPS in min fps FEAR bench and the Conroe 2.93 vs the Fx-62 gains 27 fps in the min fps FEAR bench.That tells me that the increase in clock for clock performance on conroe scales much higher in performance then AMD FX-62. The FX-62 on Toms was running at 200 MHz more and only 70 MHz difference in the 2.66-2.93 conroe. Also keep in mind Toms review uses a RX1900 and Anand uses a 7900GTX, some performance difference there on those 2 video cards as well, also note Toms uses a 1280x1024 resolution using high settings.
It seems impressive that although each benchmark isn't identical that the conroe can increase in performance in a single benchmark just by going to the 2.93 from 2.66, very impressed how well it scales in comparison to FX-62.
So a 2.66 core 2 duo going to 2.93 core 2 duo extreme offers this much gain vs almost same clock increase on Fx-62 OC to 3.00 GHz is great, assuming im clear on the 2.66-2.93 is only a increase in clock and nothing else? Can you let me know if its just an increase in clock from 2.66 to 2.93 or is it clock and cache or something else in there? Im just trying to make sense of the fact that briefly comparing TOMS results to yours im seeing a better increase in performance when clock is increased on conroe over FX-62, thanks. If this is the case I can't wait for conroe 2.93 core2duo extreme so I can get my OC on this bad boy.
Which leads me to this. You guys at Anand had a chance to test this OC'ing the FX-62 by any and comparing? It would be great if you can elaborate on the difference in OC performance on each.
ashyanbhog - Wednesday, June 7, 2006 - link
according to Anandtech, the conroe is expected in Q306, how many more days before AMD 65nm parts start rolling out? hmm not muchhttp://www.anandtech.com/cpuchipsets/showdoc.aspx?...">http://www.anandtech.com/cpuchipsets/showdoc.aspx?...
cant wait till somebody gets their hand on a 65nm Athlon and benchmarks it..... should see how Intel's oh so wornderful processor fares then! AMD was able to extract good performance jumps when moving from 130nm to 90nm, if they r able to do the same when moving to 65nm, they have Intel in a spot again!
130nm to 90nm gains
http://www.techreport.com/onearticle.x/7417">http://www.techreport.com/onearticle.x/7417
zsdersw - Wednesday, June 7, 2006 - link
Apparently you don't realize that the speeds of Core 2 Duo at launch are not the only speeds that will be available as time progresses. Intel isn't just going to leave Core 2 Duo at the speeds available at launch.
Accord99 - Wednesday, June 7, 2006 - link
So far, AMD only got an extra 400MHz/15% out of the transition out to 90nm. Conroe is expected out in late July, AMD's 65nm parts may be at the end of the year, and at lower clock speeds than the fastest 90nmp parts.termix - Wednesday, June 7, 2006 - link
400MHz for SC, there are gona be 3GHz Opterons. But count DC chips too. Its 2 times bigger , so 2.6GHz vs 2x2.8GHz for avaible parts...Makaveli - Tuesday, June 6, 2006 - link
impressive not really, just shows all the benchmarks we were seeing before were close to reality. minus the 40% faster when it closer to 20% faster on avg.Does this matter to me no. Will I still be buying a dual core opteron to put in my 939 board yes. I'll check back again in 18-24 months what is on the market then, until then just videocard upgrades for me. All u fanboys can fight over who has the better processor and larger e-penis till the next century. Does intel or AMD care probably not, why cause they make there money from the system builders.
Boys and there toys, you would think things would change as some of u guys get older. nope just like grade school.
Kudos to intel for finally making something people want to buy again.
Kudos to AMD for waking the sleeping giant.
This is good for us all, can u say cheaper cpu's.
Now i'm going to bed, Gnite all and may u all wake up 2morrow and see there are more important things in life!