Information

Dear forum members,
We are delighted to inform you that our forums are back online. Every single topic and post are now on their places and everything works just as before, only better. Welcome back!



Discussion

Discussion on Article:
A Glance into the Future: Six-Core AMD Istanbul in a Desktop Platform

Started by: East17 | Date 10/19/09 02:17:30 AM
Comments: 9 | Last Comment:  08/31/11 06:41:29 AM

[1-9]

1. 
Great article Gavric!

This goes to show that clockspeed would have been much better for AMD than the huge cache and even the 6 cores if we're talking about the desktop/notebook market. On the servers market, I guess that the 6 MB L3 Cache was quite necessary and the 6 core are really proving their worth even against the Nehalem based Xeons. On the desktop side however, give me a 3,8 Ghz Phenom II without any L3 Cache any day against any Core 2 Duo and a 4 GHz Phenom II w/ 4 MB L3 Cache against the Core i7 series and I'll open my wallet! Still, there is a point to be made about using a DDR3 memory controller with at least 1333 MHz speeds. DDR2 667 MHz is really slow these days, especially against the Triple Chanel Core i920 DDR3. AMD really needs a new server platform. If we can get 5% more performance from using DDR3 1333 MHz unbuffered memory and 5% more from using a newer platform, we get roughly a 11% performance bump on the so called Thuban platform. So, adding 11% more performance would make a Thuban corresponding to the Opteron 2435 faster than the 3,4GHz Phenom II 965 in the following benchmarks : PCMark Vantage, 3D Mark Vantage, DivX 6.8.5, x264, CyberLink Media Show, Adobe premiere, 3D Studio MAX 2009, CINEBANCH, Maya, WinRAR, Resident Evil 5. Only in Adobe Photoshop and some games would Istanbul still be slower than the 3,4 GHz Phenom II 965 but only around 4% ~ 5% slower while the Phenom II 965 has a 30,7% higher clockspeed.
0 0 [Posted by: East17  | Date: 10/19/09 02:17:30 AM]
Reply

2. 
0 0 [Posted by: qcmadness  | Date: 10/19/09 04:03:55 AM]
Reply

3. 
This comparasion have no sense ,server CPU who made with other demands than desktop CPU.But ok and dont forget Thuban will be AM3 and AM2 compatible ,Nehalem have three channel memory ,still AMD users give less money than Intel users who must buy new motherboard with new Intel CPU ,example for Nehalem u need more memory (3 channel) for same perfomance in Maya 2009.So if AMD make Thuban for less money than Intel make Nehalem ,I dont see any reason why AMD not reveal Thuban.
0 0 [Posted by: Blackcode  | Date: 10/19/09 12:03:05 PM]
Reply

4. 
Hi,

nice idea:
Just two question

a) Concerning the Game tests:
Why are you testing with MQ ? Most Games apply additional Physics etc. calculations with (very) high quality. Thus in a CPU test, I would recommend to use always HQ.

b) Concerning Overclocking:
If you want to test your luck, use software ;-)
Because it is a nvidia Chipset, you can use nTune:
http://www.nvidia.com/object/ntune_5.05.54.00.html

Normally you can overclock ~10% i.e. 220MHz Hypertransport reference clock.

cheers

Bingle
0 0 [Posted by: Bingle  | Date: 10/20/09 01:07:51 PM]
Reply

5. 
You should really pull the review for the AMD 6 core down. There are far better boards out there for a couple of reasons:
1) That board doesn't seem to be available for purchase.
2) Its a 2+ year old board based on a 4 year old platform
3) Nvidia 3400 Pro is only HT1.0 capable. Istanbul loves HT3.0
4) Board doesn't support split power planes = higher wattage all around
5) Why didn't you use DDR2-800 Registered Ram? Kingston has been making it for over a year now. If the CPU supports it, it will run at that speed. Irregardless of what the MB manf says.
6) You could have stuck that processor by itself in a new Supermicro SR5690 DP board and it probably would have given you much more realistic results than that stone-age MSI board.
0 0 [Posted by: Hawkeye4077  | Date: 10/20/09 11:41:18 PM]
Reply

6. 
This is a horrible horrible comparison... Its only in the IT world that people can get away with something pointless and intellectually insulting as this. taking a workhorse product, sticking it in an outdated platform that wasn't even designed to handle it nor does its manufacturer give any indication that the platform even runs the CPU properly, and testing it against an entirely different class of products, completely out of its element. all in order to get a glimpse of the future.

Now imagine you were in the Auto industry and pulled this, for arguments sake, lets say you were Volkswagen and you decided to pull a stunt like this. Lets say you took a turbo charged motor out of a 2008 VW beetle, stuck that motor in a 1967 VW Microbus and then tested the hell out of it in order to get a glimpse of what the 2011 Porsche Roadster will be capable of doing.

Not only would you would get laughed all the way out of Germany if you tried something like that, your sanity would get checked on the way out...
0 0 [Posted by: Wallyb132  | Date: 10/22/09 02:28:13 AM]
Reply

7. 
I imagine that Thuban will have support for 1066, 1333 and possibly 1600Mhz DDR3 which Istanbul does not, Istanbul supports up to 800Mhz registered DDR2. This should help to make Thuban slightly faster. Hopefully AMD will keep the memory controller link at current Opteron speeds rather than drop it down to Phenom II speeds, 2200MHz v 2000MHz respectively.
0 0 [Posted by: GavinT  | Date: 10/24/09 06:26:21 AM]
Reply

8. 
i agree with Wallyb132.. please xbitlabs renew this Review.
0 0 [Posted by: 3Dkiller  | Date: 10/25/09 04:13:00 PM]
Reply

9. 
k9nu speedter(ms-9655) bios P9257MS V2.91 Istanbul CPU list
e-mail send P9257MS V2.91 ungplus@gmail.com thank.
0 0 [Posted by: ungplus  | Date: 08/31/11 06:41:29 AM]
Reply

[1-9]

Back to the Article

Add your Comment