AMD Processors
Decrease font size
Increase font size
Topic Title: Current memory technologies holding AMD64 product
Topic Summary:
Created On: 10/06/2003 06:46 PM
Status: Read Only
Linear : Threading : Single : Branch
<< 1 2 Previous Last unread
Search Topic Search Topic
Topic Tools Topic Tools
View similar topics View similar topics
View topic in raw text format. Print this topic.
 10/09/2003 02:01 AM
User is offline View Users Profile Print this message

Author Icon

Posts: 55
Joined: 10/07/2003

QUOTE (Support 1 @ Oct 8 2003, 09:40 PM) This thread is fine. What I was referencing was some of the other discussions taking place in the Gen. Technology Chat. Such as "Where are you from" or "Hello", etc. These are discussions that are not really about technology at all and just people introducing themselves. Other than getting to know one another, they really do not seve any purpose in a Technology based discussion forum.
I agree totally but you have to understand these are new forum's and it's kinda like a party in the begining everyone say's hello whats your name my name is my wife look's like and I love to overclock how about yourself etc.

XP2500-m------------------------A643000NC{No IHS}
BFG 5900-------------------------EVGA6800
530 forton------------------------530 forton
SLK 800---------------------------SLKU948
 10/09/2003 02:21 AM
User is offline View Users Profile Print this message

Author Icon
Senior Member

Posts: 296
Joined: 10/06/2003

QUOTE (Ardrid @ Oct 8 2003, 09:30 PM) I'm going to have to agree with Youri on this. The Athlon 64 and Athlon 64 FX simply do not need that much bandwidth right now. Just because the processors have an onboard memory controller doesn't mean that you have to feed them with as much bandwidth as possible. The K8 architecture is still fundamentally the same as the K7 architecture, meaning it's not starving for bandwidth like the P4 is, despite the fact that the "FSB" now runs at the speed of the processor. This is simply because the chip is only running at 2-2.2GHz and is being fed plenty of information, if not more than is needed. Even the XP doesn't really need dual-channel memory, as 3.2GB/s of bandwidth is plenty in most cases.

A prime example of what I'm talking about is a comparison between a 2.2GHz Athlon 64 and a 2.2GHz Athlon 64 FX. As you know, one is single-channel and one is dual-channel, meaning that one has double the bandwidth of the other. The latencies are also essentially identical. In all of the benchmarks, save one, which I believe was data compression, the additional bandwidth made no difference whatsoever. This tells me one thing, that the Athlon 64, at it's current speed, has plenty of bandwidth using DDR400.

That being said, I see no need for Rambus' Yellowstone technology, or any RDRAM technology for that matter, being coupled with the Athlon 64. DDR provides all of the bandwidth this processor will ever need and DDR-II will take over from there. You have to remember, theoretical bandwidth measurements are synthetic and do not always equate to real world performance. Just because the Athlon 64 FX has more bandwidth than the Athlon 64 doesn't mean that it's going to perform better at the same clock speed. And as the benches show, this holds true more often than not.

This thread does bring up an interesting point for discussion though. If the benches show that there is no real discernible difference between having single-channel versus dual-channel, than how does AMD plan on qualifying the excessive price difference between the A64 and the FX once the A64 hits the same clock speed? The only way the price difference can continue to exist is if the FX is always 200MHz higher than the A64. In that sense, you would be buying the latest and greatest before it exists in the form of the A64. Otherwise, there's no real point in buying the FX. Some food for thought...

Athlon 64 & Athlon 64 FX Comparison #1' ">
Ardrid you have made many good points,

You are correct that at this point in time XDR is not exactly necessary, but you have to relies the way memory speed has increased in the past year. The way speeds have been in the last year really do make XDR look quit attractive. I would say that in the next six months to a year you would consider XDR a viable solution. Yes at these clock speeds DDR400 is just able to feed the beast but what about when they get faster or when AMD comes out with a core revision that makes their processors even more efficient. Those situations would require more bandwidth, and lets face it DDR towards the end of its life wasn’t exactly the most compatible technology, as some memory would work here and not there. And DDRII from what I have heard will probably suffer from the same problems because JEDEC dropped the ball on finalizing specifications. Anyways I personally would rather upgrade to a technology with enough head room so that we wont have to switch standards every two or three years because the industry adopted a sub par technology. I think you are getting what I am trying to say but anyways I do not agree with you when you say that the hammers wouldn’t benefit from more bandwidth. I personally believe that if you give it to them they will utilize it and that is what XDR would do. As you can see we have different views of how the on-die memory controller and processor clock speed FSB affect the processors ability to utilize bandwidth, but that is ok and only time will tell who is right.


SolTek Mainboards for life. check them out on the web ----> HERE' ">
112018 users are registered to the AMD Processors forum.
There are currently 0 users logged in.

FuseTalk Hosting Executive Plan v3.2 - © 1999-2015 FuseTalk Inc. All rights reserved.

Contact AMD Terms and Conditions ©2007 Advanced Micro Devices, Inc. Privacy Trademark information