The new central processing units from Advanced Micro Devices are expected to deliver about 5% higher performance in certain tasks, if the early benchmarks of a new AMD Athlon 64 chips published by a web-site are correct.
AMD’s Upcoming Chips May be Faster than Today’s
According to a report at XtremeSystems.org web-site, the new AMD Athlon 64 3500+ processors made using 90nm fabrication process work about 5% faster than the same 3500+ chips produced using 130nm process technology in CPUMark99 benchmark. The benchmark measures CPU integer unit and data processing speeds. Another benchmark the web-site had posted, SiSoft’s Sandra, showcased that the forthcoming AMD64 chips are approximately as fast as the current offerings.
It is not clear which improvements have been made to the processing core, but earlier this week another web-site said the new AMD Athlon 64 would have some speed improvements compared to the current models. While AMD is now quiet about peculiarities of its 64-bit chips produced at thinner fabrication process, earlier this year representatives for the Sunnyvale, California-based chipmaker said that quite natural capabilities of future-generation Athlon 64 and Opteron microprocessors could be SSE3 technology, improved pre-fetch mechanisms as well as thermal throttling. AMD did not disclose any actual time-frames for this improvements to arrive.
AMD’s 90nm Transition On-Track
According to the recently unveiled roadmap, the Sunnyvale, California-based chipmaker plans to release a number of AMD Opteron processors code-named Athens, Troy and Venus, a Mobile Athlon 64 processor known as Oakville and a desktop Athlon 64 processor code-named Winchester produced using 90nm fabrication process in the second half of the year. The initial commercial shipments of AMD64 chips produced at 90nm nodes are officially said to commence in the third quarter of 2004.
In the first half of next year AMD will release the successor of its AMD Athlon 64 FX chip with core code-named San Diego along with a lineup of mobile microprocessors. In the second half of the year the company will release dual-core chips. AMD believes that the current AMD Athlon 64 FX microprocessor made using 130nm technology will be able to scale for at least one speed-bin required to compete successfully with the rival Intel Corp..
According to unofficial sources, AMD’s Athlon 64 FX-57 processor at 2.80GHz made using 90nm Silicon-on-Insulator process technology will be released in Q2 2005. A slightly slower AMD Athlon 64 FX-55 chip at 2.60GHz will be available in Q4 2004.
In the second half of next year AMD will initiate production of dual-core processors for various market segments using 90nm SOI process.
Representatives from Advanced Micro Devices said the company started shipments of samples products using 90nm process technology in Q2 2004 and is on track to begin volume commerical production of such devices in Q3 2004.
Officials for AMD did not comment on the news-story.
Comments currently:
8
Discussion started: 07/15/04 05:40:20 PM
Latest comment: 07/22/04 06:15:23 AM
Expand all threads |
Collapse all threads
[1-7]
1.
I love how when AMD releases a new CPU is it always faster than the last generation. Unlike intel who's new chip is always slower.
[
Posted by: Stuke
|
Date: 07/15/04 05:40:20 PM]
2.
That statement is a bit too generalised.
The reason is because of manufacturing in the 90nm process.
For Intel, its like a huge brickwall. For AMD, its a really high hill that has a brickwall at the top of the hill .
What am I getting into? SOI process used by AMD/IBM has delayed what Intel is experiencing right now.
Everyone is suffering the same problem with 90nm. Its how they approach to solve it. Or at least delayed the problem until a good solution is found.
The 90nm process has effectively slowed speed increases of ALL CPUs to a grinding halt. So, its not just Intel. IBM is suffering the same problems with production of the new 90nm G5 chip for Apple. (Notice how they have to water cool the dual CPU setup?)
I wouldn't say AMD ALWAYS releases faster CPUs, as you can see, they're overlapping two CPUs from two different manufacturing processes. (one being 130nm and the other 90nm)...Both are 3500+...
What I'm waiting for is Low Power variants of the Opterons or Pentium-M variants with Dual-core. (I'm hoping for around 2.0Ghz to 2.4Ghz).
[
Posted by: 22
|
Date: 07/15/04 08:43:00 PM]
3.
I understand the issues, certainly. What is bothersome is the nearly 3/4 of a year it is going to take to even see the very first versions of 90nm cpu's. Those improvements are welcome, but it's likely to be Nov or later before we're going to have 90nm AMD64's in qty and pci-e motherboards. In fact for the FX series we aren't going to see any of these improvements until next year.
You can well imagine now that the promise of dual cores on your desktop in 2005 from AMD is a pipe dream that's never going to happen. Ok to be fair, it won't really happy till mid 2006, but it will be for reasons they won't announce until May 2005.
For this reason I've been pushed back to the other side of the fence, dealing with something Intel for now, and hoping for a drop in 775 Pent M variant dual core in 2005.
It's funny but by the time AMD manages to get 90nm top to bottom, Intel will likely be switched over to 65nm.
Oh well - at least the FX on 130 is an ok chip.
[
Posted by: Anemone
|
Date: 07/16/04 03:36:53 AM]
4.
Don’t forget the best thing on going to 90nm will be the final price of the CPU, for amd and us.
I don’t think amd is having issues with 90nm, the problem is having the same facility to manufacture 90nm and 130 nm at the same time or place if you want to put the things that way, so transaction is much slower.
I don’t remember when, but AMD already have beaten intel on .25 or .18 with the K6 (2) processor.
I think AMD 90nm will be better over Intel and IBM one, since they will use low-k like TSMC on ati radeon products, and amd have been pointing to 25W/35W which is a good pronounce.
In clock speed increase I don’t think amd is worried with that if they can deliver 5% speed increase at the same working frequency, it's already very good. At the best of the best new P4 with 2MB cache will deliver the performance level of current P4EE processors.
[
Posted by: I
|
Date: 07/16/04 09:16:58 AM]
5.
I gave up on Intel and their 3.6Ghz Prescott back in March when they announced 64bit support would be much later. That shift from Intel to AMD has saved me a lot of heart ache from all the negative Intel announcements since March that are still coming out.
Then I dropped plans for a 2.6GHz FX-55 when I realized it wouldn't ever be 90nm and could possibly come out after 2004. I read a comparison of FX vs AMD64 recently that claimed 1MB cache over 512KB only provides 3% increase in performance (in today's appliacations) but at a much higher cost.
So now (pending benchmarks) I'm waiting for a 90nm 2.6Ghz AMD64 4000+ and mobo with PCI-Express (possibly Nforce4.) This should be available sometime between August and December. Theinquirer today announced that Toledo will be 939 pin compatible so I can still upgrade to a Toledo dual core FX-57 or better at some point down the road when the 4000+ is no longer competitive. I have no confidence that Intel will bring anything comparable to Toledo earlier and without problems. That only leaves DDR2 vs DDR. But DDR is still getting faster, so hopefully by the time DDR2 provides significant performance gains it'll be time to buy a new computer again. Maybe a newer faster hypertransport standard will be available then too.
Unlikely as it is, I still wish for a PCI-Express version of an ATI All-In-Wonder or comparable multimedia-friendly Nvidia card to be available in this time frame. Also HDTV compatability with component input/output would be really nice, but that's all probably a year or more into the future.
[
Posted by: RolandRJB
|
Date: 07/16/04 12:28:55 PM]
+ expand thread (1 answer)
- collapse thread
My mistake. A Dual Core Toledo and the FX-57 are not the same chip.
What I should've said was, "...I can still upgrade to a 90nm FX-57, dual core Toledo, or something even better at some point down the road when the 4000+ is no longer competitive."
[
Posted by: RolandRJB
|
Date: 07/16/04 01:36:39 PM]
6.
I'm still more of an Intel fan...But I'm not a fan of the P4...
I still use PIII/Celeron "Tualatin" CPUs...They're bloody cheap!
I favour the Pentium-M, I wish Intel would just throw the P4 away already! :)
Oh well, I guess I'll wait until the dual-core one comes.
It also helps because I use Linux and BSD, and I'm not interested in the "latest and greatest" games. (I'm still using a GF2 MX400 !!!)
[
Posted by: 22
|
Date: 07/17/04 10:42:45 AM]
7.
I guess if we are destined to not have the FX-57 (or whatever future # will refer to the 90nm part), it would be nice if they could incorporate the 5% improvements into the 130nm FX-55. I've made my decision on the FX, probably the 55, and the NF3 250gb boards which are showing some very nice promise.
If the 55 would show some sincere improvements, maybe SSE3 and the other items that are contributing to this nice 5% increase in the chip we see now, it would help keep people happy and interested in the 55 till we can see 90nm or dual core chips next year.
Intel, after this whole 925X and yet further delays of the 64bit technology, can put that whole fiasco in their pipe and smoke it for all I care.
However, AMD would be wise to show they aren't resting on their laurels and are still actively advancing the game.
We can hope!
[
Posted by: Anemone
|
Date: 07/22/04 06:15:23 AM]
[1-7]
Make Home Page
Add to Favorites
XbitLabs is using Twitter
Windows Live Alerts
