• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Will the athlon 64 90nm suck?

K

Khyron320

Senior member
Im confused about 90nm...

amds 90nm chips are very close to being out

is 90nm going to make amd64s run hotter or cooler?

when prescott was coming out everything i knew about die shrinks was thrown out the window.
I though die shrinks meant cooler running cpus... but instead we got the bacon cooker known as prescott
 
AMD said over a year ago they had working protoypes of 90nm opterons that were using in the ballpark of 35-55watts, less than the current ones.
 
i think generally that smaller trasistors means less resistance means less heat produced. Prescott's heat problem is due to its more transistors and the almost 50% longer pipeline I don't see any major archetectural changes in the 90nm athlons so the should be cooler
 
When IBM transitioned the G5 to the 90nm-SOI process heat dissipation went from 51W@1.8GHz to 25W@2GHz. Let's hope we see something similar for the Opteron/A64 :thumbsup:
 
Hmm, hard to say really... I suspect it won't be as good as it expected. Every Semi-conductor manufacuter is having a hard time it current leakage on 90nm process @ high clock speeds. Intel(prescott), IBM (they promised a 3.0 GHz PowerPC for by the end of this year for apple. That isn't going to happen)...

I don't see how AMD will do significantly better than the two leaders.

The next althon 64 (4000+) is most likely going to be a 90nm part. The current 130nm A64 reach their limit at around 2.45GHz to 2.6GHz on air cooling.
 
Originally posted by: anthrax
Hmm, hard to say really... I suspect it won't be as good as it expected. Every Semi-conductor manufacuter is having a hard time it current leakage on 90nm process @ high clock speeds. Intel(prescott), IBM (they promised a 3.0 GHz PowerPC for by the end of this year for apple. That isn't going to happen)...

I don't see how AMD will do significantly better than the two leaders.

The next althon 64 (4000+) is most likely going to be a 90nm part. The current 130nm A64 reach their limit at around 2.45GHz to 2.6GHz on air cooling.
Click to expand...



The problem with Intel and I think IBM is they are using a strained silicon materials in their chips that have greater leakage. AMD is not doing that and the EARLY .09 amd chips show to run cooler and faster (at same Mhz of current .13) then the current .13 chips.
 
well hopefully amd will be able to hit 3.0-3.2 range with .09 if they can do that its going to take atleast a 4.5ghz p4 to keep up and I don't think the prescott has that in her
 
Originally posted by: iwantanewcomputer
i think generally that smaller trasistors means less resistance means less heat produced.
Click to expand...
Wouldn't a smaller transistor have higher resistance in the same way a smaller wire does? But I'm not sure if DC resistance is the biggest issue in an IC anyway.
 
From what I've been reading around (and I read stuff on AMD a LOT), AMD's 90nm SOI process is currently very good, and much better than Intel's and AMD. First 90nm chips are already shipping for Mobile Athlon64s, with Desktop Athlon64s coming out in a month or so. Initial reports are that heat dissipation is comparable to, or even better, than at the 130nm process.

I expect AMD's 90nm chips to be even better than current chips, and probably be cooler/heat-efficient than 130nm ones. Heck, the first ones are for laptops! Do you think AMD would be doing this if these puppies sucked up 100W+, a-la Prescott?

Didn't think so.
 
AMD have gone the SOI route as its much more power efficient in terms of leakage. I believe this is one of Intel's main problems moving to 0.09 with strained silicon. IBM are working with AMD on SOI, and seem to be having sucess with POWER5 series chips, so i'd guestimate that the AMD 90nm will come out ok.
 
FAB 30 is doing a great job, I just hope we can 3.2Ghz out of the 0.9 micron process, Intel would really be stuck, i wonder why the next fx (55) is going to be fabed on 130 nm process, surely they'd like to save money on die space ? or are they already in production ?
 
i seriously hope and pray its as good as its cracked up to be! imagine a 25watt 90nm mobile AMD64. AMD might have a chance for a super powerful,super mobile laptop market 🙂

and maybe i could swap out the AMD64 in my laptop and gain lot of battery life 😀
 
Originally posted by: clarkey01
FAB 30 is doing a great job, I just hope we can 3.2Ghz out of the 0.9 micron process, Intel would really be stuck, i wonder why the next fx (55) is going to be fabed on 130 nm process, surely they'd like to save money on die space ? or are they already in production ?
Click to expand...

I believe that FX-55s are already being churned out at 130nm, for introduction around October 2004. San Diego, the 90nm equivalent of AthlonFX, is not slated for launch until 1H2005, according to AMD's official roadmap:

http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_608,00.html

The FX-57 will definitely be a 90nm part, to be introduced around April 2005, clocked to 2.8Ghz. The reason for not shrinking Athlon64 FX yet is that the 90nm ramp is still starting, so higher-clocked variants are possible only after a few months of volume production and tweaking. OTOH, 90nm Athlon64, both desktops and mobile, are coming out in the next few weeks, probably without any big announcements. The 90nm 4000+ Athlon64 will probably be introduced around October.

The advantage for AMD will be massive production volumes, while prices are still high (ASPs for AMD64 chips are around the $250 mark), since Intel doesn't have anything to counter with. So don't count on lower prices for high-end AMD CPUs any time soon. Or at least, for the next couple of months.

Marcos
 
Also the .09 parts are around 5% faster clock for clock. so a .09 2Ghz will be like a .13 2.1GHz. I would expect the .09 parts to scale up to around 3.2-3.4 GHz before moving onto .065. By then we will have dual core CPUs witch i believe will be pure pwnage (and Ed at oveclockers doesnt seem to agreee. Just read what he has to say about it on the front page www.overclockers.com. Pure BS if you ask me, he doesnt know squat and seems to bash AMD as soon as he gets any news from them)
 
i kinda agree that amd AND intel are designing dual cores just cause that's easier than making them faster. It's not that they can't, as ed says, it's that they will have to put a ton more research into them.

it's the basic idea of engineering: you can get the first 80% of the results for 20% of the work, but the next 20% of the results take 80% of the work. that means whoever settles for 80% of the performance out of a design will get that 80% product out on the market way before the company trying to do 100%, and already have a new prduct by the time the competition catches up.

they are both going dual core because the technology for it is already there. they just slap two of em together and sell it for twice as much, rather than spending years researching technologies that give them the same results.

in reality they're just doing it to gain a cheaper performance jump than the rest of the technology they research.
 
Dont worry 90 nm yeilds are looking good, maybe not quite up to the transition from going to the .18 micron process to .13 but good enough. Headroom for Mhz will top out around 3.2Ghz. The plan is dual core will be able to provide some manoeuvring room to counter the boys from down santa clara untill FAB 36 is up and running Utilizing the 300mm wafers and with .065 nm in full swing.

The hope is FAB 36 will be ready, which it should be, however converting FAB30 with 0.65 nm process in place @ the last minute would not go down well !
 
I tend to agree that higher clock and lower heat are reasonable expectations. Even with more defects, the greater number of chips per wafer should make up for initial problems if they run into them.

Remember too, within a year we should have Cooligy systems available - that's the company that has ties to Stanford that is doing the sealed liquid cooling with pumps containing no moving parts. CPU cooling should look a lot different in the next year or two.
 
Originally posted by: clarkey01
The hope is FAB 36 will be ready, which it should be, however converting FAB30 with 0.65 nm process in place @ the last minute would not go down well !
Click to expand...
What do you mean by that?
 
It been delayed, that can't be good. Everyone has problems with 90nm could that be a sign?

Who knows. I'm sure it's expensive with automatically aliviates my concern of such a release.😉
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top