• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

AMD going 20nm this year?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
^^^Well, not necessarily. Looks at Intels 22nm trigate compared to their previous tech. Yes, it generates less heat, however due to its smaller process tech its less capable of disipating the heat due to a smaller contact area for a heat sink.
 
^^^Well, not necessarily. Looks at Intels 22nm trigate compared to their previous tech. Yes, it generates less heat, however due to its smaller process tech its less capable of disipating the heat due to a smaller contact area for a heat sink.

I think that's just intel's mess-up with the IHS TIM being paste.

Transistor shrinks work, otherwise the would not do it.;)
 
still hope that AMD can find a way to compete in the speed wars... i'm no fanboy, i just think competition breeds faster processors... and i like fast processors...
 
Here's my two cents. We've already reached the point (28nm/22nm) where profits have decreased far below where they were per chip where they were at 90nm. (there's a slide from an Nvidia briefing out in the wild where they discuss it). Without major changes in materiels used (i.e. silicon replacement) the gains that you would expect from a process shrink no longer outweigh the costs. Intel still seems to be locked into the "faster, smaller" mode whereas you can already see AMD changing that mindset (after recovering from years of a bad "captain" steering the ship) with the shift to the APU and IMHO the even more important decision to implement major redesigns to better take advantage of available resources that are already there (Bulldozer's Module design and the upcoming resonant clock mesh). I honestly think that AMD is a little ahead of the game in their design and it will be a little while before OS's and software catch up (openCL etc). I think that's what the CEO was hinting at when he said they were no longer going to go head to head with Intel in the "speed" wars. The most important thing going ahead I think is innovation and major changes in how CPU's are designed (we can only keep shrinking for so long before the end of the road that's rapidly approaching hits). I think it makes far more sense to implement design changes to improve performance than to keep going down the costly road of die shrinks and the fabrication problems that are involved (pointing at Ivy Bridge).
 
^^^Agree entirely. They're focussing on performance in different areas and I think it'll turn out to be a winner in the end. Likewise if they can play each fab off against the other we'll likely see them getting process shrinks sooner than when they did their own manufacturing.
 
If you look at there mobile APU's you can see where they are moving forward, they have managed double the performance per watt over all on the previous generation.

That's no small feat and much more than Intel managed even with a DIE shrink.

The result is while AMD are still behind in CPU performance the gap has shrunk significantly and the IGP performance has taken another chunk of a leap ahead.

Comparing Intel's 35w to AMD's 35w is a much closer run thing now despite 22nm vs 32nm.

It looks like i will now have an upgrade path via AMD once there Desktop PB parts are available, personally i think the performance per watt will be much better than what i have now, somewhere near Ivy Bridge (per core)

Also, i happen to think nVidia's ability to keep AMD honest and on there toes in the GPU department is a very good thing, AMD are more than capable in picking up nVidia's challenges, for as long as they have that challenge their Radeon cores will keep getting more efficient, which will keep translating into better and better APU's, that in turn will keep Intel finding it increasingly difficult to keep up.
 
Last edited:
Personally I think that the days of "throw raw power at it" are pretty much over with. Unless Intel starts making some major changes they might find themselves at the end of a dead end road while the others have already taken the on ramp to the highway so to speak. It will be interesting to see what happens in the future and how all of this pans out.
 
no no, intel is king, amd is queen and VIA is just tring to survive. well in the US any way, im sure VIA has a bigger marketshare in asia market for cheaper pc's then intel or amd.
 
Personally I think that the days of "throw raw power at it" are pretty much over with. Unless Intel starts making some major changes they might find themselves at the end of a dead end road while the others have already taken the on ramp to the highway so to speak. It will be interesting to see what happens in the future and how all of this pans out.

Intel is currently the king, and they know what they are doing. anyone who thinks that AMD (who is way behind currently) will in a few years time will be handedly beating intel is just lying to themselves. Is AMD doing the right thing going the fusion route (I think so) only time will tell for sure. but there should not be any talk of Intel reaching to point of just trying to survive.

They dominate the fab world, and the x64 CPU side and the newest iGPU in Ivy wasn't too shabby and makes use of current tech better (lucid virtue tech).
 
Intel is currently the king, and they know what they are doing. anyone who thinks that AMD (who is way behind currently) will in a few years time will be handedly beating intel is just lying to themselves. Is AMD doing the right thing going the fusion route (I think so) only time will tell for sure. but there should not be any talk of Intel reaching to point of just trying to survive.

Your getting way ahead of yourself, there is no talk of Intel reaching a point of just trying to survive. :)

As you said yourself AMD going the fusion route is / or looks the right thing to do. But that's it. no one said Intel will end up in the crap because of it, just that right now AMD are doing it better in that and if that's the way of the future Intel need to catch up.

intel is king and AMD is just trying to survive.:shrug:

I don't see AMD doubling there performance per watt and advancing towards there total fusion gaol as just trying to survive.
They have a mission, an idea, its a good one and there implementing it successfully.
 
Last edited:
Intel is currently the king, and they know what they are doing. anyone who thinks that AMD (who is way behind currently) will in a few years time will be handedly beating intel is just lying to themselves. Is AMD doing the right thing going the fusion route (I think so) only time will tell for sure. but there should not be any talk of Intel reaching to point of just trying to survive.

They dominate the fab world, and the x64 CPU side and the newest iGPU in Ivy wasn't too shabby and makes use of current tech better (lucid virtue tech).

I never said that intel will be "handily beaten". Likewise saying AMD is "way behind" is also a bit of a false statement as well. What I was saying is that unless Intel starts to make some major changes in how they design CPU's (who knows...maybe they've already started that "behind closed doors" in the design sections) they run the risk of falling behind and then having to play catch up. AMD on the other hand is already signaling and implementing a shift in design mentality that might just pay off huge in the long run (or not, only time will tell). But continuing to design x86 processors and throwing raw "power" at everything is not going to take you far. We're getting closer and closer to the end of simply shrinking the die size. Unless there's an alternative plan in place as a business you're going to be in trouble as that end gets closer and closer.
 
You're all forgetting that AMD got this reduction in power useage by use of a resonant clock mesh, little else. This tech was licensed from another company and I see little reason that Intel can't also go down this path should they need to do so. This is not an AMD exclusive tech.
 
You're all forgetting that AMD got this reduction in power useage by use of a resonant clock mesh, little else. This tech was licensed from another company and I see little reason that Intel can't also go down this path should they need to do so. This is not an AMD exclusive tech.

Yes but if you read that tech it reduces power consumption by 20% at best on a good day, there is way more than that shaved off Trinity vs Llano and especially BD. Way more, where did the rest come from?
 
You're all forgetting that AMD got this reduction in power useage by use of a resonant clock mesh, little else. This tech was licensed from another company and I see little reason that Intel can't also go down this path should they need to do so. This is not an AMD exclusive tech.

We're not just talking about "resonant clock meshing" though. Starting with BD and the APU's AMD has made a complete shift in their approach at designing CPU's. Now while I personally think that all of this "Faildozer" talk you hear around the forums (mainly by hardcore intel fanboys that I like to call "spIntellers") is a load of nonsense, there are going to be some "growing pains" whenever you make such a dramatic shift. People seem to forget that for example an 8120/50 is essentially a quad core CPU with hardware multi-threading. Each "core" is not a core by the traditional design. It's a shift in engineering mentality that is a way of better utilizing the resources available on a chip. I think that's a very good approach to take as "die shrinks" become more expensive and problematic.
 
Yes but if you read that tech it reduces power consumption by 20% at best on a good day, there is way more than that shaved off Trinity vs Llano and especially BD. Way more, where did the rest come from?

Most likely from hand tuning the Bulldozer architecture. BD was rushed out using a mostly entirely automated process. Most likely AMD still has a long way to go before things are even close to how far they'd gone with their old Phenom architecture. Tends to indicate that they made the reight design choices with Bulldozer, it was just their execution that was a bit of a fail. Things can improve a lot yet ;)
 
I never said that intel will be "handily beaten". Likewise saying AMD is "way behind" is also a bit of a false statement as well.
how is saying AMD is way behind a false statement? All that is needed is to look at their performance numbers from the past few year this is perfectly clear.
 
It is what it is, BD is a failure and AMD advertizing how it's architecture works won't win any benchmarks. I don't grade a CPU by how much talk we can do about it, only on it's performance and BD is a flop.;)

If AMD is able to go to 20nm that would be successful, they would be able to save power like intel.
 
Back