In another thread I wrote about the lack of speed increases and the fact that in the last few years all of the performance increases have come from added cores, Die shrink, added cache and some optinmizing. I'm not sure if they can get more than 64 cores on a single die, but lets say for arguements sake they can get 128. That's only 5 more steps until they top out. I believe, I'm not certain, but I believe I read somewhere the smallest die possible with current manufacturing processes it 11nm. Intel is at 45 and AMD will follow soon. Next up is 32, I think. After that it's in the 24nm range. (not sure) and after that 11nm. That's only 3 more shrinks to the end of the road. And I am pretty sure they haven't found a new process yet to go smaller.

The x86 train isn't at it's last stop yet, but I think I can see it from here.....

Maybe the world will end before that... problem solved! ;P

 Considering 12nm fab plants won't be built for another 4-5 years I don't think we'll be seeing a problem anytime soon.  That's plenty of time to create new ways to shrink die sizes.

It isn't just about die shrinks, adding cores, and some optimizations. There are other ways that they have to discover to keep advancing. Take the architecture for example. Intel did amazing things with the Core 2 architecture. Also, they are implementing new materials and innovative transistor structures to further shrink the transistors.

Just look at the history and you can see them changing. Before, it was like the GHz war. Then they became super efficient with the Core 2. Now it seems almost like a multi-core war. I'll leave it up to the engineers to figure out how to continually improve. All these questions can be similar in a sense because they all ask the question, "What is next? Are they at the end of the road?" Just wait and I'm sure they'll surprise us.

Anways... at least for me. There is only so much I can do at once. Computers can multitask, but I can't. I can have a few programs running, but I can only focus on one thing at a time. The only thing I can do is act like a single-core. Do one task as fast as possible and move onto the next. Like I can't do research and write a paper simultaneously. (I think I'm getting into more of the topic posed in your other thread).

AMD had a write up a few years back that talked about a technology that they were working on that was the opposite of hyperthreading. It would make a dual core at 2GHz look like one 4GHz cpu to the OS. never heard anything else about it but the quad cores of today would look like a 12GHz single core!

http://www.dvhardware.net/article10901.html

Darn looks to be uncovered as a hoax

bob_on_the_cob:

AMD had a write up a few years back that talked about a technology that they were working on that was the opposite of hyperthreading. It would make a dual core at 2GHz look like one 4GHz cpu to the OS. never heard anything else about it but the quad cores of today would look like a 12GHz single core!

http://www.dvhardware.net/article10901.html

Darn looks to be uncovered as a hoax

Not sure but I think we didn't hear any more on this because it was exposed as a myth.

Recovering

LOL. My world ended the first time I got married!

1nteljunki1

Is it 12nm? Thanks for the correction. I thought it was 11nm. As for time, I wasn't implying this would happen next week. I know they are working on it. Actually I think they started back when Intel was in the 130nm area. It is slightly worrying that it is still being worked on with no solution yet.

RyuGTX

Correct me if I am wrong but the Core architecture is nothing more than optimizations to the x86 design. Ie, they went the AMD route and started optimizing for more work done per clock cycle and less raw speed. In the end it is still an x86 CPU. Just designed better than the last gen chips.

To everyone

My thinking with this post and the question it posed was where do they go from here and how do they get there. The time when something radically different has to be done is coming. There are a limited number of tricks left in the x86 bag. I wonder what is next. x86 HAS to die sometime. It is too fundemantally inefficient to live forever. Intel knew this and developed Itanium, but they let HP talk them into believing AMD didn't have anything with 64 bit x86 design and played that card too soon and payed the cost for it. Don't get me wrong. Itanium is a great chip but it was released too soon and it cost Intel billions. (HP crept into the dark and left Intel holding the billion dollar bag.)

nECrO1967:

RyuGTX

Correct me if I am wrong but the Core architecture is nothing more than optimizations to the x86 design. Ie, they went the AMD route and started optimizing for more work done per clock cycle and less raw speed. In the end it is still an x86 CPU. Just designed better than the last gen chips.

 

I guess it depends on how you view the word "optimization". To me, it is more like tweaks. For example, when I think of the word, I think of video cards. You have lets say a 4870 card and as time goes on, ATI releases new drivers that make it run more stable and faster. To me that is optimization; it isn't anything groundbreaking.

Sure, you could say they "went the AMD route" with more work per clock cycle and less raw speed. But that seems too general of a statement and it doesn't give Intel the credit it deserves. It is how they achieved that and it isn't the same as AMD. If it was, then AMD could easily change a few things and destroy Intel like it did a few years back.

I see your point. To me all work done with x86 by both companies is just an optimization of the x86 architecture. Whether it is in the form of software drivers or a redesign. That's not meant to take anything away from Intel's engineers. They performed a heck of a feat in their work with Core2. The fact that they had to do it in such a short time when Intel management bungled and wrote off 64 bit x86 as impossible. (Actually I saw this more as a PR mistake because they didn't want  Itanium to have cheap competitor.)

When I said "went the AMD route, it just meant that Intel did so in responce to AMD's success with AMD64. Netburst would have lived much longer had AMD not had the home run it did.

Thanks a lot Ryu, Recovering and others, for replying to my ramblimg thoughts. As always, you guys make this place worth visiting.