We should all hope they come up with something that's competitive. Even 80% of the performance for 50% of the price would be good enough.
And this is because the last 3-4 Intel CPU generations have been simply not worth upgrading to. If some real competition from AMD shows up, things will change.
AMD seems to have lost their way back when they merged with ATI and integrated their GPUs into the processors. I remember that in a restructuring they switched from hand-optimizing hotspots in the CPUs to a more abstract approach using design software that worked well in GPUs.
After that decision they fell behind in performance and efficiency. I don't know if that was the root cause but it stayed in my memory as one pivotal point.
It would be interesting to read an insider story on how that decision played out in the meantime.
It depends what you mean by lost their way. But ATI has nothing to do with it. Bulldozer was a failure, it was a perfect storm of Intel's best execution since the Pentium 4 Ghz problem and AMD's worst execution. When Intel took a misstep, it cost them 20% of market shares. When AMD took a misstep, it nearly killed them. You could argue without ATI, AMD would properly be dead by now. And AMD CPU was literally uncompetitive even with price reduction for the past 5 years. And they totally miss the boat with the desktop to laptop transition. While Intel, after hitting the Powerwall with Pentium 4 and fortunate enough to have had a team working on a architecture specially for laptop, the original Pentium M, pick up speed in that direction.
But like others have pointed out, I believe it is a operational, internal struggle, and basically a leadership problem.
> And AMD CPU was literally uncompetitive even with price reduction for the past 5 years.
I don't know about that.
The Intel core are very good processors but the entry price for the smallest one is over $100.
I can remember some cheaper product lines from Intel but they weren't that great. AMD had decent CPU for the cheap laptops and computers. Dunno how much volume and margin there is to be done on that segment.
amd banked on multi core and hsa, both way too ahead of their time when the software is still heavy reliant on fat cores, and apus while better than intel+igpu today are simply regarded as good choice for portable devices or hpcs with good cpu and gpu with low tdp...but i seem to remember that was not the point! the point was to take advantage of the fact that the gpu and cpu are closer...
amd bought ati right before the release of the r600 lineup which totally lost against nvidia 88xx cards and marked nvidia for gpu domination for years to come, so they acquired a company right before their biggest product failure, and after that they released phenom 1 which was a failure in its own right, after these 2 consecutive big lost they had to go fabless because couldn't afford to keep up the pace with technology and because of low demand they were not running full time ...this brought a lot of other complications for amd, because of missed deadlines of pp and in general not competitive to intel...
after having modestly recovered from the phenom fiasco with phenom2, and with r700 gpus, it tanked again with bulldozer after lots of delays...then released a decent version with vishera which was a good enough contender to intel ivy-bridge, except for power/efficiency because they were already behind in the pp, after that to regain margin with intel an entire pp was skipped and amd stalled in the high highpower market focusing only on low power pp with apus, going for zen, thank god GCN was good enough to make them keep a sizable market in the gpus...
I don't know if the GPU integration is actually a negative, though - Intel's doing the same thing these days and doing well enough, and they both don't waste die space on integrated GPUs for their server lines.
I think they just needed to have the next major line with enough time to try to respond to the P-M/Core switch to really knock it out of the park...and we got Bulldozer instead, where they bet big on trying to partition resources and a different notion of distinct "cores", and didn't come out ahead.
The GPU integration as such isn't a problem but they threw together two teams who had a vastly different approach. That is an organizational problem. With AMD I think that the influence of the GPU team grew too big.
Now AMD has better GPUs while intel struggles in that area.
Yes, I almost forgot the resulting Bulldozer fiasco. I believe that was a result of applying GPU thinking to CPUs.
And now Zen comes out of forming an entirely new round of people. The amount of money lost from that particular issue, if that was the cause, boggles the mind.
(It's too bad that AMD didn't wind up on top so far in the GPGPU area - if they had, all that GPU influence being a problem probably would be much easier to write off.)
Intel struggles in that area?
AFAIK the only problem Intel struggles with, GPU-wise, is that they appear to loathe the idea of making discrete GPUs, so they're constantly trying to figure out how to make a fast GPU in an acceptable on-package area.
I was under the impression that AMD's on-die GPUs, while (possibly? I haven't tested their on-die GPUs in a few generations) nicer than Intel's, were still not something to write home about.
