Actually, Semiaccurate is today suggesting that Kaveri could be launched to coincide with the AMD Developer Summit in San Jose 11-14 November 2013. So your 1Q2014 guess could well be right.

well could and actually are two diff things and as of right now karveri doesn't have a release date and according to this there was no mention of karveri in it's 2013 sechdule. So if this holds up to be legit then karveri will prob end up getting released next year.

http://www.tomshardware.c...nd-APU-Release,20888.html

Tom's says kaveri will be released on 2nd half of 2013. If that is so, extremely fast refresh for the APU's. I'm also guessing Q4 2013 for kaveri release.

i remember them saying the same thing with bulldozer and it got pushed back 3 times. so i don't know how acurate amds roadmap actually is if they are basing it on a ballpark timeframe. because now this report would seem to conflict with the report toms was saying about karveri being released this years since this report mentions no word of karveri being released this year.

Like you were there during the negotiations? Don't forget, AMD needed GPU technology for the HSA strategy. Perhaps Huang didn't share this vision. HSA should revolutionise computing. The first juicy taste will be in the form of Kaveri. When it comes up against Intel's finest in Handbrake benchmarks, there will be no competition.

I'll just say don't believe everything you read on the net about the merger stories.

HSA - CPU offloading work onto GPU compute in a unified architecture - is an innovation. Coming to a desktop near you. Its power will depend, in part, on software written for HSA but also the APU will be able to decided itself (I'm not quite sure how but probably through some heuristics), to offload some non-HSA optimised code onto the GPU. Yes, it should blow Intel out of the water in parallel processing. It is a much bigger deal than the birth of x64 instruction set. That is why AMD went all out to acquire ATI. Intel don't have a good answer to the HSA architecture at the moment - but it can do aggressive die shrinks. It's the blue hare and the green tortoise.

Unless some compelling reasons comes along where the world goes back on the tablet/ smartphone trends, PC market is not picking up. Gamers and content developers who will use a higher systems/workstations are in the minority when compared to the vast millions of users who just don't need a computer for anything other than watching media, play flash or HTML 5 games and do social networking on the go.

For this reason, you can be very sure AMD will not chase Intel on performance cores. If it happens by chance that Intel doesnt advance and AMD catches up? good for AMD then, but they are not going to actively chase Intel on pure performance. There are so many other marketing models out there proving you dont even need the fastest thing out there to make a ton load of money. AMD is changing to a company that works on design wins this ensure sales especially when the marketing and other business headaces are not AMD's problems anymore but the customer's

Take this example for instance: How people do you approximate actually cares or knows they have a Qualcomm Krait in their phones/tablets? Many wont even know what Qualcomm is. But their chips are everywhere and they are making the big bucks. Thats what AMD is focusing on now.

This doesn't mean the Client is ignored. There are high performance cores in line and on track. They will come out on time. But just don't expect K7/K8 days to be repeated anytime soon unless Intel drops the ball because there are engineering requirements and efforts to pursue such a performance deficits. This means larger teams and even multiple parallel teams working on various designs, a luxury Intel can afford but even they are dropping this model and have started this process since last year. AMD simply doesn't have that resource strength and frankly there is no need for that now given the market trends.

In this world of lower power maximal compute, trends are changing. You will even see Intel going this route where they drop the aggressive ramp of performance to go for more low power maximal performance possible.

This isn't really related to the AMD focus of this article but I'll reply.

I just find some of your conjecture difficult to believe. GPU may be worse, fine. I betcha the Xbox will be cheaper than the PS4. They'll make darn sure it runs games at 1080p or it doesn't make any sense to release it. Development libraries don't usually make it harder to code, they make it far far easier. Now for coding "to the metal", sure, but game developers are focusing on high-level code anyway. This is a design decision. I'm willing to bet that the MS libraries are able to make game developing easier while still exposing all the features of the hardware for use. This could enable more accessibility for developers and therefore more developing activity, a wider array of games, and more small shop games that make hits. That's the idea anyway. The middle-ware or game engine people can maybe have more flexibility with the Sony approach. 8GB of RAM sounds awesome, are you kidding? That's plenty to run Win7 or Win8 fast in a desktop, so I'm sure it can handle the narrower focus of a console's OS and games with plenty of margin.