This could be rumour mill at its finest, but its very interesting if it were to turn true.
The boss @ HardOCP (Link) has posted that AMD's GPU's are going to soon sit inside next gen Intel CPU's.
This could be rumour mill at its finest, but its very interesting if it were to turn true.
The boss @ HardOCP (Link) has posted that AMD's GPU's are going to soon sit inside next gen Intel CPU's.
AFAIK,HardOCP is not on ze list so post away!!
AGTDenton (07-12-2016)
He said the same thing back in May, in his rant at how rubbish Polaris and RTG were.
http://www.hardocp.com/article/2016/...ey_in_futility
Heh, hindsight is a wonderful thing:
Followed by two quarters of good market share growth... the simple reality of the situation as it stands today is that AMD has a loser on its hands ...
Intel is investing more in graphics than in CPU at the minute, AFAICT. They wouldn't be doing that if a better IGP technology* was there for the taking - that would be throwing away R&D money. I'd take that with a bucket of Maldon, if I was you...
* yes, AMD's IGP technology *is* better - Intel have to strap a huge chunk of expensive eDRAM onto their chips to get graphics parity with AMD. If they were genuinely in a position to license AMD technology or buy out RTG in its entirety I'm pretty sure they'd be doing it. IMNSHO you're likely to see nvidia's graphics tech in Intel's CPUs before you see AMD's...
Even if Polaris missed its clockspeed targets,you only have to look at the number of transistors Polaris 10 has to see its more an R9 390X level performance card - Hawaii has 6.2 billion transistors and Polaris 10 has 5.7 billion transistors. If the RX480 had shipped with stock frequencies nearer to 1.4GHZ it would have been more R9 390X level performance.
Anyway,people might have forgotten that Nvidia has its technology licensed by Intel to the tune of $200 million a year IIRC(as part of a previous settlement). From what I gather this agreement is expiring and its quite possible Intel might instead license AMD technology instead especially if AMD is willing to do it cheaper.
I'm not sure that's true. Kaby Lake's 15W GT2 without eDRAM returns better frame rate in games than Bristol Ridge's 384 shader R7 at 15W according to notebookcheck's benchmarks.
Obviously Bristol Ridge is still Tonga-derived rather than Polaris-derived and AMD's CPUs probably require more of that heat budget. Based on the cutting edge products available though Intel are ahead in raw speed - although their drivers aren't as good yet.
It'll certainly be interesting to see how Kaby Lake and Raven Ridge shape up on the desktop, as well as the die area estimates.
Looks like AMD shares have just gone up nearly 9% so the news might be correct!!
Restricting to 15W is going to play to Intel's process advantage strength. However I would expect if the full 512 shaders were enabled but clocked a bit slower, the AMD part would get a speed boost at the same power level if performance was what AMD were trying for.
Still, whilst an AMD GPU in an Intel chip is more believable than Intel using an Nvidia GPU, I suspect this is just down to patent licensing which is nothing like core licensing (as shown by AMD having a full CPU patent access with Intel for years but don't make anything like an Intel core). If that is the case, this is probably more a poke in the eye for Nvidia than a nod of approval to AMD given the bad history between Intel and Nvidia.
tbh I'm always wary of notebookcheck's bench scores.
Not only is 15W quite restrictive for AMD's 28nm APUs, a lot of Bristol Ridge laptops were rolled out with single channel memory configurations, which would totally hobble the IGP (Intel's IGP has always been less memory-dependant than AMDs). I'd want some detailed information about the setups before I took that as 'proof' of anything
EDIT: having a quick glance over their A10-9600P review (and remembering that represents the *slowest* implementation of graphics in Bristol Ridge) they used a dual channel laptop and it performed between HD 520 and Iris, which is what you'd expect. Comparing at lowest available TDPs doesn't really test the technology though - it tests the implementation and power efficiency (which is never going to look good for a 28nm p[art compared to a 14nm part). Hopefully we'll get Kaby Lake desktop v Bristol Ridge desktop reviews in January and that will give us a better idea of how well Intel's doing in the IGP stakes...
Last edited by scaryjim; 07-12-2016 at 12:11 PM.
OK,AMD share price has just broken $10 - unless it is something to do with the Zen reveal next week,it has a good chance it is the reveal is down to Intel licensing GPU stuff from AMD. This is the highest AMD has been since 2007:
https://www.google.co.uk/search?q=am...eQDFw-_nOgAAAA
Lets hope Zen is actually half decent.
Can't believe I didn't buy any AMD stock when it was down below $2 back in ~ 2009....
There are currently 1 users browsing this thread. (0 members and 1 guests)