More details of Haswell are leaked

It could be an MCM package:

http://vr-zone.com/articles/mystery-...ain/15272.html

GPU performance is meant to be greatly improved. It seems that CD from SA mentioned this a while back(not sure if the numbers are correct though):

http://semiaccurate.com/2012/02/08/h...phics-monster/

http://semiaccurate.com/2011/09/21/a...mportant-bits/

AMD is also investigating used of stacked RAM too(there are pictures):

http://forums.hexus.net/2355217-post1375.html

2013 is going to be a very interesting year for integrated graphics.

I am surprised no one is interested in this. It could mean a very large improvement in IGP performance as it helps alleviates the bandwidth issues they have!

Well the AMD bit is interesting.

But faster Intel graphics... Does that mean it hangs the machine faster?

It can't hang it any better IME, last time I used Intel drivers I found myself reaching for the reset button every few days. That was under Linux where the idea of resetting a machine is really quite alien.

So basically Intel can do what they want with their hardware bottlenecks. Until they fire every last member of their driver team, meh. Wasted silicon.

tbh CAT, it's all a bit pie in the sky until they have working silicon. IB graphics are meant to be twice as fast as SB, but that only makes them competitive with Llano, which will put them well behind Trinity. And presumably the follow-up to Trinity will have GCN, as well as being based on 28nm process so more of everything, and potentially more memory bandwidth or even stacked memory or whatever other esoteric changes AMD decide to make.

2.5x faster than IB - if it happens - would only make Haswell ~ 65% faster than Trinity. Given AMD are claiming 56% faster graphics from Llano to Trinity, it's not inconceivable that they could manage that again to the next generation. Which means you're looking at another 2 generations of processor before Intel's IGPs are even competitive with AMD - assuming the rumours are true *and* Intel can produce the silicon and drivers. It's not the first time we've heard about a new Intel graphics core that's going to be really competitive against AMD and make low-end discreet graphics obsolete, but it's yet to materialise. I'll be taking this one with a big pinch of salt...

Quote:

---

Originally Posted by scaryjim

tbh CAT, it's all a bit pie in the sky until they have working silicon. IB graphics are meant to be twice as fast as SB, but that only makes them competitive with Llano, which will put them well behind Trinity. And presumably the follow-up to Trinity will have GCN, as well as being based on 28nm process so more of everything, and potentially more memory bandwidth or even stacked memory or whatever other esoteric changes AMD decide to make.

2.5x faster than IB - if it happens - would only make Haswell ~ 65% faster than Trinity. Given AMD are claiming 56% faster graphics from Llano to Trinity, it's not inconceivable that they could manage that again to the next generation. Which means you're looking at another 2 generations of processor before Intel's IGPs are even competitive with AMD - assuming the rumours are true *and* Intel can produce the silicon and drivers. It's not the first time we've heard about a new Intel graphics core that's going to be really competitive against AMD and make low-end discreet graphics obsolete, but it's yet to materialise. I'll be taking this one with a big pinch of salt...

---

Sums up my thoughts pretty well.
I'll just add that you've got to add Intel's poor drivers to the mix as well.
I do look forward to these hopeful increases in bandwidth to the GPU parts of the CPU's as they may well then mean that the only restrictions integrated Graphics suffer compared to their discrete counterparts is execution resources rather than fighting the CPU for memory bandwidth and less execution resources.