http://www.theinquirer.net/?article=32589 Maybe... Certainly AMD aren't going to be as far behind as people first thought.. Looks like AMD do indeed have an answer to Conroe...
So other than games what other programs will this have benifits? But AMD still has a problem with issues of using DDR2 with high latancies where Intel and the "Core" dont.
Seems a bit ambiguous... So if you're running multiple single threaded apps which does it do? What about a single threaded app and a multi-threaded app at the same time? Does it pick and choose depening on what youre working on at that point in time? Certainly sounds like it isnt ideal for multi-tasking. If it switches over to single-threaded mode when you're running a single threaded app does that mean that background tasks suffer? IIRC intel have been talking about a "reverse-HT" for a good long time, I wouldn't be surprised if they unveil something similar that is hiding inside the Core2 architecture (just as they did with HT, which had been hiding inside the Northwoods for years, just hard-disabled)
latency isn't a big issue on AM2 - it's roughly the same: http://www.bit-tech.net/hardware/2006/06/14/foxconn_c51xem2aa_nforce590sli/6.html
we'll have to see when it comes out... I'm skeptical about things until I find out exactly how something works.
Ok true say, but on games and few other apps, its shown that lower latancies on the AM2 platform actually increase FPS and performance, your tests show synthetic benchies but not games.
That was the case on 939 too - lower latencies helped improve gaming performance a little. In GPU limited scenarios, latency doesn't make a fat lot of difference.
i think its also on the 939 pin chip aswell. that article said its a driver update aswell, its as if its been there all along and just needs activating.
christ...........its never easy to pick who to use. Well if your saying that then you've just turned my head upside down! first it was AMD then Intel....and now maybe AMD again. why cant it ever be simple.
I'm inclined to think that if works Intel did indeed just get owned. Nice to know theres going to be some competition, and that AMD aren't just going to fall of the map for 7 months.
Does anyone else think the term "reverse hyperthreading" needs to go? It sounds like a 3 year old thought of it tbh
An article on overclockers.com made this seem a little less exciting than the Inquirer. I think the main point is that as this requires a driver update it should be fairly easy for Intel to do the same thing. If reverse hyper threading really makes a difference Intel will just copy it within a few months (especially since they've already talked about it) and then AMD's advantage will be gone again.
It's not just a driver update by the sounds of things. It's the ability built into the whole architechture, all the driver will do will enable it. Don't be so sure about Intel copying it either, that might require an entirely new chipset or even socket. If this turns out to be any good Intel got sucker punched.
As mister_tad said it sounds like Intel have been aware of the idea for a while and if it is going to be any good i would put money on them having it in Core 2 already just turned off. Oh and one last point the Inquirer article describes a "lowly 3800+" beating an E6300. Not so impressive considering that the E6300 is ~£50 more than the E6300 is listed on overclockers.co.uk
oh thats just perfect... At the best of times i am undecisive over what to buy, with my pc even more so... I thought that the intel over amd decision was going to be easy..... damn changeable market! Its like they never heared of standards!
... Still waiting for any kind of confirmation on this of course... But if true, it's nice to know AMD havent been sitting on their behinds doing nothing since K8 (apart from AM2) should be fun.
Another thought: Crossfire regularly gives just 30% extra for a second card - and that's on graphics, which is relatively easy to split up. If there's a driver dissembling programs to see how to multithread them, surely that's gonna slow it down even more. Can we expect to see even 30% performance gains? ch424
