So the HD 4870 x2 will be released sooner rather then later8)

HD4870x2 just around the corner
http://www.theinquirer.net/gb/inquirer/news/2008/06/23/hd4870x2-corner

I love how The Inquirer uses such a high level of professional-ism in their articles.

What is it with you and The Inquirer? Same with Fudzilla?

I see nothing wrong with the inquire I think its a good source of info.

rather humorous at times.

I agree.

790GX overclocks Phenom to 3.4GHz On Air
http://www.fudzilla.com/index.php?option=com_content&task=...;Itemid=69

I don't trust Fudzilla too much, they're in my opinion a little biased.

Super XP said:
I can already pull a full 1GHz OC out of my X3350. I got a full 1GHz OC out of my brothers E2400.(on a stock HSF)

That can't even pass for English. That is not journalistic writing (which they are supposed to be doing), it is artistic writing. As such, it makes anything they publish very hard to understand let alone believe. Why Fudzilla/Inquirer/Register are still around is beyond me. They are a disgrace to the journalist community.

I do actually like flicking on to the Inq/Fud because they do get some interesting tid-bits at times - but seriously they are so biased its not even funny.

Check this out from 'Charlie' - (aka nvidia hater)
Nvidia cheats on 3DMark with 177.39 drivershttp://www.theinquirer.net/gb/inquirer/news/2008/06/23/nvidia-...3dmark-177


Sorry, but would that also mean that those with Ageia cards are also 'cheating'??

Pure rubbish.

I know Mike Magee. He is a well respected computer industry veteran. He writes for CPU Magazine, that alone tells the story.

The Inquirer, Fudzilla and the Register are based on rumour, speculation & humour. They do a fine job on reporting about new stuff. I do admit some writers have a somewhat bias opinionated writings at times, but they are only stating facts with come magnetism.

Everybody has to admit The Inquirer has written stories in the past that have mostly come true. It may not be your average journalism, but never the less it’s still a great read.

mike magee left the inquirer quite some time ago now.

4870 X2 looks like a nice part. if the price is right, it could very much be the top card for a good while as nvidia cannot die shrink the GT280 for a couple of months yet. im rather happy with SLI 8800GT though, it is at least a match for GT280 and less than half the price. 4870 X2 would have to be very special to make me tempted. 4850 is a good part, the only problem as i see it is that where it matters, between 10 x 7 and 19 x 12 its only about ten percent faster than an 8800GT, where upgrading to a 4850 isnt really much of an upgrade at all. 8800GT's are also more than 10 percent cheaper already than the 4850. you might as well just overclock your 8800GT, as many people have already bought an 8800GT that will probably save nvidias market share in that sector, ATI are too late with that card to make a big impact i would say.

as for the phenoms, AMD are still screwed. the fact 45nm phenom can actually overclock half decent isnt worth noting, it still isnt as fast per clock as kentsfield never mind yorkfield, and the core 2's are light years ahead on overclocking still. 3.4 was achievable on the now ancient 65nm Q6600 on air, and its still going to be faster than an equivalent speed phenom......

if you havent noticed, theres only the small matter of nehalem is less than 6 months. in early tests, nehalem is around 30 percent faster per clock than yorkfield. which makes it miles faster than phenom again. a 2.6ghz nehalem was pitted against the extreme 3.2 yorkfield, and it was faster comprehensively. so a slow nehalem is faster than the fastest quad core out right now. intel arent stuck with only 2.6ghz on nehalem's clocks either. this isnt the phenom debacle, intel seem to have steppings that will go to at least 3ghz and rapidly beyond if intel deem necessary.

it also seems despite AMD's experience with integrated memory controllers, intel yet again have beaten them well at their own game and their controller is miles faster than phenoms and the bandwidth available will be massive, even with 'just' DDR3 1066. Nehalem appears it can access its cache something like 50 percent faster than current phenoms, despite there being four times as much of it! AMD have come a good way in the GPU segment, but expect at least another year minimum of intel parts crushing AMD's.

Wow didn't knew Nehalem was shaping up to be so good. I might be tempted to buy it, but I will have to sell my current motherboard and cpu Since I doubt the Nehalem will be compatible with a 680i. *Sigh*

Nehalem will launch on LGA 1366.

Yes, it's a whole new socket and you'll also need to buy a whole new CPU cooler as well unless you can buy a new bracket for your current cooler. If your lucky maybe Zalman will release a LGA1366 bracket for the 9500. The LGA1366 CPUs are more rectangular in shape then the current LGA775 CPUs.

vulcan raven said:
Phenom OC's have nothing to do with whether its based on 65nm or 45nm. The problem is the current chipset that has the SB600. When you try to OC the clock generator falters for some reason. Hopefully AMD will have a fix for the old SB600 and beond so that people taht bought mobo's based on it won't have to uupgrade to a newer mobo.

The newer chipset based on the SB750 has no such problem and therefore 65nm Phenoms can be easily OC'ed well past 3.0 GHz. And the high k 45nm will only prove to be better.

Clock for clock both Core and Phenom are almost neck to neck in benchmarks with a few going to Intel and a few to AMD.

Then tell me how a 3.0 GHz Core 2 Duo beats a 2.5 GHz Phenom X3 in every benchmark. Phenom's only advantage is with a 128-bit wide SSE. If the application doesn't use that, it gets spanked.

Super XP said:

Current Intel offerings run circles around AMD chips even on a clock for clock basis.

Jim H said:

maybe not circles, but kentsfield is proven to be at least ten percent faster clock for clock to phenom overall. its a proven gap, phenom is not as fast as the old kentsfield design, never mind the yorkfield. the 2.4 phenom was up against G0 Q6600, 2.4ghz. of course the intel part is faster, cooler, uses less power, overclocks better as long as you have a decent (and now cheap) chipset. when you start to overclock them too, no doubt the gap increases between them, core 2 has so much more cache the faster they are clocked the more it will matter.

it doesnt matter what AMD do with the chipsets, if they dont overclock better than the core 2's then they will not be faster. 3.4ghz on air for a no doubt carefully selected extreme sample phenom with a crap load of vcore is still unimpressive on a 'working' chipset in comparison. as i said a low end kentsfield does that easy on standard retail 65nm parts and has done for nearly 18 months with filthy cheap 965, nforce 650, P35 sets etc- consistantly chip after chip.

and as i pointed out, you are looking at another 6 months tops and intel will give AMD such a bloody nose with nehalem on the performance front, with phenom not a year old itself after the paper launch and errata farce. AMD wont catch up with core 2 before nehalem hits. then its an even harder time for them, because intel will relegate all the core 2 parts to budget bins. they will form up as the budget parts, which are still better than AMD's top offerings. before, AMD could position athlon X2 as a budget part and compete with intel on price performance there. but what about when nehalem claim top spot, and there a tonne of cheap core 2 parts swilling around the pipes. who are AMD gonna be selling to then?

vulcan raven said:
People like Samba and Super.

Somba, not Samba.

Vulcan stop spreading your FUDBS, I can smell it from right here.

FordGT90Concept said:


Core architecture is NOT 128-bit wide?




.. hint.. Macro-Ops