Unless anything changes dramatically... the configuration is almost deffinatly 1600 shaders, 32 rop, 80 tmu at a significantly higher clockrate than the 48xx. But don't get fooled into believing the increase in hardware is going to equate to a similiar increase in performance... atleast not with contemporary games... actual increase in nominal shader processing will be closer to 60-70% than 100+%, etc.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
And the Cypress shader count is...
- Thread starter Tom|Nbk
- Start date
Yeah I have to second this. Sure their performance predictions are often out of whack, but they do seem to get the interesting details right.
I think your performance increase predictions are right if we're assuming the cores are running at the same clock speed - given that RV770 has 2.5 as many shaders and texturing units as RV670, but at more or less the same clock speed is only twice as fast, if RV870 has twice as many shaders as RV770, logically it should be about 160% as fast at the same clock speed. I think since the core speed should be a bit higher than RV770 we're looking at a 70-80% performance increase in most games.
Last edited:
Soldato
- Joined
- 11 Oct 2008
- Posts
- 3,833
- Location
- London
Eh? They both threw up a ton of random guesses and got one of them right.
I predict the new ATI cards will be between -5000% and 5000% faster than the current generation - WATCH THIS SPACE!
AMD make a point of not talking about this stuff until launch, unlike Nvidia who prefer to do all their launches on paper (much cheaper that way).
I predict the new ATI cards will be between -5000% and 5000% faster than the current generation - WATCH THIS SPACE!
AMD make a point of not talking about this stuff until launch, unlike Nvidia who prefer to do all their launches on paper (much cheaper that way).
I think we will see much bigger gains in newer games, but that won't be so apparent at launch... once games start using compute shaders, etc. I'd reckon the 5870 will outperform the 4870 by more like 300% - hard to guess too accuratly at that tho.
Ton of random guesses?
I only remember them ever stating the 4800's would have 800SP. As Light said, they're normally way out with performance figures as those normally are guesses, but they seem to get the small print right
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
Its not quite that simple realistically as theres all sorts of limitations that you can't account for. The 4870 might only have been X times faster than the previous cards because it was limited by its number or rops/tmu's rather than clock speed or shaders. They might pump a lot more rops in, but they become held back on this next card, but too low clock speeds. its a fine balancing act and one thats hard to get right.
But I was mostly posting to say, people can knock Charlie all they want, he does have an axe to grind, as should everyone really, in terms of Nvidia. So what, 98% of what he posts has been proven correct since he's posted it, most notably the fiasco in laptop parts and faulty materials used in Nvidia parts, he basically broke the story, found the evidence, proved how widely it spread and didn't get a single thing wrong.
As for Fud getting anything right, for as long as Fudzilla's been around they've been copying and pasting Charlies info right after Charlie posts his storys up so I wouldn't give Fud a single slice of credit, not to mention they put out a lot more "rumours" than the Inq did of which many more were wrong. Fud just goes around every site and posts every news story he see's as his own.
One other thing is, we really don't need Nvidia, at all, while Nvidia's plan is stupidly, to make there stuff as high priced as possible to prevent most people from upgrading, ATi's has significantly changed in the past 2 years to, sell as many cheap cards as possible, because theres tonnes more profit selling 20 million cheap cards, than 2 million expensive and hard to make cards.
If Nvidia died tomorrow, and ATi could price their cards anywhere there wanted, they'd take in LESS profits by selling their 5870 at £350, instead of £200-250, Nvidia have become irrelevant now that ATi have moved to the small core cheap and affordable segment.
You also have to realise that, unless they offer significant performance difference between generations, then again people would stop buying, they only make money if people keep buying new cards. if because of no competition their new card is uber cheap to make, but only 8% faster, no one would buy it. If they made a 300% faster card for £800, no one would be it. If they make a 60-100% faster card every 12-18 months, they'll sell tonnes, and tonnes and tonnes of them.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
It won't, new games will see a large portion of there speed gains from DX11, taking advantage of DX10.1 heavily aswell, what we'll see is, in new games, the 5870 being miles ahead of Nvidia(more than double and it will destroy the 295GTX based on the fact that with tesselation and decent multithreaded use, it will gain even more of an advantage) but the current 4, 3 and 2 series cards will all see gains in performance and move further ahead of equivilent current Nvidia cards than they are now. Same way they all get a 20% bump in Assassins creed under dx10.1.
The gap will likely widen slightly, due to the extras in DX11 over 10.1, and more efficient architechture, refinements to existing bits in the core they've learnt from, but it won't widen hugely as the old cards will be getting a decent boost aswell.
I doubt that very much, with no option but to buy AMD gamers wouldn’t have any choice, they either buy the overpriced high end card or the overpriced mid range card, with no choice there profits would increase one way or the other
If you think otherwise then you’re just deluding yourself or your totally clueless, with the comment about nvidia being irrelevant I definitely know which way I'm leaning.
The fact Intel stated they will come into the midrange and downwards very aggressively, and would have no problem under cutting Nvidia by 50%.
AMD can use its own ex CPU fabs for building cards, so they would have little trouble competing in a three way price war.
Nvidia still have to pay TSMC, so it could be a question of how will it costs Nvidia to sell a card and how much this will affect relations with board partners.
Nvidia are struggling to compete right now with just one competitor.
ATI are now about to release a range of cards that look set to wipe the floor with anything Nvidia can offer until around Fed/March. which is about the time ATI could refresh its range, and Intel are about to get aggressive in segment thats easy pickings.
Associate
- Joined
- 30 Sep 2005
- Posts
- 144
If you double all the functional units (ALUs TMUs ROPS) then performance will double, assuming you've got enough bandwidth to feed those units. So again, if AMD keep the current ratios, all that will potentially hold them back is the DDR5 memory.
What intel say and what actually happens is a totally different thing... demonstrations (by intel) of simulated larrabee performance are distinctly underwhelming by this generation performance let alone in the future... and intels approach to drivers will get them nowhere in the gaming world even at low/mainstream...
Sorry I don't see intel as a threat in the gaming market to either ATI or nVidia unless their entire attitude and methodology shifts dramatically.
And the despatcher amongst other things...
I am no fan of nvidia but don't wish them gone remember that once they were good innovators and hopefully can get rid of some of the deadwood n head office and get back up there. I do think nvidia is about to hit one of it's hardest times for a long time and thats usually when you see how good a company can be if they can innovate their way out of a difficult period we might all be better for it. If they continue down the company shenanigans route then i do fear they will quickly diminish as other companys have.
I am with Rroff on this (first time for everything 
) intel may possibly be something but it won't be this year or next and will require a complete rethink of how they do things to be a major player in the mainstream gfx market.
) intel may possibly be something but it won't be this year or next and will require a complete rethink of how they do things to be a major player in the mainstream gfx market.nVidia appear to have backed themselves into a corner a bit... they have nothing to offer the low or mainstream market that has proper DX11 functionality... nothing to replace the 8800GT which has been their mainstay lately...
TBH I do see them using their muscle to push developers into including vendor specific render paths for advanced effects that would normally just be implemented under DX11.
TBH I do see them using their muscle to push developers into including vendor specific render paths for advanced effects that would normally just be implemented under DX11.
You keep confusing Intel now with Intel of the past Rroff, the Netburst boys have gone, this isn't a half attempt to sell white box's to OEM's exercise ! they want in and are ready to compete.
Intel have unlimited resource's and are serious. Look what happened with Core 2 after the idiots got the boot. A new architecture from drawing board to end user in 18 months.
Maybe I'm not giving them their fare due but I just don't see them as good innovators at all. Back in the day they were in court for ripping off 3dfx IP and would have probably lost that battle had 3dfx not been so poorly managed and gone bust.
ATi have lead the way through DX9, DX10 and DX10.1 and now look like again beating NVIDIA to DX11. Whilst the 5900Ultra dust buster (sorry for mentioning this) did eventually morph into a half decent card in the DX9 timescale(5950Ultra) it was much better at DX8. Sure they had their Shader model 2.0 out before ATi but it was a checkbox SM2.0 win and ATi eventually gave us their SM2.0 done right cards.
Of course the 8800GTX was a hail mary for them and certainly a Voodoo, R300 beauty...
J.
EDIT: Fare = Fair (I know lol...)
Last edited:
I'm not confusing anything... intel have huge amount of resources... they are serious, they want in... but they have entirely the wrong methodology to succeed no matter how much they throw at it... their methodology works for CPUs very very well and I hope they continue to throw their efforts there.
The 5950ultra was merely acceptable... in reality it had really horrid DX8+ performance still and most of that performance came at the extreme expense of visual quality - even tho it was a monster at DX7.
SM2.0 is a bit of a thorny one... if ATI had got on the ball with SM2 earlier instead of concentrating on 3dc, etc. which really wasn't a priority at the time things would have moved on much faster in game graphics.