GTX260 SLi vs DX11

I would wait until ATi's cards come out, which will most likely force nvidia to drop prices. At that point you can decide if a second GTX260 will yield a better price/performance ratio than buying a 5870.

From what we know, the new cards are twice the output of the old.
You may find 1 single 5870 is faster than a 260 sli setup, most likely.
It they output 2 terabites, the 5850 and the 5870 slightly more.....

I'm in the same position right now too, except I'm thinking about getting another 4890. I'm just going to wait a month just to see what ATI is going to do, and see if there's going to be a price drop. Unless you really need the performance boost, you might want to wait.

The high end card that outputs 2 terra*** is probably the dual-GPU card similar to the 4870x2 though.

The GT200 series and the ATI 4x00 was an improvement over the G92s and 3x00, but it wasn't a huge improvement. I wonder when the benchmarks will come out... =/

Teraflops people.

according to wikipedia the 5870 has double the shader units, so it may actually push 2 teraflops

edit: double the shader units of the 4890

Aye, but the 4770, 4850, 4870, 4890 all have 800 stream processors and they perform quite differently. Nvidia has 1/4 the stream processors and it performs the same as well. I hope they keep the DDR5 and increase the bandwidth...

It all comes down to benchmarks... =/

Reread what I typed. They refer to both cards a multi terabyte cards

Supposedly, the multi core monster comes out later, having possible 4+ terabytes
Or even 3+
Also, how many on the 4770 for shaders?

"It they output 2 terabites, the 5850 and the 5870 slightly more..... "

if that doesn't mean "the 5850 outputs 2 teraflops, and the 5870 slightly more" then I have no idea what you mean.

Multi-terabyte what?

Yeh, I'd say it's all hype then. It's very unlikely ATI could pump out something twice as good as the previous generation.

I'd say a 30%-40% improvement max in terms of price vs performance. If it has at least a 30% improvement, I'd buy it.

It's TeraFLOPS.

http://en.wikipedia.org/wiki/FLOPS


Terabyte (not "bite") is a measurement of storage.

http://en.wikipedia.org/wiki/Byte

IIRC the 3870 was getting pretty close to 1Tflop on it's own, and the 4890 only hits somewhere around 1.3Tflops. To see them get all the way up to 2tflops would be incredible, but not unheard of (generation to generation wise I mean).

drool...    

Having a gray moment, thanks for the correction.
Its already out, part of the leaked info stated both cards to have 2 teraflop abilities
Looking at the rumored P scores on Vantage of 16000-18000, tou can guess a few things from there.
These cards are that powerful

If they seriously hit 2 teraflops, and that performance shows up in games, *** is gunna hit the fan... the fan being nVidia

nVidia is already hurting bad in the sub-$200 market, if they get creamed in the $200+ market....

"Feel the power of more than 2 teraFLOPS plowing through the most demanding games."
Look for this on the new boxes containing the new cards

you're the reason why chip makers only increase performance slightly with each generation rather than pushing it to the technological limit.  

Or maybe that they are businesses and spending 2x longer on RnD isn't financially feasible...

lol, I'd rather buy something for $100-$150 that is 30%-40% better than the previous generation than buying a card that is 100% better but costs $500.   Technological limits tend to be very expensive...

Then it looks to be a ATI win

i think its alright to go with 260 SLi but if you really concerned about performance maybe you should wait for DX11 capable card to see what kind of performance it will bring.  

damn it where are you ppl seeing these 58xx reviews? Or were they only like them 3dmark like benches and not game benches

+1 Excacly ,why not just push it for at least 70% more power it would propably stop people ranting that their cards r not running as they should in games ,the next profit for games would be lesser upgrades, I mean I think its stupid and crazy to upgrade every like 7 to 8 months

Anyone remember the hype about the 2900XT and how it was the best thing since sliced bread before it was released?, all this is so familiar lets hope it's not a deja vu situation in a few weeks time.

One site reported a "leak" of what appears to be a single run of vantage. Not a review, not even close to something that means anything.. but it is the best info we have at the moment as to how fast these suckers will be when tehy are released.

thats why i'm not saying much about 'how well this card perform against present gen card' right now. i will wait for proper benchmark / review when the real thing comes out  

OP, is power consumption a factor when choosing what setup to go with? Personally i'd wait till the first DX11 cards a released

Not that Id do it, but this could become sig worthy mm.
Its true, we dont know exactly how these cards will perform, but heres what we do know.
ATI has mentioned the perfect answer in the 4770, and wed understand after the 5 series rolld out. So lets just take a look at the 4770. Modded, the 4770 can equal a 4890 in perf. Thats at core 1Ghz and max on ram, which is low ram, being rated at 800Mhz
OK, take the same 4770, add 50% shaders to it, add the biggest change since going from R500 to R600, which includes the scheduler, run the clocks up so it has at least 2 teraflops throughput and you have a card thatll put down a 285.
Now, add possibly 33% more shaders of the 1200 or so shaders, and maybe raise the clocks a lil higher yet so you get 2+ teraflops, you see where this is going? The reason to rush to market? The reason to make such a to-do for their release? The quiet confidence?
ATI set out with the 4xxx series to rectify the R600 series, and decided that they would go after market share. Their release brought down the pricing of nVidia cards immediately upon release. The 4830 brought down the price of the 9600 immediately upon release, as did later on the lowering in price of the 4850 and the 4870.
It was easy to see that ATI more or less controlled the pricing structure except the 295, which many agree is way overpriced per perf.
So, why all the hoopla if these cards are duds? Why drastically change your approach? Why make a deal on a card that according to some is pure fail? Why the rush? The confidence?
If nothings changed, just going by the numbers we know, these cards will own the gfx perf when they arrive. If theyre better with arch changes, and using the DX11 structuring, thats just a bonus.
So, just going by what we know, this isnt pie in the sky here. And going by ATIs current market approach, why change a good thing when you have it good?

IIRC there was quite a lot of "what we know already" and "leaked info points to" and all manner of conjecture pre 2900XT release but that's by the by, I want these cards to be good, they need to raise the bar to a level we have never seen before, then Nvidia will have something to aim for.

Its too late for ai9ming, as the G300 is in the fire and partly cooked.
My point is, the 2900 was a severe change, hobbled by many things, things we wont see this time around, and can actually use the preious gen as more of an example, unlike using the 19xx cards, and comparing.
The 40nm has had trial, at least on ATIs side. Its not the leaky node that the 80nm was.GDDR5 has also been used, and has come thru with its promises, unlike GDDR4, which was basically a failure, as it wasnt used for higher scaling, and wasnt seen as beneficial overall from GDDR3, whereas, you can bet nVidia will be touting GDDR5 in thier new gen.
So, yes, you can make a few assumptions this go round, and keeping it on the conservative side, the 285 wont be the fastest card once these cards are released.
If youre a nVidia fan, hold onto your money, as Im sure, the 275s are going to drop in price, as theyll most likely be considered low-mid end

Personally I shall be hanging on to my cash until the second gen 300's at least, when there might even be a DX11 game that I feel is worth playing.

ATI has plenty of time to tweak the 5000 series, so the fact that they are releasing it with Windows 7 even though nVidia's answer wont arrive this year, speaks to a lot of confidence.

The difference is that the 5000 series wont be stunted from nVidia making microsoft change DX10 so their cards would perform better than ATI's which were prepared for real DX10, or DX10.1.

DIRT2 and Battlefield look alright for starters, therell be others as well, some patched, and more using a real DX10 engine, where compatibility is more assured by dev usage, thus those DX11 improvements.
Im hearing the texturing will be increased more and more, how that plays a part, DX11 or no DX11, may also change whether to sit or buy new, as Im sure the newer cards will bring tons of perf we dont have now.
Like we saw in the ATI cards, moving to a 1gig standard in the high end, may become more needed in the lower segments as well

Excacly, BlueScreenDeath is the reason why chip makers make small incensement in performances. We must burn BlueScreenDeath  

GPUs that are too powerful often cause the infamous blue screen of death since software can't keep up.  

HUH? Proofs? Links?

You are having a hard time choosing between a reliable $200 card or a brand new untested likely problematic card at the $500+ price point? I think it sounds like a strange thing to ask when phrased this way.

If I was functional with a single 260 right now, I would at least wait and see if this new expensive superhype card can drive the price of a second 260 down.

If the superhype card looks good and my single 260 was still working OK for me, I might wait for the superhyped price to come down.

I might even wait a couple years and by then there will be software written to take advantage of the second gen superhype cards which will then be a price-performance value and actually necessary for my new computer build with 64GB DDR8 RAM, an 8Ghz core i8 octo-core CPU, 2TB SSD and a 42 inch LED monitor.  


I guess I just dont see the point in spending premium dollars on something thats not guaranteed to provide a premium benefit. This new card coming out is not going to change the hardware requirements for the software you are currently running. Either you have a problem now that you cant wait for so you must buy the second 260 or you dont and it would be far more cost effective to wait until you replace your CPU.

No one listens to me anyway, no reason you should.  

Huh what?

Of course it will be double as fast. The 4850 is as fast as the 3870x2. With a few exceptions, graphics generally double in performance every new series.

Also, the OP's question is the exact same reason why I kept telling people to buy 2x4770's back then, instead of buying a 4890. When these cards are out you are not gonna buy another 4890 to xfire because they'll probably only be as fast as a single 5870 but not have all the dx11 goodies.

Youve really opened yourself up to some serious flaming with these statements. This is exactly the type of thing Ive been reading about. Wait for later, these cards are no good, DX11 wont come til later, W7 is no good, DX9 is safe
Did Jenson hand you the manual personally? Name 1 550$+ ATI card? Oh wait, thats nVidia territory. So, only nVidia gaurantees a premium benefit? Im sorry, but this is FUD
I'll let others decide how to deal with your words, Im done already, its just too funny

I would ear mark early early 2010 either for the GTX3XX given the financial situation at nVidia, as far as I'm aware they still haven't settled on the class action lawsuit brought against them by Apple, Dell, HP and others for their 8000 series debarcle. In their last quarter nvidia reported a $200+ million loss which when compared to ATI with a loss of $38m is quite a kick in the stomach.

Of course, part of the ATI loss may mean ramp up production on DX11 card costs as well, you never know

Well ATI made more in revenues compared to Q1, yet lost $38m so yeah you can be pretty sure that there are a few hundred thousand 5x series gpu's sitting in warehouses already.

well if its something less than 300 CAD (think 200-250 USD) for the top of line single card, then I'm more or less sold.

If not, gona go with 2x 4890 when their prices drop stone low.

If both don't happen at launch, then it's whichever hits the mark first for me lol.

Or GT300 comes in first.

Trust me, the 4890 will drop well before youll get you hands on a G300.
Theyll be more than that tho, esp CAD. With no market pressure, Im thinking 350 US for top card

I'll have one of my 4770's on ebay next week and the 2nd one will go once i have whatever 5x's I'm gonna buy.

I can't imagine any other prices except $299 for the 5870 and $249 for the 5850. The 5850 might well be the better buy at that.

What I find interesting here is, the old rumors, ya know, as fast as the 295 etc, are now being questioned
as being too low heheh

Rumors like seperate shader clock domains, up to 8 teraflops etc, and these are from fairly smart people, but to me, that looks more like a wish list

What will BSOD do with his current 260? Sell it ? He can get a 65% performance increase by adding another 260 ........

IMO you should just SLI and save.... no point in waiting for the new cards ATM if DX11 won't be fully implemented till mid next year..... none the less DX11 games are about 2% of the market.... I say wait till next year if you plan on buying a next gen card, they will be cheaper and more usefull.......

But again this is just my opinion......Since he has a 260 already it is kind of a no-brainer to get a second 260 and SLI..... unless I missed something here...