ATI RV870 Expected to Arrive in Late July

I wouldn't put it past AMD/ATI to beat nVida with DX11 capable cards considering how it was reverse for DX10 and that would leave them with the only DX11 cards out with the release of Windows 7, satisfying those gotta have it early adopters. I just hope that AMD/ATI didn't sacrifice anything in the way of performance in order to get it out the door sooner.
 
i think the only cause for concern would be if ati announces an official release date then pushes it back significantly. if past instances are any indication, ati/amd does not respond well to technological stumbles.
 
i am sure nvidia will beat ATI again.......

by renaming it is current high end part to compete with ati's low end, say rename gtx 275 against HD 5350.

win again !
 
I think its funny that so many people are going to argue which is going to be better, despite the fact that we have ZERO reason to think either way or the other.

No matter what any of us say, it is pure speculation. Yet many of you will get your e-peens bent out of shape when someone says you are potentially wrong -_-
 
If they stick with their current naming scheme, yes. If ATi is first to market you can almost bet the farm that nVidia's chip will be faster.

that is not always the case, everyone thought that hd 2900 was going to be really fast but it was barried before in terms of performance even though it was months late.

it has nothing to do with performance but everything to do with who masters the 40nm process first, and ati seems to have done that, because the next gen will not be made on 55nm, so 3 months after the hd 4770 (their first 40nm card) seems pretty reasonable to me. I am sure that gt 300 will be one fast chip, and so will rv870.
 
If they stick with their current naming scheme, yes. If ATi is first to market you can almost bet the farm that nVidia's chip will be faster.

Well There is one example where you would be out one farm...

The R300 series, the 9700 pro.
 
that is not always the case, everyone thought that hd 2900 was going to be really fast but it was barried before in terms of performance even though it was months late.

it has nothing to do with performance but everything to do with who masters the 40nm process first, and ati seems to have done that, because the next gen will not be made on 55nm, so 3 months after the hd 4770 (their first 40nm card) seems pretty reasonable to me. I am sure that gt 300 will be one fast chip, and so will rv870.

Well There is one example where you would be out one farm...

The R300 series, the 9700 pro.

Granted. There are always exceptions to the rule as both of you have stated. I won't dispute that at all witnessing the sucess of the 9700 Pro and the flop of the 2900XT and GeForce FX line. But usually the last one to market is the faster part. We shall see. I for one am intrigued especially since Windows 7 is definitely in my upgrade plans and I don't want to wait for a DX11 part if I don't have to.
 
hopefully, it wont be a re-rehash of the r700 but on a 40nm node.

actually, that's more or less ati's intention. remember, when they launched r700 they stated emphatically that this is the gpu architecture they were going to stay with for a while, and they engineered it from the ground up to be efficient and scalable. this doesn't mean that the new chip won't be a great performer. if the last couple of product releases are any indication, ati seems ready for anything nvidia has to offer.
 
agree'd renny.. and since it looks like nvidia's expecting a release date of some where around october, november for the new gtx 300 series.. i think amd/ati will push as much as possible to get the r870 out as soon as possible.. the gtx 300 series seems to be a copy cat of amd/ati's 4k series with the 256bit GDDR5..
 
i'm not counting nvidia out either. they've proven that they don't necessarily need to do anything exotic to compete with ati. ati engineers have a bit of a reputation of being gadgeteers; the ring bus technology introduced with the x1xx series cards, or the supposed 'stacked' gpu design of the flagship r800 product are perfect examples. even if nvidia's using ati's throw-away technology, i'd be suprised if the gtx 300 doesn't keep up with the ati product.
 
i'm not counting nvidia out either. they've proven that they don't necessarily need to do anything exotic to compete with ati. ati engineers have a bit of a reputation of being gadgeteers; the ring bus technology introduced with the x1xx series cards, or the supposed 'stacked' gpu design of the flagship r800 product are perfect examples. even if nvidia's using ati's throw-away technology, i'd be suprised if the gtx 300 doesn't keep up with the ati product.

I am sure they are going to come up with something as well, they have been well spanked this round (good thing they had the cash to back it up). But its hardly fair to refer to ATI strategy as "throw away technology", it was by any measure brilliant. the new seems good as well, shrink the die, fill in some of the room with more units and save in production cost with a smaller die space, release and spend the time Nvidia plays catch up improving the chip and spinning off variants. if it goes like this last round Nvidia better had copy that model. It would do better then chasing physx and cuda. this is going to be interesting, We have a good idea of what ATI is going to do, nvidia has me wondering.
 
I'd just wish that both companies would also follow their excellent cards with also excellent drivers :) That would really be helpful.

Actually, there will be some time before DX11 games arrive, so propably by that time we will get a second generation of dx11 cards, with all the bugs ironed a bit and performance improved. I don't think we will get any dx11 software before maybe next summer. I also hope for a good competition on a level of 48xx and GT200. It will only benefit us

Oh, and one thing I'd gladly see is that NV/ATI partners would this time create silent and efficient cooler on stock cards (yeah, I know that it's a wild dream :p)
 
hmm, I honestly can't complain about the stock fan on the hd 4890, it is pretty silent on idle and while gaming It is bearly audible, and I mean bearly. silent is never going to happen if we are going to manually control the fan speed, if ati locked the fan speed in bios it would sure be silent. if you wanna overclock you are going to have to spend money on after market cooling or get an overclocked version with improved cooler like the TOXIC series by sapphire. Neither nvidia nor ati is going to invest heavily in cooling solution, what are the AIB partners for?
 
The real question, is will both companies get the dx11 archetexture correct, for performance reasons.

The R300 was good for dx 8 games, but very good/great for dx 9 games, especially since it allowed playable framerates when dx9 games came out later, like half-life 2, and ut2004.

the geforce 8800 was very good for dx 9 games, but great for the dx 10 games, simply because other then crysis, it sitll has enough horsepower to run most/all new dx 10 games, at reasonable resolutions.

I just hope both the RV870, and the gtx300 series, are both very good/great on dx11 games, so we don't have to toss them out, like the fx5800/5900 when we want to run games in dx11.
 
What changes are DX11 going to make anyways. Might be like MS is rebranding DX10.1 to DX11 like nvidia did with the 8800gt to 9800gt and 9800gtx+ to the gtx 250.
 
I am sure they are going to come up with something as well, they have been well spanked this round (good thing they had the cash to back it up). But its hardly fair to refer to ATI strategy as "throw away technology", it was by any measure brilliant. the new seems good as well, shrink the die, fill in some of the room with more units and save in production cost with a smaller die space, release and spend the time Nvidia plays catch up improving the chip and spinning off variants. if it goes like this last round Nvidia better had copy that model. It would do better then chasing physx and cuda. this is going to be interesting, We have a good idea of what ATI is going to do, nvidia has me wondering.

you seem to be hearing what you want to hear. my comment about 'throw-away technology' is a reference to nvidia's engineering philosophy and in no way a slander against ati. please go back and read what i typed. nvidia will take established manufacturing and design methods and stretch it out for as long as possible. consider how long it took them to adopt 55nm manufacturing. this strategy has not hurt them in the least in terms of product performance. ati engineers simply take a different approach, they like to tinker a bit more. neither is better than the other in my opinion, as both companies are reasonably successful at what they do.
 
870 to produce 1.9 Teraflops:
http://translate.google.de/translat...-ueber-19-tflop-im-juli-09&sl=de&tl=en&swap=1

Does this signal the use of 1280 shaders and ~750MHz clockspeeds, given that this is almost exactly double the 640 shader rv740 which produces 960 Gigaflops?

Probably, They may increase the other portions of the chip a bit, but ATI's shader structure is not really too space intensive, thats why they were able to ramp up significantly from the 3800 to the 4800.
 
I don't understand the point of releasing something like the 4890 if you're going to release something even more powerful less than a business quarter away.
 
I don't understand the point of releasing something like the 4890 if you're going to release something even more powerful less than a business quarter away.

Testing ground for new shader paths that give the 4890 an improvement over the 4870 clock for clock, which is then (presumably) applied to RV740/4770, and is also likely to be carried over to RV870.
 
Testing ground for new shader paths that give the 4890 an improvement over the 4870 clock for clock, which is then (presumably) applied to RV740/4770, and is also likely to be carried over to RV870.

You could consider the 4890 an important test for allowing higher clock speeds.

By tightening up the timing inside the chip, and including better power control, and the decap layer, ATI was able to create a chip with a speed of 850, which is ~13% higher then the 4870. And per overclocking, the 4870 really only consistantly hit ~800-810ish, the new chip looks to hit 950ish consistantly.

It would be a good test bed for higher clock speeds, which is important if they keep to smaller dies then nvidia.
 
jesus man who cares if "ATI Expected to Beat Nvidia with DirectX 11 Chip Time-to-Market"???

Just give me a great card and I'll be happy, I don't care who makes it like all of these fan boys around here :mad::rolleyes::eek:
 
870 to produce 1.9 Teraflops:
http://translate.google.de/translat...-ueber-19-tflop-im-juli-09&sl=de&tl=en&swap=1

Does this signal the use of 1280 shaders and ~750MHz clockspeeds, given that this is almost exactly double the 640 shader rv740 which produces 960 Gigaflops?

I personally think there is going to be A LOT more shaders then that. There is no reason why there should not be given how small the process is. The estimate is probably based on a wicked low core clock I bet. ATI crammed 2.5 times the amount of SPs on the same process compared to the RV670. I'm betting they will do similar with the RV870 considering it is a smaller process after all.
 
RV 870 Q2, no way. Expect it sometime in Q4.

Can you give one reason why? The RV740 in 40nm is coming in a few weeks. There seems to be no reason to prevent the RV870 from coming out. I read on a few sites AMD announced that they're releasing DX11 cards in the summer.
 
Every thing I have read, has pointed to a mid/later summer launch, and they have had a working 40nm chip being qualified for quite a while, remember the guru3d preview of the 4750 board a few months ago?
 
you seem to be hearing what you want to hear. my comment about 'throw-away technology' is a reference to nvidia's engineering philosophy and in no way a slander against ati. please go back and read what i typed. nvidia will take established manufacturing and design methods and stretch it out for as long as possible. consider how long it took them to adopt 55nm manufacturing. this strategy has not hurt them in the least in terms of product performance. ati engineers simply take a different approach, they like to tinker a bit more. neither is better than the other in my opinion, as both companies are reasonably successful at what they do.

sorry if I took that wrong but I thought you were being literal when you said "even if nvidia's using ati's throw-away technology," I do agree that nvidia is not going to rest on their laurels
 
Can you give one reason why? The RV740 in 40nm is coming in a few weeks. There seems to be no reason to prevent the RV870 from coming out. I read on a few sites AMD announced that they're releasing DX11 cards in the summer.

Just rumours I picked up from reliable sources over at xs. Evidently these sites dont know the difference inbetween tape out and releasing.

Every thing I have read, has pointed to a mid/later summer launch, and they have had a working 40nm chip being qualified for quite a while, remember the guru3d preview of the 4750 board a few months ago?

rv 740 is just a die shrinked rv 770 thats supposed to act as an experimental test bed of 40nm manufacturing process. RV 870 is a highly modified rv 770 (more shaders / tmus etc) thats going to be 40nm right away...not that simple.
 
Last edited:
Yet you still have nothing to back you claim up. I'm going with what rumor mill has to say this time.
 
Yet you still have nothing to back you claim up. I'm going with what rumor mill has to say this time.

Rumour mill said that rv 790 was going to have 960+ shaders etc...

Besides then its just one rumour vs another.
 
Back
Top