NVIDIA GeForce GTX 570 in December; AMD to launch Radeon HD 6900 Series on December

[Chrono Detector]

I want a dual Fermi card, and also, I wish AMD would just release their 69xx series already, I want to know the performance of the cards. Should be interesting.

[tajoh111]

Nvidia seems to be doing a few good moves and might be on a financial roll.

http://www.semiaccurate.com/2010/11/...settle-nvidia/

I don't know how, but Intel might be settling its chipset lawsuit with Nvidia and give them a billion dollars. WTF.

With things like supercomputing and Tegra 2, and from the sounds of it, Nvidia actually giving AMD stiff competition on time this year, its no wonder analysts are upgrading Nvidia to buy again.

http://finance.yahoo.com/echarts?s=N...=NVDA;range=3m

Their stock has been going up steadily for the last 3 months and the 4th quarter results are supposed to beat early estimates.

I am happy knowing NVDA will atleast be around from for a while and AMD is not going to crush them as I had original thought when the 6xxx series was announced.

[herderien]

worse, a mac

noooooooooooooooooooooooooooo !

[Alien Grey]

are also reported in the NVIDIA GeForce GTX 570 GDDR5 memory to 1536 MB volume associated with the GPU via 384-bit bus. In this edition of the accelerator is expected before the end of this year, and the video card, probably will replace the GeForce GTX 480 model, offering about the same or even better performance for less money.

Source: http://it-chuiko.com/computers/7143-...sii-gf110.html


Could this be true?

[Metroid]

Ops. This could very well be my next card

[Dimitriman]

Source: http://it-chuiko.com/computers/7143-...sii-gf110.html


Could this be true?

Sure why not? at this point its basically disabling 32 cuda cores on the 580 as they had done to the 480 and there is your 570 with less power consumption than 480.

[saaya]

hasnt amd already said officially that adjusting the clock speeds of duzens of thousands of chips could take months not weeks to perform? why do "news" sites keep insisting that it is because it wont be competitive?

please... what a lame excuse from amd is that...
there have been half a dozen of lame excuses about why the 6900 series is delayed, one worse than the other...

cards that launch in mid december?
sounds like they will miss the holiday shopping spree...
pretty amazing that nvidia managed to get the 580 out in time tbh...

i wonder what went wrong at ati/amd...
and if the 6900 series really cant compete with gf110, then ouch... that would be a big fail from the red team...

i mean rv940 is 25% smaller than rv870 and not much slower, around 7% on average according to tpu. if your normalize the clocks (7% higher fot rv940), ati needs around 10% less transistors for the same performance with rv9xx vs rv8xx, or gets 10% more performance out of the rv9xx parts.

so an rv970 at the same die size as rv870 should be 10%+ faster, which would put it below a 480, but once you up the clocks it should be on par with a 480/570.

but rv870 wasnt very big, so it would make sense that ati felt more confident especially since 40nm is mature now, and went for a bigger chip.
i would honestly be very surprised to see ati go for the same chip size as rv870 and not more... so lets just look at how big rv970 has to be to beat a 580, and lets assume rv870 clocks, which shouldnt be a problem.

rv870 = 333M
gf110 = rv870+33% (1080 tpu av)
rv9xx@333M = rv870+10%
so we need ~23% more transistors to reach gf110 performance, on average
that would result in roughly 400M transistors... that shouldnt be a problem to make on tsmcs 40nm node these days at all...
and this is with only rv870 clocks at around 850mhz...

to significantly beat gf110 ati would need a 10%+ av lead, again at rv870 clocks that would mean ~440M transistors...
such a chip would still be ~15% smaller than gf110... and since ati knows nvidia is going for such a big chip, why would they be greedy and go for a tiny 400M or even whimpier chip? at the kind of yields tsmc has right now that would barely make a difference in cost...
and then theres scaling which isnt linear, so actually they need a bit more than 400M transistors to reach 580 performance

they have two options, go for a massive chip like nvidia, which might have issues. they would have the fastest single card, but a bit late and they would have to cut down a lot to make it a dual card, which they need to keep the ultimate perf crown.

or, they go for a medium sized chip, between rv870 and gf110, which is slightly slower than a 580 but costs 100$ less, and which combined with a second chip can easily dethrone a dual gf114 card within the 300W limit.

my guess is the latter, as its whats ati has been doing in recent years and worked well for them. what would be the point do copy nvidias strategy which hasnt worked well at all for them recently...

plus, if rv970 would be huge, we would have heard of it, im sure...
all that was leaked is that partners got disabled and fused down cards... no mention of an unusually large die size...

so my guess is ~580 performance for 6970 and ~480/570 performance for 6950, at 399$ and 299$ and at a tdp of 250W+. and thats probably whats causing ati a headache right now, heat... with higher efficiency and stuck on the same node, heat is whats limiting them, not die size...

[Lanek]

.. it's the most logical move, specially as they was know Nvidia will release something then .

As you said i don't see AMD AIM the 6970 on the GTX480 level but rather higher, and higher you fall on the 580 range... Why AMD will aim his high end gpu to the GTX480 released 7 month before .. ? It's like if AMD have position the 5870 on the GTX280 level ...

[MrMojoZ]

please... what a lame excuse from amd is that...
there have been half a dozen of lame excuses about why the 6900 series is delayed, one worse than the other...

Are you sure they actually said that?

[marten_larsson]

The only inofficial official AMD statement is that increasing clocks could take months rather than weeks suggesting the delay isn't clockspeed related. The rest of Saaya's post is pretty much in line with most expectations, a chip smaller than GF110 but equally performing giving AMD the edge in perf/mm this time too and Antilles a beast, just like 5970 when it launched.

[Dimitriman]

please... what a lame excuse from amd is that...
there have been half a dozen of lame excuses about why the 6900 series is delayed, one worse than the other...

cards that launch in mid december?
sounds like they will miss the holiday shopping spree...
pretty amazing that nvidia managed to get the 580 out in time tbh...

i wonder what went wrong at ati/amd...
and if the 6900 series really cant compete with gf110, then ouch... that would be a big fail from the red team...

Mostly nothing went wrong with AMD. Whats happening is that a lot of Nvidias fanboys are still in awe that the company managed to deliver a product on time and are too blind to see that the 580 barely brings anything new to the table in performance. A lot also still believe that AMD will have a hard time delivering a chip that is merely 20% faster than the previous generation, ignoring the drastic improvement in shader efficiency obtained from the 6800 series as well as the 4d shader rework.

What we are seeing in truth is a tiny few setbacks for AMD being blown way out of proportions by Nvidia fans hoping and preying that their beloved company doesnt end up behind for another generation.

Its really annoying when people care so much about a brand they stop using their brains.

[tajoh111]

Mostly nothing went wrong with AMD. Whats happening is that a lot of Nvidias fanboys are still in awe that the company managed to deliver a product on time and are too blind to see that the 580 barely brings anything new to the table in performance. A lot also still believe that AMD will have a hard time delivering a chip that is merely 20% faster than the previous generation, ignoring the drastic improvement in shader efficiency obtained from the 6800 series as well as the 4d shader rework.

What we are seeing in truth is a tiny few setbacks for AMD being blown way out of proportions by Nvidia fans hoping and preying that their beloved company doesnt end up behind for another generation.

Its really annoying when people care so much about a brand they stop using their brains.

I think some of these fanboys as you put it, simply want to see the 6970 out so there is competition for the gtx 580. With these delays not only is there no real competition for the gtx 580, they don't get to see the alternative choice in the form of the 6970.

A three week set back is significant because alot of people will not get there cards before christmas as Saaya has put. For some gamers that are in high school or university, having the card to enjoy for christmas is rather important because it will be when they have the most free time. In addition, having the choice to get the videocard they want and not regret it for a Christmas present is nice too.

I think many NVidia fanboys are simply frustrated by the amount of flak AMD fanboys have said about Nvidia card owners. In addition, from what I can read from your post, they are frustrated with AMD card owners simply being huffy about AMD and how AMD will send NV down like the sword of Damacles.

Look at the gtx 580 thread for example, it was full of AMD fanboy saying the card doesn't exist and the thread should be closed. To me, this is simply a sign of AMD cards don't want to hear good news about Nvidia and want any good news to be denied to Nvidia card owners.

Although the gtx 580 is not a completely new generation, technologically the gtx 480 is more advanced than cypress, it simply has so many flaws that it prevents the card from really being a practical card. The gtx 580 is the card fermi should have been and if it is competitive with AMD next generation card, then we will have a price war. The GTX 580 is the card Nvidia owners can finally have that they can own without being joked about with heatjokes and complete and utter failure in power efficiency. The things the gtx 580 brings to the table are fixes for the gtx 480 and that is exactly what Nvidia needed most. I personally don't think its performance is that great, especially with the new naming moniker, but its a bigger gain than what most refreshes from NV has had as of late.

NV owners just want a competition between the two companies where they are on even playing field. Alot of the AMD fanboys want Nvidia to fall behinds so they can laugh at them and say they are a generation behind. Much of this is unfounded as well as much of the extreme optimism for AMD future products.

Of course there are certain hype Nvidia fanboys like the banned silicon doc who take fanboyism another level up.

[Dimitriman]

I think some of these fanboys as you put it, simply want to see the 6970 out so there is competition for the gtx 580. With these delays not only is there no real competition for the gtx 580, they don't get to see the alternative choice in the form of the 6970.

A three week set back is significant because alot of people will not get there cards before christmas as Saaya has put. For some gamers that are in high school or university, having the card to enjoy for christmas is rather important because it will be when they have the most free time. In addition, having the choice to get the videocard they want and not regret it for a Christmas present is nice too.

I think many NVidia fanboys are simply frustrated by the amount of flak AMD fanboys have said about Nvidia card owners. In addition, from what I can read from your post, they are frustrated with AMD card owners simply being huffy about AMD and how AMD will send NV down like the sword of Damacles.

NV owners just want a competition between the two companies where they are on even playing field. Alot of the AMD fanboys want Nvidia to fall behinds so they can laugh at them and say they are a generation behind. Much of this is unfounded as well as much of the extreme optimism for AMD future products.

Of course there are certain hype Nvidia fanboys like the banned silicon doc who take fanboyism another level up.

Sure I whole heartedly agree about the Christmas shopping time ans the need for competition. Unfortunately AMD will be too late for good price adjustments to take place before the holiday and I am dissapointed with the three week delay. What Im simply stating is that a lot of people automatically assume that a 3 week delay means AMD is on the losing end and wont deliver a competitor to the 580.

That doesnt mean I believe Cayman will destroy the 580 at all I think it will be a +/- 5% close battle, just dont see how AMD cannot improve more than 20% over the 5870 like many say. Hell, even if they did a full blown 1600 shader 6870 chip it would be a good deal faster than the 5870 (considering 6870 is 5% slower with 1120 only). Yet they are still doing more than that so I surely bet on a very competitive Christmas.

[Picao84]

Sure I whole heartedly agree about the Christmas shopping time ans the need for competition. Unfortunately AMD will be too late for good price adjustments to take place before the holiday and I am dissapointed with the three week delay. What Im simply stating is that a lot of people automatically assume that a 3 week delay means AMD is on the losing end and wont deliver a competitor to the 580.

That doesnt mean I believe Cayman will destroy the 580 at all I think it will be a +/- 5% close battle, just dont see how AMD cannot improve more than 20% over the 5870 like many say. Hell, even if they did a full blown 1600 shader 6870 chip it would be a good deal faster than the 5870 (considering 6870 is 5% slower with 1120 only). Yet they are still doing more than that so I surely bet on a very competitive Christmas.

I for one, wouldnt belive that a "1600 shaders barts" would beat GTX580. Shaders are not everything, and its been speculated since its launch that HD5870 shaders might be underused.. in essence it has 1600, but many stay idle most of the time. Teraflops is not everything, remember that GTX580 barely has a peak performance of 1,5 TF, while beating 5870 with its 2,72 TF...
In essence, what I mean is: dont be so quick to qualify a 6870 as big step on efficiency upon Cypress.. Just as GTX580 is Fermi done right, 6870 might be Cypress done right

[highoctane]

What we are seeing in truth is a tiny few setbacks for AMD being blown way out of proportions by Nvidia fans hoping and preying that their beloved company doesnt end up behind for another generation.

Its really annoying when people care so much about a brand they stop using their brains.

Isn't this the pot calling the kettle black syndrome, on the other side of the table they're doing every bit the same with nit picking every detail of Nvidia.

Not long ago its was all about Nvidia's paper launch 580, no availability, they can't do it, can't increase efficiency, smash, bash, etc.

[generics_user]

I for one, wouldnt belive that a "1600 shaders barts" would beat GTX580. Shaders are not everything, and its been speculated since its launch that HD5870 shaders might be underused.. in essence it has 1600, but many stay idle most of the time. Teraflops is not everything, remember that GTX580 barely has a peak performance of 1,5 TF, while beating 5870 with its 2,72 TF...
In essence, what I mean is: dont be so quick to qualify a 6870 as big step on efficiency upon Cypress.. Just as GTX580 is Fermi done right, 6870 might be Cypress done right

5870 was underused because of the weak front end, 6870 has a much more powerful frontend and is able to use the shaders more efficiently (improved performance/shader in 6870 comes from the new frontend), it was already proven that while 5870 is essentially 5770 x2 it seems to be much more inefficient in terms of perf/shader

i think that 6970 will be on par with GTX580 because amd made more improvements to it than the new frontend used in the 6870, it is supposed to have more shaders than 5870 and there would be no room for a 6950 if 6970 is on par with gtx480....

[Manicdan]

I for one, wouldnt belive that a "1600 shaders barts" would beat GTX580. Shaders are not everything, and its been speculated since its launch that HD5870 shaders might be underused.. in essence it has 1600, but many stay idle most of the time. Teraflops is not everything, remember that GTX580 barely has a peak performance of 1,5 TF, while beating 5870 with its 2,72 TF...
In essence, what I mean is: dont be so quick to qualify a 6870 as big step on efficiency upon Cypress.. Just as GTX580 is Fermi done right, 6870 might be Cypress done right

i think it could
look at the 6850xfire reviews, its 20% more SPs (1920 vs 1600), but the clock speed of the 6870 is also 16% higher. so in theory a 1600ish SP barts core at 900mhz would be near identical to a 6580xifre

and the reviews are showing that being super close to a 5970 and beating a 480, so i feel confident that 1600SP @ 900mhz using the same improvements of Barts, would prove to be very deadly. (and this is all based on xfire scaling efficiency, which might be alot lower than a single larger gpu with better memory speeds)

[duploxxx]

Isn't this the pot calling the kettle black syndrome, on the other side of the table they're doing every bit the same with nit picking every detail of Nvidia.

Not long ago its was all about Nvidia's paper launch 580, no availability, they can't do it, can't increase efficiency, smash, bash, etc.

well paper launch or not they are very hard to find and only few brands whit price premium in EURO land, so I would call it a wide availability either....

the funny thing in these forums is the idea that any body would just jump for a 500$ gpu card because they need it, get real, minimal volume, those who buy now will never look at any decent review/price/performance compare or whatever... all for christmas yeah right not at that price tag.

[Picao84]

i think it could
look at the 6850xfire reviews, its 20% more SPs (1920 vs 1600), but the clock speed of the 6870 is also 16% higher. so in theory a 1600ish SP barts core at 900mhz would be near identical to a 6580xifre

and the reviews are showing that being super close to a 5970 and beating a 480, so i feel confident that 1600SP @ 900mhz using the same improvements of Barts, would prove to be very deadly. (and this is all based on xfire scaling efficiency, which might be alot lower than a single larger gpu with better memory speeds)

In Crossfire you have a duplication of ALL hardware parts, something that probably would not happen with a single GPU with 1600 shaders
Thats my point actually, that generic_user already talked about... What matter is not that shaders, but all the chip organization.

[Manicdan]

In Crossfire you have a duplication of ALL hardware parts, something that probably would not happen with a single GPU with 1600 shaders
Thats my point actually, that generic_user already talked about... What matter is not that shaders, but all the chip organization.

right and in xfire you also get crappy bandwidth cause it has to duplicate everything.

so imagine they give it some serious 7ghz ddr5 instead of that sub 5ghz crap.

gains and losses can be seen by redoing a larger chip, its not always negative. but it shows CLEARLY that the potential for serious gains of the current arch exist.

[saaya]

6870 might be Cypress done right

thats why its slower... ^^
i think you mean 6970?

and about things being blown out of proportion... christmas shopping is a LOT...
if you miss the christmas season thats not good...

[Sesto Sento]

That doesnt mean I believe Cayman will destroy the 580 at all I think it will be a +/- 5% close battle, just dont see how AMD cannot improve more than 20% over the 5870 like many say. Hell, even if they did a full blown 1600 shader 6870 chip it would be a good deal faster than the 5870 (considering 6870 is 5% slower with 1120 only). Yet they are still doing more than that so I surely bet on a very competitive Christmas.

I've defenitely to agree with this statement.

If we take into account that a 6870 gives you roughly 73% of GTX580 performance...

Source: http://www.techpowerup.com/reviews/N...TX_580/27.html

1120 shaders - 73% perf. of a GTX580
1600 shaders - 104.23% perf. of a GTX580

As more shader don't improve performance linearly, it should give you 95% to 100% performance of a GTX580....

So, now let's look to the past...

One 5870 gives you roughly 63% more performance than a 5770...

Source: http://www.techpowerup.com/reviews/HIS/HD_5770/30.html

It's expected that a High end part (6970) to give at least 50% more performance than the midrange part (6870)...

This is all my theory but i think it's pretty plausible... IMHO ATI 6970 will be in worst case scenario equal to GTX580, and in best case 10% faster...

[Alien Grey]

NVIDIA intends to use the GF110 continue to build a new core product - GeForce GTX 570, this new graphics card will have 480 stream processors, and the GTX 480 as memory allocation estimates or 384-bit/1.5GB GDDR5, This performance and the GTX 480 can be flat.

Source: http://lensfire.blogspot.com/2010/11...80-can-be.html


Another confirmation that the GTX 570 is going to have 1.5GB GDDR 5.

If it's really true than it's going to be a nice card.

[eric66]

Source: http://lensfire.blogspot.com/2010/11...80-can-be.html


Another confirmation that the GTX 570 is going to have 1.5GB GDDR 5.

If it's really true than it's going to be a nice card.

yep 570 seems to be new champ to get unless of course cayman somehow manages to surprise me which in this case is very doubtful

[SKYMTL]

if you miss the christmas season thats not good...

AMD has announced they will be ready on December 13th. To retailers and even most consumers, that's past the "Christmas shopping season" so they're already missing it.

9 working days (not including Christmas Eve) does NOT give retailers enough lead time to guarantee shipping times to and from distribution at what is the busiest time of the year.

Typically, retailers in the garment, electronics, and other comodity markets consider the second week of December to be the cut-off for products coming to market for the Christmas season.

Anything after that would be considered fresh meat for the typical "January Sales" timeframe. That IS NOT something AMD wants to get into with a brand new product(s) because it will instantly devalue it when compared to the marked-down existing products.

We have seen this in the past from other launches from AMD and NVIDIA. The option left open is to paper launch the HD 6900 series in mid-December and begin volume shipments in mid to late January in order to avoid the typical market trends that arise right before and after Christmas.