Trimlock FX5 series were bad architecturally. I honestly don't think Fermi is (in potential anyway). Maybe 3 billion trannies (no not that kind) were just a bit too many on a leaky 40nm process. Bring a few revisions and a shrink and lets see what happens. We're not here for the short term are we.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
NVIDIA GeForce GTX 480 Final Specs & Pricing Revealed
- Thread starter celeron300
- Start date
Real life and CSI aren't the same.....
Was hoping to get some details out of it. Guess it's back to two more days
Luckily I have a friend who works at CSI and he was able to use their super computer to extract the following text.
Kinda confusing but here it goes...
***WARNING***
IT IS NOT SAFE TO USE FERMI VGA CARDS AS HEATING OR COOKING APPLIANCE.
- Fermi cards require 1.21 gigawatts for idle operations. This may require you to call your electricity provider or Doc Brown.
- Using the card at 100% loads for extended periods of time may cause the your case to melt, the complete destruction of the earth's ozone layer, and/or accelerate global warming.
- CUDA core size actually varies between 480 and 512 depending on your luck.
Best used with the new AzzBeztos(tm) computer cases lineup.
*******************
Just kidding guys back to waiting till Friday
Kinda confusing but here it goes...
***WARNING***
IT IS NOT SAFE TO USE FERMI VGA CARDS AS HEATING OR COOKING APPLIANCE.
- Fermi cards require 1.21 gigawatts for idle operations. This may require you to call your electricity provider or Doc Brown.
- Using the card at 100% loads for extended periods of time may cause the your case to melt, the complete destruction of the earth's ozone layer, and/or accelerate global warming.
- CUDA core size actually varies between 480 and 512 depending on your luck.
Best used with the new AzzBeztos(tm) computer cases lineup.
*******************
Just kidding guys back to waiting till Friday
Digital Viper-X-
[H]F Junkie
- Joined
- Dec 9, 2000
- Messages
- 15,116
Real life and CSI aren't the same.....
hey now, image enhancement is real =p maybe not available to most at the level that "csi" uses but a fair bit of that stuff can be done!
Ok, well go go gadget enhancement and prove me wrong
All kidding aside, yes some things can be done. But not to read the important text on the side of that box.
Digital Viper-X-
[H]F Junkie
- Joined
- Dec 9, 2000
- Messages
- 15,116
Ok, well go go gadget enhancement and prove me wrong
All kidding aside, yes some things can be done. But not to read the important text on the side of that box.
you'd be surprised
! I could even think of a few ways to get pretty close, ,with a dictionary to compare with, it wouldn't be too hard to get something out of that picture.You make a good point, but hearing Nvidia not continue with the respin blows my mind! But its hard to argue with the 3 billion transistor being the main factor...
Must mean that nvidia's next arch is on time or maybe even early.....or maybe there never was a need for a respin and all is fine and dandy in fermi land
It could also be that the problems they are having won't be fixed by a respin.
mercyground
Limp Gawd
- Joined
- Sep 28, 2002
- Messages
- 158
A complete redesign.
oh and now the date slips again. April 12th now instead of the 6th for mass availablity.
Are we sure that the Phantom dudes arent working for NV PR now?
oh and now the date slips again. April 12th now instead of the 6th for mass availablity.
Are we sure that the Phantom dudes arent working for NV PR now?
.....or maybe there never was a need for a respin and all is fine and dandy in fermi land
i think you are catching on
Exactly what kind of problem could they have that wouldn't be helped by a respin?
Actual architectural or design problems. A respin couldn't fix the 2900XTs broken AA, for example.
i think you are catching on
You're delusional if you think all is fine and dandy in Fermi land. Even if the card itself is solid when its released, its waaaaay late. Fine and dandy would have meant it was on time, which it isn't.
Digital Viper-X-
[H]F Junkie
- Joined
- Dec 9, 2000
- Messages
- 15,116
Late ? According to whose timetable? Yours?
It is ready when it is done .. Tomorrow
late according to the original it will ship with Windows 7, and late according to the fact that their competition owns 100% of the DX11 market
So they missed their target clocks, missed their target core count, according to even the more pesimistic benchmarks it is even with the 5870 in games, and it's architecturally broken? If it's architectually broken and preforming at least even with the 5870 when it missed it's targets, what the hell does that say about the 5870?
udneekgnim
Limp Gawd
- Joined
- Mar 10, 2010
- Messages
- 256
Fermi has a higher die size and TDP than the 5870
Digital Viper-X-
[H]F Junkie
- Joined
- Dec 9, 2000
- Messages
- 15,116
It says that ATI built a smaller, more efficient chip then nVidia? lol
Fermi is nearly DOUBLE the size of a single 5870 chip, which also means it's nearly double the price per chip(probably more when you take yields into account with the larger chips getting hit harder)
since the 3800/4800 AMD changed their game plan, you do remember that part right? They would no longer build big single 300W gpus, instead they would build smaller easier to manufacture lower power chips, and just use multiple gpus to compete in the high end, while they concentrate on the mainstream
Cannon_Fodder
Limp Gawd
- Joined
- Jul 21, 2007
- Messages
- 249
Rationalization! The solution to all of life's worries!
I didn't say that at all. I'm just saying that a respin can't work magic and it can't fix everything. If Nvidia isn't doing a respin, then it strikes me as pretty clear that they don't think a respin will fix the problems they are having.
late according to the original it will ship with Windows 7, and late according to the fact that their competition owns 100% of the DX11 market
.. and it launches tomorrow ...
. . . Nvidia has not lost much .. it has mostly affected them and their partners
The only way we were hurt by Fermi's lateness is that there was no competition in DX11
i really like DX11's tessellation features and the few new games that have it look really good
The other way to look at it is that Nvidia thinks that they have a very good chip that does not require a respin
flippin_waffles
Limp Gawd
- Joined
- Jan 9, 2006
- Messages
- 178
I'm curious what the history is on past launch dates. Isn't it historically the case that products launched on Friday are typically unimpressive, and so it misses the media cycle for 2 days? From what I remember, companies tend to like their products to have as much press coverage as possible if it is going to be well recieved.
Digital Viper-X-
[H]F Junkie
- Joined
- Dec 9, 2000
- Messages
- 15,116
.. and it launches tomorrow ...
. . . Nvidia has not lost much .. it has mostly affected them and their partners
The only way we were hurt by Fermi's lateness is that there was no competition in DX11
i really like DX11's tessellation features and the few new games that have it look really good
The other way to look at it is that Nvidia thinks that they have a very good chip that does not require a respin
Nvidia lost ALOT of potential Income, which went to their competitor instead.
Keyword = Potental
Subjective = a LOT
During that 6 month period Nvidia did not lose overall market share. Amazingly they just about sold out their entire GT 200 line in preparation for their new product.
Digital Viper-X-
[H]F Junkie
- Joined
- Dec 9, 2000
- Messages
- 15,116
obviously you think like a consumer and not an investor, investors don't like lost "potential" ♦
.. and it launches tomorrow ...
. . . Nvidia has not lost much .. it has mostly affected them and their partners
The only way we were hurt by Fermi's lateness is that there was no competition in DX11
i really like DX11's tessellation features and the few new games that have it look really good
I know, I've been enjoying tessellation for 6 months
Only if your a fanboy. Even the greatest GPUs like the 9700Pro had refreshes. Fermi not getting a refresh doesn't mean its THAT GOOD. It really means its THAT LATE.
I know, I've been enjoying tessellation for 6 months
Only if your a fanboy. Even the greatest GPUs like the 9700Pro had refreshes. Fermi not getting a refresh doesn't mean its THAT GOOD. It really means its THAT LATE.
i believe Nvidia is a fan of itself.
... Me? i have a Diamond HD 5870 and a PowerColor ECS+ is otw for CrossFire
Let's take that "no refreshes" and run with it .. if it is true because it is *late*, what do you think Nvidia will do?
Use some business logic.
Digital Viper-X-
[H]F Junkie
- Joined
- Dec 9, 2000
- Messages
- 15,116
i believe Nvidia is a fan of itself.
... Me? i have a Diamond HD 5870 and a PowerColor ECS+ is otw for CrossFire
Let's take that "no refreshes" and run with it .. if it is true because it is *late*, what do you think Nvidia will do?
Use some business logic.
funny, now you want to use business logic, but when it comes to nVidia losing out on 7 months of dx11 card revenue, you want to ignore it
and btw, it's called cutting your loses, if Nv knows that Fermi / GF100 cannot compete on a price/perf ratio, and a Fermi refresh would have to compete with Cypress Refresh. So it's an uphill battle now.
nothing ? 5870 is old arch and is based on so called floppy r600 arch so there is nothing good about fermi matching 5870 performance lol. i don't know how much they missed clocks but i doubt its more than 50 mhz so that makes fermi pretty much another flop 'much bigger core draws much more power and probably is much hotter and is 7 month late
'i believe Nvidia is a fan of itself.
... Me? i have a Diamond HD 5870 and a PowerColor ECS+ is otw for CrossFire
Let's take that "no refreshes" and run with it .. if it is true because it is *late*, what do you think Nvidia will do?
Use some business logic.
I think Nvidia decided to cut its losses and simply move on to its next generation (which has been in the works for at least 2 years now - just because Fermi was delayed doesn't necessarily mean that the successor is delayed). I think that they have decided that the potential gain of a respin and refresh will simply not outweigh the costs of doing it along with the time it will take. In the 9 months that it normal takes for a refresh, ATI's 6xxx series will be close to coming out as will Nvidia's next generation.
And even with a respin the GTX 470/480 are still huge chips. A respin won't solve the problem of a massive die being expensive.
i will agree to disagree with this. What would they do to plug the holes in their DX11 line?
- they sold out their higher end GT200s
There is nothing wrong with a massive die. Nvidia is aiming for a General Purpose Processor that also does amazing graphics. That is the direction they chose and is not a "problem" for them.
You are wrongly concluding that GTX 470/480 won't be competitive in price/performance. Clearly Nvidia is planning to take back the (single GPU) performance crown and launch their new architecture.
To Nvidia, Fermi is a success and their foundation for their future. They picked that direction with G80.
and they do. in the tests i've seen the 480 handily beats the 5870 & is in turn beaten by the 5970
i haven't personally seen any data regarding the 470s tho
and they do. in the tests i've seen the 480 handily beats the 5870 & is in turn beaten by the 5970
i haven't personally seen any data regarding the 470s tho
Me neither on the GTX 470
OK, so what did i say that was wrong?
If this list is correct and you are right, then Nvidia takes back the (single GPU) performance crown and has managed to do a decent price to performance ratio for their GPUs bringing *choice* back to the market.
- HD 5970, $600
- GTX 480, $500
- HD 5870, $400
- GTX 470, $350
- HD 5850, $330
isn't the above what we expect?
Me neither on the GTX 470
OK, so what did i say that was wrong?
If this list is correct and you are right, then Nvidia takes back the (single GPU) performance crown and has managed to do a decent price to performance ratio for their GPUs bringing *choice* back to the market.
- HD 5970, $600
- GTX 480, $500
- HD 5870, $400
- GTX 470, $350
- HD 5850, $330
isn't the above what we expect?
you didn't say anything wrong, i was agreeing with you
and yes, that list looks to be correct and is the exact reason AMD has no reason to lower their prices IMO
coolyou didn't say anything wrong, i was agreeing with you
and yes, that list looks to be correct and is the exact reason AMD has no reason to lower their prices IMO
Yes, agreed. There will be no price war this time. Both companies are going to improve their bottom line and leave their partners to do the discounts and the competing for the consumer's dollars.
AMD looks to be bringing (even highly) overclocked 5870s into the mix right away (i am getting PowerColor ECS+ HD 5870 soon to bench and to add to my current Diamond card for 5870 CrossFire). Soon i think Nvidia will do the same but a bit more conservatively (as is their history).
i tend to think that the new GeForce will get faster as Nvidia matures their drivers on this new architecture and the performance gap may widen in their favor. AMD did an amazing job on their drivers with 5000 series but large performance increases have probably peaked with 10-3a (imo)..
Commander Suzdal
2[H]4U
- Joined
- Jul 29, 2004
- Messages
- 2,134
Fact #1 is irrelevant to the performance of the card.
#2 is not a fact. Is basic logic really this hard to grasp? You even include the words "from what has been leaked" right there in the sentence. Leaks are not facts. Rumors are not facts. Tested, repeatable results from reputable sources are facts.
#3 is a retread of #1, therefore equally irrelevant (and now redundant).
See you on Friday.
i will agree to disagree with this. What would they do to plug the holes in their DX11 line?
- they sold out their higher end GT200s
Uh, what? A refresh/respin would give us a GTX 475 and 485, not a GTX 460 and lower. Nvidia needs something different altogether for the midrange and lower parts.
There isn't anything "wrong" with it, no, but it is a problem - its expensive and hard to manufacturer. If it wasn't a problem it wouldn't be late, now would it?
Will Fermi take back the single GPU performance crown? Likely. Will that make it competitive in price/performance? Not at all. Those are two *very* different things. For example, the GTX 280 claimed the single GPU crown - but it sucked in price/performance. Fermi is not shaping up to be price/performance competitive.
To Nvidia Fermi is not a success. They missed not one but TWO major market opportunities. They missed the Win7/DX11 launch AND they missed the holiday season whereas their competitor nailed both. Fermi is absolutely not a success. Not necessarily a failure (that remains to be seen), but it certainly isn't a success.
Me neither on the GTX 470
OK, so what did i say that was wrong?
If this list is correct and you are right, then Nvidia takes back the (single GPU) performance crown and has managed to do a decent price to performance ratio for their GPUs bringing *choice* back to the market.
- HD 5970, $600
- GTX 480, $500
- HD 5870, $400
- GTX 470, $350
- HD 5850, $330
isn't the above what we expect?
If the $450 price tag on the GTX470 put online is to be believed, then Nvidia is not opting for hte price to performance bracket at all in that case