NVIDIA GeForce GTX 480M is out

That's odd. I know ATI is a Canadian company, but I'm surprised that they went all out with the naming scheme. Vancouver, Whistler, Robson, as in Robson Street in Vancouver. Granville Island, Pretty funny.

 

The way things have been leaking out so far, GPUs with the codename Southern Island were described as a half-gen card while we wait for TSMC and GF to get their 28nm fab ready for a separate GPU family codenamed Northern Island...or Hecatonchires.


If the article is to be believed it would put a much simpler order to the codenames:

Northern Islands are the mobile versions of the half-gen 40nm GPUs.
Southern Islands are the desktop versions of the half-gen 40nm GPUs.
Hecatonchires is the full next-gen 28nm GPU architecture.

 

Indeed but this is the side of Asus i don't like.. using lousy 3D NVDIA tech.. wouldn't mind if it was ATI but seriously , which nerd is going to take 3D glasses out on the go and play games?

 

INTRODUCING: the first 480 SLI notebook, also the first laptop ever to feature a chlorofluorocarbon compressor to cool it! WE CALL IT: THE MINIFRIDGE.

 

I'm pretty sure Nvidia came out and said that it wasnt as powerful as the 5870...

 

To all of these comments about the GTX 480M burning a hole in your table, here is a nice response by Neil from the Kobalt forums

When you consider the high build quality and excellent cooling of the Sager / Clevo laptops, we really shouldn't have anything to worry about. Even back when the cooling systems in some of the 15'' only had a single fan, I was still seeing great temps. Yes the GPU's were not as powerful as the ones used in the Dual fan design but they were still powerful and over heating in other models simply because their cooling design was not as efficient.

 

True, how hot the card is is really not important as long as the cooling system can handle it, even a 260M can overheat in a G51, or a 330M overheat in a MBP if the cooling system is insufficient.

 

A lot of people have been exaggerating the heat this card produces, which I think has gotten rather old already. I doubt the manufacturers and resellers will be stupid enough to use this card in a laptop that can't handle it.

Nonetheless, the heat output does serve to limit overclockability, which this card could possibly offer quite a lot of. I wonder if any laptop will be able to handle a GTX 480M clocked at the level of the GTX 465. Since the GTX 465 has a 200W TDP, I doubt it, although the 200W figure for the GTX 465 makes the 100W figure for the GTX 480M seem somewhat miraculous.

 

Agreed... I don't think anyone here was dumb enough to assume the temps will be unmanagable, of course they would never be released if they weren't. The larger issue is how limited is overclockablility of these cards? We can assume without overclocking the 480 performance ratio over the 5870 will be pretty marginal. WITH overclocking this card could destroy the 5870, but the larger question is can any chassis handle the extra heat with the overclocking?

 

Not quite marginal; probably ~20% in DirectX 10 games, and as much as 40% in tessellation-heavy DirectX 11 games (at least until ATI starts using some of their SPs to help with tessellation).

Nonetheless, if the GTX 480M's overclockability is highly limited by heat, then overclocking will let the 5870 catch up a lot of this advantage.

 

They said they made a mistake on GFLOP performance, and corrected themselves. The first figure they gave was 897 GFLOPS (I think), then they retracted that and said they made a calculation error and it was actually 598 GFLOPS. A single Mobility Radeon 5870 is approximately 1.2 TFLOPS performance.

The thing about this is you can't really use this statistic to judge which one will be more powerful in game, it just doesn't work that way. I'm willing to bet the 480M is 15 to 20% more powerful "in game" than the MR5870.

 

More like 100% performance difference in tessellation heavy benchmarks.

But then, it's not a fair comparison since ATI didn't use their shaders for tessellation yet. However, in dx11 real games though, I would still expect performance difference to be around 40-50%.

 

I said tesselation-heavy games, not benchmarks. Nvidia's own marketing data shows figures between 20% and 50% for DirectX 11 games, and 50% in Unigine - not 100%.

When ATI uses their shaders for tesselation, this should go down, though I can't say how much.

 

another couple of post and nvidia thread and i will discover that nvidia 480m is better than nvidia 465 for desktop! lol

 

No way that will happen, considering the 480M is the same hardware as the 465 but with much lower clock speeds.

 

lol ......

 

i agree.. notebookcheck.com had a review on GTX480M and basically said it was a lower GTX465.. howver i'm not impressed by the benchmarks.. it barely beats a G73 with a desktop CPU.. in real life , i see it not being more than 5% better at most...

 

If cooling is that much of a problem, why not cram the GPU somwhere in the middle of the motherboard, and have a heatsink run horizontally through the notebook, that way air can be bought in from the relatively thick side of a notebook and sent out the other end; or do fans not work like that?

Either way, I'm in the boat of "Who needs this much portable performance?" Nice to have but realistically, what are you going to do with that much somewhere you couldn't have a desktop... but they wouldn't bother making such notebooks if they knew no one at all would buy them.

 

The last performance gains always been very small, and scale exponential in price... this is what enthusiast have to suffer, its the way its been for a long time, nothing new here, same with intel xtreme line, still if you want the fastest gpu then Nvidia will give you that, just gotta pay. ATI always have offer much better value/performance, but for some people thats not enough.

 

How does that explain the 285 GTX being higher in price then the 5870 and yet have lesser performance? This is not a GPU wide thing, this is an Nvidia pricing thing.

Notebookcheck: Test NVIDIA GeForce GTX 480M "Fermi" im Clevo D901F

In case anyone hasnt seen what benchmarks we are referring to.

 

5% increase in performance. A lot of those benchmarks showed them practically equal in a lot of games. Are those benchmarks legit? That is disappointing, and certainly not work the extra $600 price jump, and the extra heat and power consumption.

 

Notebookcheck.com Clevo D900F GTX480M with Intel i7 940 Desktop Benchmarks
- I will be very surprised if the G73jw uses the GTX 480M. Personally I think it's pretty much guaranteed. I just don't see Asus supplying a 300 watt powerpack and using the extensive, heatsink encompassing the entire GPU design. And I don't think Asus would sell such an expensive notebook, minimum $2,500.
- So I suspect with a weaker GPU, for those waiting for Nvidia, I scratch my head wondering why now...

Using my normal playing clocks. My gaming overclock is right now 800/1100, all the time.

The 3DMark06 no comparison since it uses a desktop CPU.

But the Vantage which is a far better benchmark for DX10 GPU, 9277 vs my 9026. That's about 3% increase in performance.

THREE PERCENT.

Bad Company 2
Notebookcheck GTX 480M high, HBAO on, 4xAA, 8xAF 37 FPS
My G73 HD5870M, HBAO on, 1xAA, 16xAF 45-60 FPS

Dirt 2, Nvidia wins by 6%
Notebookcheck GTX 480M gets 46.8 in DX11 Ultra


CRYSIS ODDITY??? HD5870M wins by 25%
Notebookcheck GTX 480M Very High, 0xAA, 0xAF 16.1 fps


Unigine 2.0/2.1, Nvidia wins by 35%%

Notebookcheck GTX 480M (Tessellation (normal) | High | all on | vsync, AA and AF respectively off) 33.7

 

lol, time to break out my yardstick . I don't know if I'm suprised it's already out or not.

Q

 

Its probably the don't want to melt the hands of its consumers. The GPU and CPU are both hot hardware. I figure they cram it near the back, it helps vent the air and minimize the heat towards the user.

 

From the notebookcheck benches, all I can think is.... how disappointing.

The (english) article: Notebookcheck: Review NVIDIA GeForce GTX 480M "Fermi" in the Clevo D901F

Some of the benches are just sad:




Those are your typical games from 2010, and metro is sort of what 2011 and beyond holds


Then there is the temp they recorded, while it isn't really that high, the card also doesn't look like it is stressed (note the lack of mem usage), and it doesn't say what they used to load the card with. Keep in mind this is in the biggest lappy chassis we've seen yet, with 3 fans:


There are even a couple benches in the link where the 5870 actually beats the 480. I found this quote in the article interesting too, because this is what the future of the cards hold (next year) once you get past the initial pissing contests.

Welp, looks like ati has my vote, and soon my cash too.

 

Nvidia i love you great work:! nice surprise discover is even worse than ati
i don't know if you can compare i7 720 vs i7 940

 

This is so weak and disappointing. I wonder if ATI will have a new card come out soon. Be nice to get desktop 5850 performance in my w870cu soon...

 

It was about expected since they aren't much different than their desktop counterparts. Certain games will always favor nVidia, while others ATI (typically 7/10 to nVidia, 3/10 to ATI).

 

I don't know if you can say that a 10% average increase in some games for over double the cost of a 5870 is 'to be expected', but I digress...

I think what is catching people off guard is that the 480 sucks at what it was touting as its' strong point. We already knew it had insane power requirements and that it would be expensive, and we already "knew" that it would beat a 5870 in most standard cases. But toss any DX11 or tesselation into the mix, and it is not even close to meeting expectations - a 35% increase is a far cry from 500% / 5x that their marketing was spinning. If metro and batman are any indication, you won't even be able to use their precious physx if you actually want to play the game.

 

The picture gets worse now that a NBR user has put up benches with a desktop CPU paired with a single 5870... http://forum.notebookreview.com/sag...rks-d900f-running-mobility-radeon-5870-a.html

At this point I don't even feel mad at Nvidia anymore for its ridiculous price/performance ripoff, I feel almost sorry for them. Its like a man thats been taking repeated blows and is down on the ground and took one last desperation knockout blow but missed. Someone needs to pick him up, dust off his shoulder and get him back into the fight against ATI so ATI doesnt become complacent eventually too.

 

Since when did anyone actually trust the accuracy of anything posted in Notebookcheck?

Also, here is another opinion from a trusted reseller

Edit:

More benchmarks here

 

Let me get this straight, you're asking us to trust a reseller who posts vague benchmarks with no real numbers who has several up-coming notebooks selling only the 480 Variant of the Nvidia cards over an independant reviewer?

 

This is actually not what I wanted to see, I was hoping for Nvidia to atleast slightly outperform ATI and force people to make a "difficult" decision. Honestly, now there is no decision to be made...none. I sure hope ATI takes this HUGE opportunity to hit Nvidia while they are down. It would be a darn shame if ATI sits back; and I am also hoping this is a huge wake-up call to Nvidia, time to turn it around.

 

No, I am saying I take what Notebookcheck posted with a pinch of salt.

More results here

 

Besides what I've already seen, only will wait to see user benchmarks, as there are definitely a few NBR members who will buy the Clevo D900F.

Also Hot Hardware are definitely biased, paid off by Nvidia. Only a reviewer on Nvidia's payroll would benchmark the GTX 480M the way they did. Head to head best possible? They should have used a Alienware with 920 extreme + HD5870M or Clevo with same specs but with HD5870M. Next almost all their benchmarks utilize PhysX.
- So in the end, sorry, lame benchmark comparisons.
- Reading that review, my impression is someone in Nvidia marketing department wrote it, and sent it o Hot Hardware to publish.
- For example 60 minutes battery life on G73jh? I get over 2 hours battery life. Where did the other 60 minutes?

 

$945 USD.

GG Nvidia. I'm wearing red next time.

 

Maybe it's too early to give up on the green goblin there, Kevin? Don't they always post better drivers a few months after?

 

If we're going to think about it like that, the 5870s might be getting that driver updates to use the shaders for tessellation. I also can't wait to see how much better the crossfired ones in the M17x get with the next release.

I was still considering the 480m until I just saw what Kevin posted lol.

I'm not paying a thousand dollars per card.

 

And are we not to take the fact the Hothardware 480 benches are done with desktop CPU vs a 5870 with a laptop CPU with a pinch of salt?

The benches seem to just reinforce the notebookcheck numbers, look at a fair comparison of your link against this link http://forum.notebookreview.com/sag...rks-d900f-running-mobility-radeon-5870-a.html as I posted before.

The gap when the 5870 is paired with a desktop CPU is completely closed in non-tesselated games (in fact seems to favor ATI in some games), and the gap is narrowed to about 20% in tesselated games. Also take into account ATI is developing drivers to utilize its shaders more effectively for tesselation and we may see the tesselation gap close as well. Anyone who thinks the 480 is worth $600 more is absolutely insane and is literally throwing their money away.

 

Claiming a comparison of $2,400 desktop vs a $1,500 LAPtop to be close comparison = Stupidity or shameless Nvidia marketing. Not sure which.

Also would like to emphasize. 5150Joker has proven that you can purchase a Alienware Cross-Fire HD5870 With the RGB LED Display for less than a Clevo D900F with GTX 480M.

 

lol, I don't know anything about hothardware to be honest, just copied that link from some one else to pass on as more info and hadn't read all of it yet. However, I won't jump to any real conclusions until I see a benchmark of the 480M done from an owner on this forum (if there are any).

I still find it very hard to believe that Nvidia would spend all of that time to release a new GPU only for it to still be crushed by the existing one.

 

Well it isn't crushed. It is definitely faster and more powerful than the previous GTX 285M.

The tessellation is also better than the HD5870M. Just not 5X better, just 35% better.

In game we have seen that games with lots of texture, the HD5870M WINS! HD5870M has higher texture fillrate than the GTX 480M does.

The one thing we do know for sure is that overall GTX 480M is better, but at the expense of no longer being a mobile GPU. I do not consider the GTX 480M a mobile GPU.

 

You sure can. But that doesn't mean the 5870M is going to do well on highly tesselated games. Besides, it's been out for months. On the other hand, it's almost a custom for nvidia to get a better driver out this way.

 

The price is what kills me. That said, I will end up with a W880CU, because ATI's next card will be >75W, thus requiring the new chassis.

If the GTX 480M can come within 10% of GTX 465 clocks, I'll be back on board. Via the barebones route, you can build a W880CU w/ ES 920XM for $2300, so it really isn't a bad deal.

I'm not buying until Q4 anyway, so let's see what the red team has on deck by then. 1440 shaders or bust.

 

Let's not get nuts here, nut While the 5870 does indeed beat the 480 in a few games, it doesn't crush it - on the flipside, the 480 certainly doesn't crush the 5870 in anything either. What I find hard to believe is that they are seriously selling the card for almost $1000, over $600 more than a 5870 - without having at least 60% better performance. Even $100 for a 10% increase is unacceptable by me personally, but $600 for 10% is unacceptable to anyone who doesn't live off some trust fund. Like ziddy said, you can just go purchase a damn CF/SLI kitted out setup for less cash and have it be more powerful while using less power and putting out less heat. Now THAT is nuts.

Like I said before though, I really do believe that nvidia is (going to be) losing money on the initial fermi lineup, and to cover costs will end up rebadging the fermi core over the next few years (see: G92). Meanwhile ATI will be producing chips with brand new architectures for (hopefully) a better price/performance point.

 

That is really an expensive card. But IMO nvidia is banking on that GPU computing thing on that card, and is hoping somebody will make advances so that certain things like video encoding can run faster on the GPU. Certainly not worth that much money, but I can see why they'd put it out for more than ATI's top of the line.

On the ATI side, while they do have better architecture, they've proven years after years that they're a bit lacking on the software side, i.e. drivers and such. Hopefully that'll change, but I'll believe it when I see it.

 

yes this is the main point, far better choose crossfire or sli solution. less temperature and watt also less money

 

There won't be less temperature or wattage with double the GPU, but yes, it would be faster and cheaper.

 

Wrong. ATi Stream Processors have double the GPGPU power the GTX 480M does.

And the HD5870M probably has around 2-3X more DirectCompute power than GTX 480M does also.

 

I don't know if this has been posted already or not. I found it today so I decided to share it with you guys

NVIDIA GeForce GTX 480M, Fastest Notebook GPU Yet - HotHardware