Pic of GTX 480M + 3DM06 Test

I do as well, but the cpu (whatever it may be) in the G73 is really limiting it. Other than that, the only real lead the 480m has in in tessellation heavy benches, which may not last if ATi does in fact use it's shaders in addition to the separate tessellation processor.

 

Look at the Vantage (GPU only) score. 480M should be much faster but that stuupid 980X is dragging it down! :laugh:

 

Expected scores, small difference in vantage, but 30-40% difference in tessellation heavy games (100-150% here mostly because of the CPU here and ATI not using shader for tessellation).

 

i was really expecting something earth shattering, i suspect not many ppl will be ditching their 5870s for 480.

 

I'm sure in time we will have some more numbers that may help us more on seeing if there is a true benefit of the 480M over the 5870, it's too early wouldn't you say?

 

Well, it's still 30-40% difference in tessellation heavy games though, and nvidia cards usually scales better in SLI than ATI in XF. There's also the CUDA and PhysX. It might just be worth the price, considering that 5870MR didn't have a large margin over the 285M as well, though the price was great.

 

I wouldn't ditch a 5870 for it, but it'll be a huge jump over my current 260M.

 

Yes, not much point if you already have the 5870MR, but if you are buying a new notebook, it might just be worth it if you like an nvidia GPU.

 

i'd like to see how big the power supply is for this thing.

 

Yeah, and I'm going barebones + an ES CPU, which actually comes out cheaper than a factory machine with 5870.

 

donald,

wicked laptop it is. How does it compare: GTX 285m to GTX 480m?

how does GDDR5 help the 480m over the 285m?

Jason

 

Sager has not released any benchmarks comparing the 285M and the 480M.

However we know that the 285M is only marginally slower than the ATI HD 5870 in most applications, so the comparison will be similar.

The 256bit 285M w/1GB GDDR3 has equal bandwidth to the 128bit ATI HD 5870 w/1GB GDDR5.

So, the 256bit GTX 480M w/2GB GDDR5 has not only more memory, but significantly more bandwidth.

 

With the added heat and cost, the expectation for this card is tremendous.

I hope this doesn't simply become a numbers game, with little or no effect on actual performance in the real world (something like 40% increase at least).

Again, we've come to expect something like 15-20% increase for every succeeding generation of GPU cards. The GTX480m sort of breaks this formula by being as expensive and hot as it is. Whether it can actually deliver is another story. It is definitely version 1.0 of something that purports to be superior than the current offerings (meaning ATI 5870 and GTX285m).

But I'm keeping faith with NVIDIA unless they disappoint otherwise.

 

k9hydr4, did you look at the benchmarks in post #88 on page 9 of this thread?

 

If he did, he should largely discount them considering the platforms that were tested ( i know you didint run the tests) But those benchmarks are not really a great indicator of how much better it will be, however what it is is the best case scenario for a single GPU set up.

 

Exactly, from what I see the only score that counts in those benches is the GPU (only) Vantage score and keeping in mind that 4/5 runs on a system with i7-720QM+5870 will reach above 8k in that score - Nvidia doesn't have much to boast. The rest of the tests are so greatly affected by the CPU power that they are totally irrelevant.

 

Yeah, its like testing two different engines in two similar cars, but one has a super charger and nitro. None the less, the benchmarks are interesting, and id like more.

 

One thing thart interested me was they(Clevo/Sager) used a G73 instead of the W870CU to compare it to ? .... Clearly thats the target they are chasing after but i seriously dont think people interested in the G73 would be eyeballing the NP9280 considering the price differences... Night meet day...

 

It may have been all they had on hand, but all things being equal a g71 is pretty similar in performance to a 870cu.

 

Yeah its more of a why would they have a G73 on hand to begin with ? ...

It would be hard to believe they did not have access to a W870CU as well but hey i could be wrong ....

Very true they are equal (G73 vs W870) but its funny they would then pick that machine to go head to head against the 480m desktop CPU machines...

Clearly they have it out for Asus and the G73 i can say that for sure ..

 

i dont think they have it out for asus, they did typical benchmarking /marking stuff, show the new card in the fastest possible system and pit vs its rival card in a average system, you know one thats not 8k. At this point i'm thinking for us single gpu folk, this will end up being close to the HD card, heck my 5870 gets a better vantage score in 3d than the nvidia did.

 

As I've wrote before in other posts, I expect the 480M to offer about 45% of the performance of the desktop 480.

 

Here is another review coming up:
NVIDIA GeForce GTX 480M Powered Notebook Sneak Peek - HotHardware

I think that comes close to my prediction. This also confirms previous speculations that the reason Nvidia went with a 100W card is because they simply could not make a 75W one which would beat the MR5870.

 

I suppose that's justifiable. if the laptop could take more powerful cards, why settle for less?

 

Cause making 100W notebook cards really limits the number of notebook models that can carry it. Not to mention that the laptop will become more expensive due to the cooling system required (for instance the W870CU has to be revised in order to take the 480M).

 

...and what will happen to these atlas-sized systems (X7200 and 880) a year from now when the cooler GPUs (from die shrinks) become available?

Also, not everyone can afford nor need the Intel extreme processors that makes these beasts run so bloody hot.

 

One year from now it will still be pwning faces everywhere, just that it might not be the fastest thing on the planet anymore. Like the D901C, it's already so many years, but a D901C with desktop CPU and 9800 GTX SLI is still performing better than most mid-high end notebooks.

 

Which makes plunking down $3000 for a base x7200 pointless from this perspective. At this size, I'm thinking 3-4 mobile GPUs. Maybe we should create a poll as to who will be actually buying these systems- gamers or power users that require desktop workstation performance.

I'm more concerned with the GPU/CPU components driving the trend towards larger notebooks-

To me it seems it should be the other way around. A year from now, we should be able to put dual GPUs in an 8760-sized (or smaller) form factor.

 

wow, looking at that review last posted, pretty damn fail at under 10k vantage with a extreme cpu....in my view at this point its only saving grace is if it over clocks like crazy.

 

I said it before in another thread...

If a notebook can be designed to cool a GPU with a 100W TDP then a GPU with a 100W TDP should be designed to make use of that notebook.

Die shrinks will make GPUs run cooler but in, and of itself, a die shrink does not increase performance. The performance increase comes from the increased clocks and cores that smaller die and cooler temps make room for...which in turn bumps the TDP back up.

A 28nm Fermi will be cooler than 40nm Fermi but if it's more than 10-20W cooler then Nvidia is wasting all the designing Clevo went through for the GTX 480M.

 

And I will say this again--- I understand sacrifices were made to accommodate this much power in notebook form factor.

Point is, what of the rest of us folks who doesn't need the Intel extreme processor (they go for $1000 alone), but are now stuck with boat-size systems (may not be with x7200, but the 880 more likely).

 

Buy a different notebook with a lower model GPU.

What you seem to be asking for is essentially that performance be brought down so TDP is more manageable in a smaller chassis for the sole reason that you want to say that you have the top model GPU.

 

Not saying that all- I'm simply not sold on the promise of GTX480m.

I do hope Clevo keep the 860/870 around.

 

Phinagle...Exactly!

The point is you can't have both the highest performance laptop in the world and make it in a small form factor with a low TDP.

This is the time and place to separate the men from the boys, the women from the girls, the power user from the casual user, and yes...maybe even the starving students from those who are gainfully employed.

If you can't afford the most powerful laptop in the world that is OK. But there is no need to bash its size, cost and TDP just because you can't afford it. You know very well that those who can will not settle for 2nd best.

So let's get on a more positive note with comments more in line with "it doesn't suit my purposes", "I simply can't afford it", "It's too big to lug around campus..for ME" instead of trying to pretend that everyone has the same needs as you do, or the same budget.

Bottom line is, those whose needs it fits and can afford it wouldn't have anything else.

 

^^^^There you go. I agree...

 

it almost seems like hes trying to sell something....

 

^^^^
Yes, he does. But these retailers are right, though- These systems will sell like hot cakes (no sarcasm intended) to it's target market. Poor mooks like me will just have to wait until the tech catches on and trickles down to the masses.

 

Well if we cut down the marketing campaign and focus on another key point:
Like with OC'ing you usually have to pay with a lot of heat and stability issues to gain those 5% extra in performance, not to mention the price of the extreme components; in the same way many look at the high end laptops. Should I pay 2x for a few more FPS in my favorite games when they still run perfectly smooth on a Core2Duo+MR3870/8800GTX? The software is still far behind in utilizing 4-cores (not even talking about 6 or 8),
consoles are far from being upgraded.
What remains? "Bragging Rights Inside" label
And we shall pay for that! Whoa!

 

GPUs are clearly the limiting factor in frame rates these days.

 

Depending on a game. Poorly coded ones like Crysis are not included All the rest (at least from my library) always stay above 30-40FPS @ max settings. (4870 CF). And between 100 and 170FPS I see no visual difference.

 

No need to bicker. There's something for every segment: desktop CPU/dual GPU (x7200), mobile CPU/dual GPU (x8100), single enthusiast GPU (880/870/860). All are as thin as can be and reasonably priced. Buy what fits your wants and your budget. There's something for everyone.

 

haha ^

Honestly there will be lower chips in the segment below the 480M. 480M is just the beginning. Much like the 260 vs the 280 days, if you need lower power, there will be something there to satisfy.

But there is a very real buying segment who wants a notebook (even a large one like the M17x) that can manage high amounts of GPU power. The 5870 single cards came out first and folks bought them. Now the Crossfire is available and folks will buy that and get more power with less battery life. That's a trade the customers want to make for the power gained.

Same will be true for the 480M 470M and 460M of which you will eventually see all, I'm betting. In this case the flagship 480M is just getting intro'd first. And if a notebook can "handle" a 100w TDP chip, or two of them, then there will be power using customers who desire the product.

Again, I suspect that a 480M with 1-1.2GB of memory will output somewhat less heat and is likely going to be how the SLI models come. If you put 2GB of memory on a 5870, you'd watch the heat soar. And please try hard to remember that 100W TDP is NOT the power consumption of the chip. It is the heat output only. It's still a concern from a cooling standpoint, but power consumption is a different measurement.

 

Well, took their chances, and in some ways it will work out - not without cost though; they probably could launch cut down versions (460M, 470M, etc) later to compete against 5870 directly - but the extra performance hopefully will be there, for those willing to pay more, and accept more heat.

In the end there will be more options, that I'd prefer much, than say another G92B rebrand.

 

I am pretty sure that a 75W 460M will be inferior in terms of performance to the MR5870 which is what I am trying to point out through all my posts in which I criticize the 480M. The only place where the 460M might shine is in tessellation intensive titles.

Still, if I ought to update now I still go with the 480M . But I prefer to wait for Sandy Bridge and maybe even 28nm mobile cards which will come out later next year.

 

Unless it's as fast as a desktop 5850 I can't get excited about it.

 

I think we got so used to the current cards (280/285/5870) that anything less (460) seem like insult.

But I guess the return to mid and top-of-the-line card configurations to differentiate product lines was inevitable. Performance gap between the two will probably be along the lines of 260/280 for single card configuration.

The 480 is not perfect, but I guess NVIDIA had to "just do it". We all hope this leads to faster, cheaper (heh), cooler (?) cards.

 

According to those leaked slides the GTX 470M is the next card down from the GTX 480M. My guess is that the 470M will use the GF104 GPU that the desktop GTX 460 uses, with 240 cores and a 192bit memory bus and assume the relationship that the GTX 260m/280m had for performance. Talk is that something about the GF104 makes it better at overclocking than the GF100 core.

The GTX 460M could use the GF104 GPU too with maybe GDDR3, lower clocks and/or some cores disabled. Otherwise it'll be the GF 106 GPU that was spotted in an ASUS G53....giving it anywhere between 96 and 240 cores (my guess is 144) on a 128 bit bus.


For the Red Team I'm going to go ahead and guess that Blackcomb will be ATI's 100w card. Well probably more like 80-95w (for the entire card). You don't put a 256 bit bus on the same 40nm fab without adding heat unless you cut SPUs and/or lower clocks...and that would be counterproductive.

 

I am looking forward to that 90W TDP Ati card .

 

At least with the desktop cards, the problem is that ATI tends to score higher in synthetic benchmarks but nVidia gets better FPS in majority of games (7-8 out of 10). Also, scaling for Crossfire still lags behind SLI (on average 50-60% vs. 70-80%).

 

Yeap, generally an ATI card which delivers the same gaming performance scores about 10% faster in synthetic benchmarks, but the gap has been closing.