GTX 480M 2GB GDDR5 DX11 PhysX 100W GPU?

If you were able to get the upgrade kit with the card and heatsink, have the bios that supports it and you have the AC adapter 220W 20V, 11A or better PSU for it, I don't see any reason it would not work.

 

As far as we know, Nvidia hasn't changed anything, in regards to MXM. The only issue is likely the OEMs providing ample cooling solutions. As we know, regardless of the MXM standard, OEMs have never employed universal designs. So some machines will be able to handle the 480M as currently built, while some will need to respec.

But even now, the 5870 doesn't exactly "fit like a glove" in the Clevo systems, difference being that they didn't deem it necessary to redesign the internal. But it's fair to say that some tweaks would have helped. The W880CU could've had this card in mind as well.

We'll find out when we get a look at the internals of the W880CU, where this extra cm comes in to play.

 

I notice to select the GTX 480M in this models configuration, you are looking to pay £353 extra which brings up the price of any configuration substantially. I see this as a great shame since these laptops are already so expensive.

 

It's not Clevo's fault, blame Nvidia. But yeah, that greatly reduces the numbers of future owners, IMHO.

 

Just a thought... I wonder if you can use mobile I7 with CF 5870 and SSD to reduce the amount of heat and power it consumes. Then overclock it.

 

Yesterday, I saw that the new gtx 480M offering up 897 gigaflops but i return on nvidia website today and oddly it offer only 598 gigaflops. So this card will be less powerful than the quadro FX 3800M (648 gigaflops) and of course the ati 5870M (1120 gigaflops but it's another architecture). We must see the benchmark for confirm that but how disappointing this new fermi GPU!!!

 

That a mistake, because the math doesn't add up.

 

I'm guessing its a mistake as well, its only slightly faster than the 285 in terms of gigaflops.

 

No, it isn't a mistake. The desktop GTX 480 only has 1345GFLOPs as compared to 1063GFLOPs in the desktop GTX 285 - the 598 gigaflop figure is perfectly in keeping with the GTX 480M's differences in shader count and clock speeds compared to the GTX 480.

The fact of the matter is that Nvidia's cards have been behind ATI in raw computational power for a while, but raw computational power just isn't a good indicator of gaming performance.

 

I wonder how it fares against an oc'd 5870m. I can game comfortably on 800/1100 clocks with temps staying in the 70s. Probably have more head room but don't want to push it.

 

the word on the street is the 5870s will be beat down quote badly.

 

So if I carry on my refection: the ati 5870 offer 2720GFLOPS and the 5870M offer 1120GFLOPS so it represent about 41.1%. For nvidia gtx 480 has1344.96GFLOPS and the gtx 480m 598.4GFLOPS it's represent 44.5%. So finally the GTX 480M will beat the 5870M of 13%. It is less than I expected for a price of 800$ while the 4870M cost about 400$. You have all the prices here : Clevo x7200 .
And It should be remembered than the ati has a TDP of 60W and the GTX 480M 100W. It also important for a laptop. And if you would like more power I guess it possible to overlock the 5870M more easily than the nvidia with 100W

 

Sigh... Do you understand how wrong you are?

1. GFLOP means shat, nvidia always had a lower GFLOP compared to ATI, but it doesn't mean that nvidia doesn't perform better sometimes, in fact nvidia is pwning faces during 8000-9000 series. If raw computational power means anything, human brain would be the most powerful computer ever seen and ever be, but if you can't make use of it fully, it means nothing. It's all about the design.

2. TDP is wrong, the TDP ATI gave for 5870MR is for the chip ALONE. While for nvidia, the TDP is for the WHOLE CARD. a 5870MR with 60W rated TDP draws more power than 285M with 75W rated TDP on load. That should tell you something.

3. The 465 can match a 5870 desktop in tessellation enabled dx11 games, a 480M is underclocked 465, which will match around a 5850 desktop in terms of dx11 performance, 5870MR crossfire is 5850 desktop performace, now think about that.

4. Yes, the price is high, the only big problem.

 

no ones going to know for sure until we see benchmarks.

 

I think lepierre is trying to suggest that GTX 480M vs MR5870 will be quite similar to GTX 480 vs HD 5870 (though with an extra 5-8% advantage in the mobile case), and I agree with him in that regard.

 

except that gigaflop is not everything, changing GDDR3 to GDDR5 doesn't affect gigaflop as well, but you see a huge performance difference from 5850MR with GDDR3 and GDDR5. Gigaflop doesn't take into account of the design.

 

You can't use raw processing power and memory bandwidth to compare across architectures, but it's a sensible way to compare two cards with the same architecture.

Radeon HD 5870 - 2.72TFLOPS, 153.6GB/s
Mobility Radeon HD 5870 - 1.12TFLOPS (41.2%), 64GB/s (41.7%)

GTX 480 - 1.345TFLOPS, 177.4GB/s
GTX 480M - 0.598TFLOPS (44.4%), 76.8GB/s (43%)

As you can see, both cards can be quite sensibly looked at as scaled down (quite a lot) versions of the high-end desktop versions - the MR5870 is ~41.5% of a destkop 5870, and the GTX 480M is ~43.7% of a desktop GTX 480.


In the case of the 5850MR with GDDR3 vs GDDR5, you can easily conclude that performance for the GDDR3 version will be anywhere between 45% and 100% of the GDDR5 version, because memory bandwidth is slightly less than half while the core clocks are the same. On the other hand, the HD 5730 is roughly half of an HD 5850 because it has half the shaders (with similar clocks) and about half the memory bandwidth.

Once again, the only reason it makes sense to look at things in this way is because the architecture is pretty much the same between these cards. No-one here is saying that ATI's cards will be more powerful than Nvidia's because they have more gigaflops, or Nvidia's cards are more powerful than ATI's because they have more memory bandwidth.

 

You can't see it this way even in the same architecture. Because the GFLOP is a calculation using cores and clocks, but how it is reduced is another story.

For 5870MR, the reduction is mainly in the cores, the cores are cutdown cleanly by half from 1600 to 800.

While for 480M, the number of processing cores remains at 352, down from 480, which is not really cutdown by much, even though the clocks are cutdown more. However, clock most of the time doesn't affect the performance much, as you can see from ocing your 5870MR from 700 to 800 or even 900 core clock, the bench doesn't increase by much.

 

I'm pretty sure 65w is for the whole card. If you look at AMD's website they rate the MR5870 as 50w, I'd assume that is for the chip alone. Given that we always hear 50w/65w for the MR5870, I'd think it's relatively safe to assume the 65w figure is for the whole card. Especially if you factor in that the 480M has twice the memory bus and twice the amount of vram, plus the fact that this architecture is naturally more power hungry than the ATI architecture, it certainly adds up to being 65w whole card vs 100w whole card.

 

Well, then how is this
Power, Efficiency, And Battery Life : AVADirect's W860CU: Mobility Radeon HD 5870 Vs. GeForce GTX 285M
explained.

The power draw of 5870MR is about the same as 285M on load, and actually higher on idle.

 

First of all, I'd like to see you back that up with data.

In any case, clock speed has just as much impact as extra shaders do. If increasing the clock speed of the 5870MR doesn't help much, then the reason for this is probably that it is bottlenecked by memory bandwidth, in which case overclocking the VRAM would help performance, while extra shaders wouldn't help much either.

 

I thought the idle power drain is a known issue on Clevo models with the MR5870, and is attributed to the bios. I thought I've read somewhere that the MR5870 isn't down clocking itself at idle as much as it should on Clevo's where as in the M17x it doesn't have this issue.

 

Yeah, the idle power of the MR5870 would likely be lower than the GTX 285M when downclocked properly. In any case, if the MR5870 system consumed 4W less on idle, we can probably be happy with a 70W figure for the TDP of the MR5870, as opposed to 75W on the GTX 285M - which means the GTX 480M still consumes ~40% more power than the MR5870.

 

Firstly, 465 GTX has around half the GFLOP of 480 GTX, but it matches the 5870 desktop in dirt 2/battleforge etc dx 11 games. It is half the performance of 480 in vantage, but in tessallation enabled games the GFLOP start to lose it's meaning.

2ndly, it's not 4W less on idle, it's 4W less on full GPU load. It doesn't matter if it's bios problem that it doesn't downclock properly in idle because power consumption in idle is not important and we are talking more on max power consumption here. Also, it's 4W less on load in a situation of 118W to 114W, the difference is so small it's statistically insignificant. We can actually safely say that 5870MR has the same power consumption as 285M, at 75W. Now, a 33% power increase might seem significant, but if you put it at just a 25W increase, it doesn't seem so much all of a sudden. As said previously, a 20W 8600M changing to a 38W 4670 MR it's almost 100% difference, but when we look at raw numbers, it's just an 18W difference.

 

The high power draw at load can also be attributed to the bios and it's issues with the MR5870. This Tom's Hardware test can't be looked at as a reliable source of information on comparisons between these two cards given the circumstances. We need to see more data from non affected laptops.

 

The raw power of the GTX 465 is actually around 64% that of the GTX 480. Guess what? If you look here, the DirectX 11 performance of the GTX 465 is also roughly 60% that of the GTX 480, and the Vantage score is also around 60%. Coincidence? I think not.

I meant the 4W difference at full load, not idle; sorry about that. I disagree with the 4W being statistically insignificant, however, because the HD 5870 having lower power consumption is the best explanation for the 4W difference in total system power consumption. It's far more sensible to attribute the difference to the difference in cards instead of ignoring it altogether.

If ~30W doesn't seem like much to you, then I don't know what would be, because that's almost the full power consumption of a Core i5 under relatively heavy load.

Why would the BIOS affect power draw at load?

 

From notebookcheck:
Notebookcheck: Review MSI GX740-i7247LW7P Gaming-Notebook
GX740 at 129W.

Notebookcheck: Review Deviltech Devil 9700 DTX (W860CU) Notebook
860CU with 280M at 125W.

From this, it's mostly possible that 5870MR has the same power consumption as 285M at 75W.

Hmm.. For the performance, ok I didn't really notice that. However, the high performance difference between 480 and 5870 in directx 11 games tells me that the difference in mobile versions won't be a small 10%. 10% in vantage likely, but in real world dx11 games it will be much larger.

 

Those are two completely different laptops, so the difference there is a lot less meaningful than the Tom's Hardware figures. Unlike with the Clevo test where the rest of the components were the same, the difference here could be attributed to quite a lot of other factors.

 

Notebookcheck: Review Asus G73JH (ATI HD 5870) Gaming Notebook

G73 as well, at 127W.

 

I can't think of anything off the top of my head but if the bios has issues with this particular card, then maybe there is the possibility it is affecting it in more ways than just at idle.

 

When every notebook with 5870MR is having around the same power consumption at load, even the asus one where the card is built by Asus themselves, it means that it's not just a bios issue but that is how the card is.

 

Again, yet another different laptop. There are plenty of factors that could affect power consumption. For example, the difference between a 15.6" LCD in the Sager and 17.3" in the two you linked could account for a few watts. For that matter, the G73 also has two hard drives. Then again, the tested model also uses an i5-520M instead of an i7-720QM. There are just too many differences for a reasonable comparison, and so the Tom's Hardware figures are still the most meaningful.

I do agree that the BIOS is unlikely to impact load power consumption, but failure to downclock definitely has a big impact when idle.

 

The 4W difference tested in the 860CU is statistically insignificant, because it's less than 4% difference, and it could totally be attributed to non-constant factors like chip quality etc.

 

4% difference in total system power consumption, not the video cards alone. Variance in other components could definitely cause such a difference, but without further evidence you can't say anything about it. The test would definitely have been better if Tom's Hardware had used the same system for both tests and just swapped out the video card, though. For example, it's entirely possible that the difference between the two cards under load was actually more than 4W, because the CPU in the 5870 system was more power-hungry.

As such, the most sensible choice without further evidence is still to attribute this 4W difference to the video cards. Sure, there's a high level of uncertainty in this figure, but the uncertainty in assuming the two cards have equal power consumption is even greater.

 

The reason I say it's insignificant is because the quality variance of the GPU chip itself is an uncertainty just like how no 2 CPU chip is the same, there could be some 5870MR chips that are more power efficient and some 285M that is more power efficient. The point is that the test is based on a single 285M and a single 5870MR which could not statistically justify the 4W difference. Even if you would to use 2 5870MR doing the same power consumption test there would be minor 3-4W differences.

 

Yes, but you have even less justification for the cards being equal than for a 4W difference. What's to say that the cards being compared weren't a "bad" 5870 vs a "good" GTX 285M, and the actual difference wouldn't be more like 10W?

Again, while the level of uncertainty is quite high, 4W difference is more likely than 0W.

To put it another way, let's say both systems would almost certainly have at most 5W of variance up or down. Consequently, power consumption under load would be 109W-119W for the 5870 system, and 113W-123W for the 285M system. This would mean the difference is anywhere between 14W less for the 5870 and 6W more for the 5870, but the fact is that 4W less for the 5870 lies at the center of this range.

So, it would be more accurate to say that, if the 75W for the GTX 285M is accurate (and meaningfully represents consumption under load), then the TDP of the 5870 ought to be between 61W and 81W. Of course, TDP is not the same thing as power consumption under load anyway.

 

GTX 285M and HD 5870M have the same TDP requirements which is 75W for the whole module.

But i agree on the power consumption thing

 

Where did you get that info?

 

The official figure is 50W, though it's obviously not 25W less power-hungry than the GTX 285M. I don't know where you would get 75W from, though.

 

Yeap, and the PCB usually adds another 5W, maybe 10W at max. So the 5870MR has a TDP of around 55-60W. We already went over this in other threads.

 

From my understanding ATI states the tDP of the GPU only which is 50W.
GDDR5 takes 10 - 15 W for 1 GB
max 10 W for the PCBA etc.

which would make for me something between 70 - 75..

Does that sound about right?

Also I know from a fact that the gaming boxes today are designed for 75W TDP of the Module, so why to waste available TDP and provide a lower clocked version?

 

It's more like in between 60w and 70w total. And why would a manufacturer waste available TDP and provide lower clocked versions? Probably to increase battery life and over all power efficiency.

 

Not really. It's because it can sell more of these cards. Having a lower TDP implies that more laptops can carry the card because of less heat problems and possibly cheaper cooling solutions.

If I recall correctly the, 280M was only available in Clevo and Alienware notebooks while all other manufacturers opted for the 260M which had a 65W TDP.

Yes, you are right about what you wrote earlier, as far as I know the whole TDP of the 5870MR is 65W.

 

Yup, sounds pretty accurate - the fact is that because ATi don't release specific card TDP we can only speculate, but there is very little between the 5870 and 285M power wise.

Even the 5W possible difference is going to depend on individual load of the GPU and RAM.

 

In other related news: Fudzilla - Geforce GTX 465 has 200W TDP

 

200W Woah!!!! article states thats 50W less from 480. Thats insane.

 

Please start a new thread when this thing is actually released so people don't have to dig through 1000 posts for some actual benchmarks.