Pic of GTX 480M + 3DM06 Test

If this holds true when the GTX is out, it is consistent with 15-20% vendor performance evolution rule- If you were on the selling side, you don't want to give your customer the 50-100% improvement right away, since it would become the expected norm, and there will less room for future sales strategies.

Of course, until the card is out, and end users are reporting their bench/game results, it's still mostly speculation.

 

Do you guys game?
Do you realize the difference between a desktop 940 and mobile 820QM regarding FPS in games? It could be more than 20% on ultra/high settings. In addition, ATI cards scale better with a strong CPU.
Now, if you forget the tessellation benches for a sec - the rest will be from 0-20% gain. BTW, both Risen and NFS Shift scale nicely with a sheer CPU power.
Here's your 20% - gone

Let's wait till some real tests.

 

Im getting a new heatsink in about a week, and i got a 5870 pulled out of a defective w870cu pretty cheap (a bit under 200€ , so when i manage to mod the heatsink and put it all together in my d900f as it should be i will post the benchmarks of the mobility 5870 with the i7-950 in my d900f, so those should be a good comparison to the 480m + i7-940...

 

425mhz core clock. 850mhz shader clock. These clock rates heavily influence 3dmark06 framerates. You following me yet?

Don't forget, it's 2010 now. 3dmark06 is dead.

The 480M is twice as shader heavy as the 5870. Not to mention has considerably more bandwith. DX11 games will continue to shine on Fermi tech.

Far as followers of brand spankin' new tech go, I'd think they'd also only be interested in the latest and greatest software.

 

^^^^
Which games utilize DX11 and tessellation?

EDIT: Never mind-

http://en.wikipedia.org/wiki/List_of_games_with_DirectX_11_support

 

That would be awesome!
Can't wait to see your tests!
+1 rep to keep you going, don't stop

 

i still cant find on that norebook check site where to find what cpu was used lol /cry.

 

Here you go
Scroll down to 3Dmark tests and click More Details on the right side to see what specs were used.

 

ah thanks.

 

The thing is, because the overall score combines CPU and GPU results in an unknown manner, it's hard to predict how it scales with the performance of either component. On the other hand, what I've seen so far suggests that the CPU and GPU scores do indeed scale pretty much linearly with CPU and GPU power within the same architecture.

Well, Notebookcheck has actual in-game benchmarks for the 480M; the only issue is that it's only a single source so far and not a particularly reliable one at that.

I'm not sure what you're saying, but most current games are capable of 60fps+ even with relatively weak CPUs. Although it's not a perfect comparison, if you only look at the results for very high settings the results will still mostly reflect GPU performance.

 

Let me provide an example:

P8700(@2.5GHz)+280M gets ~7fps in Total War on ultra + 1200P. OC'ing the CPU to 3.06GHz boosts the FPS to 25 average.
In numbers - more than 350% gain
Do you get my point?

Another example: (same config CPU @2.5) in America's Army 3 - ~30FPS, OC'ed to 3.06 - 45FPS.
50% gain, just by providing more MHz.

That's just one extreme example telling you that you can't rely on those tests.

Another thing you might wanna consider is that ATI cards scale much better with a strong CPU , and that's why I'm so anxious to see Hellcry do the tests on the same system with a 5870. Should give us quite a bit more info to think about.

 

I'm not sure what you mean when you say ATI cards scale better with a strong CPU. The GPU and CPU perform different tasks, after all - can you link me to some benchmarks?

As for that 350% gain you mentioned, there's something really strange going on there.

You're right in saying that we need to see a comparison between the two cards with the same CPUs used, but I stick by my ~20% prediction.

 

I claimed that the ATis scaled with cpu power when I found that a T6600/3670 was using a lot more of the cpu than a T5750/8600m GT in the same game.

Joker has also found that the Vantage gpu (note: not the overall) score changes by thousands going from an i7-720qm to an i7-920xm (comparison is with the same gpus of course).

 

Does the d900f support the 5870 in the BIOS?
Didn't think that was possible without flashing it.
Sorry to get off topic but that would be pretty cool if you ask me

 

This...IIRC stock 720 with 5870x2 scored around 10.8-11.8k. 920's with 5870x2 got 12.8-13.8k stock vantage.

 

Nothing strange, just an extreme example of a heavily CPU-demanding game on a system with a weak CPU + a powerful GPU.

 

No, a 350% gain from a 22% increase in CPU clock speed is definitely strange. If it was a 20-25% gain it would be normal for a CPU-demanding game (the 3% extra gain there is statistically reasonable), but as it stands there's either something wrong with the test, or there's something very wrong with how the game is coded.

That's not really surprising since it's a crossfire configuration - that's a lot more GPU power than a single card. Besides that, if you ran Vantage on one of the higher settings, that difference would be a lot smaller.
Also, those are GPU scores, right? Just making sure.

That's far from rigorous enough to properly justify that idea. You didn't even test the 3670 with different CPUs, and you don't state anything about the testing conditions or the period of time for the CPU usage measurement.

 

Unfortunately, I can't provide a documented data to prove my point, since the system used for those tests (M17X-R1 280M SLI vs 4870 CF) was shipped back to Dell.
But., I'll look for Lewdvig's comparative testing on the similar system 280M's vs 4870's in a wide variety of games, to see if there are similar results.

1) I did see a 350% gain in Total War, 50% in AA3, 30% in Dragon Age and a few others that I don't remember.
My point is simple: if I can find an extreme situation to show you something that normally is not realistic, so can Nvidia/ATI/etc for promoting the new tech. That's marketing and nothing fancy. Tessellation is their only strong point thus far, but that's another story.

 

so Nvidia is finally back to get the cake. Good. Can't wait to SLI these babies.

 

Indeed, all of Nvidia's benchmarks have focused on DX11 for obvious reasons. Even their DX11 performance is underwhelming considering the price of the card, too.

 

Not to mention the extra power draw.....really these benchmarks are just like the ones ati released showing it crushing the 285, but in reality it was on abit faster, same case here, but its silly that they are upping the power draw and with a huge price tag, i cant see anyone upgrading from a 5870 to the new 480.

 

Maybe upgrading from a 5870 to a GTX 480M is a moot point but for those of us that don't have either it's good to have GPU's that are almost equals.
At least now we have a choice.

 

I have the 05 bios and it recognises it without a problem, goes into windows and runs just fine. I just need to modify the heatsink, since the 5870 has different positions of those chips and coils and i need to cut away 2 pieces of the heatsink to cool everything properly (thats why i ordered a new one, so i can keep the original one intact, should the 5870 ever fry).

 

What really interests me is the oc capability of the GTX 480M. As we all know the MR HD 5870 can be overclocked quite high (btw, does anyone know the highest stable overclock?), I think oc to 850 MHz core should be no problem. On the other hand I can imagine that nobody will be able to oc the GTX 480M without breaking 95°C on load. In tesselation games the GTX 480M might still perform better but the difference could be something like 5 - 10% in games overall, which is just not worth the price.

 

You'll have to keep us up to date on how everything works out.

 

If i dont sell it before that happens.. now it seams i finally got a serious buyer..
The irony

 

Yes its quite the Irony having to wait so long to sell it.

 

WHAT!
you going to do all that work and then sell it?
What are you thinking of getting as a replacement?
At least run some basic benchmarks before you sell it

 

I dont even have the heatsink yet, and if i the buyer decides to go with the 280m (which he seams to prefer as it seams for financial reasons) and pay before i get the heatsink i will ship it before i even get a chance to run those benchmarks i wanted and ill have to sell the 5870 with the modified heatsink seperately.
I am almost done building my new desktop (got almost all parts - i7-980x, hd 5870, 6gb ram, 160gb x25-m g2, all water cooled).
Now with the money from this one im looking to get a smaller (15-17 inch) or at least ligher and not so thick notebook, that is also quieter, has no intake vents on the bottom if thats possible, since i want to be able to have it on my lap without worrying of it overheating, but also has enough power to let me game on it too if i want.
Im currently looking at:

1. the g73-jh A1 (would change hds to ssds);
+/- lighter and not so thick, but its still pretty large,
- looks ugly to me - the whole design and those blue lights (but i can get past that, performance, usability and silence matter more)
- i dont like the idea of having to take the whole thing apart when i want to upgrade the cpu (and cant upgrade the gpu) or to reaply better paste on them - i had to do that with my compaq 6720s and it wasnt fun... took me almost 3 hours to upgrade the cpu (for my gf)...
+ pretty good cpu and gpu for the price
+ as i hear a great led backlit screen
+ as i hear its silent, but i have to hear it with my own ears before i buy (im VERY sensitive to noise and i live in a room that is REAL silent - there is virtually no background noise)
+ no fan intakes at the bottom, supposed to suck in air from above (at least thats what everyone says..)
+ backlit keyboard and tilted angle of it (i do like to game at night and sometimes i just miss the old backlit keyboard of my previous M17, and i got used to the tilted angle since im using the zalman nc-2000)

2. Clevo W860cu or W870cu with i7-820qm, 8gb, mobility 5870, ssd g2 160gb..
Why 860cu or 870cu? Becouse the 870cu is supposed to be quieter as i read the 860cu is damn loud, though its larger..
+ smaller and ligher (860cu more as 870cu)
- as i read, pretty loud..
+ customizable to my liking, good components
- fan intakes on bottom...
+ all components accessible by taking off the back cover..

Im really out of ideas what to buy as a portable replacement.. none just fit what i want.. i still do some gaming, and when im away from home i want to be able to take something with me.. I just might have to wait for something better to come along (drooling looking at those japanese water cooled laptops..)..
If anyone has a better suggestion that would fit my needs (as many as it can), please do tell;
- 15-17 inch, the smaller and ligher, the better
- as quiet as it gets (unaudible would be great)!
- can take it on my lap without blocking the fan intakes (putting the cooler on my lap to prevent that is not a solution, its just uncomfortable and not practical)
- still good components so i can actually game and keep this one longer than i have my notebooks so far (im pretty spoiled by what ive had so i dont think i could go back to c2d and anything below 5850)..
- not so important, but nice; good look (to me good is plain), backlit keyboard (ya i do like that), easy access to upgrade cpu/gpu and clean the heatsinks

 

Not really... im just a student and i dont spend money on anything else (no goin to the movies, no splurging for any parties, no late night drinking out.. i just prefer a good glass of wine or beer while watching a movie or listening to some metal at home, and i dont have to buy that booze myself either since i can take it from home - always some good stuff there)

If thats true i just might hold on until its released.. ill go look around what news i can find about it..
Still, is anything like that out now?

Edit: Seams nobody knows yet which gpu it will actualy have, just that it will be a dx11 capable nvidia - fermi card, but i doubt they could fit in the gtx 480m, so probably a gtx 460m or 470m..

 

More than likely no 480GTX the power it needs is to much, ideally a 460GTX would be a good compromise.

 

Whell im still looking around at the offerings.. Nothing that would really fascinate me though.. And the more i read about the g73jh, the more faults it has i learn about..

 

Yeah same with me not into 17" inch laptops and unless the G53 is specced decently I'm going to wait for SANDY BRIDGE.
My 260 GTX is still good enough for now but you know how it is always the need for the latest and greatest.

 

Ya if i only knew what the g53 would look like and what it will actually have inside now, that would immensley help my decision whether to wait or not..

 

i'm rock solid stable at 880/1080 on my 5870, which is about a 30% oc?

 

In which laptop is this and what are your temperatures (idle/load)?

 

W860CU, idle 56c, as i recall under load 75c or something, it wont push much past this but i have zero artifacts and can take furmark for hours and hours, if only my CPU was overclockable -=]. Going to be interesting to see how the 480 OCs, my guess is with the new power consumption and cooling needed, not so good.

 

Oh, I bet this is someone from Nvidia or from those resellers focusing on this 480M, lol. They would buy your system to deprive us from knowing the gruesome truth.

 

Very nice oc, Fzhfzh. 910 MHz would be a 30% oc which is awesome and should be stable as the RV770 is normally clocked at 850 MHz anyway. I think memory could be puched to 1200 MHz as well without problems, the only limiting factor would be temperature.

And this will be the main problem with the GTX 480M, bringing the GTX 480M to stock clocks means a 42% oc. This would fry every system.

 

Here you go guys - official benches by Notebookcheck.

 

Those temperatures seem fine lol. As far as performance, some games it ties with the 5870, others it's a ways ahead.

I wouldn't mind getting one, but I can't stand Nvidia's prices.

Edit: after looking at more benches there, I've concluded what I already knew. Nvidia owns ATi at AA/AF. That's why I liked the 8600m GT more than the mr3670. It can actually anti-alias.

 

Actually that is a big disappointment. 9% more performance? The MR HD 5870 can easily be oc by 30% which should yield at least +9% performance and then it would be on the same level as the stock GTX 480M which with 84°C at stock clocks during games will not be much of an overclocker.

The GTX 480M only shines when tesselation is used heavily, but then again it will be too weak to play those tesselation heavy titles in FullHD and max details with good fps. I think this problem can be overcome with SLi, but still there is no relation whatsoever between performance and price
.
I expected more...

 

Where did you read 9%?

Also, considering stock 5870s are hitting 90 Celsius, I don't see 84 Celsius as a negative. Also, some of those 5870s are undervolted.

Direct links to benchmark pictures.

Far Cry 2

Batman: Arkham Asylum

Just Cause 2

Stalker: Call of Pripyat

Battlefield: Bad Company 2

Dirt 2

Metro 2033

 

I read it in the review:

What you linked are the benchmarks which Nivdia conducted... hardly any evidence. Both ATI and Nvidia publish own benchmarks which try to establish their card as the new savior. Those are not representative.

I was also obviously referring to MR HD 5870s in Clevo laptops. I actually like Nvidia cards better, but 800 € for a 10% performance gain?

 

My bad, I got those mixed up with the ones conducted by notebookcheck.

So it looks like it's only a really good improvement for Dirt 2.

 

Yep, plus the temperatures... a tricky subject.
Nvidia never shows the memory temps and given a 2GB GDDR5 those should be way above the 84C on the core. If you compare core temps vs core temps (480M vs 5870) the difference might be stunning - up to 15C. Still, we don't know how the sensors were calibrated.
But generally, when I had 280's the keyboard was much warmer than with the 4870's (same system and same sensors readings). So, there's still a lot that we don't know.


<<< Just noticed, that 84C reading isn't even the max value! This could mean that games like Crysis and Dragon Age would bring the temps to some unpleasant levels despite the famous Clevo cooling.
Also, have you seen any Furmark+Prime tests with temps? Normally, Notebookcheck does it for every tested system

 

I was looking for Furmark, but couldn't find it (glad to hear it wasn't just me getting lost in a sea of German words lol).

I'll probably go with the M17x and the 5870s unless something happens at E3.

 

FurMark: Graphics Card Stability and Stress Test, OpenGL Benchmark and GPU Temperature | oZone3D.Net

 

This is really a big dissapointment from nvidia (their benchmarks are as usual pure lies)..
Performance in most games is just marginally better than the 5870 - up to 9% (dirt 2 being an exception here, but that game just favours nvidia cards), their main selling point "tesselation heavy" means crud when you get 12 fps on a dx11 game with max setting, so you cant really use it, i dont see it overclocking very high with the temperatures shown, since the d900f already has one of the best cooling systems you can find in a notebook and its already struggling to keep it cool at stock (the w880cu is really going to struggle if you ask me), the power draw is also ridiculous considering how little the gain is and so is the price.
+ we still havent seen the performance of a mobility 5870 paired with a desktop cpu as the gtx 480m is - tomorrow if everything works out ill post them up (i already modified the heatsink, but i ran out of pads, so i ordered them and im getting them tomorrow - the old ones are ruined and not usable anymore)

 

@ Paladin,

We need the actual Furmark test results from that review
We know where to download the software from, thanks.

Go for it, m8!
This could really help clarifying the situation.