AMD 7970m vs GTX 680m

Yes. And likely 1600x900 or full HD resolutions. I dont know for sure.

 

Notebookcheck.com made tests of 7970M
Computer Games on Laptop Graphic Cards - Notebookcheck.net Tech

Strange for me that 7970M at Ultra settings beats Nvidia 675M almost everywhere, but it often fails at MED/HIGH settings

 

What conclusions can we draw fromm this?
It seems little biased for me for then I checked 675M benchies, in there 7970M performed much better than in said test.

 

Where on earth did that guy who is running notebookcheck get his 7970M numbers from. I see AMDs own desktop test system is listed under benchmarks and Alienware M17x R4 which he haven`t even tested yet.

Maybe he goes around in forums like this one and copy/paste whatever he can find, multiply with a made up constant and then calling it a day?

 

Maybe difference between Alien r4 and desktop system was appeared due to CPU difference ( 3720QM mobile vs 2600K desktop ).
But it seems that laptops with 7970 and Ivy will not beat desktop system . Alien and Racer show the same results.

I will test Eurocom Racer 2.0 (3720QM + 8gb + 7970M ) when I will receive it . ETA = 18th may.

 

I find it odd though. Maybe he have tested the Alienware but not put the review out yet. Anyways, here are the scores, Ultra with many goodies enabled in 1080p, High is in in 768p.

7970M vs 675M:

Alan wake:
Ultra: 43 vs 27, 59% better
High: 80 vs 49, 63% better

Anno 2070:
Ultra: 47 vs 33, 42% better
High: 74 vs 66, 12% better

Skyrim:
Ultra:52 vs 41, 27% better
High: 72 vs 71, 1% better

COD 3:
Ultra: 70 vs 62, 13% better
High: 96 vs 95, 1% better

Battlefield 3:
Ultra: 35 vs 24, 46% better
High: 49 vs 54, 9% worse

Batman:
Ultra: 49 vs 35, 40% better
High: 71 vs 69, 3% better

Deus Ex:
Ultra: 76 vs 51, 49% better
High: 90 vs 131, 31% worse

Dirt 3:
Ultra: 47 vs 44, 7.0% better
High: 62 vs 119, 48% worse

Crysis 2:
Ultra: 51 vs 36, 42% better
High: 103 vs 98, 5% better

COD, Black ops:
Ultra: 92 vs 87, 6% better
High: 108 vs 106, 2% better

Starcraft 2:
Ultra: 84 vs 62, 35% better
High: 93 vs 97, 4% worse

Metro 2033:
Ultra: 28 vs 17, 65% better
High: 73 vs 49, 49% better

Mafia 2:
Ultra: 83 vs 66, 26% better
High: 104 vs 98, 6% better

HD 7970M is in average 35% better in Ultra than GTX 675M/580M
(HD 7970M is in average 4% better in High settings than GTX 675M/580M)

 

In Ultra settings yes .

But please check other settings ( low/med/high ) - 675 beats 7970M in many cases for 20%+ . That is very strange .
If 7970M is totally better than 675M it has to win everywhere . If it wins in Ultra it should wins also with lighter settings...

 

I don`t know why you are interested in High settings at 1366x768 though since nobody will be playing with this GPU at that low res/settings, but I added it anyways.

 

It is just strange thing .
Logic is very simple : if the card1 is completely better than card2 it has to be better everywhere and it should shows the "rising trend" in FPS from low setting to ultra . We see that in some settings Amd was beaten by Nvidia but in ultra setting it is always the winner.
Maybe it is fake for ultra settings?

 

I would guess that notebookcheck has got a m17xr4 in for testing.

Very strange to see the results you mention, forcer99.

Either way, it seems like the 7970m is far from having the performance of 6990m Crossfire, which some enthusiastic 7970-owners have suggested...

BF3 Ultra: 6990m is 33% behind, meaning the 7970m is 50% faster.

It seems like 6990m crossfire still is about 30-40% faster than a single 7970m.

It would be nice to get some more confirmation though.

 

I'm thinking it may because card underclocks itself when its not necessary to be pushing itself to the limits.

 

some ppl had fps problems with WoW and other games.
http://forum.notebookreview.com/alienware-m17x/561350-official-alienware-m17x-benchmark-thread-part-4-a-281.html#post8516500

here was their solution:

maybe in lower settings the job is taken over by ivys hd4000 -> therefore the lower fps.

just a guess

 

Yeah, it being better than 6990M CF or 580M SLI was BS from the beginning. Looking at the few game benchmarks where all three are gathered at notebookcheck, reveals that 7970M is falling short. Anandtech put the 7970M 25% over GTX 580M, too low in my opinion, I always thought 7970M was somewhere around 35-40% over 580M. 3DMark11 and other benchmarks could be very deceiving and don`t give the full picture on how good the GPUs perform in real gaming.
-----------------------------------------

The results at lower settings could very well be because of AMDs PowerGating. The GPU shuts of cores when doing light tasks to save power and to reduce heat. It could be the reason why it gets so low FPS in 768p yes, like Rudlin mention.
I don`t think IGP is enabled since it shouldn`t score nearly as close those numbers with the HD4000.

 

I bought racer 2.0 and it is impossible to put SLI/CF in 15'' . So 7970M is the best choice for me.

 

Sounds like early driver teething issues to me.

Looking at 7870 launch benchmarks it is 30-40% ahead of the 560ti at 1080p.

Given the much smaller cut in clock rates for the 7970M you would then expect the difference to be 50-60% between the 580m and 7970M.

Regardless once we get our retail cards we can run some user benchmarks

 

Thank you Cloudfire . I think you are really unveil that secret.

 

That is wrong, it is 22% ahead of 560 TI at 1080p if not less. And 20% ahead all resolutions combined. And we already have FPS results from notebookcheck. 7970M is 35% ahead of GTX 580M. End of discussion.

+50-60% is totally unrealistic. Stop hyping up the GPU

http://www.techpowerup.com/reviews/AMD/HD_7850_HD_7870/26.html

 

well, in 3dmark11 it is exactly %60 on my computer, but in gaming performance I don't know as I don't have it anymore

 

3DMark11 and Vantage is not accurate, its just a synthetic test. We have gaming performance now, both from the review linked to above and notebookcheck.

 

that is not simple logic, its simple assumption. Performance is not proportional per resolution and settings, nor across architecture. Had every single thing remain the same, maybe you would have a point.

When requirements are very low, one GPU can have much higher raw performance but then fall flat faster at more demanding settings.

These cards are useless at 1366x768 resolution. Until you use ultra settings and high res, its when these cards shine.

At low resolutions even midrange GPUs can fight back with the higher end ones.

And even then, some games skew scores, like CoD when most GPUs perform the same. Or World of Warcraft.

 

Slick dude played Battlefield 3 on ultra with max resolution and he gets 50 + fps on a single 7970m so I think Notebook check may have the wrong benchmark.

 

I have all respect for you, but the fact that 660 Ti is 22% behind 7870 mean that 7870 is 30% or something ahead of 660 Ti. I hope you understand some math failure in your post.

 

The same goes for Skyrim, people are getting 80 fps in 1080p/ultra when notebookcheck shows 51 fps. That's 40-50% lower than what people are reporting here in their M17x. It really seems to me that half of the notebookcheck numbers are made up.

 

I actually think you math is off. If the chart was in FPS, and the 7870 scored 100FPS, and the 560 ti scored 78FPS, you'd be right. But, they already converted it from FPS into percent. So, the 7870 is still 22% faster.

 

Thats not correct. The 560 TI is 22% slower than the 7870. If the chart is relative, as it states, it means that the 560 TI performance is 78% of the 7870.

Would you argue that the 7870 is only 89% faster than the GT 520?

Spoiler

The 7870 should be around 850% faster that the GT 520...

ZOTAC GeForce GT 520 1 GB Review | techPowerUp

A lot of people on this board seem to ignore the math when dealing with percentage increase/decrease.

Edit: Did the math. The 7870 is 28,2% faster than the 560 in the chart. 78 * 1,28 ~= 100

 

The 7870 @ 1Ghz is nearly on par with a 7950. That's all that matters.

 

omg seriously?

7870 is (100-78)/78 * 100 = 28.21% faster than 560Ti.. (that is if that chart is correct)

 

Thank you

 

Can you anyone post accurate benchmarks for gaming on the 7970m?

 

I don't trust the notebookcheck benchmarks, something is wrong.

 

LOL well ok then, 7870 is 28% over 560 TI. I always hated those lists Still doesn`t change the fact that 50-60% is pure nonsense. Overclocked it may be close to that, but on stock that is waaay too much dreaming

7970M as good as 580M SLI and 6990MCF? Yeah right. I see notebookcheck haven`t rolled down the 7970M after doing his benchmarks, so it should be further down the list.

EDIT: Nevermind me. Believe what you want. I am not going to discuss anymore other than posting news for the next days (if there are any). Using too much time in this thread. LOL

EDIT #2: The numbers from notebookcheck could be wrong like you guys say. I thought it was fishy from the very beginning. Like he found numbers here and there, included his own factors and such to find numbers lol. But who knows

 

Notebookcheck numbers are most likely wrong. The numbers reported here (NBR) by different people prove that. According to some benchmarks 7970m is almost as good as CF/SLI previous generation.
I know you don't want to believe in this, but it's just stupid to deny obvious things.

 

Why everyone bothers with notebookcheck stuff?
Iirc it's not the first time they got the benchmarks wrong, it happened with other cards as well (reporting lower performance than it is in real world performance).
I think they're scores are made upon assumptions on how the card stack up compared to others.
Then they simply add or subtract fps to games. Which is unprofessional and stupid.

For what is worth I think that the 7970m stacks up directly into gtx570 desktop performance. Maybe it can reach a 580 thru overclock (assuming the 580 runs at stock clocks), but it's nowhere near the power of 2 gtx 560 Ti.

If math isn't an opinion, 2 desktop gtx 560 Ti are about 20-22% faster than a single gtx 570 (assuming sli is scaling almost perfectly)

I don't understand all the arguing. Fact is that we got finally a card that has gtx 570 performance in a notbook. And those who had/own such card in a desktop build knows how much does perform. It eats everything you throw at it comfortably with framerates above 50 and beyond.

 

Would you like to substantiate this claim? Whenever I purchase a new laptop, if it has a dedicated video card, I always run benchmarks. The results of my benchmarks have always, more or less, equaled the notebookcheck average score.

 

Notebookcheck may have tested the Alienware out, posted his results to keep his ranking fresh, but without the review. It may be here pretty soon

 

Also a possible notebook GPU from Nvidia. Only around 130W and same performance as 7870. Confusing though because this is GTX 660, NOT TI. GTX 660/GK106/130W. GTX 660TI/GK104/150W. Both perfect candidates for a mobile GPU

GeForce GTX 660 To Cost between $299 and $329 | VideoCardz.com

 

Ok so if a single 7970m is as powerful as the 6990m sli, a 7970m cf is two times powerful than the 6990m sli, does it mean we can play Crysis 1 or 2 or Metro maxed out with 60fps on a laptop? Is it almost time for my dreams to come true????

 

Here, fixed that for you. All the key words are in bold.

 

There isn't such thing as 6990m SLI, other than that - yes, your dream comes true as soon as you get it in your laptop.

 

Let's do a little reading:

notebookcheck 7970M:

3DMark11 Performance: 5801
3DMark11 Performance GPU: 5510
3DMark Vantage: 21351

ithuiyi 7970M(OC) in M15x:

3DMark11 Performance(GPU?): 5886

Table21 7970M in M15x:

3DMark11 Performance: 5318(probably early drivers)

GeoCake 7970M(heavily overclocked):

3DMark Vantage: 24757

Pau1ow 7970M in M17x R4:

3DMark11 Performance: 5860
3DMark11 Performance GPU: 5546
3DMark Vantage: 21890

This last result isn't overclocked, which to me, seems like the most accurate. That's fairly accurate, within a few percent. I'm sure there have been a few cards that notebookcheck has had inaccurate scores for, the 7970M doesn't appear to be one.

 

I am waiting for Mythlogic or Eurocom to start offering upgrades on the 7970m cf on my x7200 Clevo unit. I cannot wait.

 

I don't think anyone is arguing with the 3DMark scores on the notebookcheck. The problem is with the real in-game fps. It's much lower than what people measured here.
3DMark scores were posted much earlier (~2 weeks ago), while game benchmarks just now.

 

As said before. Just look at HD7870 reviews and take -15% off them.
Cant go much wrong by doing so.

 

Per Slick dude- Crysis 2 on DX 11 with high texture pack installed on ultra settings on 7970m: not below 40 fps.Ranges between 50 to 55 fps.

 

This is what I do, lol. No point needing to go looking for the benchmarks.

 

That's a bit inconsistent, don't ya think? You believe the benchmark results, but not the game results?

 

Drivers used: 8.951 Beta2

.

 

One thing we have to take into consideration is that although AMD drivers have not matured for the new products, Nvidia drivers have matured quite a bit too since the numbers for comparisons were taken.

 

Yes this is a speculation thread isn`t it? The only key here is that you don`t seem to understand that

 

And this means?......

If notebookcheck used the same driver for gaming tests and benchmark tests, there should be no doubt on the gaming tests validity.