AMD 7970m vs GTX 680m

Nope because

 

Well it's not because, it's due to no new info.

 

Only that it's not really so very next-gen...

 

If this 680M was $100 cheaper than the 7970M, I'd strongly consider taking it over the latter.

Or is it because that's how we all feel about Nvidia?

 

I did some personal benchmarks for you guys. The following game runs were done at 950/1450 except for The Witcher 2. 1080p resolution, no v-sync.

Crysis 2 - Maxed out Ultra settings, DX11, Hi-res texture pack
Run: From the start of "Second Chance" to the end of the firefight in the enemy encampment, killing everybody, blowing up barrels
Min - 25 fps
Max - 61
Avg - 42.8

Battlefield 3 - Ultra preset, FOV 70
Run: From the start of "Operation Swordbreaker" to the end of the firefight in the parking lot
Min - 28
Max - 89
Avg - 47.3

CoD: Modern Warfare 3 - Completely maxed out
Run: From the start of "Black Tuesday" until I enter the Stock Exchange
Min - 65
Max - 249
Avg - 123.5

The Witcher 2: Enhanced Edition - Maxed out except no Uber-sampling. Had to run at 920Mhz to be stable.
Run: New game from when I first gain control of Geralt in the tent until after the first fight after climbing the siege tower
Min - 0?
Max - 83
Avg - 48.2

 

Cloudfire, that picture says it better than anything I can type.

 

I missed this and have to comment.

The 9600M GT is weak as hell @ 1280x800, with any modern game, and you know it. If the GT 540M, which is nearly 3x faster than it, is only good for medium settings @ 1280x800, what is your 9600M GT really doing, in a 2012 release?

Keep it real dude, or just stop posting in here.

 

yeah well, it depends on what kinda games hes throwing at it right? minesweeper at 1280x800 maxxed out shouldnt be any problem

 

and jaybee:

I can run pes 2012 at 60fps, fifa 12 at 35fps maxed without any aa and fifa 11 maxed without any aa was 60 fps at 1280x800. I only played the demo on this as I have it on ps3.

I have dirt 3 which I run everything on high I believe but with shadows on low which runs at 35-50fps at 1280x800. I recently started playing batman arkham asylum at 1280x800 and it plays at 35fps on all high settings at 1280x800. Test drive unlimited 2 runs I believe on medium at 35fps, my cpu struggles on that game. I can run trackmania nations forever with everything maxed apart from anti-aliasing and the 3 post priocess options at 60-100fps. I even have rfactor which easily runs at 60fps maxed out.

A gt 540m is not much better really then my 9600m gt, most of the power is gained by the cpu performance and the hugh power consumption of the 540m. I reckon synthetic benchmarks 30% difference at most but the 540m is like a 35w card and mine is a 23w card and overall probably double my fps but 50% of that is because of the cpu speed difference. My gpu is more powerful then the ps3 graphics card as well as on par with a go 7950gtx.

 

Diablo III Benchmarked - Notebookcheck.net Reviews

The 7970m is 38% faster than the 6990m @ 1080p High AA on and 31% @ 768p High AA on.

I guess we will see whether this card turns out 40% faster, like Kevin and Meaker say they predicted.

 

9600M GT is a decent card but games like PES are very scalable. Arkham Asylum is about three years old and ran nicely maxed out even on my 8600M GS. Try running more recent and demanding games like Crysis 2 or BF3 on higher settings, you'll feel the pinch then for sure. Like when I tried playing Crysis 2 on my 8600M GS I got something like 14 FPS on the lowest settings and 1280x800.

 

Crysis 2 9600M GT - YouTube
Crysis 2 with Nvidia Gforce 9600 M GT(Laptop) - YouTube

crysis 2 running on similar system. In terms of sm3 score my gpu must be similar to a gt 525m which is basically an underclocked 540m. One is running maxed the other otn gamer which I believe is low. The 9600m gt was probably the first mid range card that could run most games easily on high at 30fps+ at 1280x800.

 

as an owner of both cards, it's probably closer to 50%

 

7970m reviewd in the m17x:

Test Alienware M17x R4 Notebook - Notebookcheck.com Tests

About 50% faster than 6990m/685m across 18 games, the article claims.

 

I don't have the time to crosscheck now, but it can't be faster than 50% of both since the 675m is faster than 6990m

 

Not by much at all at stock and after a 50% difference it will seem even smaller.

 

Hmm, seems like Enduro isn't working?

 

it does work, just not as efficient as optimus. seems like u have to tell it manually which gpu to use for what program...

eidt: it also looks like ull have to deal with a significant amount of performance loss as long as enduro is activated!

 

That kinda sucks! Doesn't it have a list of applications (executables for games) that require graphics acceleration, as the nVidia drivers do?

 

It does, but apparently the list isn't very big so you have to manually set the GPU for a lot of games.

 

it supposedly does, but the efficiency with which it actually chooses the correct gpu for each application is just around 60%

ok so ive checked all the listed gaming benchmarks myself and this is the conclusion *drum roll*

7970M is 47.7% faster than 6990M across 17 benched games
7970M is 45.2% faster than 675M across 15 benched games
6990M is 32.3% slower than 7970M across 17 benched games
675M is 31.1% slower than 7970M across 15 benched games

i included both the "faster than" and "slower than" statements just so that we all are on the same page here@percentage calculations. so to all the nvidia/amd fanboys out there: discussion is over, deal with it

oh btw, all benched games were on full HD res with all options maxxed out

cheers

 

Allright, I will double check. I won`t bother with lower resolutions than 1080p since nobody will be playing with this GPU lower than that anyways....

Dirt Showdown (2012)
7970M 48.7 FPS
580M 34 FPS
Difference 43.24%

Diablo III (2012)
7970M 100.8 FPS
580M 71 FPS
Difference 41.97%

Risen 2: Dark Waters (2012)
7970M 43.1 FPS
580M 29 FPS
Difference 48.62%

Mass Effect 3 (2012)
7970M 59.6 FPS
580M 60 FPS
Difference 0%

Alan Wake (2012)
7970M 44.7 FPS
580M 28 FPS
Difference 59.64%

Anno 2070 (2011)
7970M 55.4 FPS
580M 32 FPS
Difference 73.13%

The Elder Scrolls V: Skyrim (2011)
7970M 55.6 FPS
580M 33 FPS
Difference 68.48%

CoD: Modern Warfare 3 (2011)
7970M 90.7 FPS
580M 65 FPS
Difference 39.54%

Battlefield 3 (2011)
7970M 34.7 FPS
580M 24 FPS
Difference 44.58%

Batman: Arkham City (2011)
7970M 57 FPS
580M 34 FPS
Difference 67.65%

Fifa 12 (2011)
7970M 228.8 FPS
580M 200 FPS
Difference 14.4%

F1 2011 (2011)
7970M 65 FPS
580M 47 FPS
Difference 38.30%

Deus Ex Human Revolution (2011)
7970M 86.7 FPS
580M 46 FPS
Difference 88.48%

Dirt 3 (2011)
7970M 64.1 FPS
580M 44 FPS
Difference 45.68%

Crysis 2 (2011)
7970M 54.5 FPS
580M 35 FPS
Difference 55.71%

Mafia 2 (2010)
7970M 88.8 FPS
580M 62 FPS
Difference 43.23%

StarCraft 2 (2010)
7970M 99.2 FPS
580M 60 FPS
Difference 65.33%

Metro 2033 (2010)
7970M 26.9 FPS
580M 17 FPS
Difference 58.24%

896.22%/18 = 49.79%

7970M is in average dead on 50% faster than GTX 580M in 1080p

 

^^Awesome. And these are all maxxed out graphics settings at 1080p? AA/no AA? I take it Mass Effect 3 has a framelock at 60fps ?

 

Yes to all of your questions. Everthing on max. It can play all of them too in Ultra except 30+FPS in Metro 4XAF+AAA. A beastly card indeed

Just look here and hold the mouse over the FPS results and you will see the settings
http://www.notebookcheck.com/Test-Alienware-M17x-R4-Notebook.74805.0.html

Meaker was dead on with his prediction of 50%+ compared to 580M. Well done

 

hmmm interesting to see how the 675M steadily performs several percent faster than the 580M. is that difference just due to later testing, thus more current drivers?

cudos to meaker for his super-accurate prediction! so tell us now, oh meakerish one, whats the 680M gonna be like?

 

Fantastic! Simply (in the words of Russell Peters) MINDBLASTING!!

 

Well I didn`t crosscheck with 675M. And I won`t bother since it should be virtual identical with 580M. New drivers are probably the reason the difference between 580M/675M yes. I don`t think the 580M results I used is the latest ones, but in all fairness the 7970M is on fresh drivers as well. It will only improve

One important thing to remember though The 580M results are with a bit weaker CPU so the difference are maybe a bit skewed. But good luck finding exact same setup to test.
@Prasad: Everything is maxxed with the settings in all the games. I edited my post. There are a few there without AF but I think you cannot enable them anyways in the games

Meaker thinks 680M will score P9500 I`ve heard

 

Yo cloudfire were u overclocked on those benchmarks or stock clocks? looking great so far, cant wait to finish the waiting game with my NP9170

 

@ cloudfire now you are just winding him up

 

I compared the review posted a page ago in this thread and the 580M results at notebookcheck. Its not me who have done the test. Everything was tested on stock clocks

 

Probably the better drivers and slightly better CPU.

 

Hah, would be awesome if it did lol.

 

if 680m is scoring 9.5k I am buying a np9150 asap that little thing with 9.5k 3dmark11, like dreams come true

 

Awesome.. dear god yes.

I would literally work my off to get a laptop with it! I swear to god, with such power:time we are going to get ultrabooks running Crysis 1 at 60 FPS vSync @ 1920x1080.. please nVIDIA make the impossible happen

 

Just one technicality. We should take out Mass Effect 3, because V-sync had them both capped at 60fps, making the data invalid. Fair?

The average is 52.7%, minus that. But what's a few percentage points among tech heads?




The correct answer is EVERYTHING.

 

Yeah shure, thats fair I guess

 

Google Translate

Crazy low power consumption: 166.3w max or 108.6w probably running 3dmark tests. I am definately getting simething like a 7770m.

 

Just correcting, it is 7970m

Anyway it seams that enduro is not that good, not only you have to manualy select which programs use the gpu, it also slows performance compared to it switched off.

 

yep, the guys in the review kept suggesting to switch enduro off completely when benching/gaming via an alienware implemented keyboard shortcut (reboot required)

 

This review is a little bit weird. Looking at R3 review with 580M Review Alienware M17x R3 GTX 580M i7-2820QM Notebook - Notebookcheck.net Reviews and comparing power consumption, temperature and noise I get that
R4 vs R3
108.6W/166.3W 162W/196W
34.2C/38.3C 30.9C/33.8C (correcting for room temp)
42dB/50.3dB 43dB/45.9dB

So R4 with 7970M uses 20%-50% less power under load, but is 4-5C hotter and has about the same noise level (or louder?).
I'm not sure how much sense it makes. Does it mean that Dell screwed up the cooling on R4?

 

Dell did an incredible job at cooling as it is a lot more powerful 50% better at games and 90% better at synthetic benchmarks, I just checked and its even thinner and it runs quite cool 80 degrees is incredible for such performance.

 

Funny how for Metro 2033, from High -> Max, fps dropped from 70+ fps down to 30.

That's one hell of a GPU intensive game that's two years old.

 

I really find it hard to believe the power consumption the 7970M is reporting.

It's still at the top of the heap.

 

Could you explain your reasoning Kevin ?

 

Yes, but it's also poorly optimized and deserves a decent amount of patching...

 

You're forgetting the surface area of the GPU die, 580M/675M is almost twice as bigger, meaning more contact with the heatsink, better cooled.

 

I wonder if any of you guys can shed some light on what's happening with this card with Crysis 2 for me. I played a quick runthrough of one of the beginning levels and the game plays and runs very smoothly, around 40+fps with everything maxed out. This game has been crashing frequently and when I restarted the FPS of the same section seemed to lock up at around 30 fps max, dipping down to mid 20's. When I check with MSI afterburner the temp is around 60c but the GPU usage is never above 80%?

 

if u got an r4, try to switch off enduro by using Fn+F7+reboot for gaming,that should help the fps.

makes sense,same issue with IB,where ppl initially thought that the higher temps were due to the TIM paste intel was now using between the chip and the heatspreader,but actually turned out to be more accumulated heat due to more transistors on same area --> less cooling capacity. hope that trend doesnt continue with the next shrink....

Sent from my GT-I9001 using Tapatalk 2

 

Here is a quote from a guy who knows his stuff in this forum. He thinks the 680M from this picture, is 75-85W. And that the P4900 (if true) could be higher with the final sample and the right drivers. Looks like we might see a 100W 685M after all...

JUST what I was looking for, a 75W GPU. That is the maximum for me
All the resellers in this forum knows that this GPU is coming this "summer" btw. No solid date but atleast we know its pretty near.

 

funny thing was nvidia expected amd to release a 75w 7970m then they instead released a monster 100w card and the 680m being the flagship card, nvidia must have panicked and I reckon now the 680m will be 5% better then the 7970m and nvidia will say they win. No wonder why the kepler gpu's look mid range with the gtx 660m.