AMD 7970m vs GTX 680m

Awwwww.... sidewinDr, you beat me to it. =(

 

Ninja!

 

You've both messed up, because those are desktop 7970s, worthless to a laptop discussion and this thread/forum altogether.

 

Touche. Need to lurk more I guess.

 

Seems like some benchmarks are out now:
NVIDIA GeForce GTX 680M - Notebookcheck.com Technik/FAQ

What a disappoinment.

 

The GTX 680M is tested.
Real bad results compared to HD 7970M!
The HD 7970M is faster and cheaper!!!

 

I just hope those are not the oficial benchs or... it is a fail...

RogerCD beat me to it... I was just comparing to see if those were not from another card... just hope lol

 

Majority of games are 1 or 2% difference.

Performance per $ fail for nvidia.

 

On notebookcheck.net they say 680M is DX11.1 & 7970M is DX11.
What is the difference.
And from the looks of the performance charts from NBC.net 7970m is doing better than 680m Including BF3, Crysis 2 MaxPayne 3 & Skyrim etc
NVIDIA GeForce GTX 680M - Notebookcheck.net Tech
I really hope AMD make the final sprint on the drivers and win the race

+1 Jen-Hsun Huang for your balls of steel to ask $300 extra over 7970m.



P.S Is it just me that has a bad feeling AMD is gonna raise the price of their 7970m because they beat Nvidia's 680m?

 

OMG Nvidia got served!! Glad that I picked the red flag. Too bad the 7970m has a lot of driver issues...

 

The gaming performance seems really close for both the cards based on Notebook check's review. But the strange thing is Max Payne 3 where the 7970m supposedly delivers 61.2 to 42 from the 680m. 19fps is huge. Well I am guessing drivers.

Well for me it has always been the 7970m right from the moment its price was known and more importantly its performance was revealed. I never had the extra $$ to get a SSD where the hell am I going to get the 300$ extra for a Nvidia .

As for now the 7970m will do for me. People who can afford shall buy the Nvidia as it does offer CUDA, 3D vision and Physix if at all that means something to them. Well apart from that I dont see much to separate them but may be the ones buying it might explain. Good Luck to all.

 

There's no way those numbers are correct.

46% slower in max payne 3?
8% faster in Dirt showdown
27% faster in diablo 3
4% slower in skyrim
3% slower in bf3 ??? lol
3% faster in arkham city
15% slower in crysis 2 (this one stumps me)
21% slower in starcraft 2
11% slower in metro 2033

8% slower in unigine lolz....

for reference 7970 NON GHZ edition vs. 680 vanilla = same framerate in max payne 3.

 

There is no official driver for the GTX 680m at the NVIDIA site, maybe we still need to wait for the official driver to be released to see the true performance of the card.

 

That's what everyone is saying about the 7970M...

 

But these are latest benchmarks right? Well to discredit them just because they are not up to our expectations is not fair now is it?

Well lets wait for other reviews to show up as well then we might have concrete data to work with.

And what does the comparison of those desktop cards prove here? Neither of these mobile cards are based on those desktop cards? Hence what does their performance signify here?

 

It means Kepler "underscales". In other words, the way Kepler was made, when the clocks are "too low" the entire architecture falls apart and its performance is crippled.

It must be a driver issue.

 

Yeah both 7970m & 680m don't have official drivers. With the reputation of Nvidia's good drivers and AMD's dodgy drivers 7970m still performs better @ stock clocks. Remember 680m has GPU Boost so there might be a lil OC in 680m's results which widens the gap even more on 7970m vs 680m.

Or maybe we should wait who gets the last laugh. As NBC.net's results might not be accurate.

 

But if it is as you say and the architecture design is such that the performance curve does not reach optimum levels on lower clocks then I doubt they can settle this with driver updates. Any ways, all this discussion is just for the fun of it and unless we have solid proof we have noting to conclude. I guess this thread will continue on speculation for some more while (until proper drivers are here or until both the companies come up with a new card ).

 

No 7990M ahead. The .inf of the actual AMD beta driver shows 8800M.

 

It's funny because the 7970 Ghz edition is faster than the 680 and the 7970M is looking to be faster than the 680M....

Hahahhhhhhhhaaa. *Looks @ his 7970M CF M18x R2*.

AMD back in the lead?

 

Aahh So you have a CF 7970m

How are the drivers holding up? I heard the drivers are terrible for the CF. Its no surprise when the 7970m single GPU in itself is struggling to a certain extent.

 

I'm getting rid of it because of the drivers. My plan was to go 680M SLI the second I heard the 680M has 1344 CUDA cores. Now that I see the benchmarks I don't know what to do. I'll probably stick to my Aurora R4.

 

i wouldnt trust those benches from notebook check, they dont even show the specs of the machine they tested it with, so how can we know that they are correct.

 

Zero
I would wait for the oficial 7970m drivers if i were you intead of junping ships this soon.

 

aduy check again, they do show the specs.

Schenker XMG P702 PRO

3720QM

8192 MB

720 / 1800MHz

4096MB GDDR5, Optimus

 

nvm

10chars

 

The CF drivers generally take a longer time to settle down than the single GPU solutions. At least the fact that these cards wont suffer micro-stuttering is a good thing as there are not any games that would push the CF to deliver around 30fps or less.

 

they can still suffer from Micro stutering
since micro stutering isent dependent on fps
you can se it more easy on low fps but they can still be there
my m8 got cf 7970 he is toping out most games he still complain about MS
if im not wrong
the MS come from Frame timing not frame per second
its when one frame pops up slower then a nother
the higher fps you get the the faster the slow frame pops up but its still slower you would still get MS at 200 fps but not as wisable as at 30

 

Of course micro stuttering is independent of fps and I did mean it in the way that the lower it gets the more visible it becomes. People generally dont play below 30 and hence the border line games will be the ones to be affected the most. Hence I chose 30 fps

 

very interesting results from notebookcheck why not trust them? theyve always been accurate so far when it came to mobile gpu performance in games!

all in all, it all comes down to: 680M being 4% slower than 7970M across 17 gaming benchmarks at maximum settings. enough reason for me to go through with my 7970M order

cheers

 

Miss understud you then sorry for that you whant to have it at 60 above 60 the ms is non segnificant below 60 then it starts being unplayable whit MS
but thats my poit on the ms every body have lower or higher censitivnes to MS

 

I will wait for some user reviews first but... it looks like I will go with the 7970M as well.

This is the end for Nvidia. It was nice knowing you guys.

 

Looks like an epic fail from NVIDIA...

 

Do you expect them to release everything?

It's all competition, if AMD gives none, why should nVIDIA make another product when they already got one out there..

 

The card is fine, but the pricing sucks, especially with the 7970m in the wings

 

Well i am happy with the results it means i made a good choice in the end.

If i had bought the 680m i would not be unhappy since 4% slower is not much .

But the 15% faster then the 7970m predictions LOL

Having the top card even if it just for 6 months feels good

It is close to what people feel having a machine or motor they think is top line.

Lol ego

 

The price looks just ridiculous comparing to 7970. How can they charge so much more if the performance is at the same (or even worse) level?

Oh, I forgot about that nvidia have Physx and Cuda features. That's why it costs more. (Sarcasm)

 

Well, that's the thing. If Nvidia pricing was identical to that of AMD, then would be fine. 4% difference is very small and could be offset by other factors, like better driver support, PhysX and so on. Making the two cards pretty much equal.

But plus $300 for what?

There is still hope if they drop the price to match that of the 7970M.

 

Yeah, nVidia is being idiotically overconfident at this point, if they except any sales atleast from the informed/knowledgeable part of the market... I'm so glad now that I got the 7970M! And LOL @ the 15% higher performance for 680M predictions crap!

 

It all rests on how ocing goes.

 

That's disappointing. But are there still issues with the 7970m as far as hiccups and overheating? I really want to get a Sager NP9150 but some of the temperatures and niggling issues scare me.

I think nVidia has better driver support, but not $300 better. And yes, overclocking is still a question.

 

Kinda shows how this thread is really swamped by amd fanboys. Why has no one posted this yet?

http://www.digitaltrends.com/laptop-reviews/alienware-m17x-r4-review/

Looks like 680M pwnage to me in terms of performance. The price, however, is an entirely different story.

 

Amd will win on OC.

Also to the above poster, just look at notebookcheck, most games 7970m is faster then the 680m with a few being slower.

 

Well either the Notebookcheck test had some serious flaw or Nvidia have outright lied about the performance. I calculated all available scores and found the difference between the GPUs.

GTX 680M got less FPS even in Battlefield 3 and only slightly better in Batman. These games are traditionally where Nvidia shine with their GPUs.

Nvidia promised up to 30% better performance than 7970M or 15-20% better performance in Average over 7970M.

Call me deeply disappointed

 

7970m being 2% faster then the 680m, that is pretty much a tie in my eyes.(We would need many more games tested to verify who is faster in a such small percentage).

Edit: isn't it about 40% slower in Max Payne ?

Edit2:Funny thing is that Notebookcheck thinks 680m is faster on average...

 

thats just cuz they havent updated the benchmarklist yet according to those newest results.

btw,didnt u forget a game cloud? i remember there being a total 17 not 16 games tested...

and yeah max payne 3 should say 31% slower for 680m or 45% faster for 7970m

Sent from my GT-I9001 using Tapatalk 2

 

That was Mass Effect 3, which doesn`t go higher than 60FPS so there is no point including it

 

Guys c'mon something isn't right. There's no way the 680m gets 43fps on ultra with MP3.
A desktop gtx570 can max out the game easily above 80fps.
I smell Optimus issues. You know the problems with enduro and hd7970m on Clevo machines? Same issue.
I want to see a review of an Alienware machine with igpu disabled.

 

Yes it looks that something is wrong with the scores

They are definitely far from Nvidia`s own benchmarks
According to them, 680M should score 20% better in Batman (we saw 4%), 20% in Crysis 2 (we saw 15% worse), 30% better in Dirt 3, 8% better in Battlefield (we saw 4% worse)...

 

Just go with whatever you like. I prefer to support open standards and Nvidia does their hardest and spend insane amounts of their budget on marketing to promote their proprietary tech. That's how I see it.

Who cares if it gets slightly higher in a synthetic benchmark. I certainly don't care if I get 50 or 70 FPS in a single player game... Just stupidity to agonize over that.