News about the upcoming GTX 680M

I may give it one quick test. Don't expect benchmarks or anything. I'll stick it in, boot it up, install some drivers, check fans, make sure it gets to the full clocks while moderately stressed and I'll report back with a GPU-z screenshot and graphs for those who want to check temps.

 

That would be perfect. Thanks.

 

Didn't even realize how new the GTX 680m was when I made my order, ETA (or should I say EDD, estimated delivery date) July 12th. Living in California has unexpected perks.

 

Now my system comes with the base 3610QM, should I wait for the 3960XM or go 3920XM? I'm putting my 2920XM up for sale instead of what I initially planned.

Looking at Intel's roadmap, there is litte evidence that there even will be any 3960XM at all seeing as the name is so close to their 6-core flagship. I would love to have something like a 3930KM

 

3940xm is the likely refresh.....see the spoiler - intel website with footnotes on the hyperlink but I couldnt find any clear details....

Spoiler

 

Oooh how interesting. Now the waiting game is afoot...

 

I'm pretty sure the 3940XM will come soon... it already showed up in a benchmark which a German Clevo reseller recently posted, though I think no one really noticed the CPU

 

Personally I'm waiting for a 6 core mobile CPU haha

 

Mine is in production now and hoping it ships sometimes this week.

 

Oh wow! Hopefully I can score one soon. Getting some extra dough in August to splash out!

Looking around, I may be the only person to own a R4 with WiHD as I have my own module that I can install.

 

Sorry about the small pictures. Some detection problems with 3DMark (Its the 680M) inside a MSI (Firebat)





Extreme tessellation, 1080p with 4xAA and 16x Antistrophy

 

Are those stock or...?

 

Yes they are from what I can gather

I still don`t get why Notebookcheck found out that the 7970M is performing better in games, especially when looking at the extreme preset which is 1080p
1870 vs 1636 in GPU score @ Extreme. Thats a 14% difference

http://www.eurogamer.net/articles/df-hardware-radeon-7970m-alienware-m17x-r4-review

 

Cool. Obviously the CPU is helping slightly as well

 

But not with GPU score right?

 

Well maybe by one or two points from my experience but not that much. Good to see some numbers flowing in.

 

So what is your take/explanation why 14% better score at 1080p with 4XMSAA result in that 7970M get 12% more FPS in Battlefield 3

Review GeForce GTX 680M vs. Radeon HD 7970M - Notebookcheck.net Reviews

I don`t get it

 

Well I just leave that down to either bad reviewing or just because game engines are no longer about raw GPU power and hardware architecture now affects performance. I remember before some game engines would absolutely shine on one team but be useless on another team despite have similar computing power. I'm not getting too worked up about it. I got the 680m because it was offered to me for free and because I wanted the 3D.

 

For free?

Maybe you are right, maybe its the architecture although Kepler don`t seem to suffer with desktop models. I am just trying to learn and understand.I guess we will see what is what when the next review hits us or when someone from this forum recieves it. I know several are recieving the 680M the following days

 

Not really for free. As in a replacement to my current system. I wouldn't dream of spending that much money myself.

 

what is the range of FPS do you get on bf3 maxed out on Multiplayer maps? then compare it to notebookcheck's "average"

 

I believe the CPU speed has more to do with a gpu score than most people think. I can run my desktop with its stock cpu and gpu clocks and then OC the cpu to 4.65GHz keeping the gpu clocks stock and there will be a significant diff in GPU score. To think differently is a bit odd to me because the cpu feeds the gpu, I wouldn't know how 3D Mark11 could discount that and score gpu only.

Since widezu69 has a 3610qm and 680m coming on a M17x, I am very much looking forward to his results to compare to my 3610qm and 7970m M17x. Keep in mind, there should be very different Combined scores to.

@widezu69, please run some tests on that 3610qm before you switch out with that smorgasboard of cpu choices you have Congrates on the system and also to you Cloudfire.

 

watched ur videos of 7970m on youtube. can you add a bf3 multiplayer gameplay? with fps showing?

 

@widezu69- also very curious to see how the upgraded CPU changes your performance- pre and post swapping- might be time for an upgrade who knows.

 

Thanks. Did this morning, last 8 of this 12 min video is MP Caspian Border Conquest (51/64). OSD is there to view and it was recorded with a cam, so no screen recorder drawbacks or excuses, what you see is what I got.

http://forum.notebookreview.com/alienware/670572-all-alienware-laptop-gaming-videos-thread.html



Here is my example of how much a cpu speed can affect a gpu score on 3D Mark11. It's my desktop since I can't OC the 3610qm, but the principle is the same, all we need is people to test keeping their gpu clocks the same and only changing cpu clocks from lowest to highest.

Stock GPU clocks 1123MHz/1110MHz
4GHz CPU: GPU score 18533
4.65GHz: GPU score 18908 +2%

OC GPU clocks 1247/1233
4GHz CPU: GPU score 20640
4.65GHz: GPU score 21001 +1.75%

Of course scores can vary a little each way test to test on 3DM11, but here we're seeing a 2% increase in gpu score by taking an already beefy 4GHz 980x which would feed SLi gpus nicely and still feeding them more by going to 4.65. Curiosity is getting to me to switch my OC profile in the BIOS and see what the percentage would be if I was only 3.4GHz vs 4.65, because 3.4 doesn't feed SLi as well as 4, so we may see that percentage difference really stretch beyond linear.

Now on our lappies, me with a quad 3.1GHz and someone with a 3720 or 3820qm OCing to 4GHz, well we may have some significant difference to keep in mind. So an M17x-R4 3610qm/7970m vs 3610qm/680m would be a reasonable comparison to actual gpu vs gpu and not letting the cpu skew the competition. Of course a 4GHz 3720qm/7970m vs 4GHz 3720qm/680m would also be perfectly valid.

Man I love math. How this 3DM11 score translates into games will be up to the game. For example, I see Dirt Showdown averaging say 25% cpu usage, whereas BF3 gets closer to 50% cpu usage. So if a game relies on more cpu usage, the faster cpu you have to feed your hungry gpu, the tastier your gpus will find the binary

 

Only thing I can add to this conversation regarding 3dmark11 scores and CPU/GPU usage is that I read recently. That the physics tests are run on the CPU only, even on NV cards. With CPU's getting so powerful these days Physx-on-card is becoming less of an advantage anyway, and having a more powerful CPU will affect the overall benchmark a fair bit?

As Q says, it's in games FPS where the real data shows.

 

so how much would the 3720qm give FPS boost over the 3610qm? barely 5 fps?

 

I don't have a clue . Not played with CPU's (yet!). I'm guessing but in-game that sounds right.

 

I highly, highly doubt that even from the 3610qm to the 3920xm you will see a boots in gaming performance as our GPUs are already a bottleneck and modern day CPUs are hardly stressed by gaming.

If all you want is a higher number after you run a bench mark... And that 2 seconds of glory is worth the cheddar - go for it.

From what i see these CPUs are developed for professionals - and only professionals with professional programs will see a difference between them.

All in all I regret spending the extra 150$ for the bump to the 3720qm.

 

since widezu got the 3620 and my system the 3720, i guess we can try to compare if theres a difference at all (rest of the system should be identical as far as i can remember).

OT: in production

 

There is an easier way to compare.

Tge person with a 3720qm just downclock your multipliers to the clocks of a 3610qm and run a non multiplayer game.

That way you will see the same image in the same location with the most accurate frame rate.

Which i did to test my overclock - i tested my cpu at stock speed, then overclocked to 4.0ghz - and saw an amazing 0% boost in framr rate running skyrim and metro!

 

Review of the Alienware M17x R4 with 680M is up

Alienware M17x R4 (2012), Ivy Bridge and Kepler Refresh - HotHardware

 

Yup, I can see this as the outcome. As I mentioned, just because 3DM11 gpu score can be affected by cpu speed, doesn't mean it will translate to gaming. I was pointing out what I've read so much when ppl talk about the gpu score not being affected by cpu spd or next to nothing. It can be a significant factor and if one is trying to gauge gpu vs gpu using 3DM11, the cpu has to be taken into account as the gpu score alone cannot discount it. Sure the gpu score is a tighter representation of a gpu compared to the P score, but it is affected by a cpu.

I almost went for the 3720qm because I use some of those programs, but I wasn't considering it for games. AllNAll I just felt the 3610qm was enough and pocketed the $. Benchmarkers however, will def want the OC ability that the 3610qm doesn't have.

Here you tested the principal and found bupkiss on 2 games, although more games tested would be a good idea, maybe more cpu reliant games. I myself find a difference in gaming on my desktop running 3.4GHz than my usual 4GHz. So it all depends on the system. Based on your test, we can go forward with the idea that a 3610qm is plenty to feed a single 7970m and likely a 680m, but what if someone has an M18x, they may find the scenario more like my desktop where increased cpu speed will feed SLi/Xfire better.

The joys of hardware, we know we love it!

Good link, thanks. That Cinebench 11.5 cpu test is a nice little representation of a cpu for imaging and such, they're getting 6.79 and the 3610 gets around 6.25. Not exactly $150 worth to me, but I've always found that bench consistent and true to what I do with full cpu loads.

The P6133 is honestly about what I would expect from a 3720/680m, I'd be guessing what the diff is if it was a 3610/680m since it's only a 0.5GHz cpu diff, but I'm sure it's something, <0.75% likely. Kepler is more geared to the gaming spectrum instead of compute, so it should clearly take the edge in 3DM11, but I would like to see the Combined test score.

Anyway, all this info is coming as users get their hands on the orders placed. I consider this top end gpu round a win for everyone as each camp has picked a side to direct their hardware towards. Nv likely has the game edge + 3D, AMD has the compute win and not a slacker in gaming.

Review is a bit of a fail for not having a 3720qm/7970m to test though.

 

hey wizedu, did you try 680m on your m17x r3?

 

Dude I haven't even got the R4 yet

 

"hackness" back in the Sager forum have recieved his 680M.

http://forum.notebookreview.com/sager-clevo/673583-p150em-upgrading-675m-680m-4.html

New review of the 680M.

http://forum.notebookreview.com/gaming-software-graphics-cards/675454-new-gtx-680m-review-truth.html