AVADirect Clevo P170EM Part 2: GTX 680M Grudge Match
by Jarred Walton on October 15, 2012 6:50 PM ESTClevo P170EM GTX 680M Gaming Performance
We don’t have any major qualifications to go over before we hit the GTX 680M benchmarks, but we did end up running our gaming suite multiple times. The initial shipping driver was a 295 Series NVIDIA build, which is now getting somewhat old. We tested our core gaming suite with those drivers before updating to the 306.23 drivers, and then more recently the 306.97 drivers (there was almost no change in performance between the .23 and .96 drivers, if you’re wondering). Unlike AMD’s Hotfix driver, NVIDIA’s 306.xx driver update is more of a mixed bag—a few games improve in performance, but several titles in our test suite actually drop a bit. Of course, the GTX 680M already had a commanding lead, so losing a few FPS isn’t likely to hurt too much. Here’s the list of notebooks we’ve lined up for the graphs this time (and note that all of the laptop names link to the specific reviews):
| Notebook Configuration Overview | ||||
| Laptop | CPU | Graphics | Storage | Battery |
| AlienwareM17x R4 | Intel i7-3720QM | GTX680M/HD4000 | Hybrid (Intel SRT) | 90Wh |
| AlienwareM18x R2 | Intel i7-3820QM | GTX680M-SLI/HD4000 | SSD RAID | 97Wh |
| ASUS G74SX-A2 | Intel i7-2630QM | GTX560M/HD3000 | SSD | 90Wh |
| Clevo P170EM GTX 680M | Intel i7-3720QM | GTX680M/HD4000 | SSD | 77Wh |
| Clevo P170EM HD 7970M | Intel i7-3720QM | HD7970M/HD4000 | SSD | 77Wh |
| iBUYPOWERCZ-17(MSI GT70) | Intel i7-3610QM | GTX675M/HD4000 | SSD | 60Wh |
| Razer Blade 2012 | Intel i7-3632QM | GTX660M/HD4000 | Hybrid (Intel SRT) | 60Wh |
| Samsung Series 7 | Intel i7-3615QM | GT650M/HD4000 | Hybrid (ExpressCache) | 77Wh |
This is a grudge match between AMD and NVIDIA, and now that we’re sporting the same hardware the only thing separating us from the mobile gaming performance crown is a few pages of benchmarks. We’re going to skip straight to the stuff that presumably everyone is here to see: gaming performance. If you want to see how the P170EM GTX 680M stacks up against the HD 7970M in general applications, we reran all of our general application tests and basically came up with a tie, so we’re going to skip those this time around (though they’re available in Mobile Bench).
Mainstream 1600x900 Gaming
Right out of the gates, NVIDIA’s GTX 680M takes a commanding lead in our Mainstream testing (1600x900 with ~High details). The margin of victory is as much as 58% (Civilization V), though we do have a couple of games that are basically tied (Skyrim goes to the HD 7970M by a couple percent while Shogun 2 favors GTX 680M by just over 1%). With our less demanding settings, NVIDIA leads by an average of 15%—right about what you would want from an 11-15% increase in overall notebook cost. But then, it’s unlikely anyone would plunk down roughly two grand (give or take) just to run at moderately high settings and 1600x900; let’s see what happens when we turn the settings dial to 11.
Enthusiast 1920x1080 Gaming
The gap at our Mainstream settings was moderately close, but at Enthusiast detail we end up with several serious blowouts, and unfortunately there’s not a single bright spot for AMD. Batman is a 38% lead for NVIDIA, Civilization V is 33%, and Portal 2 is 30%; those are balanced by a tie in DiRT 3, a scant 7% lead in Skyrim, and an 11% lead in Battlefield 3. Incidentally, Alienware’s M17x R4 shows generally similar performance, with some variations likely caused by differences in the drivers and other components.
As for Shogun 2, unfortunately there’s a driver bug of sorts that prevents us from even testing out the “Very High” preset with AMD’s 7970M—basically, the game looks at the capabilities of the Intel HD 4000 iGPU, which doesn’t allow a few key selections, even though the game is actually supposed to be looking at the HD 7970M. We did run 1080p with identical “nearly max” settings on both GPUs, however, and at least at those settings we wind up with a blowout for NVIDIA: 46.2 FPS compared to 31.3 FPS.
NVIDIA basically takes a clean sweep of our regular gaming benchmarks, but we’re not done with the gaming discussion just yet. We decided to look at some other more recent releases to try to better characterize gaming performance.
Facebook
Twitter
RSS
58 Comments
View All Comments
Harmattan - Wednesday, October 17, 2012 - link
I've now owned a p150hm with a 7970m, an m17x with a 7970m and 680m, an m18x with dual 7970m and another (my current) m18x with dual 680ms. After having spent many hours with each config, , my initial assertions stand...-if you're going single card and Clevo, get the 680m: although the drivers are almost comparable between the two cards, Enduro power management (and the inability to turn it off) is a whole heap of wrong.
-If you're going single card and AW, get the 7970m: since Enduro issues are not there, the price increase of the 680m is not justified.
-If you're going dual cards on either Clevo or AW, go 680m. 7970 CF drivers are a mess in many games and I experienced many issues (even after replaced cards)
All in all, I am happier and had less problems with the 680m on all configs. When the 7970m ran well, it is comparable, but that is rare.
The above points are just based on pure gaming performance at stock. If you're OC-ing, compare the desktop performance of a 7870 vs. GTX 670 - both cards are beasts, but the GTX 670 is +/-20% faster in most situations (and runs cooler).
TokamakH3 - Wednesday, October 17, 2012 - link
Would that fly in a desktop video card review? "Well, the Nvidia card is $574 while the AMD card is $300, but since the desktop you'll put it in will be $1975 vs $2249, there's only a 14% price difference, so we'll only look for a 14% performance difference"
That really makes no sense at all.
TokamakH3 - Wednesday, October 17, 2012 - link
Haha, dead horse. Should read before commenting, would delete the post if I could.cartmanasan - Thursday, October 18, 2012 - link
Don't know why no one covers professional applications like Maya, 3DSM etc. nVidia has got worse performance than their own previous cards. Have seen this going around in many forums. Gaming notebooks are well placed to be workstation/DTRs. No Mention of GPU Boost technology of nVidia as well. A comparison of overclocking would have been better too.transphasic - Saturday, October 20, 2012 - link
The reason AMD GPU's score far better than the Nvidia offerings this time around on business applications, is due to Nvidia's new Kepler-based designs are more gaming-oriented in it's focus, as opposed to the older Fermi-based chips, so that is the reason why, from what I have seen as an explanation from NBR.com for AMD to win this particular battle.Kepler-based GPU's are more for gaming performance, but the AMD fares better on business-related programs and applications.
tential - Monday, November 19, 2012 - link
I think I'm a little late to the party but the reviewers opinion in the Conclusion is just that. An OPINION. I usually skip to the conclusion pretty fast because I'd like to hear their opinion. The raw numbers is usually not enough to make a decision. If so, I'd had said "The 680M is faster but I think the 7970 is good enough for me." Now after reading further and factoring in driver support as well as the other issues with AMD, I think it'd be worthwhile to spend the extra money.I understand the gripe of using the GFX performance as the only thing that matters when considering the upgrade however but I don't think that's what the reviewer was doing. A laptop is used for many other things other than just gaming. Even a gaming laptop is. Hell, my gaming laptop is currently not even used for much gaming as I do a lot of other things on it. I think it's important for people to take the reviewers opinion in context though and to know that you have to make your own opinion. It's obvious that if you do a lot of other things other than gaming than maybe you can't justify the upgrade. However, if you primarily use it for gaming maybe you can. I think an update article in a couple of months would be nice though.
Also:
IMO, some things I'd like Anandtech to start including though would be an adjusted FPS page. This page would should FPS but would weight FPS above 60 lower. I really don't think there is a point in saying "Wow this card runs a game at 120 fps and this card runs it at 100 so the first card is what you should get!" Why not get the card that runs it at 100 fps when all I need is 60 FPS anyway right? I know this can be used to say that a card may run future games better but sometimes I'd like to know just about current games.
Hrel - Wednesday, January 2, 2013 - link
How did you measure GPU utilization? As far as I know FRAPS and the like only give FPS and admin tools don't have an option to add GPU to the list. So how did you do it? What did you use?Maybe it's in the article somewhere, but I'm at work and can't comb through the whole thing.
Hrel - Wednesday, January 2, 2013 - link
Every time you talk about what a hassel it is to carry around a laptop I imagine a little hobbit stuggling to pick up something (a large watermelon) just to have Argorn come and pick it up with one hand. It's such a none issue for me I would never even think to talk about it. Yet you go on like anything but an ultrabook just isn't portable. Lol. No point really, it just amuses me.My laptop bag is generally 30+ lbs and I never have any kind of issue carrying it around all day at a convention or airport or whatever. I'm curious if you're in the minority on this or I am. I mean, does it really matter if the thing weighs 6lbs or 11lbs? I can barely even tell the difference.