My Nvidia GTX 880M Test Run Review

That looks sweet brother Fox, glad everything worked for you. The Amazon seller (TigerDirect) offered a refund even when I told them the unit was partially working. I told them I'd be happier with a replacement unit so a second one is on the way.

Also finished the bottom cover mod (cutting holes), and combined with the U3 fan mod, I managed a 7-10C drop in temps across the CPU and both GPUs. Easily the best $100 I've spent this entire year. Will post some pictures and detailed results later.

EDIT: Pics and results up!

You know one thing that never ceases to amaze me is just how perfect of a fit the Rosewell 120mm fans are for the U3. I mean, the screw diameter is exactly right, and the placement of the screw holes on the fans are also exactly right and not a single mm off, AND there's just enough space to fit 3 on the U3 and with enough ground clearance. I mean, it's almost as if the U3 was designed with those Rosewell fans in mind, but CoolerMaster decided at the very last second people would prefer "quiet" over "cool", so yanked those awesome Rosewell fans and put in those dinky, borderline useless 80mm fans.

 

Hey there,

is anyone with an Alienware17 -system able to give me the DELL OEM Hardware ID?
I tried a google-search but the results are not really satisfying -.-

(it basically should be one of those 4 shown in the picture: )

Thanks

 

I think it may be the 05AA. The 05AB is the Alienware 18 hardware ID, so you can scratch that one off your list of possibilities.

 

Hello people
I'm one of the unfortunate guys who thought it'd be in for a treat by getting the 880m to run all my games at MAX and enjoy a high FPS... Unfortunately I didn't research enough and now I'm stuck with it.

I got a question about temps as I tried running a few game benchmarks and under max load I go as high as 89 C. My laptop is fairly new (2 months), no dust, there is 3 cm space between my table. Room temp was about 20/25C max.

FF14 RR: 1980x1080 at Max settings gave me: 85 to 88 C
FF14 Legacy: 1980x1080 : 89C
Re6 at max with vsync off : 84/85 C
RE5 Max DX10 ; 75/80 C
I have yet to run 3DMark (still got issues due to some errors)

Info: Drivers 337.83, stock clock, maker vbios.

Thank you and let me know if you need more details

 

What laptop is it?

 

Added in the post and sig.
(I'm in Japan too!)

 

Looks like a rebranded MSI GT70.

 

Cool fella, where abouts in Japan are you?

As for your temps, I have a feeling those temps may unfortunately be close to the norm in that machine. Can you manage to run a 3DMark 11 benchmark and give us the temps?

 

Hi guys
Yes it seems my laptop is exactly a rebranded MSI .. I ran 3D Mark 11 and the results online stated "MSI" as the manufacturer. So much for the "Japanese-made".

I hit 91 C on the benchmark but it was hovering between 85-91. I can post a log when I'm home if needed.

I'm based in Tokyo, been living here for 5 years now

 

Sexy, thanks!

 

Sooo... I am browsing around at the 3DMark site and decide to look up GTX 880M single card rankings. Of course, Johnksss is at the top on both Firestrike and 3DMark 11 (by a wide margin I might add).

Looky here at the 880M single card Firestrike scores.

GTX 880m Firestrike - Search

Filter the GPU to "1" and look at #4. Yep, that's me.

But... I am #27 on 3DMark 11 scores. I have a higher overclock on the GPU than a lot of the others. Anybody want to clue me in? I am thinking it might be the power limitation of the 230w brick is causing my CPU to back down? I have a 330w brick that will be here on Monday.
GTX 880M 3DMark 11 - Search

 

The answer is pretty simple. Your CPU doesn't have enough juice to achieve good results in 3DMark11. It's absolutely impossible to get a high overall score in 3DMark11 without a very potent CPU. Same is true with Vantage. These are true overclocker benchmarks, but Fire Strike is more of a gamer's toy benchmark.

You do not need to have a powerful CPU to get nice Fire Strike scores, which really diminishes the value of Fire Strike as a test of overall system performance. It's really nothing but a GPU test. Fire Strike scores can turn out nice with a really wimpy CPU. For example, even an old laptop with a 920XM CPU running at 3.0GHz that has been upgraded with a modern GPU can achieve nice Fire Strike scores. As another example, Severna's P570WM with the hex core 4960K get's an insane Fire Strike physics score, but it is almost completely ignored and his overall scores are just barely (a few hundred points or less) any better than my Fire Strike overall scores with a quad-core mobile CPU.

 

I see. So I would need to overclock my 4810QM to get a better 3DMark 11 score. Man, that benchmark really pushes the hardware. With the max overclock I can currently achieve with my GXT 880M I am pulling 287w at the wall. I which case, I think really do need a 330w brick to overclock both the GPU and the CPU.

Thanks for the reply dude. :thumbsup:

 

Yeah, you would need a physics score like the one John had in 3DMark11 to get an overall score remotely close to the same. That's why 3DMark11 and Vantage are so awesome... they test the entire system very thoroughly, not just the GPUs.

Fire Strike will give you a good view of the physics score if you look at that all by itself, but for some stupid reason they give it hardly any weight in calculating the overall score. That's really unfortunate, and it is also misleading. It makes a weaker system with a wicked GPU seem almost as good as one that is a monster in every way imaginable.

Here's a side-by-side example of what I was referring to. You will see a gigantic difference in physics scores and hardly any difference in graphics and overall score between my M18xR2 and Severna's P570WM. His GPUs are overclocked higher, so there is a small difference there, but Fire Strike basically ignores his far superior 4930K physics score.

Compare: Fire Strike Benchmark - M18xR2 3920XM and 780M SLI versus P570WM 4930K and 880M SLI

That's why I say Fire Strike scores are misleading and it is not really a good benchmark to use when comparing one system against another. If you do so, you must ignore the overall score for the results to be of any value.

 

Then again most games tend to be mostly GPU bound so extreme is exactly more what it aims to be, a representation of AAA 1080p high settings performance.

 

All this talk (and a few private messages later) got me thinking about the 230w power brick shipped with my NP8178 (P170SM-A) . Here goes...

On my stock Chicony power brick the AC input is rated at 100-240v at 3.5a 50-60Hz. The DC output is reated at 19.5v at 11.8a, 230w. The input side is a bit dubious because the current can't be the same at all voltages. I have no idea where they came up with the 3.5a rating, nor am I positive how the 50Hz vs. 60Hz affects power ratings as I haven't researched it. The important things to know are the rated output and the efficiency. The output rating of 230w lines up perfectly with the stated voltage of 19.5vdc and 11.8a. (19.5 x 11.8 = 230.1)

I looked at the label of my power brick and it has an Efficiency Level V.
According to this web site, that means my ac adapter is 87% efficient. Maybe... but lets look at the numbers.
Notebook and other small devices power supply efficiency | www.qianqin.de

Taking the input power at my house (120vac) and dividing it by the output power (19.5vdc) we get 6.15. The inverse of this is which is .1625. Take this times 11.8a and we get 1.9175. 1.9175 times 120 = 230w (or roughly 1.9175a at the wall) which which is what we would pull from the wall if the power brick was 100% efficient. But over course it isn't, so the designers had to allow for the inefficiency of not only converting from AC to DC but stepping down the voltage as well.

At 87% efficiency the max I should be be pulling from the wall with a 230w power brick is 264w. (the inverse of .87 is 1/.87 = 1.149 and 1.149 x 230 = 264) I have been pulling 278w so I am exceeding the rating of my power brick if it really is 87% efficient. Assuming that I was limited to 230w output then my power brick is really only about 83% efficient when pulling 278w out of the wall... but I don't really know.

 

Ratings must offer a maximum, the max current draw the device will pull is at 100V most likely and it can over a short period likely draw 3.5A at that.

It wont be fully efficient at 100% load, a loading graph looks like this:



(From anandtech's PSU review here: AnandTech | SilverStone NightJar NJ520 Power Supply Review)

And that curve is exceptional from a PSU that is over built.

I would love to have the equipment to characterise all the major PSUs at present.

I would guess your PSU is peak 87% efficient with a 4% roll off to 83% at max load and that would mean 278W = 230W supplied (Wow I was surprised how well that actually worked out).

 

I got my 330w power brick the other day.. That allowed me to overclock the GPU and the CPU.

3DMark 11: 10013 This puts me at #2 behind Johnksss
NVIDIA GeForce GTX 880M video card benchmark result - Intel Core i7 4900MQ,Notebook P17SM-A
Search GeForce GTX 880M

Firestrike: 6620 I am #4 on this one.
NVIDIA GeForce GTX 880M video card benchmark result - Intel Core i7 4900MQ,Notebook P17SM-A
Search GeForce GTX 880M

Firestrike Extreme: 3332 #2 behind Johnksss
NVIDIA GeForce GTX 880M video card benchmark result - Intel Core i7 4900MQ,Notebook P17SM-A
Search GeForce GTX 880M

 

I'd like to see the AW benchmarks thread open up to Clevo /Sager. Or perhaps a new official benchmarking thread to accomodate all makes.
If you guys are interested I might see if its possible when Brother Fox gets back.

 

Could be fun to organise something like that, get people competing

 

That's called hwbot and has been there for ages.

 

Lol I meant here but I see your point.

 

Good lord some of those scores are crazy. Of course it's desktop stuff...

 

Clevo owners have been posting in the Alienware threads the whole time. I don't think any 880M thread even was started in the Clevo forum lol

Sent from my HTC One_M8 using Tapatalk

 

Yes, but we are generally well behaved. Some of us are even house trained.

 

Main reason why this is not a good idea. To many people get very opinionated and bran specific when it comes to this. And that is just not how benching/benchmarking or any form of it goes like this. It's based solely off of parts installed. Some parts from other manufactures will do better than others of course, but it's still parts. Not clevo vs alienware vs msi vs asus. When we can get past this as a community, then we can host a benching thread for all makes and models. Then again, that's already in place as svl7 suggested earlier. And if you use their banner on your profile, then everyone can see where you rank. And as for desktop cpu category. It should be pretty easy to tell who has top cpu for a mobile platform.

Just by using this page here.
Overclocking, overclocking, and much more! Like overclocking.
You only need to put the cpu and video card and it will bring up everything fitting that spec for the test you selected.

But with anything. It involves you signing up for free. And for those benching with free versions of the software. You would need to post a screen shot per the rules since your link would not be valid for very long.

And that's it.

 

Fine. I am going to overclock my Dell Mini 9. It's a handy little machine.

 

Already did that.

http://hwbot.org/submission/1053933_johnksss_cpu_frequency_atom_z520_(1.33ghz)_1844.1_mhz

 

Hahahahaha!

I bought mine for motorcycle trips. It fits neatly in my tank bag.

 

when i did mine i used one of those ceiling hanging lamps with a dimmer switch lol i use the dimmer switch haha on my fan mods.

 

I never got brand loyalty, I always enjoyed a place like futuremark (when it was alive and well) where you could just bench off against each other, slowly pushing forward and ending up just that bit further than you thought you could go (that's what happened with my 9800pro).

 

I don't get it either. I do like specific models for their distinct festures, but that's as far as I go. I have a Ford car and a Chevy truck. Motorcycles are two Hondas and a Kawasaki. No cruisers.

 

i am not much of a tech-guy so please excuse my simple question: how am i able to install nvidias 327.23-drivers if i am using a 880m? i read something about a modified ini-file of some sort, but i am not quite sure how i have to proceed installing these older drivers. And is there a possibilty of "flaggin" the 880m even though i would be using these "pre-destructive" nvidia-drivers? thanks!

 

"falggin" ? You would need to place the modified INF in the directory that the drivers are unpacked to when you try to install them for the first time.

 

He means this 100% undocumented, underhanded, and unethical flag which basically make the 880Ms run even worse than an iGPU.

 

Is there any known way to stop the core clock running at 993mhz whiles just browsing around the web and my desktop?

 

No. I heard you can disable Shadowplay but that didn't fix it for me.

 

Didn't think so. I may just flash my stock vbios back. I don't fancy knocking a years life of my cards just for a few extra FPS since none of us are really competing in the bench results at the minute.

 

There was a thread started in the Clevo forum... P570WM with 880s @ 1180 lol

http://forum.notebookreview.com/showthread.php?t=758274

 

Hi.

I'll be buying a notebook from xoticpc. The Sager Np 8528-S. (i7-4810, 16gb ram, 1 TB HDD 7200rpm, 250gb samsung evo SSD, and of course GTX 880m).

Should I let them overclock it for me. Do you think there will be that much of a difference (since the 880m is new and maybe they can't overclock it like older GPUs)? Do you think the notebook will heat up at dangerous levels? I'm getting the IC diamond, the copper cooling upgrade, and a notebook cooler. In terms of power, I get a 240w adapter instead of the 180w that comes with the notebook? Or will it be fine if I stick with the 180w?

 

I would contact them and ask them about the overclocking personally. You need a modified vbios in order to overclock these cards at all and they don't have hardly any room for overclocking at stock voltage either and their heat output is extremely high with stock clocks. You will probably be wasting your money if you get an overclock option. If they *do* overclock it, its going to be hot. These cards already run in the 80s in a cool environment and when you run a modded vbios to take out the boost clock and force them to 993MHz? 93C in no time.

If you don't push the voltage on the card, the adapter that comes with it will be fine. Any voltage adjustments will likely need a bigger adapter.

 

Will the ic diamond and the copper cooling upgrade be not enough to counteract the heating problems that overclocking will bring?

 

The copper cooling upgrade is a big waste of money. All it does is slow down how fast your cards heat up, they still hit the same temperature. They're just copper heatsinks they attach to the back of the core plate, totally not worth it and I actually lowered my temps when I took mine off.

IC Diamond doesn't do wonders either. You'll need to redo the stock paste job anyway with your own and I highly recommend getting Gelid GC Extreme for it, the stuff performs as well as IC Diamond basically without being impossible to apply or scratching your die.

Sent from my HTC One_M8 using Tapatalk

 

Xoticpc will apply the paste. Anyways, I'm still waiting for their reply. I asked how much performance boost I'll get if I overclocked it.

If the temp reaches 90+, what problems will it pose?

 

High temps cause the card to wear out much quicker and if you hit 94C, the core gets throttled which drops your performance dramatically in games.

The stock paste job is not adequate. I ordered from Xotic as well and when I opened my machine and saw the paste job one of my GPUs had such a large puddle it got all over the VRMs around the die and made a huge mess. Redoing the paste dropped the temps by 2C idle and 4C load which is significant.

Sent from my HTC One_M8 using Tapatalk

 

I guess I'll just have to wait for their reply and decide if it's worth. Thanks!

 

The copper cooling can make a difference IF you have a fan blowing cool air over them directly. Basically the extra copper increases the heat capacity of the heatsink, but all that heat will just be stored there if there's no mechanism to dissipate it.

 

Futuremark is alive and well my friend. Remember it's more than just laptops which some of us do bench.

It's more of a read between the lines thing when people post why they can or can't beat something.

Run balanced mode and allow for c-states to be enabled. Your card will down clock to it's lowest state when not in use.

Yeah, you have quite a bit of ways to go for that...more like running in the 115C category for a long time. Card will not let you get there. Something else will give out long before the core does anyway.

And they still dont beat 780m's. So no one wants to keep benching. You will see more cpu benching that gpu benching out of the hexaore clevo setups.

Boost can be ran or taken out. And cards will down clock if you dont force them at 993 core all the time.

Not gonna happen, that automatic downclocking will keep you from this. They are using basically the same thing as the intel cpus do for this.



With out the right power and knowledge, no one is going to be doing anything near a 780M. Even though the 880M "can be" a beast of a card. (Personal Opinion Only)

 

I've changed the preferred gpu to 'Auto Select' and I've changed the power management mode to; Adaptive. I've not seen my core clock drop in speed at all yet John. Is there any other options I have to change? Thanks for the help so far.

 

Yeah I am curious to know that one too because my cores DO NOT drop to 2D clocks with the modded vbios but do with the stock vbios.

Sent from my HTC One_M8 using Tapatalk