*OFFICIAL* Alienware 17 R2/R3 Benchmark Thread

Wow, even the AW techs could not repro, you must be a genius, or fixing the numbers, according to Frank the W***? You are aware that you're going to have to carry a heavier bag now? Major disaster. /sarcasm_off

Good work man, thanks. Did you get the power cord sorted?

 

Haha thanks Mick, not yet. I'm using my 17xr3 psu for now will call today and get them to send the right cable.

 

Good luck, we know Frank's minions are monitoring this thread so it should go fine. If they ask you for pic's then interference is being run...

Edit, my rep has not replied and we had convo's going about the new model's specs. My conspiracy radar is affecting my posts now

 

So is the 4GB 980M functioning like a 3.5GB video card? Can anyone confirm if it uses all 4GB without horrible frame latency?

 

Some Dawn of War 2 Benchmark

Dawn Of War 2 240 Watt by just1tree posted Jan 31, 2015 at 10:46 AM

Dawn Of War 2 180 Watt by just1tree posted Jan 31, 2015 at 10:46 AM

omparisons

 

That's better than 50% increase in average FPS! Now you'll be asked for CPU and GPU graphs during the run to identify which and when, like it matters. The result of poor performance can be easily fixed, as you have done

Thanks again

 

I have been trying to figure a way outside of just flat benchmarks for in game fps, but these benchmarks show such a difference it's not really necessary to show in game fps tracking.

 

Fraps for just FPS or msiafterburner for all your ingame stats. You can see the GPU throttling if you select it as a display item. Voltage is also a good one to monitor as voltage drops also indicate a throttle.

Go to settings then monitors and select the ones you want by checking them and the 'show...' check box.

 

@Alienware-Frank, @AW Khan, @Alienware-Wasserman, @Alienware-L_Porras:

I'm trying to consolidate all this info into one post so far to make it easier on the alienware guys. So here is what I gathered, if I missed anything let me know:

just1tree's first benchmark:

P10102, graphics 12226, physics 7642, and combined score 5553
CPU Max Temp: 76, GPU 66 during 3dMark11
Config is 4710hq, 980m, 16gb ram, and 1tb hdd.
GPU ran at 540 mHz

3DMark Vantage:


With 240W:
3dmark11 P10036, graphics score 12193, physics score 6523, Combined Score 6611
3dmark Vantage: P26837, graphics score 30739, cpu score 19435
Gpu ran at 1126 mHz



3DMark11 with 240 PSU with new drivers from laptopvideo2go 347.12:


Vantage Score: p30328, gfx: 33315, cpu: 23901

Driver 347.24
240 watt psu: P10824, g 12482, p 7851, c 7580
180 watt psu: P10497, g 12476, p 7504, c 6599

just1tree's talk with the tech benchmarks (Heaven benchmark?):
180w psu: 62.4 fps, Score 1571. Min fps 13.3 Max fps 94.8
240w psu: 86.1 fps, Score 2168. Min fps 24.7 Max fps 146.7

Mr.Fox's video of power draw on the 980m:


Mr.Fox's video of power draw on stock AW17R2 & Resultant benchmark:
http://www.3dmark.com/3dm11/9349774


just1tree's Metro 2033 bench:
180 Watt:

240 Watt:

just1tree's Metro Last Light bench:
180 Watt:

240 Watt:


just1tree's Dawn of War 2 bench:
180 Watt:

240 Watt:

And for a rough idea of an equivalent laptop's score

Clevo with comparable specs (3DMark11) which runs on an 180W PSU:

Hope this adds a bit more clarity.

 

Good job. A sledgehammer of a post (and only to crack a 'nut') haha

 

@bnosam

Don't forget to mention that the Clevo P650SG uses a 180w PSU.

 

Added that, thanks!

I took the liberty to tag some of the local alienware team members too, for easy finding of my post
Big thanks to the guys who did the benchmark and just1tree for pretty much having his rig be our guinea pig!

 

It's all good, would be nice if AW sent me a free ssd for doing their work

 

@AW Khan

You heard the man!

 

Several should test With Sky Diver in 3DMark instead of 3DMark Fire Strike and 3DMark 11. This benchmark test pushes the processor more.

 

SO has anyone in this thread had any success with overclocking their GPU?
I picked up a 17r2 with a 970m and cannot get it to OC with afterburner at all (when I try to apply clocks they just reset to stock)
I tried to OC to nvidia inspector and it seems to do something but I used a 10% OC (+135mhz) and and not really seeing the increase I would expect

Oc'ed

http://www.3dmark.com/fs/3964963

Not OC'ed

http://www.3dmark.com/fs/3965167

the results are really consisted tested many times at this point but im' only getting a 3-5% performance increase when I would expect closer to 10% which makes me believe that the OC in nvidia inspector is not really achieving anything.

At first I Thought that it might be a lack of power holding the oc back but I tried using the 330watt psu from my AW 18 and it made no difference. So my thought is that the increase from the "OC" is actually just from the fact that inspector is forcing a powerstate and preventing any throttling but not acually increasing clock speeds

I tried to observe my clock speeds in afterburner but Afterburner seem to only be reading the idle clocks (405 mhz).

If anyone here could shed some light on the issue or knows how to get afterburner to play nice with the 970 in the AW 17 any help would be appreciated,

 

That's weird, on the 3DMark test pages it shows that both your stock & OC'd frequency was the same 924Mhz. Try using GPUz (the latest version) to check your clock speeds when you run the tests, by viewing the Sensors Tab after the tests have finished.

 

Yeah, that clock speed is the 2D/low utilisation speed. nvidiaInspector will show you the 'P' state. P0 is full power and they go down from there, again you can see all the clocks in NvidiaInspector - overclocking. It will drop back in windows to something like 75mhz (P8 on mine) so you need to run a benchmark in a window, like heaven 4, or use afterburners on-screen-display. If you still see only the 405mhz then either it is a power issue OR a faulty card.
Did you fully power cycle your machine when you tried the 330w PSU? The bios has to see it for the OS to use it...
3D mark uses bios/vbios data for reporting speeds, it's not realtime or actual achieved.

 

Please, go back and read my posts and let me know where I said I could not reproduce what you guys were seeing? Or claims of fixing numbers according to Frank? I'd appreciate it.

I am simply replicating each user's experience for a multitude of tests.

 

By not admitting that you see the problem or if it is a problem from your perspective. It is very simple to repro and even easier to fix. There is no need for a multitude of tests - unless supplying the higher PSU is not an acceptable answer? Did you get these devices certified with a 180w PSU? Has your regulatory affairs department stomped on a bigger PSU?

I do not accept that you need more data. Simply more time to come up with an acceptable answer without increasing cost or compromising certifications. Was this some esoteric fault that happened at random your requests for more data are valid.

Can I take it you have reproduced it (reliably) and are going to fix it - or say, as many technical experts think, 'working as designed'.

Thank you for taking the time to respond but as usual, semantics and vagueness rule your statements. There was also a level of sarcasm in my post which I apologise for, when I first took up a VP position in the US I was warned that Americans do not recognise it and will take you at your word! Shame on me.

 

I'm being vague because I can't share anything conclusive until I'm done. When I am done, expect me to confirm or deny your claims with evidence to support my findings. I don't like sharing pending data. Its inconclusive. Additionally, the multitude of tests I'm asking is to analyze multiple angles on our products, games, and useage, not just the PSU. I analyse the data and the findings thoroughly for multiple different scenarios/issues and for historics. Also, I absolutely hate the phrase "working as designed". I've said it as a joke whenever I see a bug in anything on my phone, games, car. Don't expect me to say that.

 

OK, you're turning me, so we'll see. I know the restrictions corporate America places on it's staff when making public statements, I have my own bunch of NDA's making me cautious of expanding on just what I do/did.

At this point I'll wish you luck. I'll simply be returning my machine if I am refused the PSU (assuming I see the problem of course...)

 

Still doing testing on my OC (Has anyone been able to get their 17 to using afterburner ?)

but on another note does anyone know where I might be able to find inf modded versions of the latest drivers from nvidia the will work with 17rs because the shipping drivers are complete (344.73) are complete and utter dogshit.

 

Have you tried laptopvideo2go? Click on the driver you want to try and you will be offered the driver (unmodified), or the INF.

Does the NV released one give the 'hardware not found' error? If so you might have to read up on how to mod the file yourself .

 

hi i am new here and thinking about getting the aW 17 r2 too. but my question is now, you cannot use the official drivers (347.25) directly from the nvidia homepage?

 

I can say for sure that the drivers off Nvidia's site do not work they give a hardware not found error

 

I guess a user will reply with fact but this often happens when a new machine is released. The NV Verde (notebook) driver program is opt-in. The OEM's send NV the machine ID and any cards they use or intend to use. Not on the list? Not supported.

Often the originating OEM (Dell in this case) will not want NV to support a new model straight away. They need to make sure there are no problems and having different video drivers in-use confuses the issue.

HTH

 

you learn something new every day. thank you for enlightening me

 

You're very welcome. I suggest you also read the disclaimer on the notebook drivers page. I know we all just click through this stuff but this one is very clear about responsibility and use .

 

I really dislike this aspect of owning a laptop for sure. When I first got my m17xr3 the driver was not updated directly from Dell in easily over 2 years. Games would show that they would benefit from newer drivers, and with the 6990m and switchable graphics it was just the worst experience getting a driver update to work properly. The best solution in that case was to finally disable swappable graphics and the experience improved significantly. To me with the 980m being a newer architecture, driver updates are going to be improving performance as games come out. Not keeping up to date could really hold these new machines back.

 

I do agree. If you have moved to laptop gaming from the DT world it's a real shock that all you thought you knew is no longer valid.

Dell's reticence on driver updates is simply a CYA exercise for the support team. If they publish an updated driver then it is assumed to be fully supported and working. Once NV add the machine ID then it's up to you and not dell's responsibility.

The first job that happens when you call support for a video related issue. Roll back to the approved driver. It gives them a known starting point and can eliminate or narrow down a driver problem.

Laptopvideo2go are usually quick on providing an updated INF, but if it does not work then maybe look into submitting your machineID to them for inclusion?

 

True, Dell & other laptop manufacturers often rarely update their GPU drivers. However, I've always used drivers directly from NVidia. The A17 R2 will be supported soon enough on there, and it should be safe to use those drivers downloaded from the NVidia website too. (I tend to download all my drivers for the different laptop components from the websites of the manufacturers of that component. I don't think I have any Dell drivers on my laptop now - apart from the freefall sensor for the hard drive!

 

Hey guys... something interesting. I did more testing with the AC adapters (180W, 240W and 330W) and it appears that performance has less to do with the capacity of the AC adapter than what the BIOS *thinks* it has connected in terms of an AC adapter. If you shut down and connect, then power up the BIOS "shakes hands" with the AC adapter and attempts to control (or so it seems) how much power the laptop is allowed to use. However, if you cold boot with a 330W AC adapter and then hot swap the AC adapter, the performance is more or less the same with 180W, 240W and 330W. I did about 12 3DMark11 runs (3 with each AC adapter) and it seems pretty consistent. So, getting a 240W or 330W AC adapter is a good thing not solely for the sake of capacity, but so the BIOS does not try throttling things (especially the CPU) to stay below a threshold. Here are some examples.

Spoiler: AC Adapter Performance Testing

Booting with 180W AC adapter connected:


Booting with 240W AC adapter connected:


Booting with 330W AC adapter connected:


Second run with 330W AC adapter connected:


Hot-swap 330W down to 240W AC adapter:


Hot-swap 330W down to 180W AC adapter:

I think our friend @AW Khan may also find it interesting. I tagged him, so maybe he will see this and investigate further.

 

Mr. Fox, that's very interesting! The biggest difference between all the different adapters is the Combined Score - that increases dramatically from 180W through 330W, and your hot swap 180W got the same Combined Score as the 330W adapter. Good job on working out that it's the BIOS limiting the power that can be supplied from the AC Adapter - kudos! One anomaly I noticed, why is the 180W Physics score so good?

 

That makes perfect sense to me and is a little worrying. If the PSUs draw is limited by it's rating/bios and not circuitry built in, then you may be pulling way more W than the unit was made to support. Have you noticed it getting any hotter?

 

No, it does not seem to be getting hotter. If anything were to happen, I would think hot-swapping to a lower capacity AC adapter from the higher capacity unit on a frequent basis might kill (or shorten the life span) the lower capacity unit, LOL.

I could not tell for sure because just a few runs with each adapter is not enough to be 100% conclusive, but it seems like the 17 R2 is less stable with CPU and GPU overclocked with the 180W and 240W. It makes sense that it might more stable with the higher capacity AC adapter, but 3DMark11 never once crashed with the 330W adapter, but it did several times with the 240W and 180W AC adapters... but, this was a small testing sample.

 

That...actually makes perfect sense. It would explain why the Clevo P650SG with the same specs and a 180w PSU performs better. The problem isn't the PSU - it's the BIOS.

I had found it hard to believe that the Clevo was so much better designed it only needed a 180w, or that the PSU shipped with the Clevo is so much more efficient than the one that comes with the Alienware.

Now, it may not change the fact 180w is (maybe) sufficient for a stock GTX 980M but not for any kind of OC'ing.

 

That's some more good news. As long as it's (180w) not getting too hot after a hot swap then at least it's safe. Maybe that is something dell want to check out before melting plastic all over the carpet haha..

 

Probably because it did not throttle, has ample power when the GPU is not leeching away part of the juice and the machine was cold. Toward the end of all these back-to-back tests it was exceeding 90°C due to repetition.

 

Absolutely. In fact, I think this down and dirty test should increase concerns about having sufficient power for overclocking with anything less than 240W. This was all with a stock vBIOS so the core and voltage overclock are almost too small to mention. I cannot see any way to flash this BGA vBIOS yet. Hopefully, svl7 might figure out a way to do so.

I mean, who wants to hot-swap AC adapters, LOL. If you have to buy a 240W or 330W to trick the BIOS, just leave it attached. No point in hot-swapping them and overloading the low capacity adapter.

 

This seems to indicate that the BIOS locks the power draw from the PSU at a % of its rated capacity.

If the 180w can perform about as well when the BIOS "thinks" it is supplying 330w, then that indicates the 180w PSU can supply enough juice to maximize the performance of the 980M @ stock.

However, when the BIOS sees a PSU rated for 180w, you see throttling due to a lack of power. Which would indicate that the 180w PSU (which was shown to be capable of fully powering the 980) is now being prevented from supplying its stated capacity.

If this is the case then why exactly IS the BIOS putting a (lower) limit on the 180w PSU?

 

Purchased ready and waiting .

This will be for safety and a requirement. If a PSU OEM states the safe maximum load then dell are simply making sure this is not exceeded. It could actually be 175w inside the code, allowing for a safety factor?

Edit: forgot to add, battery melt-down/explosions and PSU meltdowns are not that uncommon in the laptop industry. Recalls are hugely expensive as well as the bad PR. Anyone that signed off on an out-of-spec component, or supplied one gets to pay!

 

Who knows... they limited the Alienware 18 severely in the power handling realm also. Somebody must be scared of overclocking too much. I really hate it when companies try to control how people use their products. I've been poking around in these forums as an enthusiast 6 or 7 years and I do not ever remember seeing an example of someone killing an Alienware laptop from too much overclocking. Probably happened once or twice that we don't know about, but as you guys know I overclock the crap out everything that I can find a way to overclock and have never killed one yet. It's kind of like the stupid argument that guns kill people and should therefore be banned when it's actually people that kill people.

 

Wouldn't the PSU simply have a hardware limit on what it can provide?
I'm not a software engineer or anything, so I wasn't aware the BIOS is responsible for ensuring a PSU isn't pushed past its "safe levels".

 

Yes, it does have a limit, and a circuit breaker that protects it from sustained operation above rated capacity. The BIOS should not be handling that in my opinion, but it kind of seems like Dell may have decided to add that "feature" to the BIOS. The AC adapters can take care of themselves. If you exceed their capacity they will turn off. That's why I built my dual adapter setup for the M18xR1/R2 and that's why Clevo offers that option on SLI laptops.

 

It will shut down but it will be above the rating allowing for manufacturing tolerances. (made up numbers here) So the PSU is rated at 180w. It works fine up to 240w so any units made that are not as good quality can always be said to be 180w, it's a minimum statement of the PSU OEM and a maximum draw of the user OEM..

 

So I have done some more testing

1. even after reboot 330 watt psu makes no difference in performance for the 970m
2.Tried to monitor gpu clocks with afterburner without OC applied via nvidia inspector afterburner shows max clock as 540 mhz (this is definetly a reading error if anyone can figure out how to read clocks accurately in afterburner I would love to know how)
3. when I set a manual clock with inspector afterburner seems to be able able to read the clock speed of the gpu however I do not believe that it really is because....
4. Overclock from inspector does not actually seem to be increasing clock speed and infact I can set the clock to whatever I want and afterburner will read that clock speed just for kick I tried setting gpu clock speed to 5ghz and according to afterburner my gpu was hitting 5ghz while completing a full 3dmark run; however this "OC" does not result in any performance change over my previous OC which... in conclusion leads me to believe that All that OCing through inspector does is force the gpu to stay in it's max P state and that any observed performance gains on my end were do to the gpu not being allowed to throttle.

soo if anyone else here has any actually success in Overclocking the gpu on their 17r2 I would love to know how it was accomplished.

 

Just grabbing at straws here but maybe try PrecisionX/16. It's evga's version of afterburner and uses the same OSD but you are seeing some strange clock speeds.

Nvidia Inspector has been rock solid for me, on yours it seems to not be communicating correctly. Have you installed HWInfo?

What does Heaven 4 show for the GPU clock?

Hopefully someone has had success with this...

 

Ok so afterburner does not work to monitor clocks but the system monitor in Inspector does afterwatching it my clock go up to 1038 with out the OC and also to 1038 so oc is definetly not working but I was able to figure out where the performance increase is coming from I was an idiot when I made the Batch file I put in the wrong memory clock and was oc'ing the memory by 600 mhz so mystery solved there. but still no headway on figuring out how to oc the GPU on the 17r2

 

I'm amazed at the difference you score with the 330 hot swapped down to the 180 watt. I'd like to figure out why your gfx score is 1000 more from where my system is with the 240 watt. I wouldn't think the 4980 vs 4710 would effect that score.