*OFFICIAL* Alienware M18x Benchmark Thread

Still, I don't see the FPS stats in the summary:

 

Hit the "show last runs" button

 

damit, sux to be an eternal noob, lol.


Here's the average of 3. Strangely, the first run is always a few fps lower.

 

Yeah first runs always tend to be interfered with disk caching.

 

Glad to help haha

Did not have to do that until I was using Steam install of Crysis, previously I had the disk install

 

So, based on tomshardware tests the 6990M's are ~7% faster in Crysis.

 

Tom's tests were not too re-assuring. I think there may have been some driver issues in their tests as the 6970s out performed the 6990s regularly haha

 

true. Now, we only need someone with the 580Ms and latest drivers to enter the scene

 

I always do a minimum of 3 runs on Crysis because the first run is always lousy.

Here is a series of runs I just completed testing my vBIOS hack. There are runs in 32bit, DX9, Very High and one run on Crysis64 DX10. My defective cards crash running DX10 and Crysis throws a BSOD or black-screen crashes most of the time. With the fixed clock speed hack on these defective cards, I was able to pull off a single run of 64bit DX10 without AA. The run I attempted with 64bit and 2XAA crashed (see screen shot) and the second attempt froze my system with a black-screen crash. Without fixing the clock speeds in vBIOS, Crysis will not launch under DX10 with this set of cards.

I cannot run the native Crysis64 benchmark batch file at all because it uses only DX10 and instantly freezes my system with a black screen on every attempt.

Spoiler

Series of Crysis runs (stock GPU clocks 680/900) - Average 60.07 FPS Best Run

Spoiler

9/18/2011 5:46:00 PM - Vista 64
Beginning Run #1 on Map-island, Demo-benchmark_gpu
DX9 1920x1080, AA=2x, Vsync=Disabled, 64 bit test, FullScreen
Demo Loops=3, Time Of Day= 9
Global Game Quality: VeryHigh
==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 40.14s, Average FPS: 49.82
Min FPS: 31.69 at frame 137, Max FPS: 73.50 at frame 1590
Average Tri/Sec: -31633970, Tri/Frame: -634941
Recorded/Played Tris ratio: -1.44
!TimeDemo Run 1 Finished.
Play Time: 35.18s, Average FPS: 56.85
Min FPS: 31.69 at frame 137, Max FPS: 73.95 at frame 106
Average Tri/Sec: -35877244, Tri/Frame: -631075
Recorded/Played Tris ratio: -1.45
!TimeDemo Run 2 Finished.
Play Time: 35.17s, Average FPS: 56.86
Min FPS: 31.69 at frame 137, Max FPS: 73.96 at frame 1581
Average Tri/Sec: -35887496, Tri/Frame: -631167
Recorded/Played Tris ratio: -1.45
TimeDemo Play Ended, (3 Runs Performed)
==============================================================

Completed All Tests

<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

9/18/2011 5:46:00 PM - Vista 64

Run #1- DX9 1920x1080 AA=2x, 64 bit test, Quality: VeryHigh ~~ Overall

Average FPS: 56.855
_____________________________________________________________________

9/18/2011 5:50:11 PM - Vista 64
Beginning Run #1 on Map-island, Demo-benchmark_gpu
DX9 1920x1080, AA=No AA, Vsync=Disabled, 64 bit test, FullScreen
Demo Loops=3, Time Of Day= 9
Global Game Quality: VeryHigh
==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 36.90s, Average FPS: 54.20
Min FPS: 33.94 at frame 154, Max FPS: 80.86 at frame 1604
Average Tri/Sec: -35629068, Tri/Frame: -657371
Recorded/Played Tris ratio: -1.39
!TimeDemo Run 1 Finished.
Play Time: 33.27s, Average FPS: 60.11
Min FPS: 33.94 at frame 154, Max FPS: 84.40 at frame 1660
Average Tri/Sec: -39356156, Tri/Frame: -654757
Recorded/Played Tris ratio: -1.40
!TimeDemo Run 2 Finished.
Play Time: 33.32s, Average FPS: 60.03
Min FPS: 33.94 at frame 154, Max FPS: 85.32 at frame 1652
Average Tri/Sec: -39287608, Tri/Frame: -654465
Recorded/Played Tris ratio: -1.40
TimeDemo Play Ended, (3 Runs Performed)
==============================================================

Completed All Tests

<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

9/18/2011 5:50:11 PM - Vista 64

Run #1- DX9 1920x1080 AA=No AA, 64 bit test, Quality: VeryHigh ~~

Overall Average FPS: 60.07
_____________________________________________________________________

9/18/2011 5:53:50 PM - Vista 64
Beginning Run #1 on Map-island, Demo-benchmark_gpu
DX10 1920x1080, AA=2x, Vsync=Disabled, 32 bit test, FullScreen
Demo Loops=3, Time Of Day= 9
Global Game Quality: VeryHigh
==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 52.59s, Average FPS: 38.03
Min FPS: 22.87 at frame 137, Max FPS: 44.55 at frame 1759
Average Tri/Sec: -10913112, Tri/Frame: -286954
Recorded/Played Tris ratio: -3.19
!TimeDemo Run 1 Finished.
Play Time: 48.44s, Average FPS: 41.29
Min FPS: 22.87 at frame 137, Max FPS: 45.78 at frame 1579
Average Tri/Sec: -11236931, Tri/Frame: -272177
Recorded/Played Tris ratio: -3.37
!TimeDemo Run 2 Finished.
Play Time: 48.72s, Average FPS: 41.05
Min FPS: 22.87 at frame 137, Max FPS: 45.78 at frame 1579
Average Tri/Sec: -11185045, Tri/Frame: -272495
Recorded/Played Tris ratio: -3.36
TimeDemo Play Ended, (3 Runs Performed)
==============================================================

Completed All Tests

<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

9/18/2011 5:53:50 PM - Vista 64

Run #1- DX10 1920x1080 AA=2x, 32 bit test, Quality: VeryHigh ~~

Overall Average FPS: 41.17
_____________________________________________________________________

9/18/2011 5:57:58 PM - Vista 64
Beginning Run #1 on Map-island, Demo-benchmark_gpu
DX10 1920x1080, AA=No AA, Vsync=Disabled, 32 bit test, FullScreen
Demo Loops=3, Time Of Day= 9
Global Game Quality: VeryHigh
==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 46.30s, Average FPS: 43.19
Min FPS: 16.96 at frame 146, Max FPS: 52.85 at frame 1785
Average Tri/Sec: -13163944, Tri/Frame: -304758
Recorded/Played Tris ratio: -3.01
!TimeDemo Run 1 Finished.
Play Time: 42.18s, Average FPS: 47.41
Min FPS: 16.96 at frame 146, Max FPS: 52.85 at frame 1785
Average Tri/Sec: -13751633, Tri/Frame: -290030
Recorded/Played Tris ratio: -3.16
!TimeDemo Run 2 Finished.
Play Time: 42.05s, Average FPS: 47.57
Min FPS: 16.96 at frame 146, Max FPS: 53.32 at frame 1755
Average Tri/Sec: -13823729, Tri/Frame: -290612
Recorded/Played Tris ratio: -3.15
TimeDemo Play Ended, (3 Runs Performed)
==============================================================

Completed All Tests

<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

9/18/2011 5:57:58 PM - Vista 64

Run #1- DX10 1920x1080 AA=No AA, 32 bit test, Quality: VeryHigh ~~

Overall Average FPS: 47.49
_____________________________________________________________________

9/18/2011 6:01:45 PM - Vista 64
Beginning Run #1 on Map-island, Demo-benchmark_gpu
DX10 1920x1080, AA=No AA, Vsync=Disabled, 64 bit test, FullScreen
Demo Loops=3, Time Of Day= 9
Global Game Quality: VeryHigh
==============================================================
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 50.18s, Average FPS: 39.86
Min FPS: 11.58 at frame 139, Max FPS: 53.32 at frame 1747
Average Tri/Sec: -12135135, Tri/Frame: -304460
Recorded/Played Tris ratio: -3.01
!TimeDemo Run 1 Finished.
Play Time: 45.17s, Average FPS: 44.28
Min FPS: 11.58 at frame 139, Max FPS: 53.42 at frame 1763
Average Tri/Sec: -12824463, Tri/Frame: -289620
Recorded/Played Tris ratio: -3.17
!TimeDemo Run 2 Finished.
Play Time: 45.07s, Average FPS: 44.37
Min FPS: 11.58 at frame 139, Max FPS: 53.42 at frame 1763
Average Tri/Sec: -12894703, Tri/Frame: -290601
Recorded/Played Tris ratio: -3.15
TimeDemo Play Ended, (3 Runs Performed)
==============================================================

Completed All Tests

<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

9/18/2011 6:01:45 PM - Vista 64

Run #1- DX10 1920x1080 AA=No AA, 64 bit test, Quality: VeryHigh ~~

Overall Average FPS: 44.325

Screen shot of benchmark crash with CPU-Z and GPU-Z Specs

Spoiler

I also have issues with games that run under DX10 constantly crashing or causing a BSOD on this pair of cards.

 

I'm doing the benches with the exact settings as tomshardware did.

For Crysis they ran DX10, 4AA, 1080p, All very high.

Now, I finished Just Cause 2 becnhmark and got 59.89FPS average (41.3 according to Tomshardware) which is 12% faster than their 580MSLI result.

Again, everything is stock, didn't touch the CCC settings, didn't mess with drivers/services/processes.

 

Yea can't tell if thats a fair example fox .... if it then thats quite the jump to the 6990s

 

Amazing ....

 

Here's the proof:

Now, am I doing something wrong? These are the max settings available.
Also, it seems that the game actually runs better but is capped by 60fps.

 

At the time I ran those benches it was to test what I was able to do with my screwed up cards. I did not notice that Aiki provided a link for Tom's benches until after I posted my results and folks started commenting about it. Sorry if that created any confusion. To be sure, we would never want to use these screwed up cards for a comparison against anything... (1) because they're malfunctioning, and (2) because I have modded the vBIOS just trying to limp along and still get some gaming done until I get my new exchange system. I have them running permanently at 680/900/1.1v whether on AC or DC, at idle or under load.

Nice job on the benches, Aiki. +1 Rep

That's true, but they also ran High, No AA as well (FPS in blue). Did you test those runs for comparison also? If so, were the results similar? My assumption is that the tests by Tom's Hardware are using the 32bit Crysis benchmark because it does not mention anything either way, unless I missed it somewhere.

 

Indeed ... this is some good stuff... starting to call Bull on Toms ...

 

As they say, "If you want something done right, do it yourself."

 

I did not look close enough. Take a peek at the "Test Settings and Benchmark Page" for details.

Aiki - you ran the 32bit test. Tom's ran the Crysis 64bit executable. For Just Cause 2 they ran medium detail with 8xAA. I have always run higher FPS on Crysis64 versus Crysis32 and you have higher settings on Just Cause 2.

Crysis Patch 1.2.1, DirectX 10, 64-bit executable, benchmark tool
Test Set 1: High Quality, No AA
Test Set 2: Very High Quality, 4x AA

Just Cause 2 Version 1.0.0.2, Built-In Benchmark "Concrete Jungle"
Test Set 1: Medium Details, No AA, 8x AF
Test Set 2: Highest Details, 4x AA, 16x AF

 

I will run their medium benches later.

For now, another interesting run. Notebookcheck claims 73FPS in Unigine for the CF vs 87FPS for the SLI. I ran the test at exact same settings and got 82.7FPS which is only ~5% lower and given the difference in tessellation blocks this is a very impressive result.

 

The so called professional sites like Anandtech, Tomshardware are all the same. They aren't always honest/professional when it comes to reviews. There are business interests involved when it comes to showing certain products in a certain way.

Its all in the business.

What we can do is read a lot of these reviews to get a better idea of where the real performance might be. But nothing better than benching it yourself.

 

Aiki, I think your results would be better by an even larger margin if you ran the same settings for Crysis and Just Cause 2. :wink:

Never believe everything you read or hear when a hired gun is on the job. In reality, things are often not as they seem to be. I see that all the time in litigation, with hired "experts" involved.

 

Does this mean I have to install a 64bit version of the game? Forcing the 64-bit in the benchmarking tool doesn't work - "Unhandled exception error..."

Will run the JC2 with those settings now...

Edit: same thing 59.89FPS. It is capped at 60FPS, so it is pointless. FRAPS also shows a consistent 60FPS. Is there a way to force the true numbers to be shown somehow?

 

Aiki - I think scook mentioned this already... do you have the Steam version of Crysis or the regular distro? The Steam version does not support Crysis64 according to scook.

If you do not have the Steam version, right click the executable, select Properties and Compatibility. Change the setting for all users to Administrator and change the drop-down to Vista SP2. This should correct that error. (This is the only way I can get Crysis64 to run under Windows 7.)

 

It is a Steam version

 

If you have a Bin64 folder in your Crysis installation directory, then you may have the 64bit version. Look for a Crysis64.exe file and apply the settings I mentioned in the post above. If you not do not have that folder and executable file, then you're out of luck on that.

For Just Cause 2, try turning off V-Synch. That's probably why your FPS is capped. I looked at your screen shot and that appears to be turned on.

 

Lol, true. Here you go:

 

Outstanding! I'd give another rep point, but...

Did you look for the Bin64 folder and Crysis64.exe file yet?

 

there's only Bin32

 

Aiki for just cause you need to run 8xAA .... but you 4xAA 107fps crushes Tom 0xAA 60 some fps already ...

 

I think something is fishy here with the benchmark. There can't be such a huge difference. Maybe I'm doing something wrong. I simply ran the in game benchmark and set everything to the max.

 

The settings run in Tom's benchmarks are...

Just Cause 2 Version 1.0.0.2, Built-In Benchmark "Concrete Jungle"
Test Set 1: Medium Details, No AA, 8x AF
Test Set 2: Highest Details, 4x AA, 16x AF

...and Aiki had the same settings, still stomping on Tom's benchmarks royally.

 

For a reference, here's the maxed out JC2 run:

Maybe there's a difference between the Steam version and a retail one? dunno but it looks very strange.

 

That or tom and perhaps notebookcheck dont load CAP profiles ??? looks like you are crushing toms scores... if so would look bad on the 580s ... BUT maybe the 580s were done poorly as well ... lets get some more from you aiki and see if xen or others can do the same

 

Shaden - I think your calling BS on Tom's benchmarks is right on target. Comparing Aiki's and my own results to Tom's really brings the validity of their testing into question. Both of us crushed their results to smithereens in Just Cause 2.

Given that I have the humble 2720QM CPU and malfunctioning 6970M GPUs, I still pulled off substantially better results than Tom's more robust test configuration. Take a look at my results...

Spoiler

I basically tied with the 580M SLi config equipped with a 990X CPU, which does not even seem feasible. This is not a good reflection on NVIDIA, on the $7038 Eurocom Panther 2 laptop, or the reliability of the testing and review by Tom's Hardware.

 

what's even more impressive is that the both of you have super duber cards that perform better than the desktop counterparts even at lower resolutions!!

Benchmark Results: Just Cause 2 (DX11) : Nvidia GeForce GTX 590 3 GB Review: Firing Back With 1024 CUDA Cores

The 6990 and 6970 are better than a desktop 6990! wow!



Yeah Tom is full of it for sure. haha.

 

Yea thats pretty messed up ....
Well thats why we need to do back to backs, with the same CPU and same settings in game.

 

Maybe it's because I was using the demo version. Will get the full one tonight and try again

 

I doubt that... maybe your drivers are somewhat tweaked and Tom's are riunning as they should?

Truth be told, your numbers should never be higher than his numbers Tom has at the same card, the 990 will smoke your 2960xm (and my 2920xm hehe) in any benchmark with the same card. So any numbers that look remotely the same are not legit (that alone double lol).

 

OT: Did ye forget today be international talk like a swashbuckler tide?
Ya bilge rat who ortin' t' be keel hauled!

 

Why are pirates so cool???


...because they just Arrrrrr...

 

I am pretty sure Aiki hasen't tweaked his drivers ... although I am wondering what the difference is as well. I don't think the 6990s should be performing like they are, but on both Crisis and JC he has broken Tom's scores.

Might it be CCC setup ?

Or might it be Tom not running CAP profiles ?

It Might be several things. What would be best is if Aiki could run these benches and someone with the same Processor could do the same with the 580s ...

 

I know Aki doesn't... do we know the ATI driver default settings are the ones ran on the other Benchmarks? That's my point... is tesellation on? etc,

All I know is that those Heaven 2.5 and Just Cause 2 numbers are stupid high.\


It isn't just Tom, google any Just Cause 2 benchmarks and you'll find that the numbers posted here are #1 in every benchmark (forget mobile cards... even 580gtx and 6990)

 

maybe clevo cards are better than dell

 

Yes and better than their desktop counterpart?? come one, I'd love that to be true, if that's the case I would have no problem trowing my 580m out the window for the 6990 clevo... but we know that isn't true... Not saying they can't be faster just no where near those numbers.... a 6990m on Clevo or Dell still a 6990m...

That said, Clevo 580m does not have throttling, evidenced by the insane GPU scores they get and OC at 820mhz+....

 

Oh I do agree Xen ... I fully agree with you.. Just wondering what the issue is....

 

me too lol, that's crazy.

 

I'm getting the same at stock clocks, Notebookcheck settings and 11.8 cap 2,

 

very nice, Ok I'd say that pretty much confirms Aiki's case there....

That puts our scores 13% above Tom's ....

Xen Can you run a stock Unigine ? no CPU or GPU oc ?

 

I have tested a lot of things on my dual 6970Ms and no matter what you did, the non-OC/OC performance is a good deal bit lower than what others were reporting across the board on game titles and benchmarks. The differences are considerable for similarly clocked almost similar configs, even on highly GPU bound games.

Just like steppings on CPUs with completely variable performance from the retail version. There could older steppings out there that don't perform as well as the more recent abundant cards people have in their systems.

I am beginning to see why my unit or at-least it's cards that could have been stripped down from another unit, since everything else was fresh PCB out of the fabrication, that was returned to Dell.

It has nothing to do with stability, they are rock stable. Its also not a case of a bunch of 'bad' cards. But depending on the stepping performance on very old steppings can vary by a lot.

In the GPU's case if one were testing the GPUs and needed to test certain sections of the logic I can cap the timings on the Cache clocks, increase decrease cycle times on dispatcher, change buffers on the ROPs, transcendentals can be turned off/on in certain SPs, you name it.

Now before anyone calls it unethical to put in older steppings into the market. Its AMD/Nvidia who decides what the standard performance should be, anything higher than what they set internally is just bonus for the consumer. So if it met that performance requirement then its fit to be in a retail system. No decent chip goes to waste that way. But yeah its whole new can of worms from the consumer's perspective.

 

Wait a second... for some reason mine defaults to 1080 with different settings.... I guess I never saw that these runs were done in 1024.... I'll give it a go. If I remember correctly Tom's was a 1080 run as well.... I should check.

 

Hey Xen ... we were looking at NotebookCheck not Toms for the Heaven Bench ...

Review GeForce GTX 580M SLI vs. Radeon HD 6990M CF - Notebookcheck.net Reviews

and our tests are set up the same ... but we are getting over 83 FPS insead of the 72 posted here ... would still leave the 580s ahead ... but far more slightly then they suggest.