*OFFICIAL* Alienware M18x Benchmark Thread

For reference, anyone trying to bench Crysis from steam install will have to do 32-bit...just a heads up

 

ARGH - I may have time after work tonight, but I'm not sure my current benchmarks would be of much value for the sake of comparison. (That is why I am posting older results.) As I mentioned, my currently installed video cards are defective. I don't think I can even launch Crysis right now without a freeze or BSOD. I will look in my benchmark folder, but I don't believe that I have saved any old results from Crysis 32bit. I generally only ran the Crysis 64bit benchmark and gamed in 64bit, so I did not really pay much attention to what the 32bit benchmark performance was.

If you want to see some screaming Crysis benchmarks as a target to hit, look for posts in this thread by BDRacing about 6 weeks ago. He was in the 100FPS range with "High" and "Very High" settings.

 

Those 100 fps benches were lower settings than what we are running. The 60-70 range is what will be more common for people running the max settings

 

Sounds like you may have the same issue I had with my 6970's. When the slave card hit 3D clocks, it froze up. Reason why I RMAed then paid Phektan the differance for 6990s.

 

Yes, that's one of the issues. Sometimes the second GPU remains stuck at the power-saving idle clock speeds and voltage (100/150/0.9v) instead of ramping up to full speed. Also, they will not synchronize properly. GPU0 will run stable at a higher speed than GPU1 will. I can manually set them independently a different clock speeds using Sapphire and things work fairly well, but not properly.

I think the problem is possibly bad video memory on GPU1. The cards are running nice and cool, so it's not heat related. At any rate, I should have replacement cards here in the next 48 to 72 hours based on the info I received this morning from my L3 rep. (They were out of stock.)

 

you benching right now or at work dr650se?

 

At work at the moment. Might do some background benching when I get home. Set it, and let it run. Visit with the wife in between runs, and watch some t.v. But even that doesn't win me any points

 

Crysis 2 ultra light overclock

 

can you set the mem clock to 1000 and see if it makes any difference?

 

1 frame

 

I do this often ... it scores me a few points

 

ok i just ran that test with your exact settings with 825 core / 1000 memory and i only got 61.3 average fps. how are you getting 15% higher fps?

 

ARGH - are you running the Alienware OS installation? If you haven't done a clean install, you probably are not ever going to have benchmark results that reflect your system's full potential.

I was able to pull off Crysis benchmarks with 6970 CFX and CPU at stock clock speeds in both 32bit and 64bit DX9 modes for you. Bear in mind this is with faulty video cards and this was the first time in 2 weeks I have completed a Crysis bench without a BSOD. I tried to run the benchmarks in DX10 three times and my system freezes on every attempt. So, this may not be worth much for the sake of comparison in the greater scheme of things but I told you I would give it a shot... here you go, buddy.

As you can see, there is essentially no difference in FPS performance between 32bit and 64bit. I prefer to run Crysis in 64bit mode because I have no flickering with 2X AA and 32bit always has some degree of flickering with or without AA.

 

very high settings do not work under dx9. you need dx10 unless you hacked the config.ini file or that 3rd party benchmark is not accurate in it's setting adjustments. that is why i prefer using the built in benchmark utility that ships with the game.

there are also config tweaks that skew the benchmarks and also messing with catalyst settings. i am trying to see stock vs. stock gpu's here, not how much extra fps you can squeeze by messing with settings.

 

None of that really matters if the systems you are comparing are operating under the same conditions with identical settings. Everything on my system settings are stock right now, so if you ran the 3rd party tool with the settings displayed in my screen shots you would get the GPU-to-GPU comparison you are looking for.

I am unable to run Crysis64... keeps crashing to a black screen with a windows that says "Crysis stopped working" every time I attempt a run. That has been the case since the day this set of video cards was installed... I did not have that problem with the original video cards.

Attached is a 32bit DX10 run I pulled off this morning using the native Crysis GPU benchmark tool at stock GPU clocks. I see no way to adjust the settings using the batch file, so my assumption is that in-game settings are used. In-game, everything is set to "High" under the Advanced tab. So, if you set your in-game advanced settings to "High" you should be seeing the head-to-head comparison you're so eagerly wanting.

I tried a second benchmark run and got a BSOD at the start of the third loop with these defective cards. After the BSOD the benchmark refused to launch. Disabling CFX allowed the game to launch again, but with like 20FPS. After re-enabling CFX and rebooting, my temp sensor readings are no longer registering on GPU0 and only GPU1 will ramp up under load. (I sure will be glad when my replacement GPUs get here. I hope to hear from my L3 rep that they are shipping out today.)

 

Lol. disable tessellation on CCC

 

run the game with it's 32bit mode from bin32 folder and change in-game settings. Then exit game and run the benchmark file. The benchmark will run the settings you selected in the game.

Edit: ok so tou ran it with all high settings instead of very high. I will match settings and reoprt the results.

 

ggrrr, hate it when people do

 

and I love when people don't...... :laugh:

 

Can't believe it! The 2960XM's landed! Too busy at the moment to mess with it, will do tonight!

 

all high no AA;

[/URL][/IMG]


all high 8x AA;

[/URL][/IMG]


.....

either the 6990m is not any faster than 6790m or you guys are messing with the catalyst or drivers haven't yet mature for the 6990m

 

and your score without cheating?

i want to for once get something comparable to know my cards aren't defective lol.

 

I do not expect much over the 2920xm's.....since they will take the chips they would bin as 2920s and sell them now as 2960s.....(note the 2 not being offered at the same time)

 

Exact same drivers.....look at the inf file. The 6990m's and 6970m's are not differentiated at all

It is not cheating, just a way to get a higher score. You are welcome to do the same and make it then comparable. I can promise though....the 6990s will outperform the 6970s no matter how much you want them to be equal

 

Damn right, it was night and day comparing my 6970 system to my 6990 .... but then, my 6970s were garbage... but I think at stock they should have been the same as anyones

 

lol not cheating....

 

there you go stock clocks 715-900 no cheating

Edit : that was with the 2720qm

 

Is there a rulebook I was unaware!


........off to run tell the people at hwbot...

 

Benchmarking to achieve a high score doesn't have rules, LOL.

ARGH - looks to me like your cards are working correctly. If there were something wrong with your video cards you would be having significant problems like I am experiencing. You'll always find slight differences, even between identical hardware. I know some of the 6970 and 6990 will run at higher clocks than others will.

 

Generally - yes, should be the same, but there is a hope of getting a 200mhz higher stable on all cores, especially without an unlocked BIOS. In addition, I'd like to test the temps.

Most importantly, we will be testing the new HS with the beast soonish.

 

so you'll be putting your custom quad piped cpu heatsink up for sale soon after testing?

 

interested in what it does with stock 2 pipe heat sink

 

Yes, but first we have to run series of tests and see how much better it is over the stock HS.

Will know in 3-4 hours

 

Anxious to hear about both, Aiki...

 

Ok, got it in.

The stock clocks are 34-34-36-37. Idle temps ice cold. Will do a few benches shortly...

Edit, stock clocks Wprime32, nothing fancy but what's important are the temps and testing environment. Room 25C, no cooler pad, no elevation. throughout the testing I'll try to be as real life oriented as possible.

 

Define ice cold? And that is with the quad pipe heatsink right?

 

You're testing the new i7 2960 CPU, correct? As scook asked, is it also with the new 4-pipe?

 

No, everything is stock - stock 2-piped HS.

 

What are idle temps? I idle at about 38c with my 2920xm with fans at 3800 rpm

 

I idle @ 40C with fans off.
Here's a Wprime1024 run. The fan only reached 75% temps peaked @ 80C

 

With fans off I hang around 44c

So impressive

Wonder what sort of voltage it is pulling

 

Idle temps comparison is a tricky subject, each testing environment is so different. I can only compare it to the 2 other CPU's I already tested under the same conditions and even then it's not accurate. One thing is for certain, it idles at lower temps than my 2630QM and when at load it's about 5-7C cooler than the 2920ES (2.4GHz).

Guess it is time to see some mild OC'ing results.

 

skip the mild business and go for the overclocks we all really care about haha

 

Care to provide the settings you wish me to try (stock BIOS for now)?

 

Well I assume clock for clock it will be the same as the 2920xm....so I want to see how it does at exceeding the overclock of the 2920xm otherwise nothing special.....try to get 46x or more on all 4 cores

 

he only needs to run 45x on all 4 cores...

 

We can already do that. I want to see better

 

Wprime32 @ 40-40-43-45. It does stay at 3.9Ghz on all cores so far, no bclk or any tweaks, just a single run.

 

3.9 GHz was the 80W TDP on my 2920xm when I had VID at 25

Try dropping VID and see what happens

I was able to do 4.2 GHz on all cores with stock voltage

 

And I was able to get to 3.7 max with the ES on the exact same settings.