M17 OC Benchmarking - The Next Generation!

i have 2 M17 under my naem i guess i should call them huh ?

 

Really? I thought Alienware cancelled upgrades for the M17/4850s...

going back on-topic, good luck with your 20k goal. How are the temperatures with the 4850s? Because the M17's cooling system isn't that great with 1 fan cooling two GPUs.

 

I just got them today and whatever I say doesn't count because I'm testing outside in the cold weather.

For anyone interested there is no issue flashing these cards.

 

I tried changing the voltage on the cards to equal the 4870s and BSOD.

 

Go ! Go! Go Sensie DRAKE !!!!!!!

 

The mxm 3.0 adds voltage and stability

 

Yeah and now the little light bulb went on regarding why some people only use one dimm.

 

lol......

 

which voltage values did you change? Did you change all the ones under high performance? because you don't have to do that to get the voltages you want. Could be causing the blue screen as I had some similar issues when voltmodding the W90.

 

things are at a stand still drake?

 

Hey Mandrake, is it possible to just get a baseline GPU benchmark, no OC?

 

Yes, default was 1.05 and it BSOD at 1.10. Changing it to 1.07 was OK.

Watched a movie with the wife.

For which program?

 

Vantage I'm guessing


answer should be 9.5-9.9k

 

That was easy.

 

Vantage or '06. Just the GPU. Gracias.

 

http://forum.notebookreview.com/attachment.php?attachmentid=42261&d=1261273452

stock gpu oc CPU

 

I saw that, but assumed it couldn't be right. My 3870s with no OC score:

3DM'06
SM2.0: 6717
HDR/SM3.0: 7458

Vantage
GPU 6655

And I love ya, Moo, but I need the benchies from Mandrake.

 

ill stay out but we shall see what happens

tooodles

 

Moo's GPU score for Vantage was correct, it depends on your CPU though.
As I remember:
P8400 (2.2Ghz) GPU score = 8.8k
X9100 (3.325.Ghz) GPU Score = 9.8k

You can roughly scale it as approximately 100 GPU points per 100Mhz of core speed, so if the cards keep scaling higher, and Mandrakes 3.8 is stable, I'd guess he'll hit about 10.2-3k with the cards at stock, perhaps a little more.

They're good cards, not QUITE as nippy as some things as the GTX280M SLI, but they made up for it elsewhere, and were quite a chunk faster than the GTX280m in some stuff!

 

I never doubted Moo. I mean heck, check out his Rep Power! I was just looking for something directly from the horse's mouth, since he's got the same machine with the cards that I plan on buying soon.

 

That's default? Here is my heavily OC'd CPU with default GPU when I did it with the 3870s.

By the way do you know the default voltage for the 3870s?

 

If you can get hold of the 4850s like Mandrake, I'd advise sticking with the M17! The price difference to the M17x isn't worth it if you can get hold of the 4850s for a few hundred dollars, the price difference unless overclocked is really not that staggering!

 

Yeah, those runs are with no OC on the GPU. I don't remember what the voltage was, sorry man. That was last December, shortly after I got this beast.

Would it be too much trouble to get a no OC run from you on those 4850s in '06 and Vantage? Just run the GPU portion of each. Sorry, but I'm fiending to see them. What I'd love to see is some runs on the Crysis benchmark tool. But I know you don't game much.

 

Once I figure them out I will. I'm battling with inconsistency right now.

 

Of course man, I totally understand. I'm just so excited after being told my M17 would have a GPU upgrade, and then being told it wouldn't, to finally have the option for one.

 

If it makes it easier for you, I played through the majority of Crysis on the 4850s, unfortunately I couldnt get the benchmark tool to run and its only recently I figured out what was wrong.

If it makes you feel any better, DX9, 64bit High settings were playable at 1920x1200. Average fps was anywhere between 50-25 dependant on the area of the game.

For me with the X9100 and stock GPUs, in 06, it was about 15.5k score, 9.8K in vantage GPU. I only overclocked them briefly for testing (which netted 10.6k GPU IIRC) and didnt use these settings overall.

My X9100 was at 3.325Ghz, as I mentioned before the crossfire setup seemed to scale heavily with CPU

Did you read the benchmark thread I did with the P8400 and then with the X9100, as it might answer some of your questions on performance? I ran 3dmark vantage, 06, tropics, SF4, DMC4 etc
http://forum.notebookreview.com/showthread.php?t=404008

First lot of weaker scores are with the P8400 [so ignore that bunch], and the second lot of benchmarks are with the X9100
You have a very very similar machine to the one I had, so you should see very similar results if you get the 4850s (hell the Last Remnant test for example makes my GTX280ms I have now look rather inferior)!

 

I'm reinstalling the OS right now.

 

Did the runs also, with my new 4850 cards.
Since my CPU still sucks don't expect anything impressive:

3DMARK06:



VANTAGE:



Using mobility modded 9.11 drivers without any OC.

 

Yeah your cpu score is the same as mine and I'm using a qx9300. After reinstalling it made no difference. I'd don't know if this is because of the 4850s or if there is an issue with my qx9300. I'm going to swap procs with my M17x (T9900) and see if it makes a difference.

 

looks like you really need to be running the 9.12 drivers.

 

From what i heard the 9.12 drivers are poor and offer graphical glitches in games such as CoD 4: Modern Warfare 2.

 

What do you guys think? The score on the left is the M17 and right is the M17x. Both CPU OC'ed to 3.5Ghz with stock GPUs. Normal? I'm thinking no.

 

The CPU score should be higher on both, probably around 3.5k.
Did you have too much stuff running in the background?

 

are they modded or straight from amd.com?

they look to be damn near identical.. with respect to the speed of the cpu....that's if they scale together.

 

The CPU score discrepancy is bothering me.

 

this is one of those times where you should be able to run the cpu test for about 5 seconds and know if it's going to be good or bad...

do a few test cpu runs only and see what it comes back with.

could also be superfetch kicking in during testing....

 

Wow, my stock 3870 x2 is faster?




oops. nvm, you are using single 4850.

 

Yeah it's obvious with CPU Test 2.

 

Anyone with a QX9300, T9900 or X9100 with 4850s have any stock 3dmark06 runs they can post?

 

I am using Dual 4850's.

Not sure why i am 1k below you in 3dmark06 but from what i heard 3dmark06 is more cpu dependant and your CPU is faster.

 

Nah, I was comparing 3dMarks06, and I am surprised that 3870 x2 is still very comparable with 4850x2.

I guess 4850x2 is not much faster than 3870x2.

 

How about you run vantage?

[edit] i actually found out that the 9.11 drivers are extremely laggy...everything in windows had delays. Installed 9.12 and the lag is gone .

 

you'll really need to do gaming benchmarks..3870s did damn well at 3dmark06 and they still do.

yeah, tried to tell you...they did so well they put me 5th.

and they gave mandrake a 14.2k gpu....so yeah...im gonna go out on a limb and say....they are a fair improvement over 9.11...

 

You are right John, thank you .

Updated to 9.12 drivers and got over 10k GPU score in vantage.

 

Nice, thats only 300 points away from my GTX280M, and the CPU deficit would cause that more or less.

 

great job...now go in and set everything to performance..and your score should go up some.

 

did the 9.12 hotfix really fix the MW2 issue ?

 

P7358 with X9100 @ 3.7Ghz and GPUs 60Mhz over.

http://img38.imageshack.us/img38/7510/30499217.png

 

info i got back today after speaking with someone who i can not name said this is one of the reasons "another company" did not release them.. just fyi... he would not go into to much more detail except he said they simply didnt see much if any real advantages to the 4850's so i may for sure grab me a m17 with 3870's now.. and just forget the 4850's imo

 

well the only thing with the 3870 vs the 48xx is the 512vram and the cripple aa runs.