Portables4Gamers Comparison: HD3870 x2 vs 8800M GTX SLi

So basically Portables4Gamers, a french site for performance notebooks compared the 8800M GTX SLi with the ATi Mobility 3870 x2. Basically they compared the Alienware M17x with the OCZ Whitebook/Arima W840-DI. The results are quite interesting to say the least (note: link is translated from french to english, courtesy of Google):

http://translate.google.com/translate?u=http://portables4gamers.com/?p=5142

discuss.

Edit: this holds some relevance as the 8800M GTX = 9800M GT.

 

Lol @ Forceware 174.82! I was wondering how the 8800s performed so poorly in the DMC4 benchmarks.

 

why did they use very old driver for 8800? 174.82? now is 180.70.

 

Maybe those are the drivers it came with (for the m17x). But I'm actually searching for any considerable differences that drivers may make in performance. So far, I got a UT3 (though it wasn't tested by P4G) comparison with the different drivers from NVidia and their effect:

http://www.alcomsplace.com/ut3_bench_forceware_drivers.htm

 

It's up to games that you run. I think UT3 need not powerful vga to run so nvidia (also ATI) must not push up performance for this game. Also some drivers don't push up performance for all vga. you should see the release note of drivers to see the games that got more efficient from those driver.

 

Also, newer drivers don't automatically mean better performance for every game and in general.

 

It looks like sli wasn't working properly in some test. Which is very typical of a out of the box alienware system, hence why there is a sli guide in that section of the site. Devil may cry does usually run better on the radeon system but not 40+ fps faster.

 

then how come none of you are doing numbers like them? do the french know something you guys dont??

edit: just checked orb and it's about 10 machines doing better than what they have posted for 3dmark06 sm2.0 & 3.0 scores with lower cpu scores....

 

You mean for the games? Minus Crysis, I don't think any of the games they tested have been tested here by NBR members.

 

dmc4 has a demo with the benchmark program why not test it for us "pretty please". I'd prefer to see you post this, because compared to some others with these systems you show a interest in actually helping the nbr community. Here are some links to the demo as well as the benchmark for dmc 4.
http://www.fileshack.com/file.x/12272/Devil+May+Cry+4+Benchmark+Tool
http://www.fileshack.com/file.x/12271/Devil+May+Cry+4+Demo

 

im downloading that demo as we speak bwhxeon.
9800's still out performed it in dual core mode at world in conflict.

 

Something to keep in mind for dmc4, back when brainer was testing it, he had to make a sli profile for his 7950's to work properly. I'm not sure if this has been remedied in more recent drivers or not.

 

we shall see in a bit....it's at 86 percent.

 

Sure, downloading now (but I'll probably post tomorrow because 1) my internet connection is slow and 2) it's 11:30 here

 

it comes with it's own benchmark tool....

well, that was interesting....
my lowest was scene4 at 97.87 and highest at 133.93 scene1

wonder what test they we're using....
ok, never mind...i see what it is now...

 

Downloading as well. 1 hour and 39 minutes remaining, let's see how well my 8800m GTX's on SLi @ Stock paired with a Q9450 do against the system used in Portable4Gamers' test.

 

thank you aeris!!! you mah gurl, right about now!! with your 8800m gtx's

*LOL*

 

wow, 1 hour and 16 minutes remaining for me... boy Verizon sucks. Isn't it shameful that they don't have FIOS in my area yet, considering where I'm located. I'm stuck with terrible Verizon DSL that cuts off every 2 weeks.

 

You are welcome! I like helping to research for the truth too!

Aw. You are about 11 minutes ahead of me! :/ I have a local ISP's cable internet at 1.5 Mbps / 383 Kbps which is lagging a bit right now, mine cuts off every once in a while, sometimes more than 1 time a week.

 

well, that's not good! both of you have over an hour to wait... and just where are the both of you located??

 

As of right now, 59 minutes of downloading left, I am in Mexico in a town called Tecate, in Baja California (the bigger and much more densely-populated counter-part of Tecate, California (which only has about 220 inhabitants)).

 

i know where that is..

 

Ah! I just noticed the "Location: South California" under your avatar. Haha.

47 Minutes remaining... going to update my ForceWare Driver from Dox's 180.44 to Dox's 180.70 before benchmarking.

 

bench them both.

well, i sort of knew where you we're, but microsoft streets and trips showed me exactly...lol

 

Sure.

Hehe.

 

Okay, game trial downloaded and installed, doing benchmarking!

 

make sure you do all your settings first, then click apply.

 

Okay.

I am just done doing the first round of benchmarks with the drivers I have right now, I will post validation screenshots and a clean format once I am done with the 2nd round with the drivers I am about to download (180.70 by Dox).

Just for a brief comparison, I got 231.31 FPS at Scene 1 with DX 10 at 1280x720 and 134.93 FPS at Scene 1 with DX 10 at 1920x1200, while the HD 3870's got 176.19 FPS at 1280x720 and 107.79 at 1920x1200. (Posted DX 10 results because I was unsure whether the people at Portables4Gamers used DX 10 or DX 9, my results are higher anyway! )

Sidenote: I am doing DX 9 as well.

 

i knew there was something up with them test.

and everything was set to ultra high and stock cpu and gpu right?

 

Ugh, it was not set to Ultra High, but to High since that was the default in the settings of the game, stock CPU and GPU, ... was it really supossed to be set to Ultra High?!, I spent the last 3 hours doing the benchmarks.

 

yep. but it should have only taken per run 8 minutes. and only do 1920x1200.. that other resolution was because alienware couldn't do 1680x1050. so do dx10/dx9 @1920x1200 ultra high. just to be on the safe side. although they never state the settings used....

so out of that i got..i wont be buying an alienware anything if it can't do all the resolutions. which makes no sense to start with.

 

The reason why it took so long is because I did both resolutions, at DirectX 9 and 10, with both drivers... Oh well, I will run them at Ultra High 1920x1200 and discard all of the other results.

 

Okay guys, here we go!

Comparative Graph:



Devil May Cry Performance Test / DirectX 9 / 1920x1200 / Max (Super High Everything) Settings / 8800m GTX SLi @ Stock Speeds / Q9450 @ Stock Speed:

Validation Image.

---

1st Scene: 130.68 Frames Per Second. (8800m GTX SLi)

VS

1st Scene: 107.79 Frames Per Second. (HD 3870 CF)

---

2nd Scene: 90.31 Frames Per Second. (8800m GTX SLi)

VS

2nd Scene: 94.98 Frames Per Second. (HD 3870 CF)

---

3rd Scene: 159.83 Frames Per Second. (8800m GTX SLi)

VS

3rd Scene: 121.9 Frames Per Second. (HD 3870 CF)

---

4th Scene: 90.35 Frames Per Second. (8800m GTX SLi)

VS

4th Scene: 81.28 Frames Per Second. (HD 3870 CF)


Devil May Cry Performance Test / DirectX 10 / 1920x1200 / Max (Super High Everything) Settings / 8800m GTX SLi @ Stock Speeds / Q9450 @ Stock Speed:

Validation Image.

---

1st Scene: 124.76 Frames Per Second. (8800m GTX SLi)

VS

1st Scene: 107.79 Frames Per Second. (HD 3870 CF)

---

2nd Scene: 91.13 Frames Per Second. (8800m GTX SLi)

VS

2nd Scene: 94.98 Frames Per Second. (HD 3870 CF)

---

3rd Scene: 151.77 Frames Per Second. (8800m GTX SLi)

VS

3rd Scene: 121.9 Frames Per Second. (HD 3870 CF)

---

4th Scene: 90.35 Frames Per Second. (8800m GTX SLi)

VS

4th Scene: 81.28 Frames Per Second. (HD 3870 CF)

---

Final Verdict: 8800m GTX SLi wins in 3 out of 4 tests (Lost to the HD 3870 CF in both DX 9 and DX 10 Scene 2).

Portables4Gamers did not disclose what settings nor what version of DirectX they used to benchmark the OCZ Whitebook on Devil May Cry 4's Performance Test.

Test ran on a Sager NP9262 with Intel Core 2 Quad Q9450 @ 2.67 GHz / 8800m GTX SLi / Triple 200 GB @ 7,200 RPM Toshiba Hard Drives on an Intel Matrix RAID 0 array, and with 4 GB of Apacer RAM @ 800 MHz in 2 SODIMM's.

 

I've done my test. I got 1st 101 2nd 80 (so I doubt that P4G got 94) 3rd 120 and 4th 81.

 

just got back from church, gonna install and test DMC4

 

wow suprising results imo.

 

Ok, here are the results I got with everything on DX10, Super High and AA set to 8x (can anyone show me how to take screenshots in game? print screen doesn't work).

Scene 1: 79.31
Scene 2: 55.97
Scene 3: 86.84
Scene 4: 62.03

Side note: this looks like a great game, graphically and in terms of the fighting. Reminds me of God of War. Anyways, I'm going to play with the settings a bit to see of any significant effects and report back.

Edit: Results with everything the same as before, except no AA

Scene 1: 97.81
Scene 2: 79.16
Scene 3: 112.82
Scene 4: 78.89

 

Would be nice to get input from E-wrecked and Hooper, or even RobM to see if a faster processor and more cores make a difference if any.

 

I looked at more reviews on their site, and I think you're right about the SLi not working properly. They did a test with a Clevo M570TU with a Single 8800M GTX against the W840 with the 3870 x2, and these are the results:

http://translate.google.com/translate?u=http://portables4gamers.com/?page_id=4535

But then, if that's the case, then why are the Crysis results in line with an SLi'd system?

 

with same driver same base system (with no tweak, modded or oc'ed) scores should be in the same way. I've oc'ed my cpu to 2.8 ghz and got about 2fps more than stock

 

Hey, I have same problem. print screen doesn't work in this game

 

downloading now..will bench when done. 10 minutes til completed.

 

Q6600 @ 2.4 stock
8800M GTX SLI - stock
1920x1200
All Super High - no AA
180.48
Vista x64

 

Nice camera shot hehe, I used FRAPS to take mine.

 

oh snap....forgot about using them!...DUH! lol

 

at any rate, i think it is safe to say they didn't have a properly working system when they ran the benchmark.

Reliable tests eh? not a single person thought "oh this may not be right"

 

exactly what i was thinking... apparently some did.

the 8800's are right up here with the 9800's and then here comes a test and you 8800's just fell in to non existence...lol

all i got out of that review was...leave alienware alone. *LOL*
not saying the ocz didn't drum up legit numbers, but the 8800 sli sure in the hell didn't!

 

yeah, I think it's not so much SLi not working, else those Crysis numbers would be lower. Could be a driver issue that hinders SLi scaling for those games. I already pointed out how close the scores on the other three games were on the M17x compared to the numbers they got with a Clevo M570TU with an 8800M GTX.

 

kinda like the ATI driver issue with our cards? So, you guys mean to say that our 3870s will gain 50% in performance when appropriate drivers are available?! Dec 10th buddy, dec. 10th.. then we'll see what these cards are holding back.

oh, installed dmc4 bench.. bout to run.