OC'ed 3870 vs 9600GT - which is better when OCed to the MAX

[asb2106]

REDONE AS OF 3/27/08

First an overview of the rig and tests to be run

CASE - antec p180
MOBO - P5B deluxe
PROC - e6600 @ 3.4ghz
RAM - OCZ GOLD @ ~750mhz
HD - WD 250GB HD
PSU - ANTEC tru-power 550
OS - Windows XP Pro
ALL TESTS WILL BE RUN DX9 (If I get time later I will update scores with VISTA results in DX10 mode, atleast the ones that can do DX10)

I will be running a series of games and benches at 2 resolutions - 1600x1200 & 1280x1024 **For Crysis I will also run 1920x1200
The tests to be run:

3dmark06
FurBench
aqua mark 3
Crysis
Unreal Tournament 3
FEAR

CPUz and GPUz of running speeds

And the 3870

3dmark06 @ 1280 x 1024

3870

9600

3dmark06 @ 1600 x 1200

3870

9600

3dmark06 @ 1600 x 1200 4x AA 4x AF

3870

9600

FurBench @ 1280 x 1024

3870

9600

FurBench @ 1600 x 1200

3870

9600

FurBench @ 1600 x 1200 4x AA

3870

9600

FEAR @ 1280 x 960 4x AA 16x AF Shaders on everything maxed!

3870

9600

FEAR @ 1280 x 960 NO AA OR AF everything maxed!

3870

9600

FEAR @ 1600 x 1200 4x AA 16x AF Shaders on everything maxed!

3870

9600

FEAR @ 1600 x 1200 NO AA OR AF everything maxed!

3870

9600

UT3 1280 x 1024 4x AA 16x AF Maxed out 5 & 5

3870

9600

UT3 1600 x 1200 4x AA 16x AF Maxed out 5 & 5

3870

9600

OHH and I want to mention, this is my backup rig, I was gonna run these tests on my main rig, but I found it would be way to hard messing with the water cooling.

I do think however that this hardware is more than enough to give good results and not bottleneck the vid card

I want to apoligize for not running crysis, I am having issues getting it running right now. So I will keep on this and update when I can.
Now that I am done with these tests, I going to build my new RIG! q6600, rampage formula, ddr2 1066, 2x 3870, raptor. I will post CF results here for your comparisons also!

MY CONCLUSION

I think either one of these cards will perform well for you. People say that with AA & AF on that the ATI parts cannot keep up, I didnt see that too much. In some games, yes the Nvidia card will do better, but for the most part the 3870 can keep up.

Honestly I think your choice should be guided by what chipset you are using. With the prices almost exactly the same, and performance very similar there isnt much difference.

So if you have a intel or AMD chipset, I say go ATI
If you have a nvidia chipset, go with the 9600GT.

Either way, Im happy I was able to do these tests. First hand, I have been able to see how close they really are.

I CALL IT A TIE!!!

 

[yogurt_21]

cpu scores are 800pts off, run the same settings to compare.

 

[asb2106]

cpu scores are 800pts off, run the same settings to compare.

oh good call, I did, and its funny, the score was less. Ill put it on there, 1 sec

 

[oli_ramsay]

and one is on XP and one is on Vista. Dunno how much that effects scores by...

 

[asb2106]

and one is on XP and one is on Vista. Dunno how much that effects scores by...

I just noticed that myself, I am going to install XP right now and re run the test. I had forgot about that, because it does create a difference of about 500 points. I will have it up within the hour..... sorry all

 

[Solaris17]

ya when i switched to vista from xp x64 my score went down by like 2k

 

[erocker]

I just noticed that myself, I am going to install XP right now and re run the test. I had forgot about that, because it does create a difference of about 500 points. I will have it up within the hour..... sorry all

Doh! Lol, you'll get it! I love a nice simple comparison thread like this.

 

[phanbuey]

We should start taking some bets... this is gonna be a close. I put $5 on the HD 3870 w/ 0 AA... 9600GT if 4X-8X AA.

 

[DOM]

cpu scores are 800pts off, run the same settings to compare.

5870-5236=634

 

[asb2106]

UPDATED!!!!! just ran in XP, I couldnt get the mobo to stable out at 420FSB again, but its at 415.

Im not done with this card though! Im gonna try to tweak every last frame I can without a mod

I put 1.68v @ 3.78 with no go, so I ran 1.62 @ 3.73 volts through the q6600 to stabalize, So Im gonna restart and drop back to 3.5ish, this much volts scares me!

 

[asb2106]

hey peeps, check these voltages for me, would you.....

Looking at these, it seems like this may be one of the problems that I cannot OC my q6600 any more....

I had a PC power cooling and sold it for more than I paid! so I put my antec 550 back in. I didnt realize before that my voltages were so bad!!

 

[ThatGuy16]

What program is that? Everything i use shows different voltages (not believable).

I have always wanted to see what mine are in windows

 

[asb2106]

What program is that? Everything i use shows different voltages (not believable).

I have always wanted to see what mine are in windows

everest!! It is awesome, very useful, and has some great features that go with vista

** I have version 4.2

 

[asb2106]

I was gonna sell off my 9600GT when I got my new 3870, but I think Im gonna keep it around for another gaming rig! Maybe I can get the wifey to game with me! I got her to play some portal on the orange box tonight and she is starting to like it!

 

[asb2106]

We should start taking some bets... this is gonna be a close. I put $5 on the HD 3870 w/ 0 AA... 9600GT if 4X-8X AA.

thats actually a great idea,

the reason I have this 9600 is because my 3870 is dead, and waiting for RMA. I am ordering another 3870 soon too. I got the itch and found this 9600 at BestBuy for 220, figured it will keep me happy until I get my card back.

SO, when I get my 3870, I will run the tests again, with AA at different levels, at low rez, at high rez, pretty much everything. Im thinking Ill have my other 3870 soon, week or 2. So when I get it, Ill write up a great thread with a massive list of results. Im really looking forward to this!

 

[DanishDevil]

Attaboy! Can't wait to see the results.

 

[ThatGuy16]

everest!! It is awesome, very useful, and has some great features that go with vista

** I have version 4.2

I just downloaded it, it can't be right. It say's my 12v is 2.56v

 

[gOJDO]

3D Mark is not representing the real life performance! Even the 2900XT scores higher than the 8800GTX, but the 8800GTX is kicking the crap out of it in all games. Try benching the cards with some games and you'll get the answer.

 

[farlex85]

3D Mark is not representing the real life performance! Even the 2900XT scores higher than the 8800GTX, but the 8800GTX is kicking the crap out of it in all games. Try benching the cards with some games and you'll get the answer.

I would have to disagree w/ that, as many games are optimized to run on nvidia, or less often ati. Just b/c a card performs better in some games, doesn't make it the better card. If you were able to bench on a massive number of games and compare totals (a la tom's hardware), then perhaps. 3d Mark is not a perfect indicator, but a more effective one imo (all other things being equal).

I'm putting my bet on the 9600gt, 100 point spread.

 

[EastCoasthandle]

I would have to disagree w/ that, as many games are optimized to run on nvidia, or less often ati. Just b/c a card performs better in some games, doesn't make it the better card. If you were able to bench on a massive number of games and compare totals (a la tom's hardware), then perhaps. 3d Mark is not a perfect indicator, but a more effective one imo (all other things being equal).

I'm putting my bet on the 9600gt, 100 point spread.

Good response
If you don't like what the OP has done, you don't have to be here IMO.

 

[asb2106]

Which of the 2 video cards offer better IQ in:
Desktop
Movies (both dvd and online streaming)
Games

Thanks

From yesterday when I got it, I ran a few games from the orange box, ran about 6 3dmarks, and did windows re installs and quite a bit of desktop work. With what I have done I would call the IQ comparison inconclusive. I did not see any significant differences on either side. - I will be paying closer attention to this though!

For the poster below - This morning I will run some tests on the games I have (orange box, crysis, COD4, FEAR, UT3) Ill post these results, mainly so I can use them for my full test I want to run.

I am writing up a to-do sheet of all the tests I want to run for my comparisons. I want to make it very elaborate with a mix of benches and lots of games! When I get that wrote up, Ill post it here, I would like to get feed back from more tests I could run! Give me 30 mins to an hour.!!!!

 

[asb2106]

List Of Games, & Benches I Want To Run

Please let me know if there is anything I should add....

3dmark06
FurBench
Cinebench
aqua mark 3
half life 2
COD 4
Crysis
Unreal Tournament 3
FEAR

I know the list is short, please add any you think I should run with it....

EDIT: IM ording the 3870 today for delivery this week. I plan to start testing on as soon as I get it, but I do not know how far I will get. I have a trade show in Chicago that I have to be at pretty much all week. I will try to wrap up all the tests, and have good results by next weekend

 

[gOJDO]

I would have to disagree w/ that, as many games are optimized to run on nvidia, or less often ati.

That argument is too weak. There are a lot of games made before R600 and g80 were even planned. In many of these games the g80 wipes the floor with the R600. On the other hand the R580 was doing the same to G71. So the "optimized for X brand" argument is not valid in such cases. Of course there are games like Crysis which are heavily optimized for a certain architecture(in that case g80), but a good architecture(ex. R580) performs good although the game was not optimized for it.

Just b/c a card performs better in some games, doesn't make it the better card.

This is very relative. If the card is performing better in the games you are playing, then it is better.

If you were able to bench on a massive number of games and compare totals (a la tom's hardware), then perhaps.

Actually that was my advice since I consider 3D mark as non-relevant(for anything) synthetic benchmark. With benching the cards in various games you'll end up with a solid conclusion based on facts. BTW, THG VGA charts for games are so poor and the games are outdated. I won't make any decision based on these numbers.

3d Mark is not a perfect indicator, but a more effective one imo (all other things being equal).

It is inaccurate and it is misleading. That's why the 2900XTX is achieving higher score than 8800GTX, but the 8800GTX outperforms the 2900XTX in 95% of the games. Also, clock for clock, Phenom is achieving same score like Yorkfield, but Yorkfield(same clocked) wipes the floor with Phenom in every real-life application known to man kind.

 

[DaedalusHelios]

Actually that was my advice since I consider 3D mark as non-relevant(for anything) synthetic benchmark. With benching the cards in various games you'll end up with a solid conclusion based on facts. BTW, THG VGA charts for games are so poor and the games are outdated. I won't make any decision based on these numbers.

Then help to make a better chart. I wouldn't complain unless I was willing to help make a better one.


I can't find a better game resource on the internet in chart form can you?

 

[mandelore]

off topic, but asb2106, your sig is enormous, far too big.

not complaining, just you may get a pm from a mod to ask for a reduced sig size. Just a friendly heads up

ohh, and youl find 2900s beating out gtx's because of the extreme cooling used for the wr bench runs.

And as for better cards.. the coding of a particular game often dictates a cards performance. You may have a technically superior card, but if the game favours one particular style of hardware, it means sweet bugger all.