Crossfire 4850, Low 3dmark06 Score

K

Kelv

Limp Gawd
Joined
May 10, 2008
Messages
385
I just installed a brand new system and while it was working a lot better than my 7800gs, I felt that it should have been better, I finally tested it with 3dmark06 and I got a score of 11589. Here's the link.

On both the cpu tests, I never got above one frame a second, for most of the time it was 0, Is that reasonable for a e8400? Nothing is overclocked btw. I thought that a reasonable score for crossfire 4850 was in the low 20,000's.

These are my specs
e8400
gigabyte x48 ds4
4850 in crossfire
2 x 2gb corsair xms2
 
Kelv said:
Nothing is overclocked btw. I thought that a reasonable score for crossfire 4850 was in the low 20,000's.
Click to expand...

Hello no. 3DMark06 is useless as whole, and with 48x0 it is even more useless. Most single 4850 3DMarks were between 11k and 15k, which absolutely doesn't reflect the power of this little card. And i don't think that in this case CF will increase ratings, as it already cannot show performance of one card.
 
Kelv said:
I just installed a brand new system and while it was working a lot better than my 7800gs, I felt that it should have been better, I finally tested it with 3dmark06 and I got a score of 11589. Here's the link.

On both the cpu tests, I never got above one frame a second, for most of the time it was 0, Is that reasonable for a e8400? Nothing is overclocked btw. I thought that a reasonable score for crossfire 4850 was in the low 20,000's.

These are my specs
e8400
gigabyte x48 ds4
4850 in crossfire
2 x 2gb corsair xms2
Click to expand...

Not that you should care about 3dmark at all anyway.... but do you have Vantage? If so, what is that score?
 
Whats with all this 3dmark06 is unimportant. Btw I was just playing the crysis demo on all high 1280x1024, 2x AA, and I would say my average fps must have been around a 25.

For those that have the same or similar setup as I do, what are your 3dmark scores? I don't have vantage, I thought I read somewhere that it was DX10 only and I have vista.

I'm going to try reinstalling the cards using this:
1.) Start by removing the drivers first with the ATI Uninstall Utility via the Add/Remove Programs of the Control Panel.
2.) Then uninstall the current Display Adapter(s) via the Device Manager.
3.) Power down and remove second card (if applicable).
4.) Boot into “Safe Mode” and use Driver Cleaner Pro to remove misc. drivers & registry entries. Follow up by using “Search” in the “All Files and Folders” with “ATI” as the search criteria for “All or part of the file name”. This will help clear any stragglers.
5.) Boot normally and install Catalyst flavor of choice.
6.) Power down and install second card (do not install Crossfire bridge at this point!).
7.) Boot and CANCEL out of Hadware Installation Wizard.
8.) Find second card in Device Manager (the Display Adapter with the yellow triangle) and right-click to “Update driver”. Use specific location and browse to original Catalyst install folder. (Normally C:/ATI/SUPPORT/8. xxxxxxx.xxx/DRIVER/(select folder based on current OS). This should install the driver on second card.
9.) Power down and install Crossfire bridge.
10.) Copy the folder provided at the bottom of this post into your Crysis directory: Electronic Arts>Crytek>Crysis>Game and replace/overwrite everything.(Make a backup of your original, just in case!) It is just the entire "Config" folder from the Crysis SP Demo. Use you own copy if you feel more comfortable.
11.) (Tech)Power up, open Catalyst Control Center to “Enable Crossfire” and ENJOY!
Click to expand...

The first time I did it, I just installed the two cards and put in the disk.
 
n00dlejester said:
I might be hi-jacking this thread a lil bit, but why exactly is 3dmark06 not a good benchmark?
Click to expand...

First of all, you can't compare 2 computers with same GPU. You should have same CPU, same amount and type of RAM, same HDD, everything same except the GPU. Then the total score is relevant. Otherwise you should concentrate only at SM2/SM3/HDR ratings, you should not take CPU score into account at all. Then there is a problem of driver optimization - all drivers are optimized to run 3DMark as good as possible. But that doesn't reflect the real gaming experience, your games are not same all the time, compared to 3DMark demos.
 
Even for a single 4850, is 25 fps normal for all settings high 2x AA? This is on the crysis demo around the beach. Also, should a e8400 get about 0-1 frames per second in the 3dmark 06 cpu test.
 
Kelv said:
Even for a single 4850, is 25 fps normal for all settings high 2x AA? This is on the crysis demo around the beach. Also, should a e8400 get about 0-1 frames per second in the 3dmark 06 cpu test.
Click to expand...

First of all, start comparing comparable. Standard results on web are with default settings - 1280x1024, no AA, no AF.
 
Well I found out that with crossfire disabled, I still get the same performance in crysis and in 3dmark06. Crysis without 2x aa is around 35 fps average with all high on 1280 x 1024.
 
That's because Crysis is a horrible gauge of anything else but the ability to play Crysis. It has horrible multi GPU support, and according to many, horrible single GPU support ( ie, bad coding )
 
SilverSliver said:
It has horrible multi GPU support, and according to many, horrible single GPU support ( ie, bad coding )
Click to expand...

I wouldn't go that far. Multiple gpu support on my gx2 was outstanding. The current drivers for xfire on the 4850's does not work with crysis. some reviews have outstanding xfire scaling with crysis, but they have their hands on some beta drivers. in this case it's an ati driver issue, not a crysis code issue.
 
Is there a way I can disable the first card, and only run it on the second? I'm figuring that if you disable crossfire, it only runs on the first card. I want to make sure that the second card is running fine even if it's not being used.

The reason why I want to do this is because I dropped a screw on the bottom card, I was unscrewing the top one, and if that damaged the card, would the damage be obvious? Like would I see artifacts or would the card not even turn on? Or would it be impossible for me to see the damage.

I'm gonna purchase Crysis tonight, I'm expecting that it will probably run better due to patches and stuff like that.

Looking at the 3dmark06 score, http://www.overclockersclub.com/reviews/powercolor_hd4850/14.htm that site has comparable scores to what I received. Does crossfire work with 3dmark06?
 
Take the both cards out and then turn on the system with the second one only.
But I'm almost certain that your card is okay, if it's damaged I imagine it wouldn't boot or would frequently reboot your system, get artefacts, etc.
 
Thanks for the info. I just tested it without crossfire enabled, and I got a 11495, basically the same score, at least Crossfire didn't hurt the score.
 
I have an update, I switched around the crossfirex cable, I'm not sure it caused anything but I did bump up my 3dmark06 score to 12500 instead of 11500, but there is something definitely wrong.

On the first scene, in the spaceship, at the very beginning it hovers around 60-70 on a single card, but with crossfire enabled, I get about 95-100+. This is basically the same throughout the entire first scene. In scenes where the single card gets 22, crossfire gets 32, 50-70. Differences like that, which are big. But on the next scene with the wisps going through the forest, I get the exact same fps with either setup, I believe that it is the one scene at the begginning that accounts for the extra 1000 on the score. Further on, I tried the cod4 demo. My frame rate in the demo level on the road at the beginning is around 65 on both crossfire and a single card solution, and I did select the duel video card box while using crossfire.

Btw: check out this benchmark, obviously I don't have as great of a cpu, but that, nor the ram should be reducing the points by 7000.

I decided to test one of the cards in each slot, to make sure there wasn't a mobo problem, and the card got the same fps in both of the slots.

EDIT: It does make huge performance leaps in DMC 4, over 90 percent!
 
I get about 13700 with CF, about 11k without CF I believe. My CPU limits me severely.
 
i scored 18551 with system in sig on vista 64 stock cards. 4.0 wolfy
 
soulesschild said:
I get about 13700 with CF, about 11k without CF I believe. My CPU limits me severely.
Click to expand...

More ignorance abounds from me: in the grand scheme of gaming, how important is a strong processor? I've been under the impression that it's helpful to have a top of the line processor, but it won't affect the gaming experience as much as a top of the line GPU would.
 
n00dlejester said:
More ignorance abounds from me: in the grand scheme of gaming, how important is a strong processor? I've been under the impression that it's helpful to have a top of the line processor, but it won't affect the gaming experience as much as a top of the line GPU would.
Click to expand...

I believe that atm you need a 3.2 Intel core duo to keep a single gpu from bottlenecking. Just real life gaming experience mostly with cod4.
 
3DMark06 was extremely CPU bottlenecked for me as well. Bumping my E8400 from 3.0GHz to 4GHz with my 8800GTS shot my score from 12,800 to over 14,200. Not only did my CPU score go up, but my shader scores went up almost 1000 points.
 
I'm definitely going to try and overclock a little to test how much of an improvement I get. One compliment I give to intel, though this may be more of a generation improvement, the stock cpu coolers are such an improvement over what I received with my 3500+ 64. What did you use to overclock your e8400 seth?
 
Well, I'm having low performance with both my 4850's, using an Intel BX2 board. When is disable Xfire, things run smoothly. When I have xfire enabled there is a huge performance hit. I did a complete re-install of Windows XP along with the newest correct drivers and the hot fix ATI drivers.

What I've found out is, from reading forum post after forum post is, you need a non-nvidia chipset with dual 16x pci slots for these babies to come alive.

Having both of these cards in my system running on my Intel BX2 board, both slots are defaulted to 8x. One slot could be 16x, the other 4x. The reason for the bad performance.

If you are having less performance than what you think you should have, look at your board, research it and find out what your slots are doing exactly.

I'm moving over to a MSI Intel 35 board in the next few days.

Hope that helps.
 
rthefley: which bx2 board do you have now? I have the xbx2 bad axe 2 and was thinking of crossfire myself but figured it would be a waste on this board. Also don't buy a p35 board! Buy an x48 board if you are buying a new board. P35 boards get terrible performance in xfire.
 
well I have a x48 so I'm all set there, I'm going to try overclocking my cpu right now. I heard though that hopefully some new drivers from ATI will help the 4850's performance in Crysis. I know that one of the benchmarks was getting a great performance increase in crysis from crossfire, and I heard that they were using beta drivers, it shouldn't be too hard to find that benchmark.
 
When I try Vantage with mine, the graphics looked all jacked up, even using default speeds. Other demos and games look fine. I don't recall if I tried it without the crossfire, I'll have to see there. Man, wouldn't just suck if all I could do is play COD4 really really fast? :)
 
It seems like for now, 3dmark is bottlenecked by my processor. I increased it with increments of 10 mhz from 333 to 370 and every time my 3dmark score increased. Overall my score went from 12500 to 13100 by increasing my cpu score from 3.00 ghz to 3.33ghz.
 
3dmark06 default scores are VERY CPU bound because they are run at 1280 x 1024 or whatever so they're not really good indicators of the new generation of GPU's because they're simply way faster than they were before. If you can, try running it at 1920 x 1200 w/ AA and AF and you'll probably be shocked by how high the scores are.
 
With CF disabled on mine I get 12,805 in 3dm06, with it enabled I get 17,644.

My cpu runs a 400mhz fsb for 3.6ghz...
 
fasteddy said:
With CF disabled on mine I get 12,805 in 3dm06, with it enabled I get 17,644.

My cpu runs a 400mhz fsb for 3.6ghz...
Click to expand...

i havnt had time to bench my 4850 yet but im running e6600 at 3.3ghz and i came from an 8800gts 512mb and i hit 13000 ez. your cpu is faster and the card is definately faster than my previous set up. i think something is wrong.
 
Probably as we have immature drivers for the 4x series radeons at present.

No doubt things will improve with time...
 
a good indication is checking your SM scores with and without crossfirex enabled.
 
I finally decided to try a higher overclock, I went to 390 x 9 (3.51 ghz)

Got to 16951. coming from 1250, I was really surprised. In fact the jump from 3 to 3.33 had less of an impact than the jump from 3.33 to 3.51.
 
+1 more person who is severly CPU limited in 3D Mark 06,

Is the Devil May Cry a very good benchmark? I get an A+ with 1680x1024 with maxed out settings. Can't remember what I get at its default settings.

One of the problems I have with the Benchmarks [H] runs is that I can't duplicate the benchmark on my own computer and compare my scores with them. Especially sense most of todays games don't have a benchmark feature like the old Quake games use to have.
 
You should have scored more than this with your PCIe 2.0 16x/16x X48 MOBO.
20080723-3DMARK06-19710-E8400-4GHz.jpg


Kelv said:
I just installed a brand new system and while it was working a lot better than my 7800gs, I felt that it should have been better, I finally tested it with 3dmark06 and I got a score of 11589. Here's the link.

On both the cpu tests, I never got above one frame a second, for most of the time it was 0, Is that reasonable for a e8400? Nothing is overclocked btw. I thought that a reasonable score for crossfire 4850 was in the low 20,000's.

These are my specs
e8400
gigabyte x48 ds4
4850 in crossfire
2 x 2gb corsair xms2
Click to expand...
 
well look at your CPU, its overclocked a lot more than mine.

I feel no need to overclock however when it doesn't help the performance in any real gaming at all (maybe a few).
 
I scored 14400 on an old intel bx2 board with an e6600 at 3.2ghz, 2gb ram and 4850cf. Average score considering how crappy this board is and how bottlenecked my video cards must be right now.
 
You must log in or register to reply here.
Back
Top