• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

4870x2 Performance

D

Ducman69

[H]F Junkie
Joined
Jul 12, 2007
Messages
10,561
I said I wouldn't go back to ATI after having driver issues before, and NVidia always being so painless, especially my 8800GTX... but I never learn. :p But wondering if anyone here can help me out regarding lackluster performance.

Specifications are:
Vista 64 bit - Fresh Install
Gigabyte GA-G33M-DS2R
Q6600 @ 2.88Ghz (was 3.2Ghz, but just keeping case cool)
2x2GB RAM @ 800
2x640GB drives in RAID0
4870x2 @ 777/920mhz

3DMark06 gives a lowly score of 15,789. :confused:

633317514784490542.jpg


I was really expecting a score much higher than my single 8800GTX, but I'm not even hitting 16K. Ideas?
 
Oh yeah, forgot, mobo is on the latest available BIOS (8Fe IIRC). Also has latest drivers for everything else onboard updated for it, no bangs in device manager or anything like that.

And lastly I'm on the latest 8.9 Catalyst driver.

Or are people w/ similar systems also getting such low 3DMark06 scores w/ the 4870x2?

--------------------------------------------------------------------
Edit: I've been doing some more research on my own, and it appears that perhaps only ONE of my GPUs in the 4870x2 may be working, hence the low score.
http://forums.amd.com/game/messageview.cfm?catid=260&threadid=99015

I also show "PCI-E 2.0 x16 @ 16 2.0" under Bus Interface in GPU-Z, and ATI Crossfire as "Enabled (2 GPUs), Sideport Off (unsure on Vista64)".

My motherboard only has a single PCI-E 1 x16 slot, not 2.0. :(
 
You sir fail for just believing 3dmark 2006. Load up something like Crysis and try and see how well your 8800gtx does against that or other games like COD4. The problem with 3dmark its not only a gpu sensitive benchmark but also cpu. Trust me I am being nice, wait til some others read this.
 
Still hard to understand why an arbitrary score in a benchmark means your equipment is faulty, or less capable.
 
dood, your score is fine. 3dmark06 is just a synthetic benchmark and cares more about your CPU speed then how fast your videocard is. If you overclocked your Q6600 to 3.6ghz you would see 20,000.

Please dont use a 3 year old benchmark to benchmark a brand new videocard....
 
Xenozx said:
dood, your score is fine. 3dmark06 is just a synthetic benchmark and cares more about your CPU speed then how fast your videocard is. If you overclocked your Q6600 to 3.6ghz you would see 20,000.

Please dont use a 3 year old benchmark to benchmark a brand new videocard....
Click to expand...

+1

Fire up some Crysis Warhead or something.
 
Q9450@3.8GHz and a stock 4870x2 on Vista x64=21345 in 3dmark06
BUT, Crysis Warhead runs one time in 10 or maybe more starts, crashes random and so on.
 
My QX9650 at 3.6 GHz hits 21k... overclock it to 4.0GHz and its hitting 22.5k

3dMark06 is a very CPU limited program

And for reference, 3dMark Vanatage i'm getting X7700, up from X7300 with a 3.6GHz quad, so the GPU is seeing CPU limitations even at 1920 x 1200
 
DeadSkull said:
You bought the card solely for 3dmark?
Click to expand...
Yes... I bought my card to just run benchmarks all day. :rolleyes:

I run benchmarks to get an apples to apples estimate of performance. If there is a better benchmark available that you recommend that I can use to see if my hardware/drivers are running properly, please list them rather than many here getting so damn defensive.

My concern, as noted, is that I may be experiencing driver or hardware compatibility related problems, hence my worry that my second GPU may not even be running. It never seems to clock up according to GPU-Z, even during games, and my 3DMark score seems low to me. People supposedly have seen their 3DMark06 performance double by switching PCIe slots, but I don't have that luxury. Sorry for asking for help, and sorry for being a little frustrated, but $500 is a lot of money for me, and I want to verify the result of all that cash.
 
Well your quad is at 2.8 GHz and 3dMark06 default is at 1280 x 1024 resolution, so it is very CPU bound. Crank that back up to 3.2 GHz and you'll notice a score increase.

And since you don't game at 1280 x 1024, the score won't reflect real performance, which will be a huge increase over the 8800 GTX (previously I had a 8800 Ultra).
 
Code: 
    .--.._                                                                
    | _   `    ARE YOU ABSOLUTELY SURE THERE'S NO WAY IN THIS BUG DATABASE
fsc |(")       TO RESOLVE THIS BUG AS "RESOLVED/PLEASE KILL YOURSELF"     
    J' `.                                                                 
    |/,\'      BECAUSE THAT SEEMS LIKE SOMETHING THAT'S REALLY NEEDED HERE
    |\|/
 
Cool, new results do show it was heavily CPU dependent, but is there anyway to intentionally disable the second GPU so I can see if its doing anything? Others with a single 4870 and similar specs seem to be getting about that score. I just want to verify this thing is running at full capacity, and not crippled.

3DMark Score
17306 3DMarks
SM 2.0 Score
6378
SM 3.0 Score
8940
CPU Score
4481
Result name
Microsoft Windows Vista
Intel Core 2 Quad Q6600
CPU Speed 3600 MHz
VisionTek HD4870 x2

http://service.futuremark.com/home.action;jsessionid=E40224E390B882DAD6C234F38AE97CAC
 
Setting Catalyst AI to "disabled" will disable AFR, which effectively disables crossfire (or in the case of the x2, the second GPU). Problem is it also disables a lot of app-specific driver optimizations, so using that will not give you a precise apples-to-apples comparison with a single 4870.

It's still a decent way to check GPU scaling though. From everything I've read, 3DM06 is horribly CPU-bound. Try 3DMark Vantage instead.
 
Ducman69 said:
Yes... I bought my card to just run benchmarks all day. :rolleyes:

I run benchmarks to get an apples to apples estimate of performance. If there is a better benchmark available that you recommend that I can use to see if my hardware/drivers are running properly, please list them rather than many here getting so damn defensive.

My concern, as noted, is that I may be experiencing driver or hardware compatibility related problems, hence my worry that my second GPU may not even be running. It never seems to clock up according to GPU-Z, even during games, and my 3DMark score seems low to me. People supposedly have seen their 3DMark06 performance double by switching PCIe slots, but I don't have that luxury. Sorry for asking for help, and sorry for being a little frustrated, but $500 is a lot of money for me, and I want to verify the result of all that cash.
Click to expand...

Dont pay those idiots any mind, it makes perfect sense to use a benchmark program to test your hardware.

I will say that your score looks normal though. I get about 16k at 2.8, but once overclocked to 3.8 I get 21,000. In crysis, I get about a 8-10 fps increase when overclocked.
 
Take out a game.
I like HL2 EP2 because you can set mad AA, like 24X with the X2
Run the X2 with fraps.
Now put that against your 8800 GTX at it highest AA capable.

The X2 will leave it in the dust.

Crysis is not a good comparison, nvidia has a lock on that for some reason.
Warhead on the other hand might work.

Benchmarks are BS, plain and simple.
If you want to look at the gpu utilizations and clocks, use RivaTuner 2.11. All of my cores are showing utilization, where-as they did not in GPU-z.
 
Warhead seems to show off the 4870x2's potential better. I usually get around 30fps average with everything jacked up.
 
Yep, Warhead is running smooth with my 4870X2. In fact, I can play it at 2560 x 1600 with a mix of enthusiast and gamer, with 3/4 on enthusiast.
 
Sabrewulf165 said:
Setting Catalyst AI to "disabled" will disable AFR, which effectively disables crossfire (or in the case of the x2, the second GPU). Problem is it also disables a lot of app-specific driver optimizations, so using that will not give you a precise apples-to-apples comparison with a single 4870.

It's still a decent way to check GPU scaling though. From everything I've read, 3DM06 is horribly CPU-bound. Try 3DMark Vantage instead.
Click to expand...
Sweet, yeah I'll try it out... just hate paying for it, so hopefully it'll be telling on the first try. :D

I have preordered Warhead but it hasn't arrived yet, but only problemo is I never played that on my 8800GTX to get a good comparison. As even a single 4870 is a very powerful card that should leave the 8800GTX in the dust.

I still have the 8800GTX, but its kindof a pain to get it back into my case for easy checking, beyond just having to run driver cleaner and everything, as its a micro-ATX box w/ zip-tied cabling and what not. :cool:

Sounds like consensus is that 17K w/ the Q6600 overclocked high like that isn't totally out of the ordinary though, and that its just not well optimized for 3DMark06, which I'm fine with if thats the only problem. Thanks!
 
I just switched from an 8800gtx to a 4870x2 over the weekend. The way I know it is working is by the fact that I can play Crysis at 1920x1200, 8xAA, and the custom settings automatically loaded by the game because of the card I am using. I run between 20 and 40fps with these settings while my old 8800gtx could only hit those numbers at medium settings with no AA.

I would have to say Crysis is pretty much the only effective way to truly test this video card.
 
Crossfire actually scales incredibly well in Elder Scrolls IV: Oblivion, even up to 3 and 4 GPUs. Of course you need to use a high res and/or a lot of AA to really push all the GPUs, but if you get into the forest and crank it, you can definitely see excellent scaling (as much as 80% in a lot of cases). On the flip side, there are also areas in Oblivion that are tremendously CPU-bound (mainly in and around cities). Crossfire generally won't do anything for you in those areas, and can even hurt performance a little.
 
ya, Crysis warhead ran better then crysis on my 4870x2. I could max out everything on enthusiast, and do 4x aa w/ 1920x1200. Cant say it was 100% smooth as butter, but definately playable, where regular crysis would have not been so smooth.
 
Nah man, I think something is fishy also....

yes 3d mark 06 is an old benchmark.... and a not overly useful... BUT I also have used it to see my basic score... and then see how high I can get that score...

I just installed the 4870x2 also, and I am averaging 14,200!! Which is crap... the first run was 15,500 but I have not seen that score since.... then I overclocked to 4120 mhz... and hit 20,000 which I was fine with.... but once again, I can not create the same score... the scores are all over the place...

Yea, crysis and stuff run pretty good

Crysis warhead

DirectX 9.0c GAMER 3X @ Map: airfield flythrough @ 8 1920 x 1200 AA 0x
 Framerate [ Min: 24.24 Max: 62.83 Avg: 52.38 ]

DirectX 9.0c GAMER 3X @ Map: airfield flythrough @ 8 1920 x 1200 AA 4x
 Framerate [ Min: 20.28 Max: 38.64 Avg: 33.17 ]


DirectX 9.0c GAMER 3X @ Map: airfield flythrough @ 8 1920 x 1200 AA 8x
 Framerate [ Min: 19.64 Max: 37.39 Avg: 31.17 ]


but something is up.. I mean, my 9800gx2 always hits 14, 500ish... no matter how many times I run it... my x2 is getting lower then that?!

I have done the bios, chipset, uninstall and reinstall drivers, 06 etc... I even did the fan profiles.. which I highly recommend

http://enthusiast.hardocp.com/article.html?art=MTU1MiwxLCxoZW50aHVzaWFzdA==
 
Again, 3dMark06 default runs are at 1280 x 1024. At those settings, the GPU's aren't nearly being pushed enough if the CPU cannot feed them enough information. You'll notice though that if you run 3dMark06 with a faster CPU, scores jump tremendously. At a stock 3.0GHz on my QX9650, I was getting 17k with my 4870X2. Now i'm at 22k with a 4.2GHz processor. I'm sure if I did a suicide run at 4.5GHz, i'd be at 24k+.
 
so it is suppose to be slower then my 9800gx2? I don't think so... I am averaging 14,200 on stock.... I hit 15,500 the first time I even ran it but never again.... then I overclocked to 4120 and hit 20,000 but once again... never again?

regardless of what someone thinks of it... it should always turn in similar numbers
 
I've played w/ a HD4870X2 for 2 weeks. Great card, ran every game smoothly. Didn't notice any difference w/ my previous HD4870 CF set up.
 
Xenozx said:
ya, Crysis warhead ran better then crysis on my 4870x2. I could max out everything on enthusiast, and do 4x aa w/ 1920x1200. Cant say it was 100% smooth as butter, but definately playable, where regular crysis would have not been so smooth.
Click to expand...

+1 here, crysis warhead ran great!
 
3D Mark 2006 is virtually useless for testing system performance. The numbers it produces are abstract and can not be compared to real world gaming performance. It's not like you can say 16,000 3D marks equals 60FPS at 2560x1600 in Crysis or even UT3. What influences 3D Mark to a large degree does not effect games in the same way. For example overclocking a Core 2 Quad Q9550 already overclocked to 3.6GHz an additional 200MHz to 3.8GHz will make almost no difference in gaming performance. You may see a 1 or 2 FPS change in static benchmarking tests but that's about all you might see if you were lucky. Most of the time the GPU is the driving factor behind how a modern game performs on today's computer hardware. Any Core 2 processor at 3.0GHz to 3.2GHz will largely be the same as anything clocked higher. As you increase the games resolution and visual effects to higher levels this becomes more and more true.

3D Mark can't even be used to accurately gauge video card performance relative to other video cards. An example of this comes from the ATI Radeon HD 2900XT out scoring the Geforce 8800GTX and Geforce 8800Ultra's which were in reality, much faster cards than the aforementioned ATI Radeon HD 2900XT.

3D Mark can be used to test for system stability but aside from that I can't think of any other potential use for it. It's like having 8 cores instead of 4 in a gaming rig. Sure you can say, "Hey I've got dual quad core processors in my box so bow to my HUGE e-penis" but in actual games those extra cores will go unused as the resolution ramps up in those games. 3D Mark scores can give you bragging rights but anyone who really understands how these tools work and how they work won't really care and frankly, neither should you.
 
Can someone tell me ith which driver version can i play Warhead on Vista Ultimate x64?
I get constant crashes after about 5-10seconds into the game. Crashes are kernel32.dll and "driver stopped responding" related.
 
MustRotate said:
Can someone tell me ith which driver version can i play Warhead on Vista Ultimate x64?
I get constant crashes after about 5-10seconds into the game. Crashes are kernel32.dll and "driver stopped responding" related.
Click to expand...

I'm using the NVIDIA 177.92 drivers and on my 3-Way SLI setup I haven't had any crashes or problems with Crysis Warhead. I played it last night for about 3 hours straight.
 
Proxy said:
+1

Fire up some Crysis Warhead or something.
Click to expand...


+2



new cards dont shine in 3dmark, Vantage is a better comparison if you NEED a number to justify your purchase


best bet is to use some real games with free edge detect AA
 
Ice killed my 8800Ultra in Crysis, so when I fired up ice in Warhead with 1920 x 1200, all enthusiast, with AA in DX10 and found myself consistently 25+ fps, I was shocked. Lovin it.
 
You must log in or register to reply here.
Back
Top