4870x2 Performance Problems

OmegaNemesis28

Distinguished
Nov 10, 2008
14
0
18,510
I'm having problems with my 4870x2. Have been since day one of buying this thing. I've had it for 2 months now.
Originally, the drivers were a hassle to get setup with my 34" TV. Some games would not fit the entire screen, little things like that. While I was worrying about those problems, I didn't quite take note of the horrible performance I was getting in games.

I've been struggling over the past month trying to get this card to work at it's full power.
Here are my system specs....
Intel Q6700 @ 3.66ghz
ASUS P5Q-E
ATi Visiontek 4870 X2
G.Skill 2x 2GB PC2-8000 2.1V
X-Fi Prelude Sound Card
1kw Silverstone Strider PSU
Vista Home Premium 64-bit

After tons of researching, I've ruled out many possibilities to my problem.
-For one, my clocks are at the full 750/900 clocks when playing games. In all my games I've monitored with with GPU-Z.
-It's not the drivers, I've tried about 5 different ATi drivers and their hotfixes for games. Currently on 8.10
-I have my fan speed set to 60% at all times. It says in the 60-80C range while playing games.
-It is not a motherboard problem. I originally purchased the card and used it on a 680i EVGA motherboard. Then, after seeing this problem - I swapped to an ASUS P5Q-E thinking it may resolve it because of the PCI-E 2.0.
-It is not my CPU or it's clocks because my friend has a Q6600 at 3.4ghz and his card runs much better then mine in many games if not all.
-I currently have both cores overclocked to 762/910 due to other people having similar problems but being fixed after overclocking slightly.

Here are many benches from as many games as I have installed at the moment including Furmark, Vantage, ect.
NOTE: In my 8.10 CCC - I have my fan speed set to 60% and all the 3D Settings are DEFAULT.
Some scores may seem normal, but trust me - they are not always like that. Its kind of like a hit or miss thing. Sometimes it will be good, sometimes not. But I know its not the PowerPlay because the clocks are always full while playing.

Far Cry 2 Benchmark (Ranch Small) - Latest HotFix
1920x1080 60Hz ; Maxed Out - Ultra High DX10 ; NO AA
* Average Framerate: 42.68
* Max. Framerate: 61.09 (Frame:329, 6.41s)
* Min. Framerate: 31.95 (Frame:1454, 34.46s)

1920x1080 60Hz ; Maxed Out - Ultra High DX10 ; 4x AA
# Average Framerate: 53.63
# Max. Framerate: 79.48 (Frame:392, 6.10s)
# Min. Framerate: 39.24 (Frame:1809, 34.14s)
I dont understand how putting AA on increases my FPS. Logically makes no sense. Anything higher then 4x AA kills my FPS though down to 30-40.

Crysis Warhead FRAMEBUFFER Benchmark 0.31 (Ambush Flythrough)
DX9 ; ENTHUSIAST ; 1920x1080 ; No AA
==> Framerate [ Min: 13.43 Max: 53.47 Avg: 35.10 ]

DX9 ; ENTHUSIAST ; 1920x1080 ; 4x AA
==> Framerate [ Min: 8.44 Max: 30.76 Avg: 21.64 ]
Sorry, no DX10 Benchmark for Warhead. It just never worked for me. DX10 doesn't go into fullscreen nor does it like the 1920x1080 resolution for some reason, overcontrasts the hell out of it while windowed.

Half-Life 2 Lost Coast
1920x1080 16:9 ; Everything High ; 6x MSAA ; Anisotropic 16x
Avg:130.44 fps

Counter-Strike Source
1920x1080 16:9 ; Everything High ; 8x MSAA ; Ansitropic 16x
Avg:145.27 fps

Company of Heroes
1920x1080 ; DX9 ; Ultra/High Everything ; 8x AA
Avg: 57.6 FPS
Max: 60.2 FPS
Minimum: 48.1 FPS

1920x1080i ; DX10 ; Ultra/High Everything ; 8x AA
Avg: 38.0 FPS
Max: 53.5 FPS
Minimun: 8.6 FPS
Another game that does not like DX10 for me. It sets itself to 1080i for some stupid reason and its boxed up. This is one of few games that does this even if I have all the resolutions in CCC overscanned and enabled.

Lost Planet (Snow)
1920x1080 ; DX10 ; Max Everything ; No AA ; 4x Anisotrophic
Avg: 51.3 FPS

1920x1080 ; DX10 ; Max Everything ; 4x AA ; 4x Anisotrophic
Avg: 40.4 FPS

1920x1080 ; DX9 ; Max Everything ; No AA ; 4x Anisotrophic
Avg: 90.8 FPS

1920x1080 ; DX9 ; Max Everything ; 4x AA ; 4x Anisotrophic
Avg: 53.1 FPS

Team Fortress 2
I get around 40-70 FPS with everything max, 4x Anisotrophic, and 4x AA online and offline.

3DMark 06 = 15544
I've had this score up to 18.5k one time. Its different every run.

3DMark Vantage = P13906

Furmark (renamed exe)
Benchmark ; 1280x1024 ; No MSAA ; Time 60000
FPS Min : 82
FPS Max : 130
FPS Avg: 97

Benchmark ; 1280x1024 ; 4x MSAA ; Time 60000
FPS Min : 21
FPS Max : 35
FPS Avg : 25

GPUZ Core 1
GPUZ Core 2

Any help would be greatly appreciated.
 

OmegaNemesis28

Distinguished
Nov 10, 2008
14
0
18,510



My friend has a similar rig that runs about 10-20FPS better then mine at 1920x1200 with more AA then me and he has a Q6600 @ 3.4ghz. I think something may be wrong.
 

The_Blood_Raven

Distinguished
Jan 2, 2008
2,567
0
20,790
Those are about right. The only things that seem off to me is the Crysis Warhead and the Company of Heroes scores. I get very similar Far Cry marks, and yes gaining frames with AA is not all that uncommon. I would have to say, check your RAM first of all. Also could you boot up BIOS and record the voltages and tell us please. Again they seem a little low, but nothing too odd. What system does your friend have?
 

OmegaNemesis28

Distinguished
Nov 10, 2008
14
0
18,510


My Vcore is 1.53 at all times.
101v on the PCI-E.
2.1v on the RAM

My friend has a Q6600 @ 3.4 ghz, Rampage Formula mobo, same video card as me and manufacture and bought from the same place - Visiontek, 700W OCZ PSU, 4x1gb's Ballistix Tracers DDR2-800 @ 1143mhz and Vista 64x also.
 

The_Blood_Raven

Distinguished
Jan 2, 2008
2,567
0
20,790
Hmm, sounds about right I guess. I know you will hate this, but reinstall windows there are probably too many applications, hidden minor viruses, spy/mal/adwar and fragmented files on your HDD. You would be surprised what thorough format could do! I don't think there is anything wrong with your GPU, you would notice a much larger difference. Your Vcore is bit high, your CPU might be unstable or damaged. Besides a RAM issue that is about it.
 

OmegaNemesis28

Distinguished
Nov 10, 2008
14
0
18,510



Already reformated 3 times. Obviously once before because I swapped mobos.

And the Vcore needs to be 1.5 or higher for the CPU to be stable at 3.66ghz

I don't think its the GPU either, thats why I haven't RMA'd it.
 

OmegaNemesis28

Distinguished
Nov 10, 2008
14
0
18,510


It has 16x without crossfire, 8x with crossfire. Obviously, its only 1 card so it does 16x.


Are you sure he is getting better performance?

yes
 

ZOldDude

Distinguished
Apr 22, 2006
1,251
1
19,280
Dump Vista...install XP and watch the system boot faster and have higher FPSwhile haveing about half of your Ram still free while gameing.
If your going to keep Vista and expect it to run close to XP w/2GB then use 4GB of ram.
 

OmegaNemesis28

Distinguished
Nov 10, 2008
14
0
18,510



Um no.
With XP - I get half the performance as I do right now. I've tried.
 

The_Blood_Raven

Distinguished
Jan 2, 2008
2,567
0
20,790
Alright I benchmarked my system:

E8600 @ 4.3 Ghz
4GB 4-4-4-12 DDR2 @ 900Mhz
HIS 4870 X2
PC Power and Cooling 750w

Far Cry Small ranch the average of 3 tests:
1920x1200 Ultra High 4AA no Vsync of course
Max: 83.12
Average: 54.79
Minimum: 40.24

Crysis Warhead:
DX 10 enthusiast 1920x1200 no AA
Max: 42.35
Average: 28.17
Minimum: 11.56

Yours is not so far off, and my CPU is better for gaming.
 

OmegaNemesis28

Distinguished
Nov 10, 2008
14
0
18,510



But you're at a higher res, I should be slightly closer then 30FPS
 

njalterio

Distinguished
Jan 14, 2008
780
0
18,990
You have a pretty awesome gaming computer going on. I would just be happy with what you have. There really isn't anything wrong with your computer that is worth splitting hairs about. Maybe your friend has slightly different CPU settings or VGA settings or is using a different driver.
 

OmegaNemesis28

Distinguished
Nov 10, 2008
14
0
18,510



still is around 5-10 FPS loss



Um - yes it is worth splitting hairs because I paid $100 more then I could have paid for a GTX280, but Im getting even worse performance. The card is not working as it should.
 

The_Blood_Raven

Distinguished
Jan 2, 2008
2,567
0
20,790
First of all the difference between our resolutions wont have any NOTICEABLE difference in framerates. Also as I have been telling your scores line up with the benchmark scores and are better than a GTX 280 could do. The only that seems wrong is the Crysis Warhead performance which you are apparently unable to run in DX10 so it is not a good measure.