• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Bad performance/Usage spikes with R9 290 Tri-X CF.

mull54 said:
Haha the post above was supposed to include a link to the pic. But i forgot. http://imgur.com/GyP0jDp
Click to expand...
Do you have one with the GPUZ first page specs listed for both cards? I would just like to see what its reporting the system is running like (Specifically in relation to the main board).

Honestly I see a bit of spiking but not much in that screenshot. So your saying in Crysis 3 that its just struggling throughout the game to get above 38FPS range or just in specific sections? You also said that changing quality seems to offer no noticeable improvements correct?
 
GhostRyder said:
Do you have one with the GPUZ first page specs listed for both cards? I would just like to see what its reporting the system is running like (Specifically in relation to the main board).

Honestly I see a bit of spiking but not much in that screenshot. So your saying in Crysis 3 that its just struggling throughout the game to get above 38FPS range or just in specific sections? You also said that changing quality seems to offer no noticeable improvements correct?
Click to expand...
Yeah, didn't i post it to you in a recent post?

In outdoor environments it struggles to get higher than 45FPS. With 8XMSAA and FXAA, there is no big difference in performance.
 
mull54 said:
Yeah, didn't i post it to you in a recent post?

In outdoor environments it struggles to get higher than 45FPS. With 8XMSAA and FXAA, there is no big difference in performance.
Click to expand...
Yes but I want to see it while the machine is stressed (Gaming wise).

Ah ok I just wanted clarification.
 
I don't know much about Xfire, but could it be that you forgot to connect your PSU to the SLI/Xfire Power Connector?
 
christoph_789 said:
I don't know much about Xfire, but could it be that you forgot to connect your PSU to the SLI/Xfire Power Connector?
Click to expand...
Good point...
@mull54 Do you have the 4 pin Molex motherboard hookup on? Just for curiosity...
 
Last edited:
GhostRyder said:
Good point...
@mull54 Do you have the 4 pin Molex motherboard hookup on? Just for curiosity...
Click to expand...

It was not plugged in so i used the 4 pin included with my PSU(i got one from the grahpics card aswell but that one didn't fit).

No difference in performance and a saw that the VRAM usage was abow 4500MB which is very wierd.

Maybe i didn't push it on correctly. And what are the consequences of runing without the connector for 3 years?

Edit: Found this: Short answer:
No. It's not required.

Long answer:
4pin molex has 1×12V pin, 2×ground pin and 1×5V pin.
Only the 12V and ground(s) end up into the video card, PCIe card don't use externally fed 5V.

HD4850, like majority of modern video cards, that feature an onboard power connector (6pin, eg.) the 12V current they receive from the PCIe slot is only used for minor loads of which video memory power conversion is responsible for the majority. Memory on the HD4850 uses around 35W.

Basically, if the connector is not plugged the video memory on your cards run on power the ATX24 connector provides. That's less ideal as while travelling through the motherboard the current, that will eventually feed your GFX memory, will surely catch and also create some signal interferences. If you plug it in you provide an additional, lower resistance and less EMI-ridden route for the electricity between PSU and GFX meaning atleast half of the power will flow through the molex rather than the motherboard.

Summary:
It can't hurt to plug it in.

christoph_789 said:
above the first PCIex1 slot
Click to expand...
 
What GPUs did you upgrade from? Were they Nvidia?

If you haven't already, Id just wipe the OS and reinstall.
 
MxPhenom 216 said:
What GPUs did you upgrade from? Were they Nvidia?

If you haven't already, Id just wipe the OS and reinstall.
Click to expand...

I had two MSI 6950. And i did a clean install of the OS when i installed the new cards.

Just to be clear, does 25-30 FPS on 1080P, Very high settings with Very high textrues and 8XMSAA sound normal for one card on Crysis 3 opening lvl?
 
mull54 said:
I had two MSI 6950. And i did a clean install of the OS when i installed the new cards.

Just to be clear, does 25-30 FPS on 1080P, Very high settings with Very high textrues and 8XMSAA sound normal for one card on Crysis 3 opening lvl?
Click to expand...

Yes. with 4x AA. the 290 gets ~40fps.

Do 4x AA now and see what you get. 8x AA is so unnecessary.

crysis3_1920_1080.gif
 
mull54 said:
It was not plugged in so i used the 4 pin included with my PSU(i got one from the grahpics card aswell but that one didn't fit).

No difference in performance and a saw that the VRAM usage was abow 4500MB which is very wierd.

Maybe i didn't push it on correctly. And what are the consequences of runing without the connector for 3 years?

Edit: Found this: Short answer:
No. It's not required.

Long answer:
4pin molex has 1×12V pin, 2×ground pin and 1×5V pin.
Only the 12V and ground(s) end up into the video card, PCIe card don't use externally fed 5V.

HD4850, like majority of modern video cards, that feature an onboard power connector (6pin, eg.) the 12V current they receive from the PCIe slot is only used for minor loads of which video memory power conversion is responsible for the majority. Memory on the HD4850 uses around 35W.

Basically, if the connector is not plugged the video memory on your cards run on power the ATX24 connector provides. That's less ideal as while travelling through the motherboard the current, that will eventually feed your GFX memory, will surely catch and also create some signal interferences. If you plug it in you provide an additional, lower resistance and less EMI-ridden route for the electricity between PSU and GFX meaning atleast half of the power will flow through the molex rather than the motherboard.

Summary:
It can't hurt to plug it in.
Click to expand...
Yes it just essentially be a little less stressing the power delivery by splitting the methods. Some boards sometimes do not do well with higher power cards and those connectors help. Was worth a shot but normally unless you load up a ton of slots it's not necessary.
mull54 said:
I had two MSI 6950. And i did a clean install of the OS when i installed the new cards.

Just to be clear, does 25-30 FPS on 1080P, Very high settings with Very high textrues and 8XMSAA sound normal for one card on Crysis 3 opening lvl?
Click to expand...
For one card that is about average around 30fps on maxed out. I would suggest not exceeding 4x but with two cards your fps should be higher based on reviews. However this is also based off the fact your getting low fps in bf4 as well. Crysis 3 in general is just very demanding even for the best of video cards and there are some parts that just run very rigid.
 
GhostRyder said:
Yes it just essentially be a little less stressing the power delivery by splitting the methods. Some boards sometimes do not do well with higher power cards and those connectors help. Was worth a shot but normally unless you load up a ton of slots it's not necessary.

For one card that is about average around 30fps on maxed out. I would suggest not exceeding 4x but with two cards your fps should be higher based on reviews. However this is also based off the fact your getting low fps in bf4 as well. Crysis 3 in general is just very demanding even for the best of video cards and there are some parts that just run very rigid.
Click to expand...

My fps is very unstable. In BF4 indoors i get 80-100 fps but outdoors i get under 60. It's fluctuation
GhostRyder said:
Yes it just essentially be a little less stressing the power delivery by splitting the methods. Some boards sometimes do not do well with higher power cards and those connectors help. Was worth a shot but normally unless you load up a ton of slots it's not necessary.

For one card that is about average around 30fps on maxed out. I would suggest not exceeding 4x but with two cards your fps should be higher based on reviews. However this is also based off the fact your getting low fps in bf4 as well. Crysis 3 in general is just very demanding even for the best of video cards and there are some parts that just run very rigid.
Click to expand...

What about the 4500MB Vram usage. That's very weird.
 
Ain't it old friend cg-miner? :D
 
have you tried both cards seperately in first slot?
 
mull54 said:
My fps is very unstable. In BF4 indoors i get 80-100 fps but outdoors i get under 60. It's fluctuation


What about the 4500MB Vram usage. That's very weird.
Click to expand...
Odd but do you use Mantle with BF4 or DX 11? Mantle gets me very stable smooth FPS on my 290X setup.

As for VRAM usage, thats probably not accurate because many VRAM apps do not show correct usage.

You know this may just all be because Crysis 3 is very meh in performance on some levels at least for the Crysis problem. It seems that running it on any setup does run into low spots in the FPS charts especially with some specific settings turned on. As for BF4, like I had said earlier there have been people reporting problems while Intel Hyper Threading is enabled with Performance drops. You could try turning Hyper Threading off for that game and see how well it runs, I doubt you would actually get any difference performance wise in any game honestly because normally an i5 and i7 of same generation match up clock to clock in FPS.
 
MxPhenom 216 said:
have you tried both cards seperately in first slot?
Click to expand...

Actually i have not. That's my last option. What about the high Vram usage(4500MB in Crysis 3)
GhostRyder said:
Odd but do you use Mantle with BF4 or DX 11? Mantle gets me very stable smooth FPS on my 290X setup.

As for VRAM usage, thats probably not accurate because many VRAM apps do not show correct usage.

You know this may just all be because Crysis 3 is very meh in performance on some levels at least for the Crysis problem. It seems that running it on any setup does run into low spots in the FPS charts especially with some specific settings turned on. As for BF4, like I had said earlier there have been people reporting problems while Intel Hyper Threading is enabled with Performance drops. You could try turning Hyper Threading off for that game and see how well it runs, I doubt you would actually get any difference performance wise in any game honestly because normally an i5 and i7 of same generation match up clock to clock in FPS.
Click to expand...


I will try turning HT off. I will give you a screenshot of the GPU usage. So you'll see how spiky it is.
 
GhostRyder said:
Odd but do you use Mantle with BF4 or DX 11? Mantle gets me very stable smooth FPS on my 290X setup.

As for VRAM usage, thats probably not accurate because many VRAM apps do not show correct usage.

You know this may just all be because Crysis 3 is very meh in performance on some levels at least for the Crysis problem. It seems that running it on any setup does run into low spots in the FPS charts especially with some specific settings turned on. As for BF4, like I had said earlier there have been people reporting problems while Intel Hyper Threading is enabled with Performance drops. You could try turning Hyper Threading off for that game and see how well it runs, I doubt you would actually get any difference performance wise in any game honestly because normally an i5 and i7 of same generation match up clock to clock in FPS.
Click to expand...

Look at this pic. The VRAM and the spiky usage. What do you say?

http://imgur.com/jil4X79
 
mull54 said:
Actually i have not. That's my last option. What about the high Vram usage(4500MB in Crysis 3)



I will try turning HT off. I will give you a screenshot of the GPU usage. So you'll see how spiky it is.
Click to expand...
Just also include a shot of CPU-Z with the first page information so we can make sure everything is reported properly. Or just make sure its not reporting either GPU at a low PCIE frequency while gaming because AMD cards will run at 4x but with CFX running over the PCIE lanes that would be a bottleneck and could cause the problem if that is the issue.
 
GhostRyder said:
Just also include a shot of CPU-Z with the first page information so we can make sure everything is reported properly. Or just make sure its not reporting either GPU at a low PCIE frequency while gaming because AMD cards will run at 4x but with CFX running over the PCIE lanes that would be a bottleneck and could cause the problem if that is the issue.
Click to expand...

I think we posted at the same time.

Look at this pic. The VRAM and the spiky usage. What do you say?

http://imgur.com/jil4X79
 
mull54 said:
Look at this pic. The VRAM and the spiky usage. What do you say?

http://imgur.com/jil4X79
Click to expand...
Well the spiking is a bit much in all honesty but ill have to compare it to a gaming session on my machine to see how bad the spiking is for me. But as far as VRAM usage is concerned, BF4 tends to cache more than it needs at times and MSI just reports what is being called. Whether or not its being used is a different story and the same goes for Crysis 3 however Crysis 3 needs loads of VRAM especially when it concerns MSAA and the other settings. It will do the same and cache most of your VRAM if it decided it needs it (Hence why most benchmarkers only run up to 4x).
 
Is your CPU OC'd
 
GhostRyder said:
Well the spiking is a bit much in all honesty but ill have to compare it to a gaming session on my machine to see how bad the spiking is for me. But as far as VRAM usage is concerned, BF4 tends to cache more than it needs at times and MSI just reports what is being called. Whether or not its being used is a different story.
Click to expand...
Okey i see. I think the high Vram usage begun when i inserted the 4pin connector. I have looked around and i don't think i need it. I have one 8 pin and one 6 pin connector to both card connected from the PSU.

Here is the GPU-Z pic. http://imgur.com/XTzY5fx
 
You must log in or register to reply here.
Back
Top