• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

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

Is your CPU OC'd

I have the stock cooler so no, i don't want to risk anything. The CPU usage is around 60% so i don't think it's bootlenecking.
 
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
Well it looks fine to me so I guess we can rule out the board dropping the PCIE lanes to below 8x...

Like I said before with Crysis 3 its either hit or miss when it performs well because in all honesty its optimisations are lacking and many reviewers pick a special level to run their own benchmarks or something like that and you do not get that performance level to level. There are times even at 1080p on my system when I played it it spiked around in usage and FPS, its just not that well optimized sadly and its a crazy demanding game.

BF4 is a different story in all honesty and the spiking usage im guessing your referring to is multiplayer? (or campaign)? I mean at 1080p BF4 aint to bad demanding in general especially for a CFX system like yours. But in multiplayer the game sometimes loses FPS mostly due to internet and the extra junk going on because multiplayer is almost never optimized as well as the campaign and most reviews only test the campaign.

When I get off work today, ill swap to my 1080p screen and snap some pics of usage in BF4 and or Crysis 3 maxed with it set to run off 2 290X cards and see if it looks like what your getting.
 
Do you use V-sync
 
Well it looks fine to me so I guess we can rule out the board dropping the PCIE lanes to below 8x...

Like I said before with Crysis 3 its either hit or miss when it performs well because in all honesty its optimisations are lacking and many reviewers pick a special level to run their own benchmarks or something like that and you do not get that performance level to level. There are times even at 1080p on my system when I played it it spiked around in usage and FPS, its just not that well optimized sadly and its a crazy demanding game.

BF4 is a different story in all honesty and the spiking usage im guessing your referring to is multiplayer? (or campaign)? I mean at 1080p BF4 aint to bad demanding in general especially for a CFX system like yours. But in multiplayer the game sometimes loses FPS mostly due to internet and the extra junk going on because multiplayer is almost never optimized as well as the campaign and most reviews only test the campaign.

When I get off work today, ill swap to my 1080p screen and snap some pics of usage in BF4 and or Crysis 3 maxed with it set to run off 2 290X cards and see if it looks like what your getting.

Maybe i should unplug the 4 pin connector or what do you think? And do you have any example of any good games i should try?
 
Maybe i should unplug the 4 pin connector or what do you think? And do you have any example of any good games i should try?
I would say either way should not matter plugging it in or not. I only suggested because it can relieve stress off the power delivery system when in multi-GPU setups. But in all honesty its more meant for if you have 3 video cards + or if your running alot of different cards in the slots. I just thought it could be an issue if the PSU was struggling for (Insert reason here). You can unplug it if you do not want it in as it should not effect it either way as we have seen.

Other games to try? Tomb Raider, bioshock infinite, Mudered: soul suspect (Though its not demanding), or Watch Dogs.
 
Might be good to get a newer motherboard...
 
I would say either way should not matter plugging it in or not. I only suggested because it can relieve stress off the power delivery system when in multi-GPU setups. But in all honesty its more meant for if you have 3 video cards + or if your running alot of different cards in the slots. I just thought it could be an issue if the PSU was struggling for (Insert reason here). You can unplug it if you do not want it in as it should not effect it either way as we have seen.

Other games to try? Tomb Raider, bioshock infinite, Mudered: soul suspect (Though its not demanding), or Watch Dogs.

Okey thank's. The reason why i was thinking about unplugging it was because i think the Vram Usage problem begun when i inserted that connector.

Might be good to get a newer motherboard...

Why?
 
Well the first gen 1155 chipset had many glitches, its very possible you mobo cant handle them. On another note before doing that ensure all connections in the case are fully seated (all power and data cables)
 
Well the first gen 1155 chipset had many glitches, its very possible you mobo cant handle them. On another note before doing that ensure all connections in the case are fully seated (all power and data cables)

Which connections should i look for?

Edit: I looked at GPU_z and the vram Mem usage is 4200mb. That must be one of the problems. Both MSI afterburner and GPU-z indicate that, so it must be correct.
 
Before anything else, just out of curiosity what settings are your card's switches set to? My HIS R9 290 came with the switch in quiet mode by default (similar to the default setting most 290X's ship with) while many have noted this does nothing on a 290 I've found that the gpu throttled far less in the uber setting than in the quiet setting. Fan noise also went up. I have a feeling that non reference cards in uber mode have alot more change than just a 10% bump to fan setting. My default 3dmark 11 for example went from 10580 with the switch set to quiet, to 12540 with the switch set to uber. I had tried overclocking and setting the fan manually to 100% with msi afterburner, however the gpu still throttle far more in quiet mode than in uber.

Imho your gpu's are throttling and that's what causing the low performance. If your switches on the cards are set to what would be quiet mode on a 290X, switch them to what would be uber.

typical standards from 290X reviews


"Quiet Mode" – BIOS position one. Switch is in position closest to where you plug in your displays. Fan is at 40% cap.



"Uber Mode" – BIOS position two. Switch is in position furthest away to where you plug in your displays. Fan is at 55% cap

if there really is an issue with vram usage as well and drivers don't change it, switching to the other bios may fix that issue also.
 
Before anything else, just out of curiosity what settings are your card's switches set to? My HIS R9 290 came with the switch in quiet mode by default (similar to the default setting most 290X's ship with) while many have noted this does nothing on a 290 I've found that the gpu throttled far less in the uber setting than in the quiet setting. Fan noise also went up. I have a feeling that non reference cards in uber mode have alot more change than just a 10% bump to fan setting. My default 3dmark 11 for example went from 10580 with the switch set to quiet, to 12540 with the switch set to uber. I had tried overclocking and setting the fan manually to 100% with msi afterburner, however the gpu still throttle far more in quiet mode than in uber.

Imho your gpu's are throttling and that's what causing the low performance. If your switches on the cards are set to what would be quiet mode on a 290X, switch them to what would be uber.

typical standards from 290X reviews


"Quiet Mode" – BIOS position one. Switch is in position closest to where you plug in your displays. Fan is at 40% cap.



"Uber Mode" – BIOS position two. Switch is in position furthest away to where you plug in your displays. Fan is at 55% cap

if there really is an issue with vram usage as well and drivers don't change it, switching to the other bios may fix that issue also.


Thank's for you input. I had the card in Uber mode. And i'm most certain that they didn't throttle. Because the Core and Mem clock was always at the maximum MHz when gaming. Today i returned the cards, cause i belive that there is a general problem with certain R9 290 in Crossfire that AMD must fix.

So thank you everybody for you help, it's must appreciated. I will definitly stay on this forum even though i have returned the cards.

I will most likely buy a non stock GTX 880 when it arrives.

Thank's again guys.
 
I went ahead last night and pulled up some Crysis 3 at 1080p with 2 R9 290X. It played well most of the time maxed out but it still had sections that ran low so that could be the issue there. Crytek has been having trouble lately and honestly this game never really hit good levels of optimizations for the power it needs.

As for BF4, in multiplayer I did notice some drops into the 50's with two cards on Siege of Shanghai and on the water map (Name leaves my mind, the one with the storm and ship crashing). It never really always utilizes the cards sometime properly at 1080p but with Mantle and DX 11 changing back and fourth at 1080p I noticed that I droppped below 60 rarely with Mantle on vs DX 11 so you could try that. My processor was being punched alot the whole time and I am running an FX 9590 a bit overclocked so who knows.

I switched back to 4k and all three and things seemed to be used a bit more overall on both games (Crysis though was around 30-60 depending at times with 4x MSAA and Ultra settings) on BF4 the game ran about the same but with it running Ultra with Mantle around 50-70 FPS depending (Vsync off) and with it on (I use aftermarket dynamic Vsync on Radeon Pro) It kept a 60 with a few rotations around the 55 are and back to the 60.

I will say this, you maybe just hitting some limits sadly with the overall system and games. My suggestion would be to buy a AIO LC unit or something and try overclocking the CPU a bit because even if the utilization is not to max on the CPU, you still see improvements across the board.
Thank's for you input. I had the card in Uber mode. And i'm most certain that they didn't throttle. Because the Core and Mem clock was always at the maximum MHz when gaming. Today i returned the cards, cause i belive that there is a general problem with certain R9 290 in Crossfire that AMD must fix.

So thank you everybody for you help, it's must appreciated. I will definitly stay on this forum even though i have returned the cards.

I will most likely buy a non stock GTX 880 when it arrives.

Thank's again guys.
Oh well ok whatever makes you happy sorry we were not able to fix your problem. Hopefully whatever you buy next will give you less trouble.
 
I went ahead last night and pulled up some Crysis 3 at 1080p with 2 R9 290X. It played well most of the time maxed out but it still had sections that ran low so that could be the issue there. Crytek has been having trouble lately and honestly this game never really hit good levels of optimizations for the power it needs.

As for BF4, in multiplayer I did notice some drops into the 50's with two cards on Siege of Shanghai and on the water map (Name leaves my mind, the one with the storm and ship crashing). It never really always utilizes the cards sometime properly at 1080p but with Mantle and DX 11 changing back and fourth at 1080p I noticed that I droppped below 60 rarely with Mantle on vs DX 11 so you could try that. My processor was being punched alot the whole time and I am running an FX 9590 a bit overclocked so who knows.

I switched back to 4k and all three and things seemed to be used a bit more overall on both games (Crysis though was around 30-60 depending at times with 4x MSAA and Ultra settings) on BF4 the game ran about the same but with it running Ultra with Mantle around 50-70 FPS depending (Vsync off) and with it on (I use aftermarket dynamic Vsync on Radeon Pro) It kept a 60 with a few rotations around the 55 are and back to the 60.

I will say this, you maybe just hitting some limits sadly with the overall system and games. My suggestion would be to buy a AIO LC unit or something and try overclocking the CPU a bit because even if the utilization is not to max on the CPU, you still see improvements across the board.

Oh well ok whatever makes you happy sorry we were not able to fix your problem. Hopefully whatever you buy next will give you less trouble.

I am really suprised with how much you guys helped me. I have been reading and many people have problems with R9 290 in CF. Not with 290X only with 290. I don't think you guys could have fixed it because it's AMD fault. But you really tried to find the problem, and for that i am really thankful. :)
 
I am really suprised with how much you guys helped me. I have been reading and many people have problems with R9 290 in CF. Not with 290X only with 290. I don't think you guys could have fixed it because it's AMD fault. But you really tried to find the problem, and for that i am really thankful. :)
Glad to at least try, but one thing I would still say is no matter what remember that most benchmarkers/benchmarks you see on the web are done with the latest i7's (Many times the E series) and are overclocked to at least 4.5ghz. You have a great processor but I would say you might see some better returns in games with overclocking it since Sandy-Bridge was a king among the overclockers. Even if the utilization of a processor says that its not 100%, overclocking can still show decent performance gains.

http://www.techspot.com/review/787-thief-benchmarks/page4.html

It is always dependent on the game of course, but it is one of those things that can be an easy way to push a system further. Just a little food for thought for the future.
 
Glad to at least try, but one thing I would still say is no matter what remember that most benchmarkers/benchmarks you see on the web are done with the latest i7's (Many times the E series) and are overclocked to at least 4.5ghz. You have a great processor but I would say you might see some better returns in games with overclocking it since Sandy-Bridge was a king among the overclockers. Even if the utilization of a processor says that its not 100%, overclocking can still show decent performance gains.

http://www.techspot.com/review/787-thief-benchmarks/page4.html

It is always dependent on the game of course, but it is one of those things that can be an easy way to push a system further. Just a little food for thought for the future.

True that. The funny thing is that i have a good cooler and good aftermarket thermal grease but i haven't installed it yet. I will do it soon though. :)
 
It is always been like that since ages :D, but really don't take any 28nm card anymore.... my sweet suggestion...

Maybe consider yourself a proper water loop?

amdvsnvidiabyspritanium.jpg
 
Hello I think the probleme com from to the PSU 750W its to low to work with 2 CFX R9 290X
 
Back
Top