ASUS R9 290x OC - Throttling

M76

M76

[H]F Junkie
Joined
Jun 12, 2012
Messages
14,058
I recently bought this card, and found that under load, it starts throttling the GPU clock after just a minute or so.

Is that normal behaviour?
If the cooling can't keep up with the heat what's the point of making an OC card?

The GPU clock drops from 1050 to 950 and that's about the level where it can keep the card under 95C° But that's even less than stock clock speeds.

Is this the new norm, or I have a card with faulty cooling?
 
That is the problem with just about all 290 series of cards :) Which is why I returned mine and went with the Gtx 780 instead....
 
zod96 said:
That is the problem with just about all 290 series of cards :) Which is why I returned mine and went with the Gtx 780 instead....
Click to expand...

Maybe "just about all" is unfair.

I'd say it's a problem with the reference 290/x. They were really intended to have the coolers removed, not that anyone from AMD came out and said as much.

At the part where he said "95 degrees"--that meant, "please remove this cheaply made and inadequate cooler".

The AIB partners have largely solved throttling issues with both the 290 and the 290x by using larger coolers and sinks. Some AIBs have done a better job than others. The Powercolor PCS+ 290x is fantastic and basically runs rings around the GTX 780 (I've owned 1 780, and then 2 780s in Sli, then 1 290x and now a 295x2, so I have relatively wide view of reference to make these observations).

AMD has taken what they've gained and made the 295x2 which is actually quite popular considering how retarded expensive (in a single purchase) it is. Now will come the more readily marketable 295x at the 780Ti price point (such is the rumor). These are reference designs that are intended to be nearly silent at full load. The 295x2 is essentially silent with 2 Noctuas in push/pull anyway, and in real-world gaming mine does not throttle (not even close) even though my ambient temperature is frikkin 80F right now (South Texas Summer yay). I sure as hell can MAKE it throttle, but it's "target temperature" is 70C. So meh.

People make these claims about throttling "after 1 minute":
Well, what was going on for that minute? Were you running Valley at 1440p? Not a very nice thing to do to your card--yes you can run synthetic benchmarks that will cause your card to throttle very easily if it's not really well cooled, but playing Crysis 3 isn't anywhere near this continuously nasty rendering feed.

Now, the reference 290 design with the blower cooler... I took one look at the specs and said to myself, "I wouldn't even touch one of those with someone else's computer". Depending on ambient temp... yes, throttling is more a how soon than an if.

Whats do we do?
H55 with an NZXT H10 bracket. Better airflow in your case. Turn up the AC. Turn down the eye candy. Some combination of these.

TL;DL don't buy reference 290/x unless you are going to remove the cooler and use something better.
 
I guess the card in question is a DirectCU II? I didn't realize at first :eek: Still, it depends on what you are doing to it/with it.

If you can give us some indication of what kind of load you are putting it under to cause this heat then maybe we can give you an idea if it's normal or not.

Any review of the card will tell you it should be operating below 80C in heavy gaming. But there are lots of things one could do with a 290 that would make it too hot.
 
This is the card you have? - http://www.hardocp.com/article/2014/01/13/asus_r9_290x_directcu_ii_oc_overclocking_review/1#.U6cLhLGWd9M

That should not be happening. That is a bad card unless you have zero airflow in your case. Even further overclocked it should be fine still. Per above review "...clock speed is completely stable at 1115MHz with absolutely no drops or throttling." I had the non-x version, great card except for the black screen crash issue. RMA'd it and got the Sapphire R9 290 Vapor-X, amazing card. Perfect!

zod96 said:
That is the problem with just about all 290 series of cards :) Which is why I returned mine and went with the Gtx 780 instead....
Click to expand...

Good sir, please try to restrain yourself from making ignorant comments. Thanks.
 
Sabrewulf... said:
This is the card you have? - http://www.hardocp.com/article/2014/01/13/asus_r9_290x_directcu_ii_oc_overclocking_review/1#.U6cLhLGWd9M

That should not be happening. That is a bad card unless you have zero airflow in your case. Even further overclocked it should be fine still. Per above review "...clock speed is completely stable at 1115MHz with absolutely no drops or throttling." I had the non-x version, great card except for the black screen crash issue. RMA'd it and got the Sapphire R9 290 Vapor-X, amazing card. Perfect!



Good sir, please try to restrain yourself from making ignorant comments. Thanks.
Click to expand...

Yes that's the card I have. With DirectCu II cooling. The test doesn't mention if the "no drops and throttling" is in what usage? Running what exactly?

Also it's possible that the cards sent out for tests early where selected pieces, and doesn't represent the bulk.

Yes I tried with synthetic benchmarks, but I never had a card that had to lower it's performance to be able to coop even in synthetic benchmarks.

I'll have to run some monitoring tool to check if it's indeed throttling running real games.

I have a Silverstone Raven RV03 with where the card is mounted in a 90 degree configuration, so it should have the best airflow of any case out there that I know of. Also the case fans are set to high rpm.
 
I now have a Msi Twin Frozr 780 Ti. It idles about 32 and loads after playing BF4 for about 2 hours at about 66. When I had the 290 it idled at about 55 and loaded about 95 quite a difference and that was with the twin frozr version as well and it was as loud as a vacuum cleaner....
 
zod96 said:
I now have a Msi Twin Frozr 780 Ti. It idles about 32 and loads after playing BF4 for about 2 hours at about 66. When I had the 290 it idled at about 55 and loaded about 95 quite a difference and that was with the twin frozr version as well and it was as loud as a vacuum cleaner....
Click to expand...

K, but this is a thread about a throttling 290. Your recommendation is for the OP to have bought a 780 Ti instead? FTR, Twin Frozr 290 is a fairly good example of a bad aftermarket cooling solution for the 290. Pretty much any cooling solution that had already been developed for the 7000 series isn't that great on the 290. Unfortunately, that includes the DCUII. Still, the OP is either ringing its neck or it's got an issue. Needs new paste or something.

M76 said:
Yes I tried with synthetic benchmarks, but I never had a card that had to lower it's performance to be able to coop even in synthetic benchmarks.
Click to expand...

All the new cards have this "ability". You can try to get them to dim the lights in your house. My 780s would go to 100% fan (since I used MSI afterburner to let them) to try to put out every last FPS they possibly could in benchmarks with VSYNC OFF, and yes... then like any good modern GPUs--my 780s would throttle to stay at 80C.

But in games you are presumably using VSYNC or a framelimiter. huge difference when you start taking big power ratings into consideration. And your card can suck up a shit ton of power if you ask it to. It's stock?
 
Last edited:
For [long term stress testing] a GPU don't leave it running looped synthetic benchmarks with vsync off... that's totally unrealistic. What you are testing with that scenario is: how long will it take my card to reach its throttle point if it tries to completely max its framerate out indefinitely? Obviously, there is a point at which the heat developed and the cooling solution can match and that maximum FPS rate can be maintained without throttling, but with the cards that we are talking about, that usually takes water. Give me your 780Ti for 10 minutes, I'll make it throttle. It's hot here.

Benchmarks meant to be used with VSYNC OFF are short. They don't want to cause your card to throttle. They are supposed to be just long enough to give a ballpark idea of what the card is capable of.

For long term testing a gpu, use a reasonably moderate detailed looped synthetic benchmark with Vsync ON. This will still be rougher on your card than fully detailed real-world gaming sessions, generally. It does depend on how often your card gets a break during gaming, though. Generally speaking with Vsync on, you want your detail level in game to be low enough that the card is running UNDER max load almost all the time. It's full Mhz all the time but it's not needing all of its power to convert the clockcycles into rendering instructions, so it stays cooler. As an extreme example you could add another GPU in Crossfire, run at the same settings with vsync (demanding same exact amount of rendering performance only divided between 2 cards instead of all done by 1) and you would find that the cards would be running even cooler. An identical config/resolution/framerate that runs "just barely" on 1 GPU will generally run cooler and quieter on 2.
 
Last edited:
Bluesun311 said:
For [long term stress testing] a GPU don't leave it running looped synthetic benchmarks with vsync off... that's totally unrealistic. What you are testing with that scenario is: how long will it take my card to reach its throttle point if it tries to completely max its framerate out indefinitely? Obviously, there is a point at which the heat developed and the cooling solution can match and that maximum FPS rate can be maintained without throttling, but with the cards that we are talking about, that usually takes water. Give me your 780Ti for 10 minutes, I'll make it throttle. It's hot here.
Click to expand...
There was no long term involved here. It doesn't take more than a 30 seconds in furmark for the card to start throttling.

Benchmarks meant to be used with VSYNC OFF are short. They don't want to cause your card to throttle. They are supposed to be just long enough to give a ballpark idea of what the card is capable of.
Click to expand...
I don't really get your meaning. What do you mean by want? Benchmarks are there to give me an idea about how much performance my card has in it. But if my card is only capable of producing that performance for a 30 sec benchmark then all it's good for is to get benchmark results that have no relation to real world performance whatsoever.

For long term testing a gpu, use a reasonably moderate detailed looped synthetic benchmark with Vsync ON. This will still be rougher on your card than fully detailed real-world gaming sessions, generally. It does depend on how often your card gets a break during gaming, though. Generally speaking with Vsync on, you want your detail level in game to be low enough that the card is running UNDER max load almost all the time.
Click to expand...
Wait a frakking minute you're telling me that I'm not allowed to use all the processing power of the card? But that's exactly why I bought it. If I thought I don't need the power, then I'd have bought a cheaper card. I'm trying to get into 4K gaming because it looks awesome.

I think the manufacturers went way over the top with the performance race. Producing cards that are only able to go Full Throttle for a few minutes, very disappointing. What's it good if it's just for show and not for long term use?

Yes real world gaming usually means less stress on the card, but only usually, there might be a game that demands more of the card, and then what? In my opinion they should make the card bullet proof. How did the gaming community let the standards get so low? It's like the cards already come overclocked, and I mean not a good overclock but an unstable one. Just like I used to get my 600Mhz Celeron to run on 1200 by directing a room fan at the open computer case at max setting back in the day.

It's full Mhz all the time but it's not needing all of its power to convert the clockcycles into rendering instructions, so it stays cooler. As an extreme example you could add another GPU in Crossfire, run at the same settings with vsync (demanding same exact amount of rendering performance only divided between 2 cards instead of all done by 1) and you would find that the cards would be running even cooler. An identical config/resolution/framerate that runs "just barely" on 1 GPU will generally run cooler and quieter on 2.
Click to expand...

But if I'm to add a second card I'm not doing it to lower the stress on the one. I'm adding it to have even more performance and to be able to run more demanding games at higher resolutions.


____________



I tried running a game, and the card managed to do it without throttling but just barely. The GPU temp maxed out at 86C° And the ambient temp is no more than 24C° I don't find that reassuring at all.
 
What I am saying is, be prepared to have your card throttle and run fast and loud if going with the 290 series of cards, be it reference or after market cards. In his case I would say its normal for most 290 cards...And yes I can recommend a different card if I'd like thank you very much....
 
My 290 gigabyte with WF cooler never cracks 80 and runs quieter than my case fans without throttling.
 
zod96 said:
What I am saying is, be prepared to have your card throttle and run fast and loud if going with the 290 series of cards, be it reference or after market cards. In his case I would say its normal for most 290 cards...And yes I can recommend a different card if I'd like thank you very much....
Click to expand...

You sir, are completely full of shit!
 
And you Sir have no idea what you are talking about :) I had 2 290 cards. An Msi one and a Sapphire one and they both throttled were both hot and both loud so from my stand point I am correct Sir....
 
Pretty sure both AMD and NVIDIA drives reduce clock speed when you run Furmark.

I tried running a game, and the card managed to do it without throttling but just barely. The GPU temp maxed out at 86C° And the ambient temp is no more than 24C° I don't find that reassuring at all.
Click to expand...

The 290X is designed to run AT 95C and not lower. So that is perfectly fine.
 
I now have a Msi Twin Frozr 780 Ti. It idles about 32 and loads after playing BF4 for about 2 hours at about 66. When I had the 290 it idled at about 55 and loaded about 95 quite a difference and that was with the twin frozr version as well and it was as loud as a vacuum cleaner....
Click to expand...

Way to compare a reference cooled card to a much higher end aftermarket designed competitor's product..Real Apples to DumpTruck comparison there..Bad Troll Attempt -1 point.:rolleyes::rolleyes:

What I am saying is, be prepared to have your card throttle and run fast and loud if going with the 290 series of cards, be it reference or after market cards. In his case I would say its normal for most 290 cards...And yes I can recommend a different card if I'd like thank you very much....
Click to expand...

Gotta give you credit for having heart (If trolls have them that is)..But there are PLENTY of excellent 290/290X aftermarket designs that are nice and quiet..

On topic, I am amazed that nobody has asked this yet..OP, exactly which Catalyst driver are you running? If you are running anything older then 13.12, or the 14.1/2/3/4 (possibly) then you will never stop throttling..I nearly tore my hair out with the BETA 14s because they had a powertune bug that caused my WC'd card to throttle despite a GPU temp of ~42C:rolleyes::rolleyes:..

If you are running the 14.6 (the latest WHQL release) then you shouldn't be having an issue with throttling clock speeds unless something is wrong with the card. Have you taken at look at the VRM #1 temps? If not use GPU-Z and tell me what it is reporting.
 
With the 290's I recommend buying aftermarket cards with coolers designed specifically for Hawaii GPU's. The Asus/MSI/Gigabyte use coolers that are a better fit for GK110. Especially the DCII cooler where 2 of the heatpipes don't even contact the GPU.

asus290oc-9w.jpg
 
Wait....so if you post something bad in a thread you are a troll? Is that how it works....:rolleyes:
 
ccityinstaller said:
If you are running the 14.6 (the latest WHQL release) then you shouldn't be having an issue with throttling clock speeds unless something is wrong with the card. Have you taken at look at the VRM #1 temps? If not use GPU-Z and tell me what it is reporting.
Click to expand...

I'm using 14.6, but last time I checked it was still beta. I can't check anything for a few days I'm away on business forsaken to my laptop.
 
I've just checked, VRM temperutare goes up to 114, and GPU temp goes to 95 while gaming, not in synthetic tests.

Does that confirm that the card is faulty?
 
M76 said:
I've just checked, VRM temperutare goes up to 114, and GPU temp goes to 95 while gaming, not in synthetic tests.

Does that confirm that the card is faulty?
Click to expand...

If you can, return it and get a Sapphire Tri-X or Powercolor PCS+. The coolers on those are actually designed to fit Hawaii and will cool much better.

Edit: Sorry, I forgot you have a Raven case. The Tri-X and PCS+ aren't ideal for vertical mounting. I'm not sure what your best option is.
 
Last edited:
Something is off, RMA that bad boy... 290X with a decent cooler sits at 72ish loaded, 80 overclocked The only other things to look at are case airflow, shouldn't be a problem with the Raven. 4k is awesome but loads the GPU up even more. I jumped on another 290X reference as I found only one had me turning down the details. I'm going Windforce 290X on the bottom, and reference 290X on top with a G10/AIO cooler mounted. I tell you this for down the road when you want to upgrade to another, non blower 290X's in crossfire is way to much heat in the case, any case...
 
Yeah but this is exactly the situation when it's hard to RMA, because the card seems to be working normally except for a few games.
 
M76 said:
I've just checked, VRM temperutare goes up to 114, and GPU temp goes to 95 while gaming, not in synthetic tests.

Does that confirm that the card is faulty?
Click to expand...



No. It just means your VRM temps are too high. You need to change thermal pads or check the VRM block heat sinks and make sure they have good contact. I like to put fujipoly xtreme on my video card VRMS... helps alot.
 
Well the temps seem a little extreme...Could reseat everything, and apply fujipoly pads to the VRMs if RMA isn't possible. I bought a sheet of it for my Gigabyte 290X, at stock the VRMs shoot up 10 over the core. Haven't had a chance to install yet, don't bother with the top end stuff, way too expensive! I went one level down.

Just be careful as it may void the warranty if you disassemble the card to reapply thermal paste and reseat the heatsink, stupid rule just look for a sticker covering one of the screws on the back, if so I'd say RMA is your only option
 
Yes I've seen the sticker, so reapplying the cooler option is off the table. Guess I have to RMA, which means no gaming for me for weeks. I guess I'll buy a 780Ti, and sell this when I get it back. I just don't want to sit around doing nothing after spending this much money on my PC.
 
Gaming at 4k you're going to want all the VRAM you can get, 290X is better @ 4K...bluesun311 is selling an almost brand new Powercolor PCS+ 290X, way better cooler on that than the Asus, could pick that up and sell the Asus when it gets back from RMA
 
ccityinstaller said:
Way to compare a reference cooled card to a much higher end aftermarket designed competitor's product..Real Apples to DumpTruck comparison there..Bad Troll Attempt -1 point.:rolleyes::rolleyes:
Click to expand...

my r9 290 goes up to like 90 degrees while im gaming. I don't understand why you guys are calling him a troll.
 
Last edited:
psyclist said:
Gaming at 4k you're going to want all the VRAM you can get, 290X is better @ 4K...bluesun311 is selling an almost brand new Powercolor PCS+ 290X, way better cooler on that than the Asus, could pick that up and sell the Asus when it gets back from RMA
Click to expand...

I'm in Eu, so that's not an option. I got this 290x way underpriced as far as european prices go. I'd have to pay at least $150 more for a powercolor.

These two cards are in the same price range:

GIGABYTE GV-R929XWF3-4GD
MSI R9 290X GAMING 4GB

I don't know if they have better cooling.
 
Have you tried running your GPU fan speed at 100% to see if it stops throttling?
If it does stop then keep dropping the fan speed until it starts again. Then you can set a custom fan profile in MSI Afterburner.
 
Is this card a reference design with the blower? I'm not an expert with the raven case, but I'd assume the blower style would be well suited to its design. I have the Gigabyte card you listed, works great even overclocked, mine will hit 80 on a warm day, but that's at 1150/6200 and +100mv. The only issue with the windforce I've found is VRM temps get high, about 10 degrees above the core at gaming load, so 90ish. But there are ways of fixing that, got some fujipoly thermal pads to put there and bring my temps down.

I second creating a custom fan profile in afterburner, have it stay low until 55 and then just ramp it straight line to 100% at 85 degrees. This will eliminate bad BIOS fan control that could be present. Make sure the fan profile you make is turned on and working correctly.
 
I bought a 290 that unlocks to a 290x.
Have tried 2 different 290x BIOS versions and both throttle often.
This is with an Accelero Xtreme III cooler. temps are good (well below 70C).

M76 said:
I've just checked, VRM temperutare goes up to 114, and GPU temp goes to 95 while gaming, not in synthetic tests.

Does that confirm that the card is faulty?
Click to expand...

M76, not sure about faulty.
Those temps are very high, possibly due to bad case air flow.
Too little info.
 
Nenu said:
I bought a 290 that unlocks to a 290x.
M76, not sure about faulty.
Those temps are very high, possibly due to bad case air flow.
Too little info.
Click to expand...

You would all have the info you're looking for had you read the thread from the beginning.
 
M76 said:
since hot air moves up, it shouldn't be a problem to dissipate upward, especially with the huge fans blowing air up from the bottom.
Click to expand...

The trouble is you are relying on a single 120mm fan to exhaust all the hot air out of your case. Your ASUS R9 290X DirectCU II OC is dumping 265 watts of heat directly into your case. I'm not sure if a single exhaust fan can keep up with that. The reviews I see have your GPU card load temperatures in the mid 70s.
I'm curious if running the GPU fans at 100% makes any difference.
 
Last edited:
Nuby1Canuby said:
Have you tried running your GPU fan speed at 100% to see if it stops throttling?
If it does stop then keep dropping the fan speed until it starts again. Then you can set a custom fan profile in MSI Afterburner.
Click to expand...

It's hard to tell, because the gpu clock is always changed dynamically based on load I guess. So it's throttling even when it's at 70C


Nenu said:
good luck ;)
Click to expand...

Thank you.
 
I had issues with throttling on my r9 290 'not temp related stays around mid to high 60's+ I fixed it by downloading msi afterburner, going into settings. Check extend official overclocking limits, check disable ULPS, set mode to *without PowerPlay support* Increase the power limit & set a fan profile to keep temps in check. Now my core clock never drops, but the downside to this method it won't go back down to 2d clocks unless you reset on afterburner, which is fine when your just getting ready to play a game & you can save an load a profile & tada fixed.
 
You must log in or register to reply here.
Back
Top