Titan overclocks like garbage (throttle issue)

BababooeyHTJ

Supreme [H]ardness
Joined
Jan 21, 2009
Messages
6,951
In case you guys haven't checked out any Titan threads on the forums there is a pretty big issue where these cards just throttle for no reason. The core voltage drops like crazy and the clock speed drops a few bins. From what I've read this issue appears to be the norm.

It always amazes me how this sort of thing never seems to get mentioned in reviews. I wonder why. ;)

Here is a post on OCN with a user with a water cooled Titan throttling to 967mhz at 1.012v. Now that is what I expect from a true enthusiast gpu. :rolleyes:
 
It's funny because the only people that really truly care are the people benchmarking. Just play the damn games. You won't see a huge difference between the card running at 1175mhz and if it throttles down to 1100mhz.
 
In case you guys haven't checked out any Titan threads on the forums there is a pretty big issue where these cards just throttle for no reason. The core voltage drops like crazy and the clock speed drops a few bins. From what I've read this issue appears to be the norm.

It always amazes me how this sort of thing never seems to get mentioned in reviews. I wonder why. ;)

Here is a post on OCN with a user with a water cooled Titan throttling to 967mhz at 1.012v. Now that is what I expect from a true enthusiast gpu. :rolleyes:

You are absolutely correct they never cover small issues as such Reviews all over. Seems like they don't spend enough time with the product before posting the review. its US the END USERS who really discover new things/defects about these products. Never be the first one on the block to buy fresh product unless u have plenty of money laying around. But OC doesn't make a WHOLE lot of difference. Lets see what AMD will bring us next . EXCITED !!!!
 
I wouldn't call core voltage throttling 10% or more a small issue.

It's funny because the only people that really truly care are the people benchmarking. Just play the damn games. You won't see a huge difference between the card running at 1175mhz and if it throttles down to 1100mhz.

Yeah, the type of person spending a g note on a video card doesn't care if it throttles 10% or more for no apparent reason.
 
I wouldn't call core voltage throttling 10% or more a small issue.



Yeah, the type of person spending a g note on a video card doesn't care if it throttles 10% or more for no apparent reason.

That ONE guy probably just has a bad card. I've never seen anyone else's card throttle more than 100mhz from a stable overclock.
 
That ONE guy probably just has a bad card. I've never seen anyone else's card throttle more than 100mhz from a stable overclock.

Actually a lot of people are seeing similar throttling for no apparent reason. I just used that as a more extreme example. You can look through that ocn thread for more examples.



It would be great if gpu boost 2.0 worked like that in practice, it doesn't.
 
How can you say the Titan OCs like garbage?

I have my card at stock. It overclocks itself up to about 1030 during games and sometimes drifts down to 1000Mhz. That is a 23% increase over the guaranteed speeds and like I said that is everything at stock.

What would you consider a good overclock?

Maybe NV should only guarantee clocks at around 600Mhz that way when you got to 1100you would feel so much better about the card when you got the same 1100.:)
 
The cards shouldn't throttle down for no reason. There is a guy on OCN with a water cooled Titan throttling down to less than 1ghz for no reason, hes not alone either. My card throttles as bad.
 
Last edited:
The cards are only throttling during benchmarking and with overclocking.

I am not overclocking and my Titan SC boosts to 1020ghz during gaming. And stays there.
 
My cards don't throttle like that. One card stays between 1097 and 1124 and the other will stay between 1137 and 1150 all the time.
 
The cards are only throttling during benchmarking and with overclocking.

I am not overclocking and my Titan SC boosts to 1020ghz during gaming. And stays there.

Mine throttles in some games. I wouldn't have bought it knowing this.
 
Mine throttles in some games. I wouldn't have bought it knowing this.

Chances are it can be resolved in a bios or driver update. Even then, if it doesn't, I wouldn't care as I'm getting boosts above advertised rates.
 
Yes, you shouldn't care when you can buy two GTX 680 lightnings for the price of a single titan that throttles for no reason.
 
For 1000 dollars, the gpu better be performing at least as well as advertised. I would be on the phone setting up an RMA.. Maybe a bios update can fix that throttling issue but i seriously doubt any drivers would.
 
For 1000 dollars, the gpu better be performing at least as well as advertised. I would be on the phone setting up an RMA.. Maybe a bios update can fix that throttling issue but i seriously doubt any drivers would.

But it does. Everyone pretty much boosts above advertised speeds. The people complaining are the heavy overclockers.

I'm not overclocking. My card is only suppose to boost to 928, but it actually boosts to 1020. And it stays there.
 
ahhhhh okay, yea updates will help then. I didn't read the other articles yet and assumed the issue was occurring while gaming with a stable OC. sorry bout that.
 
Yes, you shouldn't care when you can buy two GTX 680 lightnings for the price of a single titan that throttles for no reason.

This option was available to everybody. I would still rather have the Titan.

Every review site said that a single Titan would not be able to compete with overclocked 680 SLI and would even be slower than the 690 in many instances.


So if you're aim is to achieve those other solutions levels of performance through OC the hell out of the Titan, I dont think its gonna be possible.
 
ahhhhh okay, yea updates will help then. I didn't read the other articles yet and assumed the issue was occurring while gaming with a stable OC. sorry bout that.

My experience almost exactly mirrors BatJoe. I only have throttling during benchmarks. Good thing I dont play benchmarks.
 
It appears to be case by case. Mine OC to 1215(boost) and stays theres except in benchmark then it drops slightly to maybe 1175
 
Early adopters, early beta drivers, everyone is panicking for no reason... the usual pattern with every new cards...

Just wait for more mature drivers and hacked BIOS. It will get better... Like it always does...
 
I really haven't seen any throttling at 1150 Core (boost) with auto fan. I am pushing 106 TDP during benching though, so I would imagine if I clocked any higher, I might. One strange thing is with manual fan, I did see throttling down to 1137 and sticking there even though I was no where near max TDP and was only at 65 degrees vs 85 temp threshold setting. Switched the fan back to auto and BAM, back to 1150 and it never moved. Card started pushing 80+ though.

Very odd that changing a fan setting that forces the card to run cooler would cause throttling.
 
Last edited:
It appears to be case by case. Mine OC to 1215(boost) and stays theres except in benchmark then it drops slightly to maybe 1175

As you've stated on a couple of forums. That isn't the case for most people it seems.
 
This option was available to everybody. I would still rather have the Titan.

Every review site said that a single Titan would not be able to compete with overclocked 680 SLI and would even be slower than the 690 in many instances.


So if you're aim is to achieve those other solutions levels of performance through OC the hell out of the Titan, I dont think its gonna be possible.

You missed my point. A lot of Titans on the market won't oc to the levels that the cherry picked review samples will.

The fact of the matter is at the moment gpu boost 2.0 does not work as advertised. Read up on gpu boost 2.0 and what its not doing and then you might see why people spending a grand might be a little upset. Had I known about how these cards work in practice I wouldn't have bought one.
 
The way I understand GPU Boost 2.0:
It allows for over-volting which will affect operating temperature of the GPU, and since the boost capability is tied to the operating temperature of the GPU, this in essence means that the boost clock rate is going to be tied to the demand placed on the GPU. Thing is, nVidia allows each manufacturer to set their own GB2.0 parameters, so results will vary from brand to brand, as well as card to card.

Base Clock is rated for 837 MHz.
Boost clock is rated for 876 MHz.

Someone with a WC'd Titan is bitching because it's throttling back to 967 MHz, which just so happens to be 10% above the default BOOST clock rating. WTF is the problem with these, exactly? Because I don't see anything wrong. The whole point of GPU Boost is to maximize the performance out of the box while providing safe operating parameters so people aren't setting their $1000 GPUs on fire, which means that nVidia did one hell of a fine job of creating a product that damn near already runs at it's operational limits while providing a longer usable lifespan. AMD has also done something similar with their Boost and GHz editions.

We should be thanking the manufacturers for stepping up and providing products with a very nice overclocking : stability : lifespan ratio right out of the cartons they ship them in. Afterall, finding this kind of "golden ratio" is what daily driver OC'ing is all about. We, as the end users, just no longer have to fiddle-fuck with our GPUs to find it.

Bottom line is this: every current Titan owner has spent the $1000 early adopter fee per card and it sounds to me there were a lot of obvious false pretenses or higher than what should have been normal expectations for OC'ing. If you are not truly happy with the performance the Titan delivers over other single GPU solutions, then return it and get your money back. It is a lot of coin, afterall. If everyone returns their Titans, it will send a rather large message to the manufacturers about what you truly want in a flagship product going forward...vote with your wallets.
 
The whole point of GPU Boost is to maximize the performance out of the box while providing safe operating parameters so people aren't setting their $1000 GPUs on fire, which means that nVidia did one hell of a fine job of creating a product that damn near already runs at it's operational limits while providing a longer usable lifespan. AMD has also done something similar with their Boost and GHz editions.

We should be thanking the manufacturers for stepping up and providing products with a very nice overclocking : stability : lifespan ratio right out of the cartons they ship them in. Afterall, finding this kind of "golden ratio" is what daily driver OC'ing is all about. We, as the end users, just no longer have to fiddle-fuck with our GPUs to find it.

Yes, because us enthusiasts who know what we are doing (usually) have always been setting our flagship GPUs on fire. Look all those unlocked MSI 680 Lightnings blowing up left and right.

GPU Boost would have been a great idea had been made optional. Intel has done it right. Nvidia hasn't.
 
Yes, because us enthusiasts who know what we are doing (usually) have always been setting our flagship GPUs on fire. Look all those unlocked MSI 680 Lightnings blowing up left and right.

GPU Boost would have been a great idea had been made optional. Intel has done it right. Nvidia hasn't.

Good gravy, read between the lines... set on fire was not supposed to be read in a literal sense so much as to be read as decreasing failure and RMA rates. Remember, it's up to the manufacturer to set their own parameters, so if MSI wants to release a fully unlocked GPU, they have zero constraints from nVidia in doing so. Now when focusing on a $1000 GPU, I'd expect there to be much less leniency. It's not nVidia's fault here...it's the brand manufacturers for releasing every Titan at nVidia specs...well, because there's no need for anything more from a single GPU that happens to be the fastest on the planet right now.

I guess I'm just not understanding the basis for all the complaining. Don't like it? Return it. Vote with your wallets so the manufacturers deliver what is desired from the consumer base in this product price bracket.
 
I guess I'm just not understanding the basis for all the complaining. Don't like it? Return it. Vote with your wallets so the manufacturers deliver what is desired from the consumer base in this product price bracket.

Dude, this thread is about the cards not working the way that they are supposed to.

Watch this video, thats how gpu boost 2.0 is supposed to function as described by a Nvidia rep. Then look at the links that I've provided about the problem. Then comment.

Educate yourself on the issue before you comment.

R Johnson from EVGA has already commented that Titan is not supposed to throttle for no reason and that its grounds for an RMA. I'm sure that you know more about the subject then he does. :rolleyes:
 
Good gravy, read between the lines... set on fire was not supposed to be read in a literal sense so much as to be read as decreasing failure and RMA rates. Remember, it's up to the manufacturer to set their own parameters, so if MSI wants to release a fully unlocked GPU, they have zero constraints from nVidia in doing so. Now when focusing on a $1000 GPU, I'd expect there to be much less leniency. It's not nVidia's fault here...it's the brand manufacturers for releasing every Titan at nVidia specs...well, because there's no need for anything more from a single GPU that happens to be the fastest on the planet right now.

I guess I'm just not understanding the basis for all the complaining. Don't like it? Return it. Vote with your wallets so the manufacturers deliver what is desired from the consumer base in this product price bracket.

How is overclocking increase card failure? You making no sense right there sorry.

I have OC nearly everything from MacBook to Desktop for years, none of them have failure because of OC.
 
This is essentially no different than how the GTX 690 was handled, but I think it’s something that’s important to note before anyone with dreams of big overclocks throws down $999 on a Titan card. To be clear, GPU Boost 2.0 is a significant improvement in the entire power/thermal management process compared to GPU Boost 1.0, and this kind of control means that no one needs to be concerned with blowing up their video card (accidentally or otherwise), but it’s a system that comes with gains and losses. So overclockers will want to pay close attention to what they’re getting into with GPU Boost 2.0 and Titan, and what they can and cannot do with the card.

The above is directly off of Anands review.

On [H] review Brent even describes the same behavior you're complaining about.


Its funny you keep telling people to educate themselves. GPU boost 2.0 is working exactly as described in the research I did.:D

Maybe NV will tweak it in the future so guys like you who dont care if they damage their hardware can go for those 1300Mhz overclocks.
 
Last edited:
Dude, this thread is about the cards not working the way that they are supposed to.

Watch this video, thats how gpu boost 2.0 is supposed to function as described by a Nvidia rep. Then look at the links that I've provided about the problem. Then comment.

Educate yourself on the issue before you comment.

R Johnson from EVGA has already commented that Titan is not supposed to throttle for no reason and that its grounds for an RMA. I'm sure that you know more about the subject then he does. :rolleyes:

It doesn't make any logical sense that it wouldn't throttle when pushed beyond certain criteria. If it's out of it's operational boundaries, it's going to throttle to protect itself from overvolt or thermal damage. Those boundaries are set by eVGA. Perhaps they did screw up the settings?

How is overclocking increase card failure? You making no sense right there sorry.

I have OC nearly everything from MacBook to Desktop for years, none of them have failure because of OC.

Great. You are a successful OC'er and haven't had any catastrophic failures. Congrats. Doesn't exclude anyone else from damaging their hardware from doing so. There's a reason why Intel has an optional protection plan one can purchase to get a free processor replacement in the case one does fry their CPU from OC'ing.
 
Great. You are a successful OC'er and haven't had any catastrophic failures. Congrats. Doesn't exclude anyone else from damaging their hardware from doing so. There's a reason why Intel has an optional protection plan one can purchase to get a free processor replacement in the case one does fry their CPU from OC'ing.

Can you simply show me an example or someone who have ever damage their card by OC?
Not including BIOS / hardware modification.

The reason why Intel does it simply wants to make more money, just like their K series.
It's pretty obvious if you haven't figure it out.....
 
It doesn't make any logical sense that it wouldn't throttle when pushed beyond certain criteria. If it's out of it's operational boundaries, it's going to throttle to protect itself from overvolt or thermal damage. Those boundaries are set by eVGA. Perhaps they did screw up the settings?

WTF are you talking about? Did you watch the video? Did you read my post? Did you educate yourself on the subject at all?

No, you have got to just be a troll. You don't know anything about how these cards work and it shows. The worst part is you won't bother to educate yourself in the least bit about how they operate yet still feel the need to comment. :rolleyes:

Can you simply show me an example or someone who have ever damage their card by OC?
Not including BIOS / hardware modification.

The reason why Intel does it simply wants to make more money, just like their K series.
It's pretty obvious if you haven't figure it out.....


He is talking out of his ass.
 
WTF are you talking about? Did you watch the video? Did you read my post? Did you educate yourself on the subject at all?

No, you have got to just be a troll. You don't know anything about how these cards work and it shows. The worst part is you won't bother to educate yourself in the least bit about how they operate yet still feel the need to comment. :rolleyes:




He is talking out of his ass.

Did you bother to read any in-depth reviews about GB2.0 and how nVidia leaves it solely to the manufacturer to set the boundaries on how it behaves? That's what I'm basing my comments on...and after watching the video, I'm inclined to conclude that perhaps there IS a problem with how these Titans are throttling. Thought I made that clear in my previous post. FFS

And no, I'm not talking out of my ass about OC'ing having the potential to destroy. Overvolt it too much and the ICs will cook leading to premature failure. Don't provide enough cooling and the ICs will cook leading to premature failure. Pretty basic stuff.
 
Back
Top