Know Your Stuff: Experimenting with Fan Curves and Overclocking

===========================================================

When it comes to customizing our graphics cards, we often think about adjusting fan speeds or tweaking power limits. But have you ever considered experimenting with fan curves? In this article, we'll dive into the world of fan curves and explore how they can impact performance and temperatures.

I've been playing around with my NVIDIA GeForce RTX 3080 Ti "Tight Necks" for a while now, trying to squeeze out every last bit of performance. As I dug deeper into the settings menu, I realized that the default fan curve was being quite... conservative. So, I decided to take matters into my own hands and create a custom fan curve.

By adjusting the fan speed limits, I was able to push the temperatures up and down. Here's what I ended up with:

Temperature Target: 85°C (now changed to a more aggressive target)

Power Limit: Still at its default setting

Now, let's talk about what happens when you adjust the fan curve. By allowing the fan to spin faster at higher temperatures, we can sacrifice a little bit of acoustic sound in exchange for increased performance.

I'm not worried about being quiet; I'd rather have more performance than a silent system. And it seems that NVIDIA was being too conservative with the default 50% fan cap. By pushing the fan speed up, we're able to reach higher core clocks and gain an average of 4 FPS at 4K.

Here's what my results looked like:

Average Frame Rate:

- Factory Settings: around 60-70 FPS

- Custom Fan Curve: around 64.7 FPS (new lowest) and 66.5 FPS (average)

Core Clocks:

- Factory Settings: starting in the mid-to-low 1600s

- Custom Fan Curve: reaching up to 2500 MHz

Average Temperatures:

- Factory Settings: fluctuating between 77°C and 85°C

- Custom Fan Curve: staying relatively consistent at around 77°C

It's clear that adjusting the fan curve can have a significant impact on performance and temperatures. By allowing the fan to spin faster, we're able to push the limits of what our graphics card can do.

Overclocking: The Final Step

-------------------------

To take my overclocking to the next level, I decided to try and reach even higher core clocks. With the custom fan curve in place, I was able to squeeze out an additional 50-100 MHz from the base clock speed.

However, as I pushed the fan speed up, temperatures began to rise exponentially. I found that running at 100% fan speed resulted in temperatures of around 77°C, which is relatively hot but manageable for this particular card.

The Results: A New Benchmark

---------------------------

By combining the custom fan curve with the overclocked settings, we were able to set a new benchmark for performance:

- Average Frame Rate: around 64.7 FPS (new lowest) and 66.5 FPS (average)

- Core Clocks: reaching up to 2500 MHz

Keep in Mind: Not Just for NVIDIA

-------------------------------

This experiment can be applied to any graphics card, not just the tight necks from NVIDIA. By experimenting with fan curves, you may find that it helps improve performance and reduce temperatures.

Conclusion:

----------

Adjusting fan speeds and using custom fan curves is a powerful way to squeeze out more performance from your graphics card. By pushing the limits of what our cards can do, we can enjoy faster frame rates and more responsive gaming experiences.

Of course, this comes with some trade-offs, such as increased noise levels or higher temperatures. But for those willing to experiment, the results can be well worth it.

As I mentioned earlier, I've installed two of these RTX 3080 Ti's in my Skunkworks system, and I must say that it's been a joy to play around with their custom fan curves. The results speak for themselves: a faster gaming experience that's worth every penny.

So, go ahead and give fan curves a try! You never know what kind of performance gains you might discover.

"WEBVTTKind: captionsLanguage: enhey Nvidia hey what's up J here yeah I was this on your website right now and I noticed that you guys did look Jay from Jay Z J's to sense a YouTube channel with 680 thousands of subscribers we met in Austin back in May during the launch of it for the 1080 deadmau5 was there I'm the tall brown-haired guy kind of handsome no that's that's Luke anyway so the reason why I'm calling is I saw that you guys launched a brand new Titan or relaunched an old Titan whichever I'm whatever you guys want to call it and I was kind of curious if you were interested in study 1 over for review oh I see ok well what if ok so how about this then I counter the no with my soul for one Titan X or at least what's left of it up in YouTube it's got to have some monetary value I see I've got kids those have any monetary value oh okay great great um ok so you should have my address and go ahead and send that over and we'll get right to it yeah alright yeah you YouTube yeah j JS JS 2 says nor ryan is with pc / that that's an entirely different site well you know it doesn't matter just go ahead and send it over and we'll get right on that all right thanks bye bye uu tuna that wasn't really all that hard wow they're fast with its unique freeform modular system the new master case maker 5 from coolermaster allows unparalleled flexibility with its adjustable internal layout and exterior customization options learn more about how you can start customizing your own case by following the link down in the description the GT x 1080 has been extremely popular so popular in fact they are not in stock anywhere they have been nothing but a logistics nightmare when it comes to keeping these things in stock so if you've ordered one or you've got one congratulations you were top dog for all of about two months because now we've got the new Titan here which is promising to be quite the behemoth when it comes to power yeah we're gonna we're gonna kind of move through this video kind of quickly here because I want to do some live benchmarking with this video and and you guys like like benchmarking haven't done it in a while and we're going to do that today but before we do that we have to talk about some of the specs because you know that if you're a specs nerd like myself you kind of you kind of get off on some of that stuff but anyway it's got three thousand five hundred and eighty CUDA core it's approximately a thousand more then we'll call little Pascal here on the GTX 1080 and it's got a fourteen hundred and seventeen megahertz boost clock which is funny because that's about as far as one of my Titan X's the Maxwell version of Titan X is somewhere in this pile here kind of maxed out about 1450 the three that are in skunkworks get up to about 1500 but this one's base clock is barely below that and it's got a fifteen hundred and thirty one megahertz boost clock but as you guys know we're gonna we're gonna push this thing as far as we can I know why some people are afraid to overclock this thing but not me not me I'm gonna I had to pay for it I just had to give a couple of my kids to Nvidia and it was totally worth it but for memory it's got 12 gigabytes of g5x running at 10 gigabit gigabit per second that is up that's some fast memory right there as 10,000 megahertz and as you guys know that's overclockable as well but a 384 bit memory bus and a 480 gigabyte per second memory bin that's pretty huge considering this is not HBM and I think that might be one of the first things that some of the people have been disappointed about is the fact that we were hoping that Titan on Pascal was going to have HBM but considering all the P 100 cores that you know they talked about earlier in the year are probably eating up all that HBM productivity and manufacturing it's not surprising that the Titan X does not feature HP m especially since they're saying that this is technically not a gaming card it was developed and made super fast for the purposes of deep learning and creating Skynet and having the end of the world as we know it ground zero B right there at Nvidia headquarters but again I digress the card itself is actually aimed more towards deep learning that is gaming even though it says GeForce GTX right on the side whatever of course it has the stuff that you would expect like pcie 3.0 HDMI 2.0 B and all that stuff so you can hook this up to your home theater of course you know this is a home theater card obviously because you know you could do HD 4k playback through HDMI which is kind of a big deal in terms of output here it's got the same stuff as the 1080 it's got an HDMI 2.0 B and it's got three DisplayPort 1.4 as well as a dual link DVI but obviously would you guys care most about it does have a backplate and the answer is yes it has a backplate but for 1,200 bucks it better have a backplate hello better have a front plate too for all that matters does it support sli yes up to two-way especially with an HB Bridge definitely recommended does a support simultaneous multi-project yes does it support Ansel yes does it support g-sync no it actually does not just just kidding yes of course it supports g-sync but it's got a TDP of 250 watts any recommended 600 watt or higher power supply to supply power to the 8 pin and 6 pin PCI Express power so that's that here's we're going to do a quick we're going to jump it to the benchmarks I'm going to show you those and we're going to hook it up on the test bench we're talk about a few things there are some things to talk about here believe it or not and we're going to get through this as best we can without it being as painful as possible transition you so if you bought the GTX 1080 simply because of the fact that it was the best of the best at the time that you got it or at least tried to get it rest in peace myself included because this guy has just been pushed out by this guy for being the fastest single GPU in the world but it's not that cut and dry guys because if you look at things like performance per dollar if you care about that even if you're shopping for the best of the best then you guys will know that there is a huge diminishing return on the Titan X versus the GTX 1080 when you compare the price of the fact that this thing right here is 1200 freaking US dollars or three thousand nine hundred and sixty two dollars in Australia or whatever it converts to be sorry guys you guys really do get wrecked with pricing down there it's terrible but yeah the GTX 1080 is still a huge huge performer when it comes to performance per dollar I say that considering it's a $700 graphics card but remember we are comparing it to 1,200 two of these is only 200 more than one of these and two of these will literally smash on one of these and in fact we're going to do a separate video about 1080 sli versus a single type next in fact I'd like to do that video today but I want to keep this video manageable we're going to keep it with one card at least compared to this one for now and we will test other SLI configs versus this guy right here but there are some things I want to talk about with this card and to do that we are going to have to go ahead and transition over to the test bench here so I can show you guys actually firsthand and not just talk about it on some of the things to expect with a GTX Titan X or just tight next because apparently they're going to maybe drop the GTX moniker in the future who knows so from left to right this is the power percentage we're leaving that at stock 100 this is the temperature this is the GPU load percentage this is the fan speed percentage and that is the core clock remember this has a 1400 and 17 megahertz base clock with a 15 31 boost clocks we want to keep an eye on that number and see what happens here so anyway gameplay is super smooth even at 4k but then again it should be as you can see we've got a single card with MSA a 2x at 4k achieving higher than 60fps that's actually kind of a big deal it's funny cuz I used to use 4k tests and I always found it very difficult to play in 4k because it was kind of laggy stuttery that's obviously not the case here as long as we stay above that 60fps then we're pretty good I've done this mission so many times this is the mission I actually use when I benchmark grand theft auto 5 because it's pretty much on a track everything is identical happens the same way every single time when it comes to the world and stuff so it's in game play stupid shove up to it it's in the game it's not a benchmark on a track I'm controlling it but it's very real-world scenario in my opinion as you can see here right now the temperatures have reached 83 to 84 that's where it's going to start to control things to try and keep the temperatures down the only way the GPU can actually keep the temperatures down is going to be by either increasing the fan speed which by default they've got it maxed out at 50% it can't go any faster than that or reduce the voltages and/or the clocks now the clocks and the voltages go hand-in-hand so those work together it's what you're going to notice here the more that I play you're going to see these boost clocks start to come down a little bit now anything over 1531 right now is considered above the boost clock so technically although we've come down from 1700 ish down to the high 15s low 16s it's not considered a thermal throttle in the sense of the way thermal throttling is intended to work yeah it is controlling the speed of the graphics by speeding up and slowing it down to control thermals but thermal throttling usually relates to the GPU trying to save itself from certain death or the CPU in that matter because it's the same concept there by slowing itself down so that it doesn't cook now we're not doing that here because we're still above the boost clock so that's just one of the things I want to mention here the same goes for AMD this is an Nvidia specific thing where people want to say oh it slowed down therefore it's thermal throttling no it's it's not thermal throttling its thermal adjusting and thermal throttling would be if we start to go down below the advertised speeds because of of thermals if that makes sense oh it's the cops the cops they know I stole this Titan X they're after me Simeon's not gonna be happy that I don't wreck this car up whoa that's that's bad oh geez so the temperatures pretty much went right up to 83 84 and as you guys saw in the benchmark chart because I did all founders Edition cards you know or reference cards for 980ti Titan X and the 1080 and the type the new Titan X they all pretty much just went straight to their max temperature threshold which is set right here as you can see it's set to 84 so what I'm doing right now is I have got Metro last light running on 4 K loop 5 times because I want to see what's going to happen to this core clock and I want to see if I can recreate what was happening during some of my benchmark testing was how far C we came all the way down to 15 31 now so all the way down to the boost clock that the factory boosts clocks we and now we're below it we went down to 14 81 for a moment this is just the beginning of the first test as you can see as our temperature is sitting here at 83 and our load is out a hundred percent our fan speed isn't allowed to go any faster than 50 or 51 percent I want to see how far that this clock will actually come down so we're gonna let the tests go five times we'll compare the average FPS then we'll make some tweaks here and then we'll see how well it actually does so the first test here at a sixty four point seven one average frame rate then the the clock runs zero that's so weird then the second test was sixty two point three then 61 point eight then 61.7 then 61 point one so you can see it was slowly starting to slow down but if we take a look at the the chart here you can see right here the core clock bounces around so much I mean so much that's that's GPU boost and the way it's meant to work but I find myself asking if this result could be better by simply playing around with the fan curve and you know it can be that's the whole point of this you know it can be so all I'm going to do here I'm going to leave the temperature target and the power limit the same but all I'm going to do is put in here a custom fan curve and I'm just going to allow the fan to go faster at the higher temperature so what we're sacrificing here is a little bit of acoustic sound so the card will be a little bit noisier but I don't know about you I'm the kind of guy that would be more than happy to add a little bit of noise to my system if it means I get more performance than worrying about it being stupidly silent and I think that NVIDIA is being waved too conservative with the 50% fan cap go faster on the fan speed and let's see what actually happens to these core clocks as you can see the way it started here we're already higher but anyway we're going to let it go and then we'll come back and we'll compare well if you can hear me over the fan it's obviously kind of noisy it's running 85% now I want to show you where on the fifth test right here and it's still starting in the 1700 megahertz range and it drops down into the mid to low 1600s and then comes back up but look at that compared to the other test which dropped down into the high 14s low 15s we're sitting almost 200 megahertz average faster speed alright so the test is done and the only difference here is as I showed you I change the fan curve nothing else so now we've got a sixty six point seven nine a sixty six point five one a sixty six point six three a sixty six point three seven a sixty six point six nine do you see the difference there we gained four FPS average increase by doing nothing more at 4k by the way for FPS at 4k gain is a lot and that was by doing nothing more than changing the damn fan curve I wish NVIDIA would just make it a little bit more aggressive from the factory yeah and then for those that are curious here is the core clocks right here on the previous test before the new fan curve this is where I did the little video there's the segment saying I'm gonna adjust a fan curve and there's the new core clock right there look how much that jumped up by changing the fan curve and then something else I want to also point out is if we look at the average temperatures here's the average temperature with the poor fan curve that it shifts with here's the new average temperature with the new fan curve it came down approximately what max was 85 that was probably just a spike so really was more like 84 down to about 77 average you can see it's 77 to 76 all the way across it's pretty consistent so yeah this applies to any graphics card by the way not just the tight necks change your fan curves do it you're gonna hurt anything oh yeah and did I mention it overclocks really good to it'll reach two thousand and fifty megahertz stable I can go farther than that I haven't spent a lot of time trying to go farther than that but then again the heat also exponentially rose we're even running this fan at 100% gave me 77 C well overclocked at a hundred percent fan speed yeah that's kind of hot but um anyway yeah check out the results of this was what it looked like with the factory fan curb and the factory clocks this is what it looks like with the overclock holy balls Falls and Batman if you've got an original tight necks like this one right here or perhaps three of them in a system you might be feeling like I don't know the best way to explain it III know how I feel but the new tight necks for the time being anyway until they come out with the Titan xx are my personal favorite the Titan X xx this thing is a freaking Godzilla yeah I find myself with this internal struggle because I'm installing two of these 10s in skunkworks which I know is more than enough I know it's more than enough but this is $1200 like I said I don't I don't need two of these I want two of these so I want 200 of these I don't need it do i I'm fine with 210 80s I don't need to tighten X's nope nope nope don't need to don't need to don't need to don't need to don't I just I don't I don't need to I don't need you you're you don't control me I make the decisions around here this is my channel hey hey NVIDIA hey Jay here so I just did that review and I decided that it's Jay Jay with Jays two cents know how many Jays do you guys actually knowhey Nvidia hey what's up J here yeah I was this on your website right now and I noticed that you guys did look Jay from Jay Z J's to sense a YouTube channel with 680 thousands of subscribers we met in Austin back in May during the launch of it for the 1080 deadmau5 was there I'm the tall brown-haired guy kind of handsome no that's that's Luke anyway so the reason why I'm calling is I saw that you guys launched a brand new Titan or relaunched an old Titan whichever I'm whatever you guys want to call it and I was kind of curious if you were interested in study 1 over for review oh I see ok well what if ok so how about this then I counter the no with my soul for one Titan X or at least what's left of it up in YouTube it's got to have some monetary value I see I've got kids those have any monetary value oh okay great great um ok so you should have my address and go ahead and send that over and we'll get right to it yeah alright yeah you YouTube yeah j JS JS 2 says nor ryan is with pc / that that's an entirely different site well you know it doesn't matter just go ahead and send it over and we'll get right on that all right thanks bye bye uu tuna that wasn't really all that hard wow they're fast with its unique freeform modular system the new master case maker 5 from coolermaster allows unparalleled flexibility with its adjustable internal layout and exterior customization options learn more about how you can start customizing your own case by following the link down in the description the GT x 1080 has been extremely popular so popular in fact they are not in stock anywhere they have been nothing but a logistics nightmare when it comes to keeping these things in stock so if you've ordered one or you've got one congratulations you were top dog for all of about two months because now we've got the new Titan here which is promising to be quite the behemoth when it comes to power yeah we're gonna we're gonna kind of move through this video kind of quickly here because I want to do some live benchmarking with this video and and you guys like like benchmarking haven't done it in a while and we're going to do that today but before we do that we have to talk about some of the specs because you know that if you're a specs nerd like myself you kind of you kind of get off on some of that stuff but anyway it's got three thousand five hundred and eighty CUDA core it's approximately a thousand more then we'll call little Pascal here on the GTX 1080 and it's got a fourteen hundred and seventeen megahertz boost clock which is funny because that's about as far as one of my Titan X's the Maxwell version of Titan X is somewhere in this pile here kind of maxed out about 1450 the three that are in skunkworks get up to about 1500 but this one's base clock is barely below that and it's got a fifteen hundred and thirty one megahertz boost clock but as you guys know we're gonna we're gonna push this thing as far as we can I know why some people are afraid to overclock this thing but not me not me I'm gonna I had to pay for it I just had to give a couple of my kids to Nvidia and it was totally worth it but for memory it's got 12 gigabytes of g5x running at 10 gigabit gigabit per second that is up that's some fast memory right there as 10,000 megahertz and as you guys know that's overclockable as well but a 384 bit memory bus and a 480 gigabyte per second memory bin that's pretty huge considering this is not HBM and I think that might be one of the first things that some of the people have been disappointed about is the fact that we were hoping that Titan on Pascal was going to have HBM but considering all the P 100 cores that you know they talked about earlier in the year are probably eating up all that HBM productivity and manufacturing it's not surprising that the Titan X does not feature HP m especially since they're saying that this is technically not a gaming card it was developed and made super fast for the purposes of deep learning and creating Skynet and having the end of the world as we know it ground zero B right there at Nvidia headquarters but again I digress the card itself is actually aimed more towards deep learning that is gaming even though it says GeForce GTX right on the side whatever of course it has the stuff that you would expect like pcie 3.0 HDMI 2.0 B and all that stuff so you can hook this up to your home theater of course you know this is a home theater card obviously because you know you could do HD 4k playback through HDMI which is kind of a big deal in terms of output here it's got the same stuff as the 1080 it's got an HDMI 2.0 B and it's got three DisplayPort 1.4 as well as a dual link DVI but obviously would you guys care most about it does have a backplate and the answer is yes it has a backplate but for 1,200 bucks it better have a backplate hello better have a front plate too for all that matters does it support sli yes up to two-way especially with an HB Bridge definitely recommended does a support simultaneous multi-project yes does it support Ansel yes does it support g-sync no it actually does not just just kidding yes of course it supports g-sync but it's got a TDP of 250 watts any recommended 600 watt or higher power supply to supply power to the 8 pin and 6 pin PCI Express power so that's that here's we're going to do a quick we're going to jump it to the benchmarks I'm going to show you those and we're going to hook it up on the test bench we're talk about a few things there are some things to talk about here believe it or not and we're going to get through this as best we can without it being as painful as possible transition you so if you bought the GTX 1080 simply because of the fact that it was the best of the best at the time that you got it or at least tried to get it rest in peace myself included because this guy has just been pushed out by this guy for being the fastest single GPU in the world but it's not that cut and dry guys because if you look at things like performance per dollar if you care about that even if you're shopping for the best of the best then you guys will know that there is a huge diminishing return on the Titan X versus the GTX 1080 when you compare the price of the fact that this thing right here is 1200 freaking US dollars or three thousand nine hundred and sixty two dollars in Australia or whatever it converts to be sorry guys you guys really do get wrecked with pricing down there it's terrible but yeah the GTX 1080 is still a huge huge performer when it comes to performance per dollar I say that considering it's a $700 graphics card but remember we are comparing it to 1,200 two of these is only 200 more than one of these and two of these will literally smash on one of these and in fact we're going to do a separate video about 1080 sli versus a single type next in fact I'd like to do that video today but I want to keep this video manageable we're going to keep it with one card at least compared to this one for now and we will test other SLI configs versus this guy right here but there are some things I want to talk about with this card and to do that we are going to have to go ahead and transition over to the test bench here so I can show you guys actually firsthand and not just talk about it on some of the things to expect with a GTX Titan X or just tight next because apparently they're going to maybe drop the GTX moniker in the future who knows so from left to right this is the power percentage we're leaving that at stock 100 this is the temperature this is the GPU load percentage this is the fan speed percentage and that is the core clock remember this has a 1400 and 17 megahertz base clock with a 15 31 boost clocks we want to keep an eye on that number and see what happens here so anyway gameplay is super smooth even at 4k but then again it should be as you can see we've got a single card with MSA a 2x at 4k achieving higher than 60fps that's actually kind of a big deal it's funny cuz I used to use 4k tests and I always found it very difficult to play in 4k because it was kind of laggy stuttery that's obviously not the case here as long as we stay above that 60fps then we're pretty good I've done this mission so many times this is the mission I actually use when I benchmark grand theft auto 5 because it's pretty much on a track everything is identical happens the same way every single time when it comes to the world and stuff so it's in game play stupid shove up to it it's in the game it's not a benchmark on a track I'm controlling it but it's very real-world scenario in my opinion as you can see here right now the temperatures have reached 83 to 84 that's where it's going to start to control things to try and keep the temperatures down the only way the GPU can actually keep the temperatures down is going to be by either increasing the fan speed which by default they've got it maxed out at 50% it can't go any faster than that or reduce the voltages and/or the clocks now the clocks and the voltages go hand-in-hand so those work together it's what you're going to notice here the more that I play you're going to see these boost clocks start to come down a little bit now anything over 1531 right now is considered above the boost clock so technically although we've come down from 1700 ish down to the high 15s low 16s it's not considered a thermal throttle in the sense of the way thermal throttling is intended to work yeah it is controlling the speed of the graphics by speeding up and slowing it down to control thermals but thermal throttling usually relates to the GPU trying to save itself from certain death or the CPU in that matter because it's the same concept there by slowing itself down so that it doesn't cook now we're not doing that here because we're still above the boost clock so that's just one of the things I want to mention here the same goes for AMD this is an Nvidia specific thing where people want to say oh it slowed down therefore it's thermal throttling no it's it's not thermal throttling its thermal adjusting and thermal throttling would be if we start to go down below the advertised speeds because of of thermals if that makes sense oh it's the cops the cops they know I stole this Titan X they're after me Simeon's not gonna be happy that I don't wreck this car up whoa that's that's bad oh geez so the temperatures pretty much went right up to 83 84 and as you guys saw in the benchmark chart because I did all founders Edition cards you know or reference cards for 980ti Titan X and the 1080 and the type the new Titan X they all pretty much just went straight to their max temperature threshold which is set right here as you can see it's set to 84 so what I'm doing right now is I have got Metro last light running on 4 K loop 5 times because I want to see what's going to happen to this core clock and I want to see if I can recreate what was happening during some of my benchmark testing was how far C we came all the way down to 15 31 now so all the way down to the boost clock that the factory boosts clocks we and now we're below it we went down to 14 81 for a moment this is just the beginning of the first test as you can see as our temperature is sitting here at 83 and our load is out a hundred percent our fan speed isn't allowed to go any faster than 50 or 51 percent I want to see how far that this clock will actually come down so we're gonna let the tests go five times we'll compare the average FPS then we'll make some tweaks here and then we'll see how well it actually does so the first test here at a sixty four point seven one average frame rate then the the clock runs zero that's so weird then the second test was sixty two point three then 61 point eight then 61.7 then 61 point one so you can see it was slowly starting to slow down but if we take a look at the the chart here you can see right here the core clock bounces around so much I mean so much that's that's GPU boost and the way it's meant to work but I find myself asking if this result could be better by simply playing around with the fan curve and you know it can be that's the whole point of this you know it can be so all I'm going to do here I'm going to leave the temperature target and the power limit the same but all I'm going to do is put in here a custom fan curve and I'm just going to allow the fan to go faster at the higher temperature so what we're sacrificing here is a little bit of acoustic sound so the card will be a little bit noisier but I don't know about you I'm the kind of guy that would be more than happy to add a little bit of noise to my system if it means I get more performance than worrying about it being stupidly silent and I think that NVIDIA is being waved too conservative with the 50% fan cap go faster on the fan speed and let's see what actually happens to these core clocks as you can see the way it started here we're already higher but anyway we're going to let it go and then we'll come back and we'll compare well if you can hear me over the fan it's obviously kind of noisy it's running 85% now I want to show you where on the fifth test right here and it's still starting in the 1700 megahertz range and it drops down into the mid to low 1600s and then comes back up but look at that compared to the other test which dropped down into the high 14s low 15s we're sitting almost 200 megahertz average faster speed alright so the test is done and the only difference here is as I showed you I change the fan curve nothing else so now we've got a sixty six point seven nine a sixty six point five one a sixty six point six three a sixty six point three seven a sixty six point six nine do you see the difference there we gained four FPS average increase by doing nothing more at 4k by the way for FPS at 4k gain is a lot and that was by doing nothing more than changing the damn fan curve I wish NVIDIA would just make it a little bit more aggressive from the factory yeah and then for those that are curious here is the core clocks right here on the previous test before the new fan curve this is where I did the little video there's the segment saying I'm gonna adjust a fan curve and there's the new core clock right there look how much that jumped up by changing the fan curve and then something else I want to also point out is if we look at the average temperatures here's the average temperature with the poor fan curve that it shifts with here's the new average temperature with the new fan curve it came down approximately what max was 85 that was probably just a spike so really was more like 84 down to about 77 average you can see it's 77 to 76 all the way across it's pretty consistent so yeah this applies to any graphics card by the way not just the tight necks change your fan curves do it you're gonna hurt anything oh yeah and did I mention it overclocks really good to it'll reach two thousand and fifty megahertz stable I can go farther than that I haven't spent a lot of time trying to go farther than that but then again the heat also exponentially rose we're even running this fan at 100% gave me 77 C well overclocked at a hundred percent fan speed yeah that's kind of hot but um anyway yeah check out the results of this was what it looked like with the factory fan curb and the factory clocks this is what it looks like with the overclock holy balls Falls and Batman if you've got an original tight necks like this one right here or perhaps three of them in a system you might be feeling like I don't know the best way to explain it III know how I feel but the new tight necks for the time being anyway until they come out with the Titan xx are my personal favorite the Titan X xx this thing is a freaking Godzilla yeah I find myself with this internal struggle because I'm installing two of these 10s in skunkworks which I know is more than enough I know it's more than enough but this is $1200 like I said I don't I don't need two of these I want two of these so I want 200 of these I don't need it do i I'm fine with 210 80s I don't need to tighten X's nope nope nope don't need to don't need to don't need to don't need to don't I just I don't I don't need to I don't need you you're you don't control me I make the decisions around here this is my channel hey hey NVIDIA hey Jay here so I just did that review and I decided that it's Jay Jay with Jays two cents know how many Jays do you guys actually know\n"