GeForce GTX Titan Thermals & Acoustics Comparison Linus Tech Tips

Linus Tech Tips ·Linus Tech Tips ·2013-05-07 · 1,266 words · ~6 min read
Floatplane YouTube

Transcript

JSON SRT VTT 88
0:05 one of the things that's toughest to quantify about the Titan is its
0:09 Acoustics and its temperature sort
0:13 of thing because the way that it works
0:17 you can actually set the temperature of your card you can go okay I want to
0:20 prioritize the temperature of my card I'm going to go prioritize this and you
0:25 set a temperature you can set your temperature at 70° it comes at 80° you
0:30 set it to 90° you can move that slider around and what that'll do is it'll
0:34 actually control itself and stay there
0:38 but then wait there's more you can actually stay there a number of
0:42 different ways you can either stay there by having your fan ramp up and down and
0:45 that's with controllable fan curves or you can stay there by having the GPU
0:51 clock itself down dynamically and reduce its own voltage dynamically in order to
0:55 stay there so the actual temperature fluctuation of a GTX Titan is is
1:00 extremely tight I mean if it doesn't
1:03 have to reach that temperature it won't if it's already got its fan speed on
1:06 minimum it's not going to just ramp up voltage needlessly in order to in order
1:10 to get there but what it will do is it'll stay at or under that temperature
1:14 and it'll do whatever you tell it to do in order to get there so what that means
1:18 is that comparing Titan against the other cards so GTX 680 didn't work this
1:23 way because it used GPU Boost 1 versus GPU boost 2 on Titan and that one was
1:28 controlled by the TD are by the total
1:32 power of the board so it was it was a power limitation as opposed to a thermal
1:36 limitation that caused the GPU to to ramp itself up or ramp itself down so
1:41 okay so that's that's totally different hd7970 Works totally differently again
1:46 where it's actually just got its own kind of profile and it does its own
1:50 thing and there's not a whole lot of uh there's not a whole lot of control it
1:54 just kind of the the the the GPU sort of is fast and then the fan speed ramps up
1:58 as the GPU heats up and it compensates um so so what we had to do was we had to
2:03 kind of decide where we wanted our Titan to be so we aimed for more of a high
2:09 performance Titan however what you could do is set the fan to stay very very much
2:13 in control and have more of a Silent Titan so we had our fan curve that kind
2:17 of went like this and then ramped up quite steeply at the end so it's kind of
2:21 a it's a middle ground but we gave it a good 50 55% fan speed at that 80° Mark
2:29 in order to keep itself cool and make sure that the turbo was going to be able
2:32 to ramp up because once you hit that temperature limit your GPU boost sorry I
2:37 called it turbo but your GPU boost is going to start ramping up so when I
2:40 compare temperatures against the reference 680 and the reference 7970 are
2:47 probably the ones to look at most carefully here uh when I compare
2:50 temperatures against these cards you have to bear in mind I could have just as easily set the Titan to run at a
2:56 completely different temperature so look at it this way all of our cards are
3:00 overclocked 660 Ti SLI 7970 680 Titan
3:05 and there was one other card what was our other card ah yes the Aries too
3:08 they're all overclocked so this is what they sound like and this is the
3:12 temperatures that they run at in order to reach the performance numbers that we
3:17 were able to reach in our other videos
3:20 so let's start with idle temperatures without further Ado uh the 7970 was a
3:25 little warmer than the others and everything else was pretty much the same
3:29 with the GTX 660 Ti is an SLI that's
3:32 probably honestly that's probably an anomaly in the reading because pretty
3:36 much all gpus that have good coolers on them idle pretty close these days
3:40 because of how advanced these video cards are and how they actually don't
3:43 deliver much power to the GPU when it's idling so it can run quite close to
3:47 ambient temperature load temperatures we see a different story Aries 2 steals the
3:52 show no duh it's liquid cooled so there's individual blocks on the two
3:57 gpus that are keeping that very very cool 660 Ti SLI pulls ahead of
4:03 everything else so this is a good opportunity to talk about reference
4:06 boards versus non reference boards uh for example r680 uses a reference cooler
4:12 which has the advantage of exhausting the air out of your chassis through the
4:16 vents in the back but has the disadvantage of not having an open
4:20 design like this MSI 660ti that allows
4:23 it to keep the GPU itself cooler so this
4:27 one kicks heat into your case but keeps your video card cooler overall this one
4:32 gets the heat out of your case but doesn't keep your video card itself as
4:35 cool so you have to make the decision do I care more about the other components
4:39 of my do I have a stuffy cramped case and I got to get that heat out of there
4:43 or do I have a big well ventilated case and I don't care and the other
4:46 components aren't going to get hot anyway so do I go with more of an aftermarket design uh moving right along
4:51 so load temperatures are tighten ran a little bit warmer than our GTX 680 but
4:55 like I said before we could have just as easily uh set it to 78° and would have
4:59 run at 78° but that's how we got the performance we did and it ran a little
5:03 bit less warm than our Radeon
5:08 7970 now temperatures are meaningless without
5:13 Acoustics so what I what we've done is we've set up our our test bench with a
5:19 microphone that is actually 12 in away from where the GPU fans were located
5:23 it's a shotgun mic so it's a directional mic and we have taken sound readings of
5:27 all the cards that we ran at idle and as
5:30 well as some reference noises so I'm going to play back those reference noises I'm going to play back those fan
5:34 speeds and I'm going to calibrate it while I'm editing so that the sound of
5:38 my voice in my ears is about like listening to me talk from you know a few
5:43 feet away and then hopefully all those other sounds will be similar to what
5:47 they would be like if you were listening to them so there you go there's our there's our Acoustics and temperatures
5:52 don't forget to subscribe to L's Tech tips from unboxings reviews and other
5:55 computer videos
6:59 oh
7:23 I think that whispering is sort of random and not always the same but this
7:28 is like kind of if you were 2 to 3 in from someone's ear
7:32 you'd whisper about like this this would be a normal conversational talking voice
7:36 just talking to someone who's about 3 to 4 ft away from you