Overclocking GeForce GTX 1080 Ti To 2.1 GHz Using Water

[LanzoCommando]

"Handles" depends on what frames you like to play with. Some of us like to stay well above 60 FPS....in the case of your 980TI I'd say above 30 or 40.

Whenever my 144hz monitor drops out to 60 hz it makes me cringe instantly. I can't really even stand 60hz anymore. I'll never go back to lower than 60 FPS...and honestly....even 60 is too low now that I've seen the light.

Irony: a recent article in New Scientist reported results from research which showed that people who use high frequency flat panel monitors become sensitised to such update rates, meaning they cannot go back to standard refresh monitors without enduring much annoyance.

Thus, the push to higher refresh monitors means that as time goes by relevant users will need ever higher refresh rates to perceive the same smoothness. It's like a drug for the rods and cones...

Ian.

LoL. I believe it! I'm telling you 60hz is TORTURE.

 

[carlosriosness]

i found a video with better data. Gamer Nexus uses an EVGA 1080 hybrid kit on the 1080ti.

https://www.youtube.com/watch?v=DdYEqTolk0Q

 

[TechyInAZ]

"Handles" depends on what frames you like to play with. Some of us like to stay well above 60 FPS....in the case of your 980TI I'd say above 30 or 40.

Whenever my 144hz monitor drops out to 60 hz it makes me cringe instantly. I can't really even stand 60hz anymore. I'll never go back to lower than 60 FPS...and honestly....even 60 is too low now that I've seen the light.

Irony: a recent article in New Scientist reported results from research which showed that people who use high frequency flat panel monitors become sensitised to such update rates, meaning they cannot go back to standard refresh monitors without enduring much annoyance.

Thus, the push to higher refresh monitors means that as time goes by relevant users will need ever higher refresh rates to perceive the same smoothness. It's like a drug for the rods and cones...

Ian.

LoL. I believe it! I'm telling you 60hz is TORTURE.

Actually it's not too bad. It depends on the game your playing.

For first person shooters, YES YES YES 60hz is torture. But for slower games like RTS, and tank games, I am able to go from 96hz to 60hz without issue.

 

[anbello262]

Irony: a recent article in New Scientist reported results from research which showed that people who use high frequency flat panel monitors become sensitised to such update rates, meaning they cannot go back to standard refresh monitors without enduring much annoyance.

Thus, the push to higher refresh monitors means that as time goes by relevant users will need ever higher refresh rates to perceive the same smoothness. It's like a drug for the rods and cones...

Ian.

First hand experience with that, and a good explanation of why I have always been OK with playing games below 20fps. You just get used to it and your mind doesn't even care after some time. Never noticed any smoothness issue (except for the times when everything lagged and dropped me below 3 fps).

(Then, I got a better PC which I used to play on locked 24fps to avoid overheating, it was a big laptop with heat issues. One day I decided to lock the fps at 60, and I really noticed the smoothness). But I'm still perfectly fine playing above 20-30fps, luckily.

 

[mapesdhs]

I don't play competitive FPS so I don't need mega fps, and atm the main game I play (Elite Dangerous) is fine at 60, though I like to ensure it doesn't drop below 60, ie. consistency is preferable to crazy maximums. Hence why I'll buy a 1080 Ti soon to move up to UHD, drop the heavy AA which should recover some performance at the same time.

Everyone's sensitivities varies. I found this when I did my dissertation on Doom. Some were content with 10 to 15fps, others preferred it way higher, and the latter of course were the people eager to buy the latest Pentium systems (original 1st-gen). I recall many a drooling Doom player delighted with their new P90 setup.

Ian.

 

[LanzoCommando]

You all are crazy talking 20, 30, 40 FPS. No F'n way. LoL

 

[ledhead11]

"Handles" depends on what frames you like to play with. Some of us like to stay well above 60 FPS....in the case of your 980TI I'd say above 30 or 40.

Whenever my 144hz monitor drops out to 60 hz it makes me cringe instantly. I can't really even stand 60hz anymore. I'll never go back to lower than 60 FPS...and honestly....even 60 is too low now that I've seen the light.

Irony: a recent article in New Scientist reported results from research which showed that people who use high frequency flat panel monitors become sensitised to such update rates, meaning they cannot go back to standard refresh monitors without enduring much annoyance.

Thus, the push to higher refresh monitors means that as time goes by relevant users will need ever higher refresh rates to perceive the same smoothness. It's like a drug for the rods and cones...

Ian.

I, unfortunately, have to agree. I'm old enough to have grown up with B&W T.V.s and seen most of the gamut at this point. I remember with my P4 build running S-Video @ 1024x768 60hz to a 27" Trinitron and using PowerDVD w/ hardware accelerations on and being amazed at both clarity and smoothness even if the extra fps were synthetic. Watching DVD's then at friends houses was usually depressing.

When I heard about 120hz I thought, why? Back in the day schools taught that the human eye could only perceive around 60fps. When I got my Toshiba Qosmio X775(120hz 3d Vision Display) I was blown away and soon after got a 120hz Asus monitor. I only recently got the 144hz model but I can say I'm unable to perceive much beyond 110-130.

60hz isn't bad when there's a steady 60fps and you've got the hardware capable of either v-sync/g-sync to prevent tearing. But when something drops to around 30-40 I start to cringe. My 1080's hold a fairly steady 55-65fps in Cinema 4k w/ v-sync and it looks pretty good to me.

 

[mapesdhs]

You all are crazy talking 20, 30, 40 FPS. No F'n way. LoL

I think you've missed the point here. This is a physical response, your brain is literally processing what your eyes are seeing differently because of what you're used to. Saying someone else is crazy because they're happy with a lower fps is not logical. As I said, the threshold is different for each individual. If you've been using higher frequency monitors then, as the research shows, that will resensitise your vision precisely to using such displays, ie. switching to higher frequency displays makes one's threshold higher, it's a vicious circle. I have no idea if the process can be reversed.

This was my research area for a while btw, human vision, VR, etc., and update rates was something I focused on, for which I had some help from the US Army Westpoint VR research people who were doing similar work for their flight simuator training systems. I was researching the game Doom and general stereo VR, but the issues were the same, both in terms of resolution, aliasing fidedlity and update rates. Then I was system and displays admin at one of the UK's leading VR centres for several years (2000 to 2004), where I was able to observe exactly how people responded to these varying factors with different types of display (HMD, CAVE, RealityCentre, etc.)

For visual simulation, it's why SGI had an oft used marketing expression, "60Hz, 30 hurts!". It's also why their older marketing material did not refer to performance as polys/sec, instead listing the no. of triangles per frame at 30Hz with a certain depth complexity (usually 4) as that was more useful for vissim designers to understand, and then later when the technology improved (InfiniteReality), they started quoting the no. of tris/frame at 60Hz. The CAVE I ran was a slight compromise, running 96Hz stereo (3 walls and a floor, each surface 10' across at 1024x768, with projection distortion to blend the images and hide the seams), one gfx pipe per surface.

If you're happy with, say, 100Hz, well, a typical bird would think that was awful, as their threshold (experiments show) is more like 200Hz, far greater acuity than humans. What's fascinating is that the backend processing of what our eyes are detecting can be changed so radically by repeated exposure. I wonder in time whether we'll see something similar with resolution, people get used to 4K and eventually don't like using "mere" HD...

Ian.

 

[vulcan78]

https://www.youtube.com/watch?v=74RF5U3-cLg

Video description:

Hi everyone, sorry the video cut off abruptly (phone ran out of memory) but this is pretty much most of what I wanted to show. I will include additional relevant information.

Using MSI Afterburner, with a custom fan curve that goes to 100% at 85C, the fan's RPM is at 78% at 68C, the hottest the card will get in this case, having removed the metal piece dividing PCI-E slot 1 and 2. If I set the fan to 70% the temps typically don't exceed 72C. The sound difference between 70 and 78% isn't worth the sacrifice in thermal performance IMHO so I just leave the fan curve like this. Yes, it's louder than 50% fan speed, but I for me this is FANtastic performance at this RPM, placing utmost emphasis on proper airflow to this card, in and out as various outlets, to include PC Gamer and HardOCP have reported max temps of 75C at 75% RPM with a comparable overclock.

To reiterate, I'm at +150 core and +500 memory with default voltage, 100% rock solid, Unigine Heaven (lol) steady.

Keeping the core under 70C keeps the core clocks from dipping under 2.0GHz. For me this overclock starts out at 2025 and then dips down to 2000 MHz at 65C and then 1987 Mhz if it goes over 70C, sometimes down to 1975 MHz.

This overclock yielded a significant gain vs. my 980 Ti @ 1500MHz core / 8GHz memory @ 1.218v:

http://www.3dmark.com/compare/fs/12009250/fs/11780853

50% in GPU test 1 and 60% in GPU test to and overall. So basically, nearly a 60% improvement.

I'm also seeing comparable gains in the "real world", at least a 50% gain in The Witcher 3, a solid 60% gain in Titanfall 2 in 2D (~90-100 FPS avg with Texture Quality on High to ~130-143 FPS avg having turned Texture Quality up to Insane, love that 11GB video buffer!)

https://www.hardocp.com/article/2017/03/09/nvidia_geforce_gtx_1080_ti_video_card_review/10

http://www.pcgamer.com/geforce-gtx-1080-ti-overclocking/

i7 4930k @ 4.5 GHz w/ 1.361v
GTX 1080 Ti FE @ 2.025 GHz Core / 6 GHz Memory default voltage
32GB Corsair Vengeance Quad Channel DDR3 System Memory @ 2400 MHz w/ 1.55v
Asus Rampage IV Black Edition
Corsair Air 540 with custom side panel fan
Samsung 850 Evo Rapid Mode Enabled
Asus ROG Swift PG278Q 2560x1440 144Hz G--Sync and 3D Vision capable display
Windows 7 Pro

The Witcher 3 Settings good for 120 FPS at 2560x1440 everywhere, including Novigrad (60 FPS 3D Vision)

Everything Maxed except:

AA: FXAA
Shadow Quality: Medium
Foliage Distance: High
Hairworks: Off

3D Vision Fix:

http://helixmod.blogspot.com/2016/06/the-witcher-3-wild-hunt.html

Oh and this 1080 Ti FEruns at 24-26C at idle at 30% fan speed (inaudible).

At 60% fan speed the card stabilizes at 78C overclocked.

At 100% fan speed the card stabilizes at 58C overclocked.

 

[vulcan78]

@vulcan78
I compared the TitanXP with the 1080 Ti in the lab, screw by screw, the 1080 is totally uninteresting in this comparison. And BTW; the linked review is more or less totally worthless, because there I missed the most the important data and any technical details about the coolers. I wrote you above, why this cards are similar (both cards are coming from PNY's same insertion line in China mainland) and where you can find the (small) differences.

If you have a problem to understand technical basics and why refurbished iron can't be so significant better, it is your problem. My one and only problem is to believe PR bubbles if I know the details and the background simply better. I have all three cards here - the Quadro P6000 with the uncrippled chip too. Same cooler under the hood (but other cover design) - similar cooling performance. And it is only one call to a factory to get closer info which parts are used.

Your video is entertaining, the typical YT posing with a lot of blah blah instead of any professional measured data. You are invited to visit my lab and my semi-anechoic chamber (room-in-a-room, calibrated microphones and professional audio-software) to understand the basics of an exact, scientific work and to compare this small Asetek midget with a real water cooling solution and high-end fans.

The small Asetek pump is hearable (high pitch), the hybrid fan too (especially in idle!) and the 120mm fan is not a pain but the typical 5 USD standard crap from China. If someone is believing this is quiet he must be simply deaf. I've tested a Kraken X40/G10 in my lab long time ago and it is - as I said - a toy. Closed loop, not more. On a R9 290 I've measured three years ago 37,2 dB(A) in a distance of 50 cm, this IS noisy, if you can have the same thing with a level below 31 dB(A). THIS is nearly unhearable in the most rooms. A water cooler is only good, if you can think that the computer is switched off.

Physics isn't a bitch. If you are satisfied with this, why not? But never write about things that you have never owned, tested or measured. If you never tried a Mercedes, you will really think, a VW Rabbit would be the best car you can buy.

the classic Nvidia aesthetic lighting flavor

To be honest, but you are the typical Nvidia marketing victim. But it is the best proof, that their PR is working well

@anbello262:
I can't read any facts and arguments from his side, only insults and malicious allegations. And I'm sure he was searching only a chance to push his funny YT video and other websites. Or it is in real a paid Nvidia PR. It is a waste of time to bring facts if a guy not even can't understand the difference between duty cycle in percent and real speed in rpm. This are simply basics. Elementary school level.

I moved from the H55 to an X41, which is much thicker and whose fan I could run at 50% RPM, which was inaudible, with load temps <45C.

But that was before, have a look at my post above, where my 1080 Ti wont get above 67C under full load sustained with +150 core / + 500 memory, default voltage. If I set the fan to 60%, which is much more bearable, the temps don't exceed 78C, which is admittedly warm and far from inaudible, if I set the fan to 100% the temps don't exceed 58C with the overclock.

I just wanted to post back because part of your argument was that 1080 Ti FE runs too hot (84C!) and needs $400 in liquid cooling. Well, aside from The Witcher 3 in 3D Vision (for whatever reason) 2D titles the clocks don't drop under 2000MHz unless it exceeds 70C, which I've seen it do with a less aggressive fan algorithm. At 70% RPM the temps stabilize at 72C. At 68C the fan RPM hit's 78% and it really doesn't go higher than that.

67C and clocks over 2000Mhz stable with a little bit of noise and I can save that $400 and put it forward my next upgrade, which I don't believe will happen for a good long while as I'm satisfied with 2560x1440.

Yeah, I can hear the fan at night with a game with ambiance such as Fallout 4 when the background music stops playing, but in the day, say playing Titanfall 2, I can't hear it over the headphones to be perfectly honest. Positioning the PC under the table helps, maybe everyone complaining about the noise has their PC mounted on their table 2.5 feet from their head.

$700 1080 Ti FE ~= $1200 Titan XP with a water-block and $400 liquid cooling set-up performance wise, acoustics aside.

Put up your benches without your shunt mod:

http://www.3dmark.com/fs/12009294

http://www.3dmark.com/fs/12011832

100% Rock solid stable, zero display driver crashes, zero artifacts.

Oh and so much for "The Paragon demo with the 1080 Ti running at 67C overclocked was pure PR"

LMFAO.

If you want I will post a video with it not exceeding 58C at 2025 core and 6GHz memory with the fan at 100% RPM. So yeah, 67C isn't even the full performance potential of this card on air.

Try that with Titan XP, as you said, and refuse to retract, "there is no difference between Titan XP and 1080 Ti FE cooling-wise, they both need liquid cooling".

"No difference" comparison across 18 games, all three cards using reference cooler:

http://www.overclock.net/t/1624521/nvidia-gtx-1080-ti-owners-thread/800_100

Tad long post, but thought I would share some benching I did most of today between 1080's and 1080Ti's both FE to see how they compare at stock and overclocked with a few Titan X Maxwell Scores sprinkled in.

System Specs used: 5960x @ 4.6 Ghz, Cache @ 4.4 Ghz, 32GB RAM @ 14-17-17-34 1T

Panel used was the ASUS PG348Q, so 3440 x 1440 with V-Sync and G-Sync off in everything.

All cards used a 1:1 fan profile

GTX 1080Ti
Stock settings average boost 1860 Mhz Core / 11,016 Memory
Overclocked settings average boost 2002 Mhz / 12,000 MHz Memory


GTX 1080
Stock settings average boost 1848 Mhz Core / 10,012 Memory
Overclocked settings average boost 2075 Mhz / 11,000 MHz Memory


Titan X (Maxwell, EVGA SC BIOS)
Stock settings average boost 1316 Mhz Core / 7,012 Memory
Overclocked settings average boost 1474 Mhz / 8,020 MHz Memory


% increase and decrease is based on my stock 1080 scores. FPS are Averages

OVERALL AVERAGES ACROSS THE BENCHMARKS
GTX 1080: BASELINE
GTX 1080 OC'd: 9.98 % increase
GTX 1080Ti: 29.5% increase
GTX 1080 OC'd: 39.3% increase
Titan X: 12.2% Decrease
Titan X OC'd: 3.1% Decrease

Overall stock 1080Ti is 30% ahead or so of stock 1080 and similar for overlocked 1080Ti vs 1080. Gap is smaller, when we strip out synthetics which will scale better then games mostly. Now to see how SLI 1080 vs SLI 180Ti compares smile.gif


Far Cry Primal - Ultra Preset + HD Pack
GTX 1080: 60 FPS
GTX 1080 OC'd: 67 FPS (11.6% Increase)
GTX 1080Ti: 80 FPS (33% increase)
GTX 1080 OC'd: 87 FPS (45% increase)
Titan X: Average: 53 (11.7% Decrease)
Titan X OC'd: Average: 59 (2% Decrease)


Ghost Recon Wildland - Very High Preset
GTX 1080: 57.5 FPS
GTX 1080 OC'd: 61.21 FPS (6.5% increase)
GTX 1080Ti: 71.25 FPS (23.9% increase)
GTX 1080 OC'd: 76.59 FPS (33.2% increase)


Ash’s of the Singularity GPU Test - DX11 - Extreme Preset
GTX 1080: 57.5 FPS
GTX 1080 OC'd: 61.21 FPS (6.5% increase)
GTX 1080Ti: 71.25 FPS (23.9% increase)
GTX 1080 OC'd: 76.59 FPS (33.2% increase)


GTA 5 - All settings Maxed but no MSAA on anything
GTX 1080: 71 FPS
GTX 1080 OC'd: 83.1 FPS (15.7% increase)
GTX 1080Ti: 88 FPS (22.6% increase)
GTX 1080 OC'd: 94.8 FPS (32% increase
Titan X: 64.34 FPS (10.4% Decrease)
Titan X OC'd: 72.5 FPS (1% increase)


Arkham Knight - All settings maxed, No gamework settings on
GTX 1080:84 FPS
GTX 1080 OC'd: 91 FPS (8.3% increase)
GTX 1080Ti: 103 FPS (22.6% increase)
GTX 1080 OC'd: 116 FPS (38.1% increase)


Tomb Raider - Ultimate Preset
GTX 1080: 98.2 FPS
GTX 1080 OC'd: 108 FPS (10% increase)
GTX 1080Ti: 131.9 FPS (34% increase)
GTX 1080 OC'd: 141 FPS (44.5% increase)
Titan X: 88 FPS (10.4% Decrease)
Titan X OC'd: 96 FPS (2.3% decrease)


Dues Ex mankind Divided - Ultra Preset no MSAA
GTX 1080: 41.3 FPS
GTX 1080 OC'd: 45.6 FPS (10.4% increase)
GTX 1080Ti: 55.3 FPS (33.9% increase)
GTX 1080 OC'd: 62.1 GPS (50.4% increase


Dragon Age inquisition - Ultra Preset
GTX 1080: 72.4 FPS
GTX 1080 OC'd: 79.6 FPS (9.9% increase)
GTX 1080Ti: 89.1 FPS (23.1% increase)
GTX 1080 OC'd: 102 FPS (40.1% increase)
Titan X: 60.5 FPS (16.4% Decrease)
Titan X OC'd: 68 FPS (6.1% Decrease)


Witcher 3 - High Post processing Preset and Ultra Graphical quality Preset - Novigrad loop
GTX 1080: 54 FPS
GTX 1080 OC'd: 60 FPS (11% increase)
GTX 1080Ti: 67 FPS (24.1% increase)
GTX 1080 OC'd: 72 FPS (33.3% increase)
Titan X: 49 FPS (9.3% Decrease)
Titan X OC'd: 52 FPS (3.1% Decrease)


Witcher 3 - High Post processing Preset and Ultra Graphical quality Preset - Skelliga crossroads trees / forests loop
GTX 1080: 49 FPS
GTX 1080 OC'd: 51 FPS (4.1% increase)
GTX 1080Ti: 59 FPS (20.4% increase)
GTX 1080 OC'd: 63 FPS (28.6% increase)
Titan X: 43 FPS (12% Decrease)
Titan X OC'd: 47 FPS (4.1% Decrease)


Witcher 3 - High Post processing Preset and Ultra Graphical quality Preset - Beauclair castle balcony loop (Long Vista)​​
GTX 1080: 67 FPS
GTX 1080 OC'd: 72 FPS (7.5% increase)
GTX 1080Ti: 82 FPS (22.4% increase)
GTX 1080 OC'd: 87 FPS (29.9% increase)
Titan X: 60 FPS (10.4% decrease)
Titan X OC'd: 65 FPS (3% Decrease)


Hitman Absolution - Ultra - 2 x MSAA
GTX 1080: 81.1 FPS
GTX 1080 OC'd: 92.6 FPS (14.1% increase)
GTX 1080Ti: 103.2 FPS (27% increase
GTX 1080 OC'd: 113 FPS (39% increase)
Titan X: 68.4 FPS (15.7% Decrease)
Titan X OC'd: 78.1 FPS (3.7% Decrease)


Shadow of Mordor - Ultra Preset
GTX 1080: 94.95 FPS
GTX 1080 OC'd: 107.19 FPS (12.9% increase)
GTX 1080Ti: 116.06 FPS (22.2% increase)
GTX 1080 OC'd: 127.06 (33.8% increase)
Titan X: 81.2 FPS (14.5% Decrease)
Titan X OC'd: 92.6 FPS (2.5% Decrease)


Total War Warhammer - Ultra Preset
GTX 1080: 68 FPS
GTX 1080 OC'd: 78.2 FPS (15% increase)
GTX 1080Ti: 96.71 FPS (42.2%)
GTX 1080 OC'd: 103 FPS (51.5% increase)


Metro Last Light Redux - All settings Maxed No Motion Blur
GTX 1080: 38.1 FPS
GTX 1080 OC'd: 41.14 FPS (8% increase)
GTX 1080Ti: 49.91 FPS (31% increase)
GTX 1080 OC'd: 56 FPS (47% increase)


Vally Benchmark - Extreme HD Preset resolution at 1440p
GTX 1080: 66.1 FPS
GTX 1080 OC'd: 70.5 FPS (6.7% increase)
GTX 1080Ti: 89.9 FPS (36% increase)
GTX 1080 OC'd: 97.3 FPS (47% increase)


Firestrike
GTX 1080: Graphics Score: 20800
GTX 1080 OC'd: Graphics Score: 23700 (13.9% increase)
GTX 1080Ti: Graphics Score: 27417 (31.8% increase)
GTX 1080 OC'd: Graphics score 29050 (39.7% increase)


Firestrike Extreme
GTX 1080: Graphics Score: 9721
GTX 1080 OC'd: Graphics Score: 11451 (17.8% increase)
GTX 1080Ti: Graphics Score: 13465 (38.4% increase)
GTX 1080 OC'd: Graphics Score: 14525 (49.4% increase)


Firestrike Ultra
GTX 1080: Graphics Score: 4917
GTX 1080 OC'd: Graphics Score: 5624 (14.4% increase)
GTX 1080Ti: Graphics Score: 6808 (38.5% increase)
GTX 1080 OC'd: Graphics Score: 7109 (44.6% increase)

 

[FormatC]

Can I remeber you, why we are discussing? You wrote, the 1080 Ti cooler is 8°C cooler than the cooler of the Titan X Pascal. This is wrong and you know this. Simply remove the 1080 Ti cooler and put it on the Titan X Pascal (or vise versa). You will see: no difference. I tried it here in a PC shop with a few ordered custom builds, simply to proof the marketing bubbles with real iron. But what I saw, while I compared a few 1080 Ti, is a very big difference between all this cards due their chip quality differences.

The second thing: advertising or paid by industry? I spent over the years a lot of my own money (not only) into the cooling development and to be honest: my incoming from my own lab (pc components, pc-audio), documentations and publications, product development (cases, air-cooling parts ) and reviews is high enough to pay all my stuff by myself AND to pay others.

I'm since my youth an audio freak, spent a lot of money in speakers and headphones. Ok, I'm also amused if I meet a "golden ear" with cables and plugs for thousands of Euro. This is really abnormal, but I can accept this. 400 Euro or more for a good cooling solution? Boring, if I can buy for this money a big piece of living quality and - fun. It isn't a challenge to change fan profiles or to put an simple pre-filled closed loop AiO into a system and the result will be never the same. Benchmarks are only secondary - I need it simply quiet. The performance increase is a nice side-kick.

Call me an idiot or better - call me an enthusiast. Because it is definitely not the same. I'm just building a cooling tower with a tube system in another rooom (my storage) to move the last mechanical things out of my office. The PSUs are also totally fanless and I'm sure, I will hear after this project the whispering of my monitors. Good for a next project to remove/decrease the "electrical" noise of components (just playing f.e. with special cables and block caps for VGA) .

And it seems, that you have some difficulties to understand the goal of this short review and my followed posts. The deal was, to get two things at the same time: a cool card on a really low noise level together with the target, to scratch on the 2,1 GHz borderline. I know a lot of enthusiasts, cooling the complete system with a custom loop. May be, it is in Europe more common to build real water coolers and to use this parts over a lot of years - but I never met such a air-cooling fanboy, unable to accept, that other people simply want to live (and work) in a noise-free environment. If you satisfied with this hair dryer - nice for you. Save the money and pay your wife/girfriend a good and expensive dinner. But don't speak for others, this will end in a total trumpification with alternate facts

 

[n0ns3ns3]

the decent liquid cooling for GPU is 100-160$ depending on the block and backplate (if any at all) you choose. The rad and pump cost is shared with CPU cooling and next GPUs.
If you have 700$ for a GPU, that should not be a problem.

Fans @ 100% is fine for server room, not for a home usage.

As FormatC mentioned, we like our systems quiet.

 

[FormatC]

Finally one guy who understood my principles. Fight hard but be quiet

This is the piece, which will be delivered next week to cool down all running systems below 25°C (ok, for more than 400 Euro):
http://www.eteknix.com/alphacool-debuts-eiszeit-1500w-compressor-cooler/

But I have to place it in another room. Perfect for all my rigs, workstations included. Up to 1500 watts and down to 20-25 degrees (more or less independend at the room temp) at this maximum heat emission is a real deal - also in summer. Try the same with a SLI of two or three Flanders Editions in summer at 30°C. This cards will glow in the night

 

[n0ns3ns3]

not all of the principles
I'm more into a "put as many as you can, into the smallest possible volume at low noise" thing.
Hence the mini ITX for me. If I had a need for it, I could put an X99 board (unfortunately only one model exists for now) with any matching CPU and any GPU. it would be still fairly quite (~1200RPM under full load) even overclocked. Main concern for CPU overclocking would be keeping the VRM cool enough.

 

[FormatC]

Good VRM cooling on mainboards is an important thing, especially for OC or builds with minmal space. This keeps all more stable, not only the temps within the specs. The chiller is a beast, I need it in first row for my workstation + render cards. But it allows me, to measure (and write reviews) with a stable water temperature. This is the requirement to get comparable deltas for things like thermal compound tests, cooler tests and so on. A scientific and exact work starts always with the best test conditions you can get. And I'm an equipment freak. And no: it's not paid advertsing - I'm sure that nobody will sponsor a 1000 Euro cooler

 

[ledhead11]

Finally one guy who understood my principles. Fight hard but be quiet

This is the piece, which will be delivered next week to cool down all running systems below 25°C (ok, for more than 400 Euro):
http://www.eteknix.com/alphacool-debuts-eiszeit-1500w-compressor-cooler/

But I have to place it in another room. Perfect for all my rigs, workstations included. Up to 1500 watts and down to 20-25 degrees (more or less independend at the room temp) at this maximum heat emission is a real deal - also in summer. Try the same with a SLI of two or three Flanders Editions in summer at 30°C. This cards will glow in the night

I'm just in the early stages of researching liquid cooling(started last summer). I've been holding off for a while now since air got me what I needed with SLI for the last 10 years. I agree that the industry is definitely moving away from it and my own primitive tests with Pascal have shown a lot of promise if can keep 'em cool and with the right kind of power. At the rate that displays are evolving past the cards that drive them its obvious other paths have to be taken.

That Alphacool is awesome! I didn't know things like that existed for custom loops. I can't wait to see your next set of benches. I read on another site about KingPin using LN2 to get to 2.5 on a Ti but I bet that the Alpha could get you really close too and much less headache involved.

 

[FormatC]

The goal for this short review was not an OC record or something similar, it was the missing part of our launch review. I had to handle Ryzen and the 1080 Ti nearly in the same time and it was simply a question of time (and sense) to make this part with a better cooler than the original. This was not written to destroy Nvidias legend of the best cooler ever, it was a very fast and easy option to show, what it's possible with a minium of work.

This Chiller compressor cooler can cool down the water in my systems systems below 25 degress and I can use it for a big loop of more than only one system. On the other hand I can control the water temperature very exactly and this alone is it worth (for exact measuring) to do it now.

 

[ledhead11]

Totally understood on the review vs. oc record. I was just happy to see someone apply this solution so quickly and with such detail. I think the Ti's are great and my hats off to NV for 'em.

It wasn't until Pascal that I could witness so many performance differences just in cooling and power. A difference of 10-15c, i.e. keeping it below 60c and ideally ~45 has very noticeable gains in OC'ing. I've seen it with my Xtreme G1's also. It seemed like with Maxwell and Fermi once you got so far with air that liquid was good for silence(I understand one of your main goals) but I usually only read 200-300mhz gain over factory OC settings from most peoples posts. Exceptions of course being the ones who went for records.

Pascal seems to have a much bigger trade off, 400-800mhz, under the right conditions that most of us should be able to do with care.

 

[CyberBlade]

What about cooling it with liquid nitrogen. After highly overclocking it.

 

[mapesdhs]

... that other people simply want to live (and work) in a noise-free environment. ...

Many years ago I visited a movie studio in London, back then all SGI (late '90s). Enabling the artists to do what they do best without distraction was the key goal for the sysadmins, etc. The main editing and animation/design rooms were all very nice, relaxing places to work; there were monitors, keyboards, mice, Wacom tablets... and that's it. All of the equipment, basically anything that made noise, was housed in a separate machine room, far away from the creative staff. The machine room had powerful cooling/vent systems and the builing had its own power delivery setup. It was an eye-opener to see such a space designed to maximise the creative flow. Zones with tech were like maintenance gantries on the Nostromo, but areas where creative professionals did their work were luxury zones.

I know a guy in the US who setup a custom loop with all the kit in his attic, so no noise in his main room. Another did it with the equipment in the basement (he lived in a colder region, the rising heat useful for the house as a whole).

There are definitely people who care about noise, and their nature varies considerably.

But don't speak for others, this will end in a total trumpification with alternate facts

100 points for that one. Still waiting for someone to be triggered by my MAGA hat.

Ian.

 

[ledhead11]

... that other people simply want to live (and work) in a noise-free environment. ...

Many years ago I visited a movie studio in London, back then all SGI (late '90s). Enabling the artists to do what they do best without distraction was the key goal for the sysadmins, etc. The main editing and animation/design rooms were all very nice, relaxing places to work; there were monitors, keyboards, mice, Wacom tablets... and that's it. All of the equipment, basically anything that made noise, was housed in a separate machine room, far away from the creative staff. The machine room had powerful cooling/vent systems and the builing had its own power delivery setup. It was an eye-opener to see such a space designed to maximise the creative flow. Zones with tech were like maintenance gantries on the Nostromo, but areas where creative professionals did their work were luxury zones.

I know a guy in the US who setup a custom loop with all the kit in his attic, so no noise in his main room. Another did it with the equipment in the basement (he lived in a colder region, the rising heat useful for the house as a whole).

There are definitely people who care about noise, and their nature varies considerably.

Much more primitive but. . .when I did my experiments with mine this winter I had positioned my case sideways in the window frame so that the case GPU fan could pull air from outside directly. Outside temps at the time were around 20-30F.

I was able to keep the stock fan settings while maintaining under 66c thus allowing 2012MHz consitently on the cores.

 

[filipcristianstroe]

We cannot truly see this card's potential even on water without having a proper custom PCB. The nvidia one is utter crap. Hopefully Zotac & Msi & Gigabyte and Evga will give us 8pin 8pin with a much better power delivery system which will allow by overvolting and proper watercooling to reach 2.2-2.3 ghz, the chip has the potential as well as this new memory, the power system..<CRAP>

 

[filipcristianstroe]

We will only see this card's true potential on water as soon as CUSTOM PCB's come out with 8pin8pin and some modded bios features which will allow serious overclocking. The chip is capable of probably 2.2 2.3ghz maxing out, but the power delivery system and overvolting...really disappointing. Wait for EVGA/ZOTAC/MSI/GIGABYTE to come out with custom PCB's to really see the card's potential under water.

 

[Sam Hain]

4k 60fps will always be multiGPU.

Not with 60Hz panels... Now, with upcoming 120+Hz HDR 4K panels on the horizon, that may be a test. However, GPU-tech is not sitting around waiting for another round of 30-40 FPS performance at 4K with these panels when they release.