Decrease your GPU temps with 10 C - is cheap and easy to do

[rusty caterpillar]

I sent expanded metal to China...or more accurately I had to deal with a customer who did this. Never had an oxidation issue in decades...but had one now. Yep...zinc coated steel arrived in China after a month on the seas as a pile of red rust. It's because they took a wooden pallet of our stuff, secured by a pair of nylon straps, and shoved it into a shipping container. Stuff that I watched customers sit under a tin roof outside, that lasted 2+ years without visible rusting, couldn't make it from the east coast of the US to China. So...no. Your argument about not needing passivation doesn't hold water....pun fervently intended.

That example doesn't apply to PC parts which are packed differently, also most oft them have sealed bags that doesn't lets salty humid air to get to it.

Ni plating requires energy and resources which needs to be spent. Resulted plated products can have deficiencies and faults which are more detrimental than slight oxidations of Cu alloys.

Using right Cu alloys in producing cold plates can have positive results in longevity before first microns of oxidations start to cumulate and be a problem.
You forget a very important factor in most scenarios the air inside a PC or laptop is very dry therefore corrosion and oxidation factors is smaller.
Are various materials and applications where NI or Zn plating is really necessary and I still don't see or believe is necessary to the cold plates in PC parts.

So...let me also answer your ignorance

My ignorance comes from 17 years experience in servicing computers in dealing with various Cu or Al coolers. Very little to none oxidation or corrosion issues. But even if it was oxidation, are various solvents that will clean most of it. On harder deposits I used sand paper, but those high deposits was coolers left outside in poorly closed boxes. Had to recover them cause the boss wanted just that

have landed on complex issue

Landed well, parachute opens successfully, and dealt with very small and rare issues of oxidation and corrosion exceptions.

My old Mugen air cooler with Cu cold plate was beaten by Prolimatech Vers B by type of cold plate and number of fins with 2C and not because was Ni plated.

Take a look at pictures bellow to see your "problematic corrosion and oxidation" on an old Pentium 4 cooler made by Intel with Cu core and a Thermalright GPU cooler full Cu.
I have P4 cooler from 2005 from a friend and the VGA cooler ~2007-2008. Both coolers sitting idle for 4 years in a humid cellar and approximately 15-17 years in a drier attic in a carboard box but no plastic bags. The cardboard box was changed at some point as was moldy.

 

[Waldorf]

@all
not sure why so many folks spend time "playing" around with fans/positioning to lower temps by 5* maybe 10*C, instead of going LC (at least for cpu),
and drop case/psu/mb/drive temps by around 30*C (every pc i build with dgpu), thus allowing for much lower temps.

and anyone with xx80/90 chip (or equivalent prive/heat), should have the funds for either a case with side mount for cpu rad (leaving top of case for case flow),
case mounting the gpu in a different chamber, or LC gpu.

i means i great if you inherited/buy something used, but on new parts/upgrades, i would think about (proper) cooling from the start...

@rusty caterpillar
maybe add "aircooled" to the title?
not saying its misleading, but i doubt it will change temps on my wb card by 1*C

 

[rusty caterpillar]

@all
not sure why so many folks spend time "playing" around with fans/positioning to lower temps by 5* maybe 10*C, instead of going LC (at least for cpu),
and drop case/psu/mb/drive temps by around 30*C (every pc i build with dgpu), thus allowing for much lower temps.

and anyone with xx80/90 chip (or equivalent prive/heat), should have the funds for either a case with side mount for cpu rad (leaving top of case for case flow),
case mounting the gpu in a different chamber, or LC gpu.

i means i great if you inherited/buy something used, but on new parts/upgrades, i would think about (proper) cooling from the start...

@rusty caterpillar
maybe add "aircooled" to the title?
not saying its misleading, but i doubt it will change temps on my wb card by 1*C

Not positioned fans, I added one fan and cut the grill of my case for the exhaust fan, those type of slats grills shouldn't be present on expensive cases.

Bellow, the grill with the lowest impact on the noise and airflow.

Tests carried by Puget Systems in 2011, yet Lian Li ignored all the data. They added EVO in the title of the new modified case, change the grills to honeycomb and they wanted 70 EU more - clowns.
I have tested and I have 5-7 C less on my GPU just cause I remove the grill.

I do have Arctic 280 mm AIO for CPU, just to try out water cooling. Not impressed but, not disappointed either.
Water block for my 1080 Ti out of the question, too expensive, maybe I'll change my GPU sometimes next year anyway, but even if I not, I'll need a Rad for LC GPU as Kraken bracket + I have to buy frontal fan/rad support for Lian Li OD 11. I believe I can modify the Kraken to bare Cu plate and to support my next GPU and add extra fins as small rads.

 

[Waldorf]

@rusty caterpillar
first part wasnt about you fixing something, but for those "wasting" time on minor changes, instead of doing something with bigger impact.
yeah, didnt mean to go water on existing card.
i dont have any "grills" on exhaust ports, anyone dumb enough to stick their fingers in, their fault.

you should seal the area around the fan, for better flow/not rebreathing exhausted air, but the biggest gains would be side mounting the rad (exhaust, not intake)
and have the top fans exhaust air from the gpu, not sure why you put the rad on top.
that way, you wont really need the gpu under water..

 

[BMfan80]

Tried this just for sh$t's and giggles, thought is was pointless.(for me at least)
Didn't do anything to a 4090 pulling 430w

 

[Waldorf]

@BMfan80
part of the problem is ppl do stuff, while having a single fan in the rear, while 3 or more fans provide intake, way too much vs being removed,
so clearly will have air going around inside the case.
main reason why i have a case with two 140 fans for (case) exhaust.

 

[chrcoluk]

When I setup my curve profiles, and maybe an undervolt for the 4080 super I will possibly get a 10C+ drop. Although not sure as the cooler is so overbuilt on this thing even at stock it wont go much more than about 20C over idle.

 

[rusty caterpillar]

Tried this just for sh$t's and giggles, thought is was pointless.(for me at least)
Didn't do anything to a 4090 pulling 430w

Good to know. Thanks

When I setup my curve profiles, and maybe an underclock for the 4080 super I will possibly get a 10C+ drop. Although not sure as the cooler is so overbuilt on this thing even at stock it wont go much more than about 20C over idle.

Manufacturer and card full name pls?

@BMfan80
part of the problem is ppl do stuff, while having a single fan in the rear, while 3 or more fans provide intake, way too much vs being removed,
so clearly will have air going around inside the case.
main reason why i have a case with two 140 fans for (case) exhaust.

Positive pressure is not that bad, specially if you have some in top or at least a PSU fan Yeah you are right 14 cm fans are the best specially when you have 2 in the rear but, very rare cases with such exhausts.
Mine have poor support for 14cm fans generally and only 12cm in the back.

@rusty caterpillar

you should seal the area around the fan, for better flow/not rebreathing exhausted air, but the biggest gains would be side mounting the rad (exhaust, not intake)
and have the top fans exhaust air from the gpu, not sure why you put the rad on top.
that way, you wont really need the gpu under water..

Thanks. I did it some time ago, not completely though. But now that I have a backplate fan it might suck some air just trough the gap right underneath my exhaust fan.
However sucking back exhausted air is usually when you have negative pressure in the case.
So I see you playing along with "minor changes" agenda.

 

[Waldorf]

@rusty caterpillar
was meant more about not having air circulate inside the case (vs straight in and out).
i just make sure its balanced when it comes to flow, does help with both chips being LC, so i dont have much heat in the case anyway.

still say you're better off mounting the rad to the side, and have the top for case flow,
i bet it'll drop 20-30C on the case, probably 10C for the gpu temps.

 

[rusty caterpillar]

@rusty caterpillar
was meant more about not having air circulate inside the case (vs straight in and out).
i just make sure its balanced when it comes to flow, does help with both chips being LC, so i dont have much heat in the case anyway.

still say you're better off mounting the rad to the side, and have the top for case flow,
i bet it'll drop 20-30C on the case, probably 10C for the gpu temps.

Thanks
So, pushing air trough side radiator from outside? I have glass as front panel which means the 2x14CM pushing air inside the case trough the rad will be the only intake. Looks a bit like negative pressure in the case. If I leave the 12 cm on the bottom to push air along with radiators fans maybe I'll brake it even. I might try your idea.
Don't know what you mean by drop 20-30 C on the case though. I mean M2 running under or around 40 C, MBO 34-39 average. xHCI 1 and 2 can get to 50 C - GPU is throwing heat on it.

 

[ShrimpBrime]

Thanks
So, pushing air trough side radiator from outside? I have glass as front panel which means the 2x14CM pushing air inside the case trough the rad will be the only intake. Looks a bit like negative pressure in the case. If I leave the 12 cm on the bottom to push air along with radiators fans maybe I'll brake it even. I might try your idea.
Don't know what you mean by drop 20-30 C on the case though. I mean M2 running under or around 40 C, MBO 34-39 average. xHCI 1 and 2 can get to 50 C - GPU is throwing heat on it.

He's talking about T-Case Delta. By dropping ambient temps inside the case from intake fans and lower ambient temps, Could lower the internal component's temps by as much as 10c.

But I suppose it would help to drop a thermometer in the case first to take that measurement.

Maybe get one of these. Indoor Hygrometer.

 

[lexluthermiester]

Hello,

Hope the video and this post will help you guys decreasing temps on your GPU. Effort is minimal. In my case went 10 C lower just with a use of a 92mm fan.

I suggest the best fan for this application to be a 92 mm fan, as my test shows is better than the 120 mm fan , haven't tried and 80 mm yet.

My setup is 7600 Ryzen, GPU 1080 Ti EVGA, Lian Li OD 11 Dynamic XL.

If you have questions pls let me know.

Is the second test I've done on this particular mod, present driver 555.85 the earlier test was on 536 driver and on that was 8-9 C lower.

In the description of the video I posted results of a test in BF V with RPM reduced on GPU fans and back case exhaust fan for who is interested.

Fans used on my GPU are NF-B9 -redux x3 + 1 NF-A9 PWM

Thanks

Having done something similar to this many times myself over the years, I can fully confirm that this kind of mod works well! Very well done Rusty!

 

[chrcoluk]

Manufacturer and card full name pls?

It is a Nvidia Founders Edition 4080 Super.

 

[Knight47]

@BMfan80
part of the problem is ppl do stuff, while having a single fan in the rear, while 3 or more fans provide intake, way too much vs being removed,
so clearly will have air going around inside the case.
main reason why i have a case with two 140 fans for (case) exhaust.

Well Op has 2 intake and 4 exhaust, no idea where the gpu is getting fresh air from since the only fan below the gpu is exhaust, he has a glass front panel, bottom is sealed, so that leaves the PCI Slot Cover Bracket holes?

 

[ARF]

@BMfan80
part of the problem is ppl do stuff, while having a single fan in the rear, while 3 or more fans provide intake, way too much vs being removed,
so clearly will have air going around inside the case.
main reason why i have a case with two 140 fans for (case) exhaust.

People should investigate the issue, because the asymmetrical positioning of intake vs. exhaust fans creates either positive or negative air pressure inside the case, which could lead to increased temperatures, also because of the air local turbulence zones. Also, the intake fans deposit huge amounts of dust inside.

I have chosen to have my case without any main intake fans, I think the only one is the PSU's at the bottom and in its own cage.
I have put one 140 Arctic as rear, and another 140 Arctic on top, both push air out of the case..

 

[rusty caterpillar]

He's talking about T-Case Delta. By dropping ambient temps inside the case from intake fans and lower ambient temps, Could lower the internal component's temps by as much as 10c.

But I suppose it would help to drop a thermometer in the case first to take that measurement.

Maybe get one of these. Indoor Hygrometer.

I might try that.

Well Op has 2 intake and 4 exhaust, no idea where the gpu is getting fresh air from since the only fan below the gpu is exhaust, he has a glass front panel, bottom is sealed, so that leaves the PCI Slot Cover Bracket holes?

Is a strange big case. I would remove the glass panel from exhaust and build my self a frame and buy a mesh. The air intake gets sucked from various parts of the case I believe, also I think there is 1 side that got a mesh and not glass.
I have 2x14 cm intake + 1x 12cm intake. 2x14cm exhaust as part of the AIO, but is a slow exhaust since it has to pas AIO rad fins before air could expelled, 1 x12 cm rear. I can't consider the 92 mm fan from the backplate exhaust, it helps in exhausting hot air from the lower chamber but, at slow rate/efficiency, only 900-1000 rpm. So is like: 3 intake and 3 exhaust.

People should investigate the issue, because the asymmetrical positioning of intake vs. exhaust fans creates either positive or negative air pressure inside the case, which could lead to increased temperatures, also because of the air local turbulence zones. Also, the intake fans deposit huge amounts of dust inside.

I have chosen to have my case without any main intake fans, I think the only one is the PSU's at the bottom and in its own cage.
I have put one 140 Arctic as rear, and another 140 Arctic on top, both push air out of the case..

If the PSU got her own cage than the fan is not an intake or exhaust regarding your case cooling. Is just an isolated intake.
As I understand from your post you have 2x exhaust fans top and rear and none intake? Than is clearly negative pressure and in my experience, brings more dust in your case.
In my old case I had negative pressure for a while, GPU was the most part cover in the dust.

Having done something similar to this many times myself over the years, I can fully confirm that this kind of mod works well! Very well done Rusty!

Thanks.

It is a Nvidia Founders Edition 4080 Super.

As long as I know those are hard to service, change thermal past or pads, fans wiring is soldered to the PCB.
As long as it got nice temps I will not worry.

 

[Knight47]

I have 2x14 cm intake + 1x 12cm intake. 2x14cm exhaust as part of the AIO, but is a slow exhaust since it has to pas AIO rad fins before air could expelled, 1 x12 cm rear. I can't consider the 92 mm fan from the backplate exhaust, it helps in exhausting hot air from the lower chamber but at slow rate/efficiency, only 900-1000 rpm. So is like: 3 intake and 3 exhaust.

Is the 12cm intake a reverse blade fan?

 

[rusty caterpillar]

Is the 12cm intake a reverse blade fan?

No, is just and old Arctic F12 PRO which is a very decent fan even is 2011 fan technology. Arctic F12 PRO is very close in tests to Mobius OC 120 mm fans . Mobius got a nasty rotor noise over 1400 RM, something I didn't encounter before in fan noises, so I just sent them back and kept my old Arctic fans.

 

[Kapone33]

Not guessing anything - Are very, very few GPU which are well air cooled from the start because of the points I mention above and some more points I will mention bellow - all have an ill effect on how well a GPU is cooled but, those factors can be avoided only if manufacturers change their ways.
Are many other factors - too small radiators or just not enough fins
-bad design PCs
- not enough textolite in the PCB or not at all
- cheap PCB even on expensive cards like 4090, easy to twist or bend again because lack of textolite a very thin layers.
Try to brake a PCB with enough textolite in it with your bare hands
- Factory OC cards which doesn't have proper rad or fans not even for standard frequencies
-lack of thermal pads between back plate and GPU PCB
- radiator plate covering only half of the memory chips - see Asus
- Quality checks non-existent
My question is not stupid and wasn't addressed to you. That's the reason they pulled out lead out of the soldering - lead is not eco. Well, more electronics will brake faster without lead elasticity in the soldering- therefore is not eco, not even remotely. Is just plain stupid and counterproductive.

In actual fact the size of the heatsinks on modern GPUs make cooling a non issue. Fan noise is common for all of them though (As Rock cards are quietest). Things like Pulse (budget vs Nitro (Premium) are not performance but indeed accoutrements like a fan header and RGB bling. Even those huge cards like the Gigabyte Gaming OC are fine for cooling.

This hack could work. I have a 40mm Noctua sitting around somewhere. I am going to try it later.

 

[rusty caterpillar]

In actual fact the size of the heatsinks on modern GPUs make cooling a non issue.

Sorry, but is inaccurate for a lot of types of new cards.
Look at the various issues in VRAM cooling at 3070 and 3080.
Middle passive plates not covering all the surface of VRAM, see some Asus cards with this issue. Maybe I'll find some pictures.
Even they have big heatsinks, draws a lot of power and those big heatsinks in some cases proved to be insufficient in cooling the VRAM.
Cheaper PCB compared with older cards and the lack of textolite in the PCB. Which leads this kind of damage which you see on This Gigabyte 4090 which is far from cheap:

You see. Textolite was developed during the cold war by USSR as component of the plating armor for T64 tanks. It was there to prevent the huge heat created on impact by allied HESH shells maybe HEATFS to. Later on was used in PCB in a lot of electronics - not only giving very good structural strength to the PCB but, also very little heat will escape/pass from one side to the other of PCB which proved to be a valuable attribute in electronics.
With enough textolite amount in GPUs PCBs you not gonna see so much ripped pads under VRAM or GPU main, or cracks in PCB when people remove their cards. This issues with weak PCBs is very often seen now days in very expensive cards like 4090.

This hack could work. I have a 40mm Noctua sitting around somewhere.

I really think 40mm is too small to have an impact on your VRAM, you need at least 80mm.
However if you map your VRAM on your GPU and see which memory chip is the hottest you can place the 40mm fan just there and test.

 

[Waldorf]

@ARF
anytime you have airflow, there will be dust, doesnt matter if intake/exhaust etc, amount of total flow is relevant, as well as dust in the home.
the best fan covers will never prevent dust getting in, short of running a hepa air cleaner in the room (did for a time), extending the time between dust cleaner by 4 times,
its down to balancing noise/cooling/dust.
still, ppl not willing to spend 10min a few times a year to clean +1.5K of parts, seem to me like those driving a +100K car that's dirty/has curbed wheels.

@rusty caterpillar
i would cut out the slot cover on the side panel (borrow a flex, quicker and cleaner), mount the aio on top area,
get an Arctic P12/14 (if possible CO, has BB) mounted below for intake.
if you get the front open for flow, set the rad (side) to exhaust, instead of intake.

one reason i will probably switch to this case, as it makes no difference if use air or water for the gpu..
what i lose from one fan less, i gain from 3 more being now 140 instead of 120

 

[rusty caterpillar]

@ARF
anytime you have airflow, there will be dust, doesnt matter if intake/exhaust etc, amount of total flow is relevant, as well as dust in the home.
the best fan covers will never prevent dust getting in, short of running a hepa air cleaner in the room (did for a time), extending the time between dust cleaner by 4 times,
its down to balancing noise/cooling/dust.
still, ppl not willing to spend 10min a few times a year to clean +1.5K of parts, seem to me like those driving a +100K car that's dirty/has curbed wheels.

@rusty caterpillar
i would cut out the slot cover on the side panel (borrow a flex, quicker and cleaner), mount the aio on top area,
get an Arctic P12/14 (if possible CO, has BB) mounted below for intake.
if you get the front open for flow, set the rad (side) to exhaust, instead of intake.

one reason i will probably switch to this case, as it makes no difference if use air or water for the gpu..
what i lose from one fan less, i gain from 3 more being now 140 instead of 120

View attachment 358319

Is that case a Lian Li? It doesn't matter back grill is a bad one, are you ready to butcher it?
I will not ever go for bottom case fans, because of the dust.
I would recommend to you Fractal Torrent the big one as an air cooled case. Nothing can beat 2x 200 or 2x 180 mm case fans in the front. For LC however, which you have and like, don't think is a very good case.

anytime you have airflow, there will be dust, doesnt matter if intake/exhaust etc, amount of total flow is relevant, as well as dust in the home.

That is correct, EMF act as a magnet for all dust particles around. PCs are underrated hoovers

However with negative pressure in the case I believe you tend to have more dust inside the case vs positive pressure. Maybe I'm wrong.

still, ppl not willing to spend 10min a few times a year to clean +1.5K of parts

Agree. I see that very often.

 

[Waldorf]

again, fans/positioning are (more or less) irrelevant for dust (or its build up), dust in the air and rate of flow is.
e.g. my last IT job had a small server (room) with its own split AC. in almost 2y i never had to clean a single fan/cooler etc, just because there was no ppl (most dirt in air is skin particles),
or outside air coming in (emissions/bio matter).

its better to control the dust in the house/room, than focus just on the case.
one reason i will go back to use an air cleaner in the room, the cleaner the room air, less dust in the case.

i would go with a passive rad outside the cas, if it wasnt for cost (360x360 rads are +2xx), or if i can find a used reserater for cheap.
with that, i was able to run 8dB blacknoise fans throttled to "barely" running, mid range rpm under load, so not much dust.

 

[Kapone33]

Sorry, but is inaccurate for a lot of types of new cards.
Look at the various issues in VRAM cooling at 3070 and 3080.
Middle passive plates not covering all the surface of VRAM, see some Asus cards with this issue. Maybe I'll find some pictures.
Even they have big heatsinks, draws a lot of power and those big heatsinks in some cases proved to be insufficient in cooling the VRAM.
Cheaper PCB compared with older cards and the lack of textolite in the PCB. Which leads this kind of damage which you see on This Gigabyte 4090 which is far from cheap:
View attachment 358301
You see. Textolite was developed during the cold war by USSR as component of the plating armor for T64 tanks. It was there to prevent the huge heat created on impact by allied HESH shells maybe HEATFS to. Later on was used in PCB in a lot of electronics - not only giving very good structural strength to the PCB but, also very little heat will escape/pass from one side to the other of PCB which proved to be a valuable attribute in electronics.
With enough textolite amount in GPUs PCBs you not gonna see so much ripped pads under VRAM or GPU main, or cracks in PCB when people remove their cards. This issues with weak PCBs is very often seen now days in very expensive cards like 4090.

I really think 40mm is too small to have an impact on your VRAM, you need at least 80mm.
However if you map your VRAM on your GPU and see which memory chip is the hottest you can place the 40mm fan just there and test.

I can't say I have experienced any of the issues you mention. I have not looked at it to that degree but every card I have taken apart in the last 12+ years has had no issues with the PCB or VRAM cooling. The only issue that I had happen over that time that was repeated was Gigabyte not putting enough Thermal paste on the GPU. It was on the Vega 64 and 6500XT. Exact same issue. Where I live those 30 cards were too expensive for the VRAM buffer provided.

 

[eidairaman1]

again, fans/positioning are (more or less) irrelevant for dust (or its build up), dust in the air and rate of flow is.
e.g. my last IT job had a small server (room) with its own split AC. in almost 2y i never had to clean a single fan/cooler etc, just because there was no ppl (most dirt in air is skin particles),
or outside air coming in (emissions/bio matter).

its better to control the dust in the house/room, than focus just on the case.
one reason i will go back to use an air cleaner in the room, the cleaner the room air, less dust in the case.

i would go with a passive rad outside the cas, if it wasnt for cost (360x360 rads are +2xx), or if i can find a used reserater for cheap.
with that, i was able to run 8dB blacknoise fans throttled to "barely" running, mid range rpm under load, so not much dust.

View attachment 358322

I would have that tower rad secured to the case to prevent it from being knocked down