Does your GPU orientation actually matter? Surprising results...

Its been debated forever... does the orientation of your GPU negatively affect GPU temperatures??

Sponsored Links

○ Join this channel to get access to perks:

○○○○○○ Items featured in this video available at Amazon ○○○○○○

••• Follow me on your favorite Social Media! •••

[Usersfssk]

I mounted my rtx 4090 vertically, turned it on and then I woke up

[ChrisDombroski]

Fun fact: convection is also a gravity driven process. Heat doesn't travel up, it disperses equally in all directions. However, as fluids get warmer, they get less dense, and are displaced upward by cooler, denser fluid. The difference in density causes a difference in the pull of gravity. (technically convection can still occur from small effects like surface tension and microcurrents, but that shouldn't be too relevant here I think)

[ClellBiggs]

I always enjoy seeing what happens when things are done to things.

[devvvinparker]

I'd love a follow up that tests the temperature difference of open air to having it pressed up against glass at varying distances. Found this pretty interesting

[Weaseldog]

I remember when a friend of mine decided to set the airflow in a full height server cabinet, to push the air from top to bottom.

I pointed out that was sucking the warmest air in the room into his cabinet.

He thought it didn't matter.

Fast forward a few years of fighting a temperature problem. He could not keep the room cool and his servers ran hot..

He reversed the fan, and all of those problems disappeared.

[Flameheadwilson]

One test you missed would be horizontal, but with the cooler above, like you might see in an inverted case. It would, in theory, put the gpu at the bottom of the vapor chamber, where gravity would maximize the fluid return. But as you said, it's peanuts compared to all the other factors in a case.

[Lif-]

As someone who, over the last 6 years or so, has been running a couple of GPU''s [not concurrently] mounted vertically - slot down - into a horizontally mounted mobo, I found your
piece on the thermal ramifications of different GPU orientations to be very interesting.
Anecdotally, I have found the above scenario to operate just fine and have noticed no adverse affects regarding performance, throttling or noise, whatever.
The fact that I insist on using a large, roomy, properly ventillated case, no doubt helps the thermal cause.
Cheers J.

[-TheLynx-]

15:00 Important note about the 7900XTX Vapor Chamber design flaw is that this applied to the reference design. Aftermarket designs were not affected by this.

[mastermind]

Someone PLEASE get Jay 40 more different-colored tape measures for Christmas.

[SkullAngel]

6:06 - Missed Jay yoga pose opportunity for the video thumbnail 😂😂

[saulelietumi]

I love how Jay casually says: It's chilly, it's winter time, so the temperature here in the room is about 21.5°C. Bro, that's my summer vacation temperature!

[NSDaishi]

The level of confidence of some of these comments surprises me but at the same time it doesn't.

[sherrydowns]

not only vertical mounting, but some people tilt the whole case totake the force off slots

[timcooper]

actual tests inside a contained case with normal fans would be interesting

[pillar]

It makes sense. Vapor chambers are essentially a vacuum environment and a vacuum is still affected by gravity. It may be a negligent difference due to the amount of matter within the chamber. I would be interested in seeing if there is any difference in water cooled cards in different orientation.
As always Jay, great video!!

[Cosmodjinn]

The problem is not GPU mounting orientation, but mounting them in one of these novel positions that leave no room for air intake. Its harder to design cases for appropriate airflow in those positions and too many just cram it against a case wall or door.

[jeffcox]

Thanks for the physics lesson. Thanks for helping me with my build a year ago, and still going strong!

[GoatLockerGaming]

(Not computer related) Outstanding to see that you bring a “home gym” to work. You’d be surprised at how well your brain works, especially during troubleshooting when you get a chance to workout. Livin the American dream there Jay!

[JETWTF]

Boxes over thermostats are just there to keep people from messing with them and are in public places. Thermostats are already made to avoid direct airflow and when properly placed should only get direct airflow from people walking past them hence why they are normally in a hallway or someplace that doesn't have an HVAC register pointing directly at them. Direct airflow onto the security box from HVAC would have the same effect as the thermostat itself without one, the air inside the box will be affected by the airflow from the register just as the air inside the thermostat would be. Either hotter or colder depending on what the HVAC is pumping out rather than after it mixed with the rest of the air and reaches the thermostat from normal circulation.

[LOLHoneybadger]

PCB heatsoak is a real thing on GPUs, and it's effect on temperatures/boost clocks cannot be understated - it made a huge difference on my 3090 Kingpin when I had it under full liquid cooling. The only way to truly factor any heatsoak out is to either let the GPU idle for a LONG time before testing again, or shut down for a consistent period of time, then boot back up. I typically did 15 minutes of cool down time with the fans set pretty high, between high-power runs, and the consistency of those runs vs running back to back with only letting it 'cool down to idle temps' before going again was noticeable.