Do Backplates Cool Your GPU?

[vanished]

23 minutes ago, kaiju_wars said:

Am I the only one who doesn't care if the backplate helps with cooling?  It helps the card with sagging and it looks better than an open PCB when in a case with a window.

Not having to worry about touching exposed components, along with the look is the main reason I would want one, but the downside to consider is increased cost, and I thought, increased temps, but glad to see there didn't seem to be any affect.  The reason I expected there would be was that the backplate on my card gets painfully hot to touch to the point where I can't keep my finger there when it's under load, which, being on the opposite side of the PCB as the hot stuff seemed odd to me  

[kaiju_wars]

2 minutes ago, Ryan_Vickers said:

Not having to worry about touching exposed components, along with the look is the main reason I would want one, but the downside to consider is increased cost, and I thought, increased temps, but glad to see there didn't seem to be any affect.  The reason I expected there would be was that the backplate on my card gets painfully hot to touch to the point where I can't keep my finger there when it's under load, which, being on the opposite side of the PCB as the hot stuff seemed odd to me  

Hmm, I think really it just depends on the brand.  Sometimes they're cheaper, sometimes not so much.

But yeah, I do like the backplate on my GPU.

[Sakib]

36 minutes ago, kaiju_wars said:

Am I the only one who doesn't care if the backplate helps with cooling?  It helps the card with sagging and it looks better than an open PCB when in a case with a window.

I care when the manufacturer claims their backplate helps with cooling when in reality, it doesn't. Just increases the cost of the card.

[Hans Power]

I hope you got the same balls when it comes to the big boys like ASUS, Gigabyte or especially MSI. They all got or at the very least had their snake oil on some of their products. Remember those red USB "Gaming" ports on MSIs Z97 Motherboards which supposedly helped with Controller latency with because of gold contacts (yeah, those did actually fudge-all compared to the normal ones). That backplate does at least something in terms of cooling as the cool spot on the PCB shows - definately more than those USB ports. It's easy to rip on a small manufacturer because good relations to those don't really matter for you in future so I really wonder if you'll point out the snakeoil the big manufacturers undoubtedly have to offer in some of their products and call them "$h!t manufacturers" as well if that happens. I don't know what it is this generation but I'm sure you'll find some snakeoil in their product line-up if you go through their marketing material. If you open these cans of worms I hope you'll do it fairly to everyone.

 

Edit: How about this SSD "Heatshield" which actually increases your SSD temps instead of lowering them on the MSI Z270 Gaming Pro Carbon?

[Squazzer]

@nicklmg

 

This is shamelessly stolen from YouTube comments, but I was thinking about the same thing:

Quote

 

Founders Edition
2:33 With BP  Fan Speed: 65%
3:08 Bareback Fan Speed: 69%

 

Adrus xtreme edition 8G
3:46 With BP  Fan Speed: 44%
Bareback Fan Speed Conveniently Left Out

 

There are a lot of variables that you forgot (chose) not to mention.

Why didn't you lock down as much as possible to see if the temp changed? You could have locked the fan-speed as an example

[Liltrekkie]

The testing method is so wrong here. 

 

Allow me to explain, JayzTwoCents recently did a video about the new EVGA icx cooler and the FTW2, in which he also tested the cards backplate cooling abilities, the thing is though, while he was using an open air test bench, he had a fan blowing air on the card as if it were a front case fan. 

 

The effect? VRM and memory temps were improved by more than  few degrees than without the fan there. 

 

So if this video were to be redone and a fan put behind the card as if it were a front fan, i do in fact believe the temps would come down as they claim. 

 

Everyone forgets that backplates are PASSIVELY cooled. They rely on outside cooling to keep them cool. Servers that use passive cooling rely on those high RPM fans to cool them. Raid cards, some GPU's even, Ethernet cards. They all require airflow to keep cool.

 

If you block a fan on a GPU even partially (think SFF build) it'll overheat even though there is a massive heat sink over with a fin array under. Why? it requires airflow. 

 

When you have flawed testing methods you have a flawed video such as this.

[vatorres]

Pretty much what @The Ran and @Liltrekkie said.

 

The cold spot shows due to high differences in emissivity, which is particularly low on high polished copper. For all we know, it could even be warmer than the rest of the backplate. Might wanna consider buying some cheap thermocouples for this kinda testing.

As far as thermals go, the lack of proper airflow control in an open test bench makes a HUGE difference in convective heat transfer rate on the copper plate, as well as the physical position of the card (vertical instead of horizontal).

 

Proper testing should be done inside a closed environment (PC Case) with room temperature and airflow control, or at the very least with a fan blowing into the card so as to better simulate actual perfomance inside the machine. The way this video was done, your claims could be considered as void as the manufacturer's.

[Pat-Roner]

6 hours ago, JohnMcBane said:

it would have been interesting to test this with evgas new card, with all the additional temp sensors on the vram and power components.

 

 

[samcool55]

Hooraay... Wrong information and flawed testing method ftw...

 

I mean come on man, did really nobody at one point took a step back and just looked at more than just temps?

We all know damn well nvidia boost 3.0 whatever thingy will just cause the card to go whatever temp it want and it will adjust fan speed and core speeds to keep it there...

[FirstArmada]

14 hours ago, 21rkosta said:

when you think this is the new floatplane video only to realize its not

Havent been active on the forums in awhile whats a floatplane video?

[samcool55]

3 minutes ago, FirstArmada said:

Havent been active on the forums in awhile whats a floatplane video?

Since vessel died they started the floatplane club, same idea as vessel (early access to video's but you have to pay for it) but only it's LTT content exclusive and you can only find it on the forum.

[Granular]

The copper square appearing darker in the thermal image than the surrounding PCB means it's dissipating heat into air faster than it's absorbing it from the PCB, which is not what you want to see.

The amount of heat transferred to air by the copper plate can be calculated by multiplying a quotient characterizing the efficiency of the plate (measured in Watts per degree C ) by the the difference in temperature between the plate and the surrounding air. So whatever the efficiency of heat transfer between the plate and the air, for maximum amount of heat energy transferred into air, you want the plate temperature to be as close to the board temperature (and therefore as high above the air temperature) as possible.

Now the plate could still be helping, provided it's somehow amazing at transferring heat into air compared to the PCB, but it's definitely limited by the heat conductivity of the pad.

[The Professor]

Signed up to say that I love the idea for this new "show". This sort of thing is what I used to love doing when our reps sent us something that was meant to be the bee's knees.

 

Sadly I don't have the time anymore to challenge the claims of someone in marketing. So from me and all the other overworked technicians out there, keep up the good work!

[Rainwalker]

But my question, and free idea for the workshop, is the theory sound? Could you actually make a well engineered backplate that dissipates heat sufficiently to have an actually notable cooling effect. Or better yet, would this backplate do the job if it weren't for that massive teribad thermal pad?

[Drak3]

On ‎2‎/‎16‎/‎2017 at 11:35 AM, Granular said:

The copper square appearing darker in the thermal image than the surrounding PCB means it's dissipating heat into air faster than it's absorbing it from the PCB, which is not what you want to see.

Actually, FLIR doesn't accurately measure temperature ranges, the materials being "measured" can skew the results due to how they reflect Infrared light. Hot and cold spots also don't correlate to anything other than the "hottest" and "coldest" areas in the image.

 

Basically, for what LTT (and JayzTwoCents) want to do, FLIR is worthless.

[VSG]

2 hours ago, Drak3 said:

Actually, FLIR doesn't accurately measure temperature ranges, the materials being "measured" can skew the results due to how they reflect Infrared light. Hot and cold spots also don't correlate to anything other than the "hottest" and "coldest" areas in the image.

 

Basically, for what LTT (and JayzTwoCents) want to do, FLIR is worthless.

The FLIR consumer/gadgets division products are useless, the research division products are not. That said, this video was so bad that I have given up hope from LTT. 

[honna1612]

They made a mistake in the video:

The cold spot on the backplate wasn't cold. Infrared is reflected by copper like a mirror so it picked up the ambient around the back-heatplate.

If it were conductiong really good it would also be hotter and not colder.

[Yoinkerman]

I would imagine a backplate pulls heat from surrounding components(vrms and whatnot), which is good for the life of the card.  which this doesn't test at all

[Drak3]

4 minutes ago, Yoinkerman said:

I would imagine a backplate pulls heat from surrounding components(vrms and whatnot), which is good for the life of the card.  which this doesn't test at all

That's the justification EVGA used for the beefed up sensor loudout on the FTW2, and it's a legitimate explanation and use of the back plate. While the GPU die temperatures saw near zero change in temperatures, other vital components, that are oft not monitored, will see some improvement.

But there's something else to consider, does the Aorus even have probes monitoring temperatures of other components on the card? Are they available for software pulling readings from them?

[Yoinkerman]

11 minutes ago, Drak3 said:

That's the justification EVGA used for the beefed up sensor loudout on the FTW2, and it's a legitimate explanation and use of the back plate. While the GPU die temperatures saw near zero change in temperatures, other vital components, that are oft not monitored, will see some improvement.

But there's something else to consider, does the Aorus even have probes monitoring temperatures of other components on the card? Are they available for software pulling readings from them?

no idea but at work stuff dies because the power delivery dies(rectifiers, then vrms, then caps, then diodes), not the chips themselves.

[captain_to_fire]

Apparently, Gigabyte responded to Luke's video.

 Which doesn't really change anything 

[srslyuncool]

it does affect VRM and mosfet temperatures quite a bit (at least it did on my r9 290)

[Zyalon]

LOL try the RTX 3090 without the backplate now