Question about VRMs

[don_svetlio]

 I've seen some cards have vastly lower VRM temps than other cards even though they are pretty much identical and it got me wondering. Why is that?
Example - under load my VRMs get to 72*C average
A friend's 280X using the same cooler gets the VRMs to 65*C.

Can anyone explain this since I'm not sure myself why that is.

[themaniac]

 I've seen some cards have vastly lower VRM temps than other cards even though they are pretty much identical and it got me wondering. Why is that?

Example - under load my VRMs get to 72*C average

A friend's 280X using the same cooler gets the VRMs to 65*C.

Can anyone explain this since I'm not sure myself why that is.

maybe he has better airflow, and what i find weird is that you have a 280 which is less powerful and produces less heat then the 280x so i would think your friends would be hotter

[don_svetlio]

maybe he has better airflow, and what i find weird is that you have a 280 which is less powerful and produces less heat then the 280x so i would think your friends would be hotter

Exactly - doesn't make a lot of sense. Airflow wise - he has 1 intake, 1 exhaust and I have 1 intake, 2 exhaust. His core is 10*C hotter usually tho.

[themaniac]

Exactly - doesn't make a lot of sense. Airflow wise - he has 1 intake, 1 exhaust and I have 1 intake, 2 exhaust. His core is 10*C hotter usually tho.

maybe its like silicon lottery but with metal instead of silicon

[AresKrieger]

 I've seen some cards have vastly lower VRM temps than other cards even though they are pretty much identical and it got me wondering. Why is that?

Example - under load my VRMs get to 72*C average

A friend's 280X using the same cooler gets the VRMs to 65*C.

Can anyone explain this since I'm not sure myself why that is.

Ignore this

[don_svetlio]

maybe its like silicon lottery but with metal instead of silicon

Could be. I mean - most cards I see have higher VRM temps but some have these anomalies.

[don_svetlio]

Maybe the vram itself is different, and thus has temperature differences on that end

VRAM?

[AresKrieger]

VRAM?

Whoops dislexia

 

 

 

 

Edit: your vrm could have more resistance thus increasing the temperature it reaches, kinda weird though since the power draw should be less and power equations are linear in nature for the most part, of course capacitors and inductors don't always follow these rules and are used in power regulation

Edited October 21, 2015 by AresKrieger

[themaniac]

Could be. I mean - most cards I see have higher VRM temps but some have these anomalies.

well at least most cards have higher VRM temps 

[don_svetlio]

well at least most cards have higher VRM temps 

That's what I've found to be the rule. Oddly enough - mine use to be lower but over time they became higher than the core temps.

[PineappleHolidays]

it varies based off chip consumption

 

your card might be oc'd more or might use more power to reach the same performance

[themaniac]

That's what I've found to be the rule. Oddly enough - mine use to be lower but over time they became higher than the core temps.

then maybe its more dependent on how old the VRM's are then?

[Vivilacqua1]

With my 290, the vrm's hit over 100c. Now with the Kraken G10 they only hit 75 at full load.

[don_svetlio]

then maybe its more dependent on how old the VRM's are then?

Could be - or the usage.

[AresKrieger]

then maybe its more dependent on how old the VRM's are then?

That could be a factor as resistance increases with age, thus increasing temps

[Vivilacqua1]

Is it overclocked at all?

[don_svetlio]

Is it overclocked at all?

stock.

[Vivilacqua1]

stock.

hmmmm.....

 

Welp, there goes my ideas.

[Mr.McMister]

There can be a fair amount of variance in component efficiency even within the same manufacturing run of a card. Some cores or the vrms themselves may be less/more efficient.

There are also 2 different basic types of voltage regulation modules. One simply sinks extra volage and dissipates the rest as heat while the other type regulates the duty cycle of a capacitor inductor loop. The former is what's used on most gpus because of size and cost, but the latter is very common on motherboards for the CPU and not unheard of in expensive large gpus.

[Kinda Bottlenecked]

What 280x is that? Custom PCB or reference? 

[don_svetlio]

What 280x is that? Custom PCB or reference? 

Both Custom Asus designs.

[don_svetlio]

There can be a fair amount of variance in component efficiency even within the same manufacturing run of a card. Some cores or the vrms themselves may be less/more efficient.

There are also 2 different basic types of voltage regulation modules. One simply sinks extra volage and dissipates the rest as heat while the other type regulates the duty cycle of a capacitor inductor loop. The former is what's used on most gpus because of size and cost, but the latter is very common on motherboards for the CPU and not unheard of in expensive large gpus.

15 mins of valley

[Mr.McMister]

15 mins of valley

I don't think there is anything particularly wrong with the card, those temps can be accounted for by component variance and the voltages look fine. I'd assume here that on the card you have the vrms don't make contact with the heatsink and are only cooled by passing air. You can try blowing out any dust from the heatsink to make sure you're getting good airflow past the vrms, but I wouldn't worry about it too much unless the max temps you see are rising over time as that would indicate a slowly failing component.

[don_svetlio]

I don't think there is anything particularly wrong with the card, those temps can be accounted for by component variance and the voltages look fine. I'd assume here that on the card you have the vrms don't make contact with the heatsink and are only cooled by passing air. You can try blowing out any dust from the heatsink to make sure you're getting good airflow past the vrms, but I wouldn't worry about it too much unless the max temps you see are rising over time as that would indicate a slowly failing component.

I actually have reverse airflow thanks to the CPU cooler, maybe that has something to do with it? On an open case I get 64*C. As for voltage, I don't think GPU-Z is particularly reliable as sometimes it reads spikes to 1.442V which should have fried my card several months ago if it were the case, right?