Raising gpu throttle temp?

[Somerandomtechyboi]

Cooler on the rx 580 2048sp is garbage so just incase i cant just ziptie a 775 copper core onto the thing the only other option would be to raise throttling temp

 

It already gets to 90c on stock 1180mhz 0.95v and i cant go lower than 0.86v so i wonder if i can raise the throttle temp to maybe 100c or 115c so i can run higher clocks, at such low voltages like <1v degradation of the gpu die should be non existent even at 115c so im not concerned about the gpu

 

 

Now how do i go about raising temp limit? Cause i cant go over 90c with afterburner nor the included oc tool in the full amd drivers, and i do not mind vbios modding, so just need to know what utilities i can use

 

Maybe i wont go all the way to 115c but main concern would be the other parts of the gpu cause obviously letting the die run at 100c will problably heat the other components up, vram is problably gonna be fine just slightly reduced oc headroom at high temp, though im kinda concerned about the vrms even though theyre supposed to be fine upto around 115c just like the vram. So any parts i wanna avoid heating past 100c?

[PDifolco]

3 minutes ago, Somerandomtechyboi said:

Cooler on the rx 580 2048sp is garbage so just incase i cant just ziptie a 775 copper core onto the thing the only other option would be to raise throttling temp

 

It already gets to 90c on stock 1180mhz 0.95v and i cant go lower than 0.86v so i wonder if i can raise the throttle temp to maybe 100c or 115c so i can run higher clocks, at such low voltages like <1v degradation of the gpu die should be non existent even at 115c so im not concerned about the gpu

 

 

Now how do i go about raising temp limit? Cause i cant go over 90c with afterburner nor the included oc tool in the full amd drivers, and i do not mind vbios modding, so just need to know what utilities i can use

 

Maybe i wont go all the way to 115c but main concern would be the other parts of the gpu cause obviously letting the die run at 100c will problably heat the other components up, vram is problably gonna be fine just slightly reduced oc headroom at high temp, though im kinda concerned about the vrms even though theyre supposed to be fine upto around 115c just like the vram. So any parts i wanna avoid heating past 100c?

You can't unless you really tamper physically heavily with it, and shoudn't because allowing too high temps will just destroy the GPU... 

(then you can become a celebrity by sending it to GNSteve  ) 

[Somerandomtechyboi]

10 minutes ago, PDifolco said:

You can't unless you really tamper physically heavily with it, and shoudn't because allowing too high temps will just destroy the GPU... 

(then you can become a celebrity by sending it to GNSteve  ) 

Destroy the gpu how?

The vram and the gpu die should be fine till 115c 

Vrms are abit of a concern

 

i doubt ill actually be able to kill the card till i get any of those 3 to 150c, and degradation is non existent with such low voltages even at high temp

[Somerandomtechyboi]

Ok looks like red bios editor can actually change the tjmax, shutdown temps, and all that other stuff so ill just see for myself what happens

 

My somewhat educated guesses would assume that 100c should be ok at such low volts and 115c abit sketchy (surrounding components connected to the heatsink may overheat, gpu die itself probs fine) so ill just go for a 100 or 105c tjmax then see where that gets me

[PDifolco]

32 minutes ago, Somerandomtechyboi said:

Destroy the gpu how?

The vram and the gpu die should be fine till 115c 

Vrms are abit of a concern

 

i doubt ill actually be able to kill the card till i get any of those 3 to 150c, and degradation is non existent with such low voltages even at high temp

Good luck, I still feel you'll be shortening the life of your card for a couple frames more...

[Somerandomtechyboi]

1 hour ago, PDifolco said:

Good luck, I still feel you'll be shortening the life of your card for a couple frames more...

Again not by an actually noticable amount, i mean what kinda degradation you think will happen at <1v? my somewhat educated assumptions would assume basically next to none even at 115c

 

i just dont focus on things that dont actually affect anything in the real world and that way of thinking dont only apply to psus. And besides 60$? Whatever i can buy 2 more if i really wanted to with my leftover money, and even if there was some noticable degradation guess what? ill just resell the damn thing and buy another one problably losing only 5-10$ in the process

 

i geniunely start questioning my own intellegence when such obvious solutions of cheaply replacing a mildly degraded part isnt noticed by most ppl, and i dont think of myself as being that particularly intellegent either =/

[QuantumSingularity]

8 minutes ago, Somerandomtechyboi said:

i geniunely start questioning my own intellegence when such obvious solutions of cheaply replacing a mildly degraded part isnt noticed by most ppl

It's not that it's not noticed, it's that very few people who don't have interest in this can be bothered to do it. Like servicing or repairing your car on your own. Some of us like doing it because it's a hobby, other people just prefer to pay someone to do it for them so they can enjoy their time in some other way. It's usually not a question of saving money, but saving time. Another example - i love everything PC hardware related, i love gaming, i love driving, i love working on my car. What i don't love is cleaning and chores in general. And even though i live in a small 69m² (nice) apartment, it still takes half of my weekend to clean the bathroom, dust/wash the drapes, clean the windows, clean the balconies... the usual things associated with living in a big, dusty city. That's why i prefer to pay the company, which sends someone at my place to do that for me and not only once a week, but every Tuesday and Friday. 

 

[Somerandomtechyboi]

1 hour ago, QuantumSingularity said:

It's not that it's not noticed, it's that very few people who don't have interest in this can be bothered to do it

Theres literally ppl that spend hours on their overclocks and they still adhere to likely bullshit safe voltages or something so clearly it isnt just a matter of not bothering to resell hardware its a matter of outright not knowing that its a possibility in the first place

 

id understand preserving a golden sample but an average sample who gives a shit, might aswell take the chance to get a better sample

 

 

The only times i dont just crank the volt would be when thermals, power consumption, or straight up running into diminishing returns becomes an issue. So that leads me to not go too hard on the cpu but rams with their low power conumption and ease of cooling? Yea ill just crank em as far as theyll go

 

Though im curious why the tjmaxes of gpus are so low compared to cpus, i mean intel has been using 100c tjmax for a long time and yet this 14nm gpu that runs <1v has a tjmax of 90c? Seems kinda odd to me aside from preserving the other gpu components

[QuantumSingularity]

GPUs have exposed dies, CPUs have IHS - Internal Head Spreader. The die in a CPU has some kind of thermal interface between in and the IHS, most commonly germanium which allows for the CPU cooler to mount imperfectly, but still be able to keep the CPU die cool. This also acts like a buffer for head absorption and that's why you can boot a CPU with no cooler for a few seconds w/o it instantly shutting down or dying. That's also why when applying paste on CPU, a blob of pate is enough and it's not necessary to spread it even across the entire IHS. If you do the same with a GPU and you have small 1mm² area that's not covered with paste, you basically have a few thousands transistors which have no heat exchange in and that area becomes your thermal throttle. Tjmax is the hottest spot on the die. 115°C is the tjmax for GDDR6X, but the GDDR5 on the RX580 is rated for 95-100°C depending on the manufacturer. That's why it has lower limit.

[Somerandomtechyboi]

4 hours ago, QuantumSingularity said:

GPUs have exposed dies, CPUs have IHS - Internal Head Spreader. The die in a CPU has some kind of thermal interface between in and the IHS, most commonly germanium which allows for the CPU cooler to mount imperfectly, but still be able to keep the CPU die cool. This also acts like a buffer for head absorption and that's why you can boot a CPU with no cooler for a few seconds w/o it instantly shutting down or dying. That's also why when applying paste on CPU, a blob of pate is enough and it's not necessary to spread it even across the entire IHS. If you do the same with a GPU and you have small 1mm² area that's not covered with paste, you basically have a few thousands transistors which have no heat exchange in and that area becomes your thermal throttle.

Exactly the reason why i just put a shitload of thermal paste when repasting gpu to make sure every spot is covered

 

Excessive? Sure but hey no overheating issues, and iirc if you just do a thin spread there might not be any proper contact in some areas so just went nuts with the thermal paste, i mean gd007 is dirt cheap like gd900, maybe like 2-3x the price but 30g tubes only going for like 4-5$ i dont mind paying the extra if it gives better cooling

 

4 hours ago, QuantumSingularity said:

Tjmax is the hottest spot on the die. 115°C is the tjmax for GDDR6X, but the GDDR5 on the RX580 is rated for 95-100°C depending on the manufacturer. That's why it has lower limit.

Welp i havent replaced the pads on the vram and theyre bone dry so hopefully theyll be fine ish even if i push the gpu die to 115c but ill just be a tad more conservative and set a 100 or 105c max since this is a system i wanna sell