Permanent Vram Damage?

[ishansardar]

Hi, I'm a new miner. I've been running a 3070 FE to mine for a month or so, today I got a Leadtek 3090 and I've put it in my mining rig with my 3070. Here's how it looks

https://i.imgur.com/0bhxqcm.png


I guess my first question is what do you guys think? are the temps, fan speed, hashrate and stuff ok? I want my card and fan to last me a lifetime or at least longer than professional miners who roi their cards in a few months with double profits and stuff but since I'm mining to hold it's different for me. I might just take them off and start gaming after a year or so or maybe less.

My 2nd question is I heard some of 3090 cards vram gets overheated easily and it gets permanently damaged, I don't want that to happen to my card. So is there anyway I can see memory temp of my card in HiveOS? I saw a image doing this, but I don't know how to

https://i.imgur.com/wTGhRCm.png


I mean I could turn my autofan settings and target memory temp into something safe but I heard putting the rig on autofan decreases your fan health as it spins according to the target temp and not in a stable static rpm. So what do I do??

Thank you

[philipma1957]

since the card cost a fortune and one way to lose you money is to fry it.

I advise 270 watts and 2150 memory

vs 300 watts and 2400 memory

check what that drops the hash to 110 hash is about 12 a day usd


115 hash is about 12.50 a day usd

120 hash is about 13 a day usd

if I were you anything over 110 is decent only 1 dollar less a day before power cost. and may save the card.

also a fan blowing on the back of the card is good.

[ishansardar]

since the card cost a fortune and one way to lose you money is to fry it.

I advise 270 watts and 2150 memory

vs 300 watts and 2400 memory

check what that drops the hash to 110 hash is about 12 a day usd


115 hash is about 12.50 a day usd

120 hash is about 13 a day usd

if I were you anything over 110 is decent only 1 dollar less a day before power cost. and may save the card.

also a fan blowing on the back of the card is good.

I'm not sure why but it seems like 3090 is more power hungry than memory hungry when it comes to hashrate. Played with memory a bit and seems like changing it to 2000 from 2400 or 2600 doesn't bring much change to hashrate but the power does. Bringing power down even by 5 makes hashrate goes down by 10-15. So as per your suggestion I did 280 on the power as 270 gets it down to 105mhs and bought down memory to 2100 and I'm getting 112-113 mhs. You think thats ok for the longevity of the card? Thanks

[philipma1957]

since the card cost a fortune and one way to lose you money is to fry it.

I advise 270 watts and 2150 memory

vs 300 watts and 2400 memory

check what that drops the hash to 110 hash is about 12 a day usd


115 hash is about 12.50 a day usd

120 hash is about 13 a day usd

if I were you anything over 110 is decent only 1 dollar less a day before power cost. and may save the card.

also a fan blowing on the back of the card is good.

I'm not sure why but it seems like 3090 is more power hungry than memory hungry when it comes to hashrate. Played with memory a bit and seems like changing it to 2000 from 2400 or 2600 doesn't bring much change to hashrate but the power does. Bringing power down even by 5 makes hashrate goes down by 10-15. So as per your suggestion I did 280 on the power as 270 gets it down to 105mhs and bought down memory to 2100 and I'm getting 112-113 mhs. You think thats ok for the longevity of the card? Thanks

yeah dropping the memory down to 2100 from 2400 will help it last longer.

I run smos so i do not know how to get ram temp with hiveos. i have run my two 3090 in windows and get better ram temps with 2050 to 2100 ram settings.

[adaseb]

I am really surprised that Nvidia GPUs keep getting issues with damaged components due to be cooling. I remember a few years back, I think it was the 1080Ti, there was a bunch of gamers that ended up frying their MOSFETs on the GPUs. Upon investigation it was found that they were being fried because there was no heatsink on them and had bad circulation in the computer case.

Basically gamers didn't have a proper cooling setup with an adequate intake and exahust fan and it led to over >100C temps which ended up damaging the GPUs. They were replaced by warranty and they used heatsinks in future models. The point being is that this was only gamers which run their GPUs to the max and not miners which have an open air rig, so I wouldn't worry too much about frying anything.

[CryptoATM]

I'd rather get the best hashrate out of my gpu to get closer to my ROI than been worry about life span, your card can't last a lifetime for mining because difficultly grows bigger every year meaning at one point it will become obsolete, avoid high overclocking and use the most stable settings possible, that's all you'd ever need

[64dimensions]

I'd rather get the best hashrate out of my gpu to get closer to my ROI than been worry about life span, your card can't last a lifetime for mining because difficultly grows bigger every year meaning at one point it will become obsolete, avoid high overclocking and use the most stable settings possible, that's all you'd ever need

There is an issue with this line of thought, do you maximize the residual value of a GPU? or cook the card? Probably the two best answers to this question are the 480 GB nitro+ sapphire card and the EVGA 1080 ti. Both have held up in value, still do a decent amount of hashing and were not too bad cost wise. The 1080ti is particularly noteworthy because it has/had good mining flexibility.

One other point to note is that higher efficiencies (W/Mhs) are at the lower MSI AB settings. There is a message there.

[batsonxl]

I'd rather get the best hashrate out of my gpu to get closer to my ROI than been worry about life span, your card can't last a lifetime for mining because difficultly grows bigger every year meaning at one point it will become obsolete, avoid high overclocking and use the most stable settings possible, that's all you'd ever need

There is an issue with this line of thought, do you maximize the residual value of a GPU? or cook the card? Probably the two best answers to this question are the 480 GB nitro+ sapphire card and the EVGA 1080 ti. Both have held up in value, still do a decent amount of hashing and were not too bad cost wise. The 1080ti is particularly noteworthy because it has/had good mining flexibility.

One other point to note is that higher efficiencies (W/Mhs) are at the lower MSI AB settings. There is a message there.

Yes exactly i still have 1070 and and rx470/480/570/580 working many years i never pushed them to max.every year i change paste and they are in good shapes.
 If you see cards burnt in black and fans destroyed it is people who are running for ROI fast i guess. otherwise cards are very solid i never had in my hand burnt mosfets how people manage burn them. once i saw memory chips were burnt into black, i wonder what kind of OC they did to get memory chips get burnt so badly.

[TheMimic1]

I'd rather get the best hashrate out of my gpu to get closer to my ROI than been worry about life span, your card can't last a lifetime for mining because difficultly grows bigger every year meaning at one point it will become obsolete, avoid high overclocking and use the most stable settings possible, that's all you'd ever need

There is an issue with this line of thought, do you maximize the residual value of a GPU? or cook the card? Probably the two best answers to this question are the 480 GB nitro+ sapphire card and the EVGA 1080 ti. Both have held up in value, still do a decent amount of hashing and were not too bad cost wise. The 1080ti is particularly noteworthy because it has/had good mining flexibility.

One other point to note is that higher efficiencies (W/Mhs) are at the lower MSI AB settings. There is a message there.

Yes exactly i still have 1070 and and rx470/480/570/580 working many years i never pushed them to max.every year i change paste and they are in good shapes.
 If you see cards burnt in black and fans destroyed it is people who are running for ROI fast i guess. otherwise cards are very solid i never had in my hand burnt mosfets how people manage burn them. once i saw memory chips were burnt into black, i wonder what kind of OC they did to get memory chips get burnt so badly.

Hi batsonxl I'm running a rx580 8gb memory gpu at 2200 Mclock, this is the only way I am able to achieve 30.45 MH on the GPU without BIOS modding, is this a safe memory clock overclocking for long term mining?

[ishansardar]

Miners or not but I prefer running the cards in a way so that it last the longest time. I mean yeah you get your roi sooner but why risk the card? specially if you can game with it after you are done getting your roi + profits. I bought a 1080Ti like 3 years ago and I gamed the hell out of it for 3 long years and I even started mining on it when I don't game and it gives me around a good 43Mhs+. So there's no point ruining your card that like. after 3 years of gaming and a few months of mining my card is still perfect and I can still sell it for like £500. So yeah I personally think keeping GPU health good is better than rushing to ROI.

[batsonxl]

I'd rather get the best hashrate out of my gpu to get closer to my ROI than been worry about life span, your card can't last a lifetime for mining because difficultly grows bigger every year meaning at one point it will become obsolete, avoid high overclocking and use the most stable settings possible, that's all you'd ever need

There is an issue with this line of thought, do you maximize the residual value of a GPU? or cook the card? Probably the two best answers to this question are the 480 GB nitro+ sapphire card and the EVGA 1080 ti. Both have held up in value, still do a decent amount of hashing and were not too bad cost wise. The 1080ti is particularly noteworthy because it has/had good mining flexibility.

One other point to note is that higher efficiencies (W/Mhs) are at the lower MSI AB settings. There is a message there.

Yes exactly i still have 1070 and and rx470/480/570/580 working many years i never pushed them to max.every year i change paste and they are in good shapes.
 If you see cards burnt in black and fans destroyed it is people who are running for ROI fast i guess. otherwise cards are very solid i never had in my hand burnt mosfets how people manage burn them. once i saw memory chips were burnt into black, i wonder what kind of OC they did to get memory chips get burnt so badly.

Hi batsonxl I'm running a rx580 8gb memory gpu at 2200 Mclock, this is the only way I am able to achieve 30.45 MH on the GPU without BIOS modding, is this a safe memory clock overclocking for long term mining?

i dont know which miner you are you sing. Without bios mode im doing 480/580 30.60mhs on phoenix miner with this : -cclock 1150 -cvddc 900 -mclock 2070 -mvddc 920 rxboost 1 -mt 1 driver version is 20.11.1 some gpu can run lover voltage.

[Commie]

My few cents:

1. It's NOT the core temperature you should worry about. Memory temp. and hotspot temp. are what you want to see. Unfortunately, there's no way to see them if you run Linux system. On Windows, run hwinfo64. 

2. If you have reasons to think your memory chips are somewhat faulty then rum MATS/MODS test (30XX cards need version 400+) and see if there are any errors. If errors do exist then the only way to fix them is to replace memory chips. Can be expensive, 1Gb chip is now $50-$100.