Bitcoin Forum
October 24, 2021, 10:29:29 AM *
News: Latest Bitcoin Core release: 22.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: Vega 64 Concerns  (Read 212 times)
sumitc1989
Newbie
*
Offline Offline

Activity: 33
Merit: 0


View Profile
January 25, 2018, 10:37:32 AM
Last edit: January 25, 2018, 11:00:50 AM by sumitc1989
 #1

Hi All,

Recently i got 12 vega 64's and mining xmr on cast miner with blockchain drivers.

I tried using every video + articles tweaks to make my vega power efficient at the same time hash more but almost every video tweak ended up dropping my hash rates from 2020 to 2000 or even further between 1990-1950. 1st card - 2010-2020, 2nd card 1990-2000, 3rd card 1950-1980 and so on.
Allthough all my cards having the same wattman settings and vega registries configured (got them downloaded online) yet all aren't staying stable at 2000+ hashes. I can understand that 1 card always performs less then the rest but considering all cards defer in hashrates by 20-50 is really frustrating to see when one pays 500$ per card.

I eventually ended up tweaking with my own settings and i achieved hashrates between 2050-2070 with lowest being 2020-2030 (last card in series) but this was possible only after increasing the power limit by 15% and now my total power draw is 1950w Cry even though the vega's are hashing above 2020.

I even reached 3000+ hashes per card but with errors.

https://image.ibb.co/fR4YMw/IMG_20180122_001815716.jpg

https://image.ibb.co/imVUuG/IMG_20180121_223423712.jpg

My wattman settings :

Clock
min1262 max1407
Volt
min1150 max1150

Memory
1100Mhz
Volt
850Mhz

Fan
min3500 max4900

Power limit
+15%

My concerns are :

1 - How long my cards with above tweaks would last as the temps are between 70-75 degrees C. Are these temps safe and normal for the vegas & HBM ? What are the safest temps for vega to last 2+ years ?

2 - Does anyone has any settings that can yield higher hashes then 2000 or 2020 + with lower tdp's ? (I tried the youtube videos settings but failed to keep the hashes stable even though tdp & temps were under control)

3 - Lastly does limiting the temperature settings in wattman to min 40 max 55 or any other figures would cause the vegas to crash/malfunction/deteriorate as it restricts it limits to go at higher temps ?

Note : I know that vegas perform better undervolting unlike other cards, but this trick just doesn't works with stable 2000+ hashing in my case.

Any suggestions are highly appreciated, pardon if my settings are noobish.

Thanks.
1635071369
Hero Member
*
Offline Offline

Posts: 1635071369

View Profile Personal Message (Offline)

Ignore
1635071369
Reply with quote  #2

1635071369
Report to moderator
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction. Advertise here.
1635071369
Hero Member
*
Offline Offline

Posts: 1635071369

View Profile Personal Message (Offline)

Ignore
1635071369
Reply with quote  #2

1635071369
Report to moderator
1635071369
Hero Member
*
Offline Offline

Posts: 1635071369

View Profile Personal Message (Offline)

Ignore
1635071369
Reply with quote  #2

1635071369
Report to moderator
1635071369
Hero Member
*
Offline Offline

Posts: 1635071369

View Profile Personal Message (Offline)

Ignore
1635071369
Reply with quote  #2

1635071369
Report to moderator
Bare
Member
**
Offline Offline

Activity: 129
Merit: 11


View Profile
January 25, 2018, 01:59:00 PM
 #2

70-75°C - 24/7??
and +15% plimit

I shed a tear when I read that ;( you're raping your cards...wearing them out hard...

I learned my lesson the hard way when one of my hawaiis died in the summer of 2016 while mining for 4 straight days around 70-75°C
thankfully only 1 card died... :/

Since then I'm taking card temps very seriously and precautions making sure my cards work in an adequate environment maybe even more than really needed just in case.

Don't ever run cards above at least 70°C core if you want them to live long and without any potential issues.
I usually run my fijis and hawaiis always around 63°C max.
right now in winter they are below 60°C at all times and that's with fans running at around 60-65% power
so yeah im obviously downclocking, -10% powlimit atm

you have no reason to be dissatisfied with the performance of your vegas especially in regard that you're running them 12 in a single rig

One of the rigs I have is a 6 card rig consisting of 4 fijis(nitro furys all) and 2 hawaiis, best performing and worst performing fijis differentiate about 4-5% in hashrates...it's pure silicon lottery man... there's nothing wrong with your cards hashing, if you apply 4-5% difference between your cards you get the same results you're complaining about.

on the other note, I noticed it's normal to have a little bit worse hashrate across all cards when running larger number of gpus in a pc than usual, although I can't confirm for sure by how much or at all... just my opinion.

so to answer your questions and few free tips:
1. Run your cards below 70°C since your clearly intend to have them working for 2+ years I suggest @ 63°C max. your cards will thank you in profits later as much as they live the longer the better

2. I don't own any vegas unfortunately but the general rule every serious miner applies is downclock/downvolt/tweak until you have stable temps, don't be driven by hashrates,
card temps are FIRST PRIORITY! then according to temps try to get as much hashes from your cards as possible!
Mining is a long-term game, remember that!

3. Probably, wouldn't know for sure, as far as I know less tweaks in wattman means more stable miner, set gpu settings in bat/config files.

Happy mining!
Shnikes101
Full Member
***
Offline Offline

Activity: 255
Merit: 109


View Profile
January 25, 2018, 03:26:43 PM
 #3

Well there's silicon lottery - every card is a little different. $500/card is great in today's market. Powerdraw of 1950/12 = 162 watts/card is not that bad - especially if that's full system #'s.

Have you tried a different miner? Temps are a little on the high side - remember Vega's will throttle when their memory gets too hot.
I'd probably raise my mem volt to 905-950 and drop my core volt to something similar. Assuming you're on blockchain drivers, I wouldn't be messing with wattman at all.


This question would have been perfectly fine in any of the threads below but you know - new thread whatever.
I'd encourage you to do some more thorough reading with the following threads/guides:

Vega thread  https://bitcointalk.org/index.php?topic=2002025.0
Cast thread   https://bitcointalk.org/index.php?topic=2256917.0
vega.miningguides.com

Also reddit - https://www.reddit.com/r/MoneroMining/comments/74hjqn/monero_and_vega_the_definitive_guide/
fanatic26
Hero Member
*****
Offline Offline

Activity: 756
Merit: 560


View Profile
January 25, 2018, 04:34:31 PM
 #4

I have a few thousand cards that run around 70c all day long. That is nowhere near the thermal limit and they will run for a long long time like that. I wouldnt worry about temps at all until they are breaking 80c

Stop buying industrial miners, running them at home, and then complaining about the noise.
Shnikes101
Full Member
***
Offline Offline

Activity: 255
Merit: 109


View Profile
January 25, 2018, 04:43:52 PM
 #5

I have a few thousand cards that run around 70c all day long. That is nowhere near the thermal limit and they will run for a long long time like that. I wouldnt worry about temps at all until they are breaking 80c

Agreed. Below 80c is fine for almost every card. However, we've found that Vega's will throttle when HBM gets too hot - even in the 70's can sometimes impact performance. Crazy, but it happens with these finicky cards was all I was trying to point out.
sumitc1989
Newbie
*
Offline Offline

Activity: 33
Merit: 0


View Profile
January 25, 2018, 07:04:20 PM
 #6

70-75°C - 24/7??
and +15% plimit

I shed a tear when I read that ;( you're raping your cards...wearing them out hard...

I learned my lesson the hard way when one of my hawaiis died in the summer of 2016 while mining for 4 straight days around 70-75°C
thankfully only 1 card died... :/

Since then I'm taking card temps very seriously and precautions making sure my cards work in an adequate environment maybe even more than really needed just in case.

Don't ever run cards above at least 70°C core if you want them to live long and without any potential issues.
I usually run my fijis and hawaiis always around 63°C max.
right now in winter they are below 60°C at all times and that's with fans running at around 60-65% power
so yeah im obviously downclocking, -10% powlimit atm

you have no reason to be dissatisfied with the performance of your vegas especially in regard that you're running them 12 in a single rig

One of the rigs I have is a 6 card rig consisting of 4 fijis(nitro furys all) and 2 hawaiis, best performing and worst performing fijis differentiate about 4-5% in hashrates...it's pure silicon lottery man... there's nothing wrong with your cards hashing, if you apply 4-5% difference between your cards you get the same results you're complaining about.

on the other note, I noticed it's normal to have a little bit worse hashrate across all cards when running larger number of gpus in a pc than usual, although I can't confirm for sure by how much or at all... just my opinion.

so to answer your questions and few free tips:
1. Run your cards below 70°C since your clearly intend to have them working for 2+ years I suggest @ 63°C max. your cards will thank you in profits later as much as they live the longer the better

2. I don't own any vegas unfortunately but the general rule every serious miner applies is downclock/downvolt/tweak until you have stable temps, don't be driven by hashrates,
card temps are FIRST PRIORITY! then according to temps try to get as much hashes from your cards as possible!
Mining is a long-term game, remember that!

3. Probably, wouldn't know for sure, as far as I know less tweaks in wattman means more stable miner, set gpu settings in bat/config files.

Happy mining!

Thanx a ton for the advice bro, I was able to maintain my temps between 60-65 by cooling the ambience in ac. So basically are temps upto 65 degrees safe for the long run or yet you recommend to lower the temps by sacrificing the hashes (undervolting+underclocking) because i was able to achieve stable hashes of 2030+ with 65 degrees max under ac keeping the same old settings of +15% power.
However i am willing to bear the electricity charges for those tempting hashrates Tongue
sumitc1989
Newbie
*
Offline Offline

Activity: 33
Merit: 0


View Profile
January 25, 2018, 07:06:17 PM
 #7

I have a few thousand cards that run around 70c all day long. That is nowhere near the thermal limit and they will run for a long long time like that. I wouldnt worry about temps at all until they are breaking 80c

How long have you been running them at those temps & which cards ?
Any malfunctions observed ?
Bare
Member
**
Offline Offline

Activity: 129
Merit: 11


View Profile
January 26, 2018, 04:52:18 PM
 #8

I have a few thousand cards that run around 70c all day long. That is nowhere near the thermal limit and they will run for a long long time like that. I wouldnt worry about temps at all until they are breaking 80c
I have a few thousand cards that run around 70c all day long. That is nowhere near the thermal limit and they will run for a long long time like that. I wouldnt worry about temps at all until they are breaking 80c

Agreed. Below 80c is fine for almost every card. However, we've found that Vega's will throttle when HBM gets too hot - even in the 70's can sometimes impact performance. Crazy, but it happens with these finicky cards was all I was trying to point out.


Good thing you said "almost" every card, because as you may or may not know every graphic card has it's own temp limits under a workload, older cards such as amd HD5000-7000 series, generally withstand higher temps limits better than newer ones because of the bigger fabrication process as opposed to today's newer cards which are made on much smaller fabrication process.

I know this because my first gpus I ever mined with were 2x hd5970, power hungry beasts back in the days, and guess what? 73-78°C core was nothing for them 24/7, mined for 2 whole years and later sold them in a fully working condition, bought one of my 390x some time later, died after 4 days of mining at those same temps.

few experts told me the same, it's the fab. process hence vegas and other new cards should be treated according to this input too...

70-75°C - 24/7??
and +15% plimit

I shed a tear when I read that ;( you're raping your cards...wearing them out hard...

I learned my lesson the hard way when one of my hawaiis died in the summer of 2016 while mining for 4 straight days around 70-75°C
thankfully only 1 card died... :/

Since then I'm taking card temps very seriously and precautions making sure my cards work in an adequate environment maybe even more than really needed just in case.

Don't ever run cards above at least 70°C core if you want them to live long and without any potential issues.
I usually run my fijis and hawaiis always around 63°C max.
right now in winter they are below 60°C at all times and that's with fans running at around 60-65% power
so yeah im obviously downclocking, -10% powlimit atm

you have no reason to be dissatisfied with the performance of your vegas especially in regard that you're running them 12 in a single rig

One of the rigs I have is a 6 card rig consisting of 4 fijis(nitro furys all) and 2 hawaiis, best performing and worst performing fijis differentiate about 4-5% in hashrates...it's pure silicon lottery man... there's nothing wrong with your cards hashing, if you apply 4-5% difference between your cards you get the same results you're complaining about.

on the other note, I noticed it's normal to have a little bit worse hashrate across all cards when running larger number of gpus in a pc than usual, although I can't confirm for sure by how much or at all... just my opinion.

so to answer your questions and few free tips:
1. Run your cards below 70°C since your clearly intend to have them working for 2+ years I suggest @ 63°C max. your cards will thank you in profits later as much as they live the longer the better

2. I don't own any vegas unfortunately but the general rule every serious miner applies is downclock/downvolt/tweak until you have stable temps, don't be driven by hashrates,
card temps are FIRST PRIORITY! then according to temps try to get as much hashes from your cards as possible!
Mining is a long-term game, remember that!

3. Probably, wouldn't know for sure, as far as I know less tweaks in wattman means more stable miner, set gpu settings in bat/config files.

Happy mining!

Thanx a ton for the advice bro, I was able to maintain my temps between 60-65 by cooling the ambience in ac. So basically are temps upto 65 degrees safe for the long run or yet you recommend to lower the temps by sacrificing the hashes (undervolting+underclocking) because i was able to achieve stable hashes of 2030+ with 65 degrees max under ac keeping the same old settings of +15% power.
However i am willing to bear the electricity charges for those tempting hashrates Tongue

keep the temps below 70°C(at low-mid 60's) at all times and you'll never have to worry about any potential problems, make sure your cards are properly ventilated(or a/c if you can but more electricity costs), make some more distance between them and lastly if needed sacrifice some hashrate (downclock/downvolt) whatever as long as you can maintain these "ok" temps!
what are few hashes more compared to the longer lifespan/health of your gpu?? is it really worth it? and to add, card scarcity is already bad as it is and you don't need yourself even more headache.

of course one more thing, never max your fan rpm, that's also dumb, you gain nothing setting your fans above 75% power(some people say 80%) than heavily wearing your fans out, so I suggest to set them max. 75% fan speed.

after you've made sure your cards are working in these conditions you won't have to worry about any component on your cards, occasionally clean that dust residue between the heatsink fins when you see it accumulating and pretty much that's it

sumitc1989
Newbie
*
Offline Offline

Activity: 33
Merit: 0


View Profile
January 26, 2018, 07:17:38 PM
 #9

I have a few thousand cards that run around 70c all day long. That is nowhere near the thermal limit and they will run for a long long time like that. I wouldnt worry about temps at all until they are breaking 80c
I have a few thousand cards that run around 70c all day long. That is nowhere near the thermal limit and they will run for a long long time like that. I wouldnt worry about temps at all until they are breaking 80c

Agreed. Below 80c is fine for almost every card. However, we've found that Vega's will throttle when HBM gets too hot - even in the 70's can sometimes impact performance. Crazy, but it happens with these finicky cards was all I was trying to point out.


Good thing you said "almost" every card, because as you may or may not know every graphic card has it's own temp limits under a workload, older cards such as amd HD5000-7000 series, generally withstand higher temps limits better than newer ones because of the bigger fabrication process as opposed to today's newer cards which are made on much smaller fabrication process.

I know this because my first gpus I ever mined with were 2x hd5970, power hungry beasts back in the days, and guess what? 73-78°C core was nothing for them 24/7, mined for 2 whole years and later sold them in a fully working condition, bought one of my 390x some time later, died after 4 days of mining at those same temps.

few experts told me the same, it's the fab. process hence vegas and other new cards should be treated according to this input too...

70-75°C - 24/7??
and +15% plimit

I shed a tear when I read that ;( you're raping your cards...wearing them out hard...

I learned my lesson the hard way when one of my hawaiis died in the summer of 2016 while mining for 4 straight days around 70-75°C
thankfully only 1 card died... :/

Since then I'm taking card temps very seriously and precautions making sure my cards work in an adequate environment maybe even more than really needed just in case.

Don't ever run cards above at least 70°C core if you want them to live long and without any potential issues.
I usually run my fijis and hawaiis always around 63°C max.
right now in winter they are below 60°C at all times and that's with fans running at around 60-65% power
so yeah im obviously downclocking, -10% powlimit atm

you have no reason to be dissatisfied with the performance of your vegas especially in regard that you're running them 12 in a single rig

One of the rigs I have is a 6 card rig consisting of 4 fijis(nitro furys all) and 2 hawaiis, best performing and worst performing fijis differentiate about 4-5% in hashrates...it's pure silicon lottery man... there's nothing wrong with your cards hashing, if you apply 4-5% difference between your cards you get the same results you're complaining about.

on the other note, I noticed it's normal to have a little bit worse hashrate across all cards when running larger number of gpus in a pc than usual, although I can't confirm for sure by how much or at all... just my opinion.

so to answer your questions and few free tips:
1. Run your cards below 70°C since your clearly intend to have them working for 2+ years I suggest @ 63°C max. your cards will thank you in profits later as much as they live the longer the better

2. I don't own any vegas unfortunately but the general rule every serious miner applies is downclock/downvolt/tweak until you have stable temps, don't be driven by hashrates,
card temps are FIRST PRIORITY! then according to temps try to get as much hashes from your cards as possible!
Mining is a long-term game, remember that!

3. Probably, wouldn't know for sure, as far as I know less tweaks in wattman means more stable miner, set gpu settings in bat/config files.

Happy mining!

Thanx a ton for the advice bro, I was able to maintain my temps between 60-65 by cooling the ambience in ac. So basically are temps upto 65 degrees safe for the long run or yet you recommend to lower the temps by sacrificing the hashes (undervolting+underclocking) because i was able to achieve stable hashes of 2030+ with 65 degrees max under ac keeping the same old settings of +15% power.
However i am willing to bear the electricity charges for those tempting hashrates Tongue

keep the temps below 70°C(at low-mid 60's) at all times and you'll never have to worry about any potential problems, make sure your cards are properly ventilated(or a/c if you can but more electricity costs), make some more distance between them and lastly if needed sacrifice some hashrate (downclock/downvolt) whatever as long as you can maintain these "ok" temps!
what are few hashes more compared to the longer lifespan/health of your gpu?? is it really worth it? and to add, card scarcity is already bad as it is and you don't need yourself even more headache.

of course one more thing, never max your fan rpm, that's also dumb, you gain nothing setting your fans above 75% power(some people say 80%) than heavily wearing your fans out, so I suggest to set them max. 75% fan speed.

after you've made sure your cards are working in these conditions you won't have to worry about any component on your cards, occasionally clean that dust residue between the heatsink fins when you see it accumulating and pretty much that's it



thanx man that was really helpful Smiley Will surely work my card temps and hashes accordingly as i too agree that life span is more precious then hashes considering the scarcity and gpu price rise.
Bare
Member
**
Offline Offline

Activity: 129
Merit: 11


View Profile
January 28, 2018, 03:13:38 AM
 #10

Alright Smiley glad you learned something, follow these rules and remember,

take care of your cards and the cards will take care of your profits for a long while, I've heard reports from some users cards mining for full 5 years! talk about treating your cards well huh?  Grin
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!