Bitcoin Forum
April 26, 2024, 10:37:04 PM *
News: Latest Bitcoin Core release: 27.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: Vega closed case  (Read 296 times)
Riptide_NVN (OP)
Member
**
Offline Offline

Activity: 294
Merit: 16


View Profile
April 13, 2018, 09:52:19 PM
 #1

I'm trying to mine cryptonight with two asus rog vegas in a closed case.  Air flow and ambient temps are fine.

Fans on both gpu 80%.  One gpu is much warmer than the other which seems common in closed case dual gpu systems.

I have tried using wattman to lower the power -25% and downclock the memory to mem_p3=800;850.  gpu_p6=1212;905 and gpu_p7=1407;950.

I have not tried (yet) to adjust lower pstates.  If I adjust the power setting down further it helps but hash bottoms out near 1080ti territory.

Am I wasting my time trying to get this card to run cooler?
1714171024
Hero Member
*
Offline Offline

Posts: 1714171024

View Profile Personal Message (Offline)

Ignore
1714171024
Reply with quote  #2

1714171024
Report to moderator
1714171024
Hero Member
*
Offline Offline

Posts: 1714171024

View Profile Personal Message (Offline)

Ignore
1714171024
Reply with quote  #2

1714171024
Report to moderator
"Your bitcoin is secured in a way that is physically impossible for others to access, no matter for what reason, no matter how good the excuse, no matter a majority of miners, no matter what." -- Greg Maxwell
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction.
cpmcgrat
Member
**
Offline Offline

Activity: 223
Merit: 21

DCAB


View Profile
April 13, 2018, 09:56:08 PM
 #2

I'm trying to mine cryptonight with two asus rog vegas in a closed case.  Air flow and ambient temps are fine.

Fans on both gpu 80%.  One gpu is much warmer than the other which seems common in closed case dual gpu systems.

I have tried using wattman to lower the power -25% and downclock the memory to mem_p3=800;850.  gpu_p6=1212;905 and gpu_p7=1407;950.

I have not tried (yet) to adjust lower pstates.  If I adjust the power setting down further it helps but hash bottoms out near 1080ti territory.

Am I wasting my time trying to get this card to run cooler?


If they are achieving max hashrate over an extended period of time (or the scaled hashrate you expect for the power settings) then most likely yes.

Member, Dero Community Advisory Board (DCAB)
P00P135
Full Member
***
Offline Offline

Activity: 1123
Merit: 136


View Profile
April 13, 2018, 10:03:45 PM
 #3

I'm trying to mine cryptonight with two asus rog vegas in a closed case.  Air flow and ambient temps are fine.

Fans on both gpu 80%.  One gpu is much warmer than the other which seems common in closed case dual gpu systems.

I have tried using wattman to lower the power -25% and downclock the memory to mem_p3=800;850.  gpu_p6=1212;905 and gpu_p7=1407;950.

I have not tried (yet) to adjust lower pstates.  If I adjust the power setting down further it helps but hash bottoms out near 1080ti territory.

Am I wasting my time trying to get this card to run cooler?



What are the actual temps?
Have you tried disabling the other GPU or swapping pcie slots to rule out added heat?  GPU could just be constantly sucking in super heated back plate air.
why havent you done a custom power play table its pretty basic and worst case scenario you ddu and reinstall drivers.
Riptide_NVN (OP)
Member
**
Offline Offline

Activity: 294
Merit: 16


View Profile
April 13, 2018, 10:47:05 PM
 #4

The warmer card runs memory temps 70-73c. Core around 60c.

I haven't set any lower p states yet but will try if that might help these load temps significantly.

And yah top card has a little over 1/2" clearance from the back of the lower one. Fan blows right on it but doesn't seem to help enough.
CryptoPlay
Jr. Member
*
Offline Offline

Activity: 202
Merit: 2


View Profile
April 14, 2018, 01:08:45 AM
 #5

You need to keep HBM bellow 70 or youll have hash drop pretty fast.

read the guide. You cant change lower P states... only P6 and P7 from core and P3 from mem.
Keep min rpm high.

https://signature.statseb.fr/sig-2065.png
Riptide_NVN (OP)
Member
**
Offline Offline

Activity: 294
Merit: 16


View Profile
April 14, 2018, 02:20:29 AM
 #6

I have the fans at 67% (2485rpm) right now.  Case fans cranked up too.  Higher than that is unbearable.  This is a workstation not a dedicated mining rig.

The warmer of the two Vega hashes 1450 h/s.  The cooler of the two 1900 h/s.  While that is not impressive it is a big step up over the 1080 I had in here before that managed 945 h/s.  Even the warmer/slower of the two Vega is a step up over that.  The 1080 was nearly silent - which was nice.

If I absolutely must remove a card I could mod the case, get a riser, and set it outside the computer.  Will that be necessary to alleviate this situation?  Maybe.

Couple photos here.  As you can see your standard 2.5 slot dual card issue going on.  The side fan does push some good air right over the gap.  Only so much it can do.   Sad



P00P135
Full Member
***
Offline Offline

Activity: 1123
Merit: 136


View Profile
April 14, 2018, 03:57:50 AM
 #7

I have the fans at 67% (2485rpm) right now.  Case fans cranked up too.  Higher than that is unbearable.  This is a workstation not a dedicated mining rig.

The warmer of the two Vega hashes 1450 h/s.  The cooler of the two 1900 h/s.  While that is not impressive it is a big step up over the 1080 I had in here before that managed 945 h/s.  Even the warmer/slower of the two Vega is a step up over that.  The 1080 was nearly silent - which was nice.

If I absolutely must remove a card I could mod the case, get a riser, and set it outside the computer.  Will that be necessary to alleviate this situation?  Maybe.

Couple photos here.  As you can see your standard 2.5 slot dual card issue going on.  The side fan does push some good air right over the gap.  Only so much it can do.   Sad





So let me get this straight, you think because your vega gets 60c that it hashes 1450h/s?  Has it ever hashed 1900+? 
Do you have max temp set too low and its throttling?
Do you know how to disable and re-enable a GPU when it starts to hash low? 
Have you even bothered to read the vega mining guide? http://vega.miningguides.com/
Riptide_NVN (OP)
Member
**
Offline Offline

Activity: 294
Merit: 16


View Profile
April 14, 2018, 04:36:06 AM
 #8

No it hashes 1450 because the miner reports that.  The cooler running card hashes 1900 again per xmr stak.

It isn't throttling. I have the power limit at -50% and it's down clocked. The card runs too hot otherwise.

The hash I am getting is pretty stable. Rebooting the machine and reapplying wattman profiles makes no difference.

I've read the guide yes.  It has some good info but is getting old.  I'm not going to use BC drivers from August of last year on a mixed use system.
P00P135
Full Member
***
Offline Offline

Activity: 1123
Merit: 136


View Profile
April 14, 2018, 04:47:12 AM
 #9

No it hashes 1450 because the miner reports that.  The cooler running card hashes 1900 again per xmr stak.

It isn't throttling. I have the power limit at -50% and it's down clocked. The card runs too hot otherwise.

The hash I am getting is pretty stable. Rebooting the machine and reapplying wattman profiles makes no difference.

I've read the guide yes.  It has some good info but is getting old.  I'm not going to use BC drivers from August of last year on a mixed use system.

What is too hot in your mind?  The only temp you have given us is 60c core which is perfectly fine.  When you start getting over 70c is when you might run into stability problems with OC'd HBM2. 
You don't need to use block chain drivers anymore with the newest AMD drivers.
Riptide_NVN (OP)
Member
**
Offline Offline

Activity: 294
Merit: 16


View Profile
April 14, 2018, 04:52:11 AM
 #10

Post number two.

Even with the power limit down -50% and memory at 800mhz. Right now hbm temps are in mid 60s.

If I put the power limit back to zero and clocks back to 945 hbm temps hit close to 80.  Fan at 100% isn't enough then to keep him temps in the 60s.
szafa
Hero Member
*****
Offline Offline

Activity: 812
Merit: 500


View Profile
April 14, 2018, 05:16:57 AM
 #11

I dont have solution about hbm memory in gpu memory.Maybe broken,maybe replace thermal gpu,radiators.
CryptoPlay
Jr. Member
*
Offline Offline

Activity: 202
Merit: 2


View Profile
April 14, 2018, 05:21:58 AM
 #12

Your card doing 1450 have HBcc disable. I think you should think about min rpm close to 3000 with closed case.
Your core must stay bellow 60 to make sure your HBM is bellow 70. This is the reason of the hash in the 1450 card....to get your hash back up, youll need to disable and re enable your card in device manager. After that, re apply the OverdriveNTool profile in both.
If still not 2k, and if you do have wattman installed, hit the HBCC switch there...only of the card with low hash. Close wattman and never open it again.
https://mega.nz/#!zYh3BCRS!fSLdmiQ7pUxzgZuISpiMaJ5RLvQoolJmHVuKXljqS0Y
I did this pack to a friend. It do have some soft power tables, 2 excell files to edit power tables and 1 startup.bat file.
Check this startup.bat and edit it with your file path. Youll need devcon and this bat in the same foldar of OverdriveNTool....


You need to read http://vega.miningguides.com/p/tuning.html   to use this stuff. Is easier if you dont do it manually. check just to enable and disable your cards automatically.


https://signature.statseb.fr/sig-2065.png
P00P135
Full Member
***
Offline Offline

Activity: 1123
Merit: 136


View Profile
April 14, 2018, 05:33:29 AM
 #13

Your card doing 1450 have HBcc disable. I think you should think about min rpm close to 3000 with closed case.
Your core must stay bellow 60 to make sure your HBM is bellow 70. This is the reason of the hash in the 1450 card....to get your hash back up, youll need to disable and re enable your card in device manager. After that, re apply the OverdriveNTool profile in both.
If still not 2k, and if you do have wattman installed, hit the HBCC switch there...only of the card with low hash. Close wattman and never open it again.
https://mega.nz/#!zYh3BCRS!fSLdmiQ7pUxzgZuISpiMaJ5RLvQoolJmHVuKXljqS0Y
I did this pack to a friend. It do have some soft power tables, 2 excell files to edit power tables and 1 startup.bat file.
Check this startup.bat and edit it with your file path. Youll need devcon and this bat in the same foldar of OverdriveNTool....


You need to read http://vega.miningguides.com/p/tuning.html   to use this stuff. Is easier if you dont do it manually. check just to enable and disable your cards automatically.



You haven't needed devcon with JJ's script for a while now.
CryptoPlay
Jr. Member
*
Offline Offline

Activity: 202
Merit: 2


View Profile
April 14, 2018, 05:38:03 AM
 #14

And you dont have fee with the old bat....
This simple script makes everything automated with i click....zero fee.

JJ not working in my rig, dont know why but it do not apply the OverdriveNTool profile and it could damage my cards like that.
Im not using it to start the miner....only to disable, enable and OT profile...

cd C:\your_folder

timeout /t 5

devcon.exe disable "PCI\VEN_1002&DEV_687F"

timeout /t 10

devcon.exe enable "PCI\VEN_1002&DEV_687F"

timeout /t 5

OverdriveNTool.exe -r1 -r2 -r3 -r4 -r5 -r6 -p1Vega5 -p2Vega3 -p3Vega5 -p4Vega5 -p5Vega5 -p6Vega5

timeout /t 5

cd C:\Your_folder\xmr-stak-win64

xmr-stak.exe

https://signature.statseb.fr/sig-2065.png
Riptide_NVN (OP)
Member
**
Offline Offline

Activity: 294
Merit: 16


View Profile
April 14, 2018, 01:41:47 PM
 #15

There is no way these cards are ever going to hash 1900+ in this computer unless I have the mining shut down entirely when I'm home.  3000rpm is STUPID loud with these cards.

I'm fine with getting each card to 1800 and being able to restart the computer and play games.  Apply profiles when I'm done and mine.  That is the goal here not max absolute hash.  The temps on the one card are making this difficult.

And to top it off HWINFO64 won't work on my system.  Crashes when detecting the video adapters.  I've unchecked the wake options.  Doesn't help.  The 32bit one works however only seems to detect one GPU.

I still don't understand what the point of the lower p states are.  They have no effect on the card while it is in actual use gaming or mining.  So why do people even bother with them?
P00P135
Full Member
***
Offline Offline

Activity: 1123
Merit: 136


View Profile
April 14, 2018, 03:38:30 PM
 #16

3000rpm is STUPID loud with these cards.


lol ok  Roll Eyes
CryptoPlay
Jr. Member
*
Offline Offline

Activity: 202
Merit: 2


View Profile
April 14, 2018, 04:24:50 PM
 #17



I still don't understand what the point of the lower p states are.  They have no effect on the card while it is in actual use gaming or mining.  So why do people even bother with them?

Agreed with you and i did not changed my lower P-states much. Just a little but your card, staying at 100% dont remain at P7 all the time. Just to let you know that.

3000rpm is to 24/7 operations AND to those who set MEM to 1100hz. IF you can keep your CORE temp bellow 62, your HBM will probably stay bellow 70 so the driver wont disable HBCC and your hash will stay up.

Once your hash goes down...You have the option to run the script(1click) or get into the wattman, hit the HBCC switch on again and reapply the OT profile.

I dont know your card but the ref design with 3200rpm is okay.

https://signature.statseb.fr/sig-2065.png
Riptide_NVN (OP)
Member
**
Offline Offline

Activity: 294
Merit: 16


View Profile
April 14, 2018, 05:03:26 PM
 #18

My 6 GPU rig downstairs with 1080ti @ 55-65% fan speeds doesn't put out the amount of noise these 2 asus rog vega produce with fans spun up to 3000rpm.  I guess some people have a different definition of loud?  YMMV.  Smiley

Using Wattman I have got a couple profiles set up that keep hash to a (for me) acceptable level.  The cooler running GPU runs memory at 1100mhz with the core at around 1357mhz.  1900 h/s.  The warmer card has memory clocked back to 945mhz and core at 1012mhz.  1450 h/s.  Power is at 0%.  Core temps on both are around 53c and HBM around 60c.  I expect as the house warms up they will climb and get up to around 65c HBM temps.

This is with the fans at 3000rpm.  I am going to make a second set of profiles for when I'm actually using the PC or watching TV in this room.  That will be with fans around 2500rpm and with both GPU significantly clocked back down.  Hash will be terrible.  It will however be better than completely turning the miner off.  Eventually I will make a third set of profiles for use when I'm playing battlefront or far cry.

I really don't care if the cards aren't hitting 2000 h/s.  I replaced a 1080ti in this computer that was hashing 950 h/s.  Which was good for a 1080ti.  These two cards even being well off the point of optimized mainly due to the heat and noise issue are still a huge step up in production over the 1080ti.

If I could just get HWINFO64 to work for me..  Right now I'm forced to use gpu-z.
CryptoPlay
Jr. Member
*
Offline Offline

Activity: 202
Merit: 2


View Profile
April 14, 2018, 07:24:33 PM
 #19

Just to let you know.
The default 2000h config is Core 1408 and mem 1100 both with voltage around 900mv.
Ive set power to -22% ...above that itll hit the hash.

Reducing your voltage(p7 and memP3) with OT will decrease the temp.
Setting min rpm to 3000 will make the fan work at 3200rpm.

https://signature.statseb.fr/sig-2065.png
Riptide_NVN (OP)
Member
**
Offline Offline

Activity: 294
Merit: 16


View Profile
April 14, 2018, 08:50:46 PM
 #20

I'm about 100 h/s short of that with those same clocks and voltages and 0% power. Not sure why it's slower maybe because I'm using the adrenaline driver or could be the 1800 intensity I'm using.

I have OT.  I like wattman better.
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!