I have a Gigabyte Sapphire Pulse RX 5600 XT (6 GB, Navi 10) that has served me well. In my advancing years, however, I'm done trying to keep AMPGPU drivers in sync with Ubuntu updates, upgrades, etc. To make my life easier, I have replaced it with an NVIDIA RTX 4070. The AMD card is free to a good home. DM me if interested.
Ideas are not fixed, nor should they be; we live in model-dependent reality.
I seem to be having a problem putting 2 AMD gpu's in a Windows box, both gpu's are model 580's but they are different makes...I installed the software once and Boinc sees both gpu's as the correct models etc but when crunching the AMD software is always crashing on one of the gpus'. Does that mean I need to load it a 2nd time for the 2nd gpu? Or can you think of something else that could be wrong? Both gpu's worked as individual gpu's when they were the only gpu in a box, either Linux or Windows. Yes there is a full slot between the gpu's and no they aren't running hot at all. There are no cpu tasks running so all 8 cpu cores are for the gpu's to use.
just a thought - have you told BOINC to use ALL GPUs ?
Usually you don't need to load GPU drivers for each new GPU of the same type.
But I found it better, that after putting in a new GPU it has helped to re-do the
load of the driver.
AND don't forget to do a complete CLEAN GPU driver install.
This is my experience on Windows and Nvidia.
What do you mean by crash ? BSOD, compute error or what .... ?
S-F-V
I built the pc out of used parts so both gpu's were in the pc when I first loaded Windows, so yes a clean install of the AMD software, but only once.
I did not think about the 'use all gpu's' line so that will go in today.
Windows does not crash, Boinc does not crash the AMD software crashes but only for the 2nd gpu, AMD puts a big red box on my screen asking if I want to report the reason for the error but there are no details and every line wants something from ME telling them why it crashed....I have no clue WHY it crashed!! The pc has 24gb of DDR4 memory and like I said is not running any cpu tasks except for 2 NCI projects, Ithena Measurements and wuprop. Ithena Measurements tests the internet from point a to point b which is then used against internet providers when they say '90% of our people have high speed internet' when in fact they don't. Wuprop tracks the hours you crunch by the app, you must run it on every pc, so if you run both cpu and gpu tasks here at Einstein you could be tracking 3 or more applications at the same time. They then give badges based on the number of hours you run each app.
To be clear the 1st gpu keeps on crunching, just the AMD Software for the 2nd gpu crashes, and no I did not load the AMD Software twice as I've haven't had to in a long time. Boinc does recognize the 2nd gpu as different from the 1st one and both are listed in the Event Log when Boinc starts.
Windows does not crash, Boinc does not crash the AMD software crashes but only for the 2nd gpu, AMD puts a big red box on my screen asking if I want to report the reason for the error but there are no details and every line wants something from ME telling them why it crashed....I have no clue WHY it crashed!!
What happens if you "rotate" the gpus? Does the crashing gpu follow the card or stay in the same slot? If it stays in the same slot try cleaning the slot by rubbing alcohol/microfilament cloth for 3 cycles. Insert the gpu three times with a fresh microfilament clean each time.
If the slot is set to "auto" try setting it to a fixed speed like "gen3" or "gen2".
HTH, keep talking to us.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)
Windows does not crash, Boinc does not crash the AMD software crashes but only for the 2nd gpu, AMD puts a big red box on my screen asking if I want to report the reason for the error but there are no details and every line wants something from ME telling them why it crashed....I have no clue WHY it crashed!!
What happens if you "rotate" the gpus? Does the crashing gpu follow the card or stay in the same slot? If it stays in the same slot try cleaning the slot by rubbing alcohol/microfilament cloth for 3 cycles. Insert the gpu three times with a fresh microfilament clean each time.
If the slot is set to "auto" try setting it to a fixed speed like "gen3" or "gen2".
HTH, keep talking to us.
Tom M
I kinda rotated them, I rebooted after adding the line to use all gpu's in the cc_config file and noticed that one of the 580's fans were not spinning so I swapped them around and the pc wouldn't boot. So I took both 580's out and put in an AMD 5600XT and it booted right up and is crunching workunits. Today I will add in one of the 580's as the 2nd gpu and see if it runs the tasks and if so let it run for awhile to ensure it keeps working. If it's still running tomorrow morning then I will swap in the other 580 and see if the same thing happens.
My thought was also a BIOS setting on the 2nd slot, so I will clean it before I restart the pc, thanks Tom!!
I have the book but don't know where to look, it has 2 of the same color pcie slots that I'm using for the gpu's and several black ones that aren't being used by me. I have put network cards in one of the black ones in the past but one of the Windows updates fixed the onboard one so it works at 1gb now.
I have a Gigabyte Sapphire
)
I have a Gigabyte Sapphire Pulse RX 5600 XT (6 GB, Navi 10) that has served me well. In my advancing years, however, I'm done trying to keep AMPGPU drivers in sync with Ubuntu updates, upgrades, etc. To make my life easier, I have replaced it with an NVIDIA RTX 4070. The AMD card is free to a good home. DM me if interested.
Ideas are not fixed, nor should they be; we live in model-dependent reality.
23.04.2024 08:49:29 |
)
23.04.2024 08:49:29 | Einstein@Home | update requested by user
23.04.2024 08:49:30 | | [proxy] HTTP_OP::no_proxy_for_url(): https://einsteinathome.org/rss_main.php
23.04.2024 08:49:30 | | [proxy] returning false
23.04.2024 08:49:35 | Einstein@Home | Sending scheduler request: Requested by user.
23.04.2024 08:49:35 | Einstein@Home | Reporting 123 completed tasks
23.04.2024 08:49:35 | Einstein@Home | Requesting new tasks for AMD/ATI GPU
23.04.2024 08:49:35 | | [proxy] HTTP_OP::no_proxy_for_url(): https://scheduler.einsteinathome.org/EinsteinAtHome_cgi/cgi
23.04.2024 08:49:35 | | [proxy] returning false
23.04.2024 08:49:36 | | Project communication failed: attempting access to reference site
23.04.2024 08:49:36 | | [proxy] HTTP_OP::no_proxy_for_url(): https://www.google.com/
23.04.2024 08:49:36 | | [proxy] returning false
23.04.2024 08:49:36 | Einstein@Home | Scheduler request to https://scheduler.einsteinathome.org/EinsteinAtHome_cgi/cgi failed: SSL connect error
23.04.2024 08:49:37 | | Internet access OK - project servers may be temporarily down.
I seem to be having a problem
)
I seem to be having a problem putting 2 AMD gpu's in a Windows box, both gpu's are model 580's but they are different makes...I installed the software once and Boinc sees both gpu's as the correct models etc but when crunching the AMD software is always crashing on one of the gpus'. Does that mean I need to load it a 2nd time for the 2nd gpu? Or can you think of something else that could be wrong? Both gpu's worked as individual gpu's when they were the only gpu in a box, either Linux or Windows. Yes there is a full slot between the gpu's and no they aren't running hot at all. There are no cpu tasks running so all 8 cpu cores are for the gpu's to use.
thanks mikey
mikey: just a thought -
)
mikey:
just a thought - have you told BOINC to use ALL GPUs ?
Usually you don't need to load GPU drivers for each new GPU of the same type.
But I found it better, that after putting in a new GPU it has helped to re-do the
load of the driver.
AND don't forget to do a complete CLEAN GPU driver install.
This is my experience on Windows and Nvidia.
What do you mean by crash ? BSOD, compute error or what .... ?
S-F-V
San-Fernando-Valley
)
I built the pc out of used parts so both gpu's were in the pc when I first loaded Windows, so yes a clean install of the AMD software, but only once.
I did not think about the 'use all gpu's' line so that will go in today.
Windows does not crash, Boinc does not crash the AMD software crashes but only for the 2nd gpu, AMD puts a big red box on my screen asking if I want to report the reason for the error but there are no details and every line wants something from ME telling them why it crashed....I have no clue WHY it crashed!! The pc has 24gb of DDR4 memory and like I said is not running any cpu tasks except for 2 NCI projects, Ithena Measurements and wuprop. Ithena Measurements tests the internet from point a to point b which is then used against internet providers when they say '90% of our people have high speed internet' when in fact they don't. Wuprop tracks the hours you crunch by the app, you must run it on every pc, so if you run both cpu and gpu tasks here at Einstein you could be tracking 3 or more applications at the same time. They then give badges based on the number of hours you run each app.
To be clear the 1st gpu keeps on crunching, just the AMD Software for the 2nd gpu crashes, and no I did not load the AMD Software twice as I've haven't had to in a long time. Boinc does recognize the 2nd gpu as different from the 1st one and both are listed in the Event Log when Boinc starts.
mikey
mikey wrote: I did not
)
Thanks for the infos
hope the use all gpus will work for you ...
S-F-V
mikey wrote:Windows does
)
What happens if you "rotate" the gpus? Does the crashing gpu follow the card or stay in the same slot? If it stays in the same slot try cleaning the slot by rubbing alcohol/microfilament cloth for 3 cycles. Insert the gpu three times with a fresh microfilament clean each time.
If the slot is set to "auto" try setting it to a fixed speed like "gen3" or "gen2".
HTH, keep talking to us.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor)
Tom M wrote: mikey
)
I kinda rotated them, I rebooted after adding the line to use all gpu's in the cc_config file and noticed that one of the 580's fans were not spinning so I swapped them around and the pc wouldn't boot. So I took both 580's out and put in an AMD 5600XT and it booted right up and is crunching workunits. Today I will add in one of the 580's as the 2nd gpu and see if it runs the tasks and if so let it run for awhile to ensure it keeps working. If it's still running tomorrow morning then I will swap in the other 580 and see if the same thing happens.
My thought was also a BIOS setting on the 2nd slot, so I will clean it before I restart the pc, thanks Tom!!
What about the Mainboard
)
What about the Mainboard (BIOS).
Does it support this GPU combination.
maeax wrote: What about the
)
I have the book but don't know where to look, it has 2 of the same color pcie slots that I'm using for the gpu's and several black ones that aren't being used by me. I have put network cards in one of the black ones in the past but one of the Windows updates fixed the onboard one so it works at 1gb now.