Hello.
In cc_config.xml I have the line <use_all_gpus> 1 </use_all_gpus>.
They are visible in the device management and BM. But the tasks do not work on them.

What to do to run tasks on them also on Nvidia Tesla K80?
In another computer I also have a Tesla K80 with Titan X Maxwell and everything works as it should.

Host:
https://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=926996
Extract from the event log:

28.04.2022 23:05:24 | | Starting BOINC client version 7.16.20 for windows_x86_64
28.04.2022 23:05:24 | | log flags: file_xfer, sched_ops, task
28.04.2022 23:05:24 | | Libraries: libcurl/7.47.1 OpenSSL/1.0.2s zlib/1.2.8
28.04.2022 23:05:24 | | Data directory: C:\ProgramData\BOINC
28.04.2022 23:05:24 | | Running under account chr80
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 0: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 1: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 0: AMD Radeon (TM) R9 Fury Series (driver version 3240.6, device version OpenCL 2.0 AMD-APP (3240.6), 4096MB, 4096MB available, 7782 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 1: AMD FirePro W7000 (driver version 3240.6, device version OpenCL 1.2 AMD-APP (3240.6), 4096MB, 4096MB available, 2432 GFLOPS peak)
28.04.2022 23:05:27 | | Windows processor group 0: 36 processors
28.04.2022 23:05:27 | | Host name: BOINCPOLAND5
28.04.2022 23:05:27 | | Processor: 36 GenuineIntel Intel(R) Core(TM) i9-7980XE CPU @ 2.60GHz [Family 6 Model 85 Stepping 4]
28.04.2022 23:05:27 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx tm2 dca pbe fsgsbase bmi1 hle smep bmi2
28.04.2022 23:05:27 | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.19042.00)
28.04.2022 23:05:27 | | Memory: 63.70 GB physical, 73.20 GB virtual
28.04.2022 23:05:27 | | Disk: 930.94 GB total, 820.18 GB free
28.04.2022 23:05:27 | | Local time is UTC +2 hours
28.04.2022 23:05:27 | | No WSL found.
28.04.2022 23:05:27 | | VirtualBox version: 6.0.14

As a "noob", I'm wondering how you are managing to install one driver that covers two different GPU-makers ?
Have a nice day !

Hello.
In cc_config.xml I have the line <use_all_gpus> 1 </use_all_gpus>.
They are visible in the device management and BM. But the tasks do not work on them.

What to do to run tasks on them also on Nvidia Tesla K80?
In another computer I also have a Tesla K80 with Titan X Maxwell and everything works as it should.

Host:
https://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=926996
Extract from the event log:

28.04.2022 23:05:24 | | Starting BOINC client version 7.16.20 for windows_x86_64
28.04.2022 23:05:24 | | log flags: file_xfer, sched_ops, task
28.04.2022 23:05:24 | | Libraries: libcurl/7.47.1 OpenSSL/1.0.2s zlib/1.2.8
28.04.2022 23:05:24 | | Data directory: C:\ProgramData\BOINC
28.04.2022 23:05:24 | | Running under account chr80
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 0: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 1: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 0: AMD Radeon (TM) R9 Fury Series (driver version 3240.6, device version OpenCL 2.0 AMD-APP (3240.6), 4096MB, 4096MB available, 7782 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 1: AMD FirePro W7000 (driver version 3240.6, device version OpenCL 1.2 AMD-APP (3240.6), 4096MB, 4096MB available, 2432 GFLOPS peak)
28.04.2022 23:05:27 | | Windows processor group 0: 36 processors
28.04.2022 23:05:27 | | Host name: BOINCPOLAND5
28.04.2022 23:05:27 | | Processor: 36 GenuineIntel Intel(R) Core(TM) i9-7980XE CPU @ 2.60GHz [Family 6 Model 85 Stepping 4]
28.04.2022 23:05:27 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx tm2 dca pbe fsgsbase bmi1 hle smep bmi2
28.04.2022 23:05:27 | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.19042.00)
28.04.2022 23:05:27 | | Memory: 63.70 GB physical, 73.20 GB virtual
28.04.2022 23:05:27 | | Disk: 930.94 GB total, 820.18 GB free
28.04.2022 23:05:27 | | Local time is UTC +2 hours
28.04.2022 23:05:27 | | No WSL found.
28.04.2022 23:05:27 | | VirtualBox version: 6.0.14

To me this looks like a problem as well:

28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 0: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 1: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 0: AMD Radeon (TM) R9 Fury Series (driver version 3240.6, device version OpenCL 2.0 AMD-APP (3240.6), 4096MB, 4096MB available, 7782 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 1: AMD FirePro W7000 (driver version 3240.6, device version OpenCL 1.2 AMD-APP (3240.6), 4096MB, 4096MB available, 2432 GFLOPS peak)

You have how many gpu's in the machine...2, 3 or 4? yet Boinc only has them assigned as "gpu0" and "gpu1" AND they seem to be cross connecting the slots! How can you have an Nvidia Tesla is in the gpu0 slot on one line and then a couple lines later have an AMD Fury in the gpu0 slot?

Each slot should get it's own gpu and start at gpu0 and go thru gpu9 if you have 10 gpu's in the machine but it seems your gpu's are sharing slots and that's probably confusing the heck out of Boinc. Are you by chance using a riser card with multiple cpu's on it that then connect to the same slot of the motherboard? If so you need to find someone else doing something similar to figure this out.

As a "noob", I'm wondering how you are managing to install one driver that covers two different GPU-makers ?
Have a nice day !

Normally no problems.

Hello.
In cc_config.xml I have the line <use_all_gpus> 1 </use_all_gpus>.
They are visible in the device management and BM. But the tasks do not work on them.

What to do to run tasks on them also on Nvidia Tesla K80?
In another computer I also have a Tesla K80 with Titan X Maxwell and everything works as it should.

Host:
https://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=926996
Extract from the event log:

28.04.2022 23:05:24 | | Starting BOINC client version 7.16.20 for windows_x86_64
28.04.2022 23:05:24 | | log flags: file_xfer, sched_ops, task
28.04.2022 23:05:24 | | Libraries: libcurl/7.47.1 OpenSSL/1.0.2s zlib/1.2.8
28.04.2022 23:05:24 | | Data directory: C:\ProgramData\BOINC
28.04.2022 23:05:24 | | Running under account chr80
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 0: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 1: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 0: AMD Radeon (TM) R9 Fury Series (driver version 3240.6, device version OpenCL 2.0 AMD-APP (3240.6), 4096MB, 4096MB available, 7782 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 1: AMD FirePro W7000 (driver version 3240.6, device version OpenCL 1.2 AMD-APP (3240.6), 4096MB, 4096MB available, 2432 GFLOPS peak)
28.04.2022 23:05:27 | | Windows processor group 0: 36 processors
28.04.2022 23:05:27 | | Host name: BOINCPOLAND5
28.04.2022 23:05:27 | | Processor: 36 GenuineIntel Intel(R) Core(TM) i9-7980XE CPU @ 2.60GHz [Family 6 Model 85 Stepping 4]
28.04.2022 23:05:27 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx tm2 dca pbe fsgsbase bmi1 hle smep bmi2
28.04.2022 23:05:27 | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.19042.00)
28.04.2022 23:05:27 | | Memory: 63.70 GB physical, 73.20 GB virtual
28.04.2022 23:05:27 | | Disk: 930.94 GB total, 820.18 GB free
28.04.2022 23:05:27 | | Local time is UTC +2 hours
28.04.2022 23:05:27 | | No WSL found.
28.04.2022 23:05:27 | | VirtualBox version: 6.0.14

To me this looks like a problem as well:

28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 0: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 1: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 0: AMD Radeon (TM) R9 Fury Series (driver version 3240.6, device version OpenCL 2.0 AMD-APP (3240.6), 4096MB, 4096MB available, 7782 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 1: AMD FirePro W7000 (driver version 3240.6, device version OpenCL 1.2 AMD-APP (3240.6), 4096MB, 4096MB available, 2432 GFLOPS peak)

You have how many gpu's in the machine...2, 3 or 4? yet Boinc only has them assigned as "gpu0" and "gpu1" AND they seem to be cross connecting the slots! How can you have an Nvidia Tesla is in the gpu0 slot on one line and then a couple lines later have an AMD Fury in the gpu0 slot?

Each slot should get it's own gpu and start at gpu0 and go thru gpu9 if you have 10 gpu's in the machine but it seems your gpu's are sharing slots and that's probably confusing the heck out of Boinc. Are you by chance using a riser card with multiple cpu's on it that then connect to the same slot of the motherboard? If so you need to find someone else doing something similar to figure this out.

This is Evga X299 Dark Motherboard (151-SX-E299-KR).
3xGPU:
- AMD FirePro W7000
- AMD Radeon R9 Fury
- NVIDIA Tesla K80
Everything is without PCIe Risers.

Hello.
In cc_config.xml I have the line <use_all_gpus> 1 </use_all_gpus>.
They are visible in the device management and BM. But the tasks do not work on them.

What to do to run tasks on them also on Nvidia Tesla K80?
In another computer I also have a Tesla K80 with Titan X Maxwell and everything works as it should.

Host:
https://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=926996
Extract from the event log:

28.04.2022 23:05:24 | | Starting BOINC client version 7.16.20 for windows_x86_64
28.04.2022 23:05:24 | | log flags: file_xfer, sched_ops, task
28.04.2022 23:05:24 | | Libraries: libcurl/7.47.1 OpenSSL/1.0.2s zlib/1.2.8
28.04.2022 23:05:24 | | Data directory: C:\ProgramData\BOINC
28.04.2022 23:05:24 | | Running under account chr80
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 0: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 1: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 0: AMD Radeon (TM) R9 Fury Series (driver version 3240.6, device version OpenCL 2.0 AMD-APP (3240.6), 4096MB, 4096MB available, 7782 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 1: AMD FirePro W7000 (driver version 3240.6, device version OpenCL 1.2 AMD-APP (3240.6), 4096MB, 4096MB available, 2432 GFLOPS peak)
28.04.2022 23:05:27 | | Windows processor group 0: 36 processors
28.04.2022 23:05:27 | | Host name: BOINCPOLAND5
28.04.2022 23:05:27 | | Processor: 36 GenuineIntel Intel(R) Core(TM) i9-7980XE CPU @ 2.60GHz [Family 6 Model 85 Stepping 4]
28.04.2022 23:05:27 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx tm2 dca pbe fsgsbase bmi1 hle smep bmi2
28.04.2022 23:05:27 | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.19042.00)
28.04.2022 23:05:27 | | Memory: 63.70 GB physical, 73.20 GB virtual
28.04.2022 23:05:27 | | Disk: 930.94 GB total, 820.18 GB free
28.04.2022 23:05:27 | | Local time is UTC +2 hours
28.04.2022 23:05:27 | | No WSL found.
28.04.2022 23:05:27 | | VirtualBox version: 6.0.14

To me this looks like a problem as well:

28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 0: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 1: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 0: AMD Radeon (TM) R9 Fury Series (driver version 3240.6, device version OpenCL 2.0 AMD-APP (3240.6), 4096MB, 4096MB available, 7782 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 1: AMD FirePro W7000 (driver version 3240.6, device version OpenCL 1.2 AMD-APP (3240.6), 4096MB, 4096MB available, 2432 GFLOPS peak)

You have how many gpu's in the machine...2, 3 or 4? yet Boinc only has them assigned as "gpu0" and "gpu1" AND they seem to be cross connecting the slots! How can you have an Nvidia Tesla is in the gpu0 slot on one line and then a couple lines later have an AMD Fury in the gpu0 slot?

Each slot should get it's own gpu and start at gpu0 and go thru gpu9 if you have 10 gpu's in the machine but it seems your gpu's are sharing slots and that's probably confusing the heck out of Boinc. Are you by chance using a riser card with multiple cpu's on it that then connect to the same slot of the motherboard? If so you need to find someone else doing something similar to figure this out.

This is Evga X299 Dark Motherboard (151-SX-E299-KR).
3xGPU:
- AMD FirePro W7000
- AMD Radeon R9 Fury
- NVIDIA Tesla K80
Everything is without PCIe Risers.

Like mikey is pointing out: How do you explain the double allocation of the two different GPUs ?

Again, how can you install two different drivers for two different manufacturers ?
Or, let me ask differently: How did you install the two different drivers for each GPU ?
I'm just curious and always ready to learn new things ...

There are even more than two drivers from different companies. There is also Intel. Now it all surprises me myself. Young is amazed. You helped like a dead man with incense. Looking for more problems that I didn't even mention. Drivers installed normally, detecting hardware, and installing themselves is what someone made them for.

Young is amazed. You helped like a dead man with incense. Looking for more problems that I didn't even mention.

Young ? What is that ? A new driver ?

I didn't mention any new problems - that was someone else.
I politely asked how that works using mixed GPU-producers - because when I tried to install NVIDIA into a mixed rig having AMD and NVIDIA GPUs it didn't succeed.
I know, probably my own dumbness.

But anyway, you are being unpolite - I wonder if you know what that means.
We don't need peolpe like that here.

Have a great weekend ...

Yes I am new-young at computers. And I don't know everything.
You don't want to help, don't write and don't raise me.
Sorry for asking a stupid question.
They also do not like pompous buffoons anywhere.
Your help tends to point out my mistakes.
I installed the drivers as normal as the installation file.
Nobody knows everything.
Or where I wrote untruth?

Hello.
In cc_config.xml I have the line <use_all_gpus> 1 </use_all_gpus>.
They are visible in the device management and BM. But the tasks do not work on them.

What to do to run tasks on them also on Nvidia Tesla K80?
In another computer I also have a Tesla K80 with Titan X Maxwell and everything works as it should.

Host:
https://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=926996
Extract from the event log:

28.04.2022 23:05:24 | | Starting BOINC client version 7.16.20 for windows_x86_64
28.04.2022 23:05:24 | | log flags: file_xfer, sched_ops, task
28.04.2022 23:05:24 | | Libraries: libcurl/7.47.1 OpenSSL/1.0.2s zlib/1.2.8
28.04.2022 23:05:24 | | Data directory: C:\ProgramData\BOINC
28.04.2022 23:05:24 | | Running under account chr80
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 0: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | CUDA: NVIDIA GPU 1: Tesla K80 (driver version 472.98, CUDA version 11.4, compute capability 3.7, 4096MB, 4007MB available, 4111 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 0: AMD Radeon (TM) R9 Fury Series (driver version 3240.6, device version OpenCL 2.0 AMD-APP (3240.6), 4096MB, 4096MB available, 7782 GFLOPS peak)
28.04.2022 23:05:27 | | OpenCL: AMD/ATI GPU 1: AMD FirePro W7000 (driver version 3240.6, device version OpenCL 1.2 AMD-APP (3240.6), 4096MB, 4096MB available, 2432 GFLOPS peak)
28.04.2022 23:05:27 | | Windows processor group 0: 36 processors
28.04.2022 23:05:27 | | Host name: BOINCPOLAND5
28.04.2022 23:05:27 | | Processor: 36 GenuineIntel Intel(R) Core(TM) i9-7980XE CPU @ 2.60GHz [Family 6 Model 85 Stepping 4]
28.04.2022 23:05:27 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx tm2 dca pbe fsgsbase bmi1 hle smep bmi2
28.04.2022 23:05:27 | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.19042.00)
28.04.2022 23:05:27 | | Memory: 63.70 GB physical, 73.20 GB virtual
28.04.2022 23:05:27 | | Disk: 930.94 GB total, 820.18 GB free
28.04.2022 23:05:27 | | Local time is UTC +2 hours
28.04.2022 23:05:27 | | No WSL found.
28.04.2022 23:05:27 | | VirtualBox version: 6.0.14

@chr80,

Does this help[? https://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=4883

Note that I did not use <use_all_gpus> 1 </use_all_gpus> Actually, I didn't know about that until today...

No Intel GPUs, but I do run NVIDIA and AMD under one roof. NVIDIA runs a single task at a time, AMD runs 3 at a time.

Good luck.

You have helped like no one has ever done before. Everyone around is only scaring and threatening.
After reading in GPU-Z how much VRAM does one WU to milky take, I successfully run 32 WU on a computer with 24 core with a good result. On three GPUs 96 tasks. A You can and more.
Thank you and best regards, God bless you.

You have helped like no one has ever done before.

Awww shucks...

Again, how can you install two different drivers for two different manufacturers ?
Or, let me ask differently: How did you install the two different drivers for each GPU ?
I'm just curious and always ready to learn new things ...

In Windows you just download and install the drivers and the gpu uses them, back in the 'old' days if you had 2 gpu's of the same kind you even had to load the same drivers TWICE so the 2nd gpu would crunch.

In the original post starting this thread I see no sign of Nvidia OpenCL drivers listed in the BOINC startup. Nvidia drivers from Microsoft do not contain the OpenCL component needed to compute BOINC project applications using the OpenCL API.

You need to install the Nvidia drivers directly from Nvidia to get the OpenCL component in the whole driver package.

Unless you intend to only crunch for projects that offer CUDA applications. They are rare among BOINC projects. Most projects use OpenCL applications.

The AMD drivers work because they only offer the OpenCL API for application usage.

---

I have reinstalled the drivers. Always from NVIDIA and Visual C ++ Redistributable Runtimes All-in-One from Techpowerup.
And I have NVIDIA OpenCL.

01.05.2022 18:22:46 | | Starting BOINC client version 7.16.20 for windows_x86_64
01.05.2022 18:22:46 | | log flags: file_xfer, sched_ops, task
01.05.2022 18:22:46 | | Libraries: libcurl/7.47.1 OpenSSL/1.0.2s zlib/1.2.8
01.05.2022 18:22:46 | | Data directory: C:\ProgramData\BOINC
01.05.2022 18:22:46 | | Running under account chr80
01.05.2022 18:22:48 | | CUDA: NVIDIA GPU 0: Tesla K80 (driver version 385.12, CUDA version 9.0, compute capability 3.7, 4096MB, 4009MB available, 4111 GFLOPS peak)
01.05.2022 18:22:48 | | CUDA: NVIDIA GPU 1: Tesla K80 (driver version 385.12, CUDA version 9.0, compute capability 3.7, 4096MB, 4009MB available, 4111 GFLOPS peak)
01.05.2022 18:22:48 | | OpenCL: NVIDIA GPU 0: Tesla K80 (driver version 385.12, device version OpenCL 1.2 CUDA, 12214MB, 4009MB available, 4111 GFLOPS peak)
01.05.2022 18:22:48 | | OpenCL: NVIDIA GPU 1: Tesla K80 (driver version 385.12, device version OpenCL 1.2 CUDA, 12214MB, 4009MB available, 4111 GFLOPS peak)
01.05.2022 18:22:48 | | OpenCL: AMD/ATI GPU 0: AMD Radeon (TM) R9 Fury Series (driver version 3240.6, device version OpenCL 2.0 AMD-APP (3240.6), 4096MB, 4096MB available, 7782 GFLOPS peak)
01.05.2022 18:22:48 | | OpenCL: AMD/ATI GPU 1: AMD FirePro W7000 (driver version 3240.6, device version OpenCL 1.2 AMD-APP (3240.6), 4096MB, 4096MB available, 2432 GFLOPS peak)
01.05.2022 18:22:48 | | Windows processor group 0: 36 processors
01.05.2022 18:22:48 | | Host name: BOINCPOLAND5
01.05.2022 18:22:48 | | Processor: 36 GenuineIntel Intel(R) Core(TM) i9-7980XE CPU @ 2.60GHz [Family 6 Model 85 Stepping 4]
01.05.2022 18:22:48 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx tm2 dca pbe fsgsbase bmi1 hle smep bmi2
01.05.2022 18:22:48 | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.19042.00)
01.05.2022 18:22:48 | | Memory: 63.70 GB physical, 73.20 GB virtual
01.05.2022 18:22:48 | | Disk: 930.94 GB total, 808.97 GB free
01.05.2022 18:22:48 | | Local time is UTC +2 hours
01.05.2022 18:22:48 | | No WSL found.
01.05.2022 18:22:48 | | VirtualBox version: 6.1.34

I have reinstalled the drivers. Always from NVIDIA and Visual C ++ Redistributable Runtimes All-in-One from Techpowerup.
And I have NVIDIA OpenCL.

01.05.2022 18:22:46 | | Starting BOINC client version 7.16.20 for windows_x86_64
01.05.2022 18:22:46 | | log flags: file_xfer, sched_ops, task
01.05.2022 18:22:46 | | Libraries: libcurl/7.47.1 OpenSSL/1.0.2s zlib/1.2.8
01.05.2022 18:22:46 | | Data directory: C:\ProgramData\BOINC
01.05.2022 18:22:46 | | Running under account chr80
01.05.2022 18:22:48 | | CUDA: NVIDIA GPU 0: Tesla K80 (driver version 385.12, CUDA version 9.0, compute capability 3.7, 4096MB, 4009MB available, 4111 GFLOPS peak)
01.05.2022 18:22:48 | | CUDA: NVIDIA GPU 1: Tesla K80 (driver version 385.12, CUDA version 9.0, compute capability 3.7, 4096MB, 4009MB available, 4111 GFLOPS peak)
01.05.2022 18:22:48 | | OpenCL: NVIDIA GPU 0: Tesla K80 (driver version 385.12, device version OpenCL 1.2 CUDA, 12214MB, 4009MB available, 4111 GFLOPS peak)
01.05.2022 18:22:48 | | OpenCL: NVIDIA GPU 1: Tesla K80 (driver version 385.12, device version OpenCL 1.2 CUDA, 12214MB, 4009MB available, 4111 GFLOPS peak)
01.05.2022 18:22:48 | | OpenCL: AMD/ATI GPU 0: AMD Radeon (TM) R9 Fury Series (driver version 3240.6, device version OpenCL 2.0 AMD-APP (3240.6), 4096MB, 4096MB available, 7782 GFLOPS peak)
01.05.2022 18:22:48 | | OpenCL: AMD/ATI GPU 1: AMD FirePro W7000 (driver version 3240.6, device version OpenCL 1.2 AMD-APP (3240.6), 4096MB, 4096MB available, 2432 GFLOPS peak)
01.05.2022 18:22:48 | | Windows processor group 0: 36 processors
01.05.2022 18:22:48 | | Host name: BOINCPOLAND5
01.05.2022 18:22:48 | | Processor: 36 GenuineIntel Intel(R) Core(TM) i9-7980XE CPU @ 2.60GHz [Family 6 Model 85 Stepping 4]
01.05.2022 18:22:48 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx tm2 dca pbe fsgsbase bmi1 hle smep bmi2
01.05.2022 18:22:48 | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.19042.00)
01.05.2022 18:22:48 | | Memory: 63.70 GB physical, 73.20 GB virtual
01.05.2022 18:22:48 | | Disk: 930.94 GB total, 808.97 GB free
01.05.2022 18:22:48 | | Local time is UTC +2 hours
01.05.2022 18:22:48 | | No WSL found.
01.05.2022 18:22:48 | | VirtualBox version: 6.1.34

Okay this is grasping at straws but you did check BOTH of these boxes in your settings...right?
Use ATI GPU
Use NVIDIA GPU

I am most pleased that here I got the most feedback from users.
What makes me more convinced that it is worthwhile to deal with astronomy, physics and MilkyWay @ home,
because you can always count on the help of other people with similar interests.

I am most pleased that here I got the most feedback from users.
What makes me more convinced that it is worthwhile to deal with astronomy, physics and MilkyWay @ home,
because you can always count on the help of other people with similar interests.

The main problem is that less than 10% of crunchers ever come to the forums in the first place and a lesser amount of them actually post and then an even lesser amount respond to other peoples posts. It's long been a Boinc 'problem' that's never been solved, ie you can't force people to participate.

After he reinstalled the Nvidia drivers directly from Nvidia, all cards are being enumerated by BOINC properly.

FYI, BOINC shows the gpu detection in the startup of the Event Log. It finds all CUDA capable cards through the CUDA API numbered gpu#0 through gpu#9 and then enumerates all gpus capable of OpenCL through the OpenCL API numbered gpu#0 through gpu#9.

So, the normal detection of gpus firsts lists all CUDA capable cards, then all OpenCL cards. Nvidia cards will be listed with BOTH capabilities IF the OpenCL component of the Nvidia drivers is found and loaded by the OS.

Intel igpu would be listed also in the OpenCL enumeration following the AMD enumeration and Nvidia enumeration.

Windows has a nasty habit of installing its own drivers even if the manufacturer drivers are loaded and overrides them. The Windows provided Nvidia drivers often do not have the OpenCL component.

There is a toggle in Windows do disallow Windows from updating user installed drivers. You should use that function.

---

After he reinstalled the Nvidia drivers directly from Nvidia, all cards are being enumerated by BOINC properly.

FYI, BOINC shows the gpu detection in the startup of the Event Log. It finds all CUDA capable cards through the CUDA API numbered gpu#0 through gpu#9 and then enumerates all gpus capable of OpenCL through the OpenCL API numbered gpu#0 through gpu#9.

So, the normal detection of gpus firsts lists all CUDA capable cards, then all OpenCL cards. Nvidia cards will be listed with BOTH capabilities IF the OpenCL component of the Nvidia drivers is found and loaded by the OS.

Intel igpu would be listed also in the OpenCL enumeration following the AMD enumeration and Nvidia enumeration.

Windows has a nasty habit of installing its own drivers even if the manufacturer drivers are loaded and overrides them. The Windows provided Nvidia drivers often do not have the OpenCL component.

There is a toggle in Windows do disallow Windows from updating user installed drivers. You should use that function.

+1