AMD Cayman [FirePro V5900 (FireGL V)]

[bt99]

Not sure this is the right place for this.

Trying to get an AMD FirePro V5900 (FireGL V) to fold, but I keep getting:

Code: Select all

20:05:16:WU00:FS01:Requesting new work unit for slot 01: READY gpu:0:Cayman [FirePro V5900 (FireGL V)] from 192.0.2.1
20:05:16:WU00:FS01:Connecting to 192.0.2.1:8080
20:05:37:WARNING:WU00:FS01:WorkServer connection failed on port 8080 trying 80

It seems to be supported and in the GPUS.txt:

Code: Select all

0x1002:0x6707:1:5:Cayman [FirePro V5900 (FireGL V)]

system info:

Code: Select all

20:05:02:******************************* System ********************************
20:05:02:            CPU: Intel(R) Xeon(R) CPU X5650 @ 2.67GHz
20:05:02:         CPU ID: GenuineIntel Family 6 Model 44 Stepping 2
20:05:02:           CPUs: 12
20:05:02:         Memory: 32.00GiB
20:05:02:    Free Memory: 29.85GiB
20:05:02:        Threads: WINDOWS_THREADS
20:05:02:     OS Version: 6.2
20:05:02:    Has Battery: false
20:05:02:     On Battery: false
20:05:02:     UTC Offset: -6
20:05:02:            PID: 9080
20:05:02:            CWD: C:\Users\christopher\AppData\Roaming\FAHClient
20:05:02:  Win32 Service: false
20:05:02:             OS: Windows 10 Enterprise
20:05:02:        OS Arch: AMD64
20:05:02:           GPUs: 1
20:05:02:          GPU 0: Bus:3 Slot:0 Func:0 AMD:1 Cayman [FirePro V5900 (FireGL V)]
20:05:02:           CUDA: Not detected: Failed to open dynamic library 'nvcuda.dll': The
20:05:02:                 specified module could not be found.
20:05:02:
20:05:02:OpenCL Device 0: Platform:0 Device:0 Bus:3 Slot:0 Compute:1.2 Driver:1800.12

[Joe_H]

The current version of the GPUs.txt file has this entry:

Code: Select all

0x1002:0x6707:1:1:Cayman [FirePro V5900 (FireGL V)]

They are part way through redoing the species numbering for AMD GPUs to make them more performance based than generational. Some issues have arisen, will pass on your report.

[JimboPalmer]

Welcome to Folding@home!

https://www.techpowerup.com/gpu-specs/f ... v5900.c577

This should (very slowly) fold. It does 64 bit (Double Precision) floating point math and supports OpenCL 1.2, the minimum requirements for F@H GPU folding.

It is 9 years old and (depending on how you count) three AMD generations back. With only 512 cores it would not be fast even if new. (80% as fast as a GT 1030?)

The newest driver I can find is from 2017 https://www.amd.com/en/support/professi ... epro-v5900

I hope they get the GPUs.txt worked out.

[bruce]

Folding slowly should be acceptable for some projects, but not all. I don't mind leaving that up to the project owner. The question is whether Joseph has blocked species 3 (->192.2.0.1)