error <cryptonight_core_gpu_hash>:751 "invalid resource handle"

[juazki]

I am trying to use an old NVIDIA Tesla C2075 with this plugin.

It used to work with the xmrig-nvidia, but now it doesn't work.

• ABOUT XMRig/5.0.1 MSVC/2017
• LIBS libuv/1.31.0 OpenSSL/1.1.1c hwloc/2.0.4
• HUGE PAGES permission granted
• CPU Intel(R) Xeon(R) CPU X5672 @ 3.20GHz (1) x64 AES
  L2:1.0 MB L3:12.0 MB 4C/4T NUMA:1
• ASSEMBLY auto:intel
• POOL Cleanup everything in GPU context when releasing it #1 stratum+tcp://XXXX:1111 coin monero
• COMMANDS hashrate, pause, resume
• OPENCL disabled
• CUDA 8.0/9.0/2.0.1-beta
• NVML 9.385.69/385.69 press e for health report
• CUDA GPU #0 42:00.0 Tesla C2075 1147/1566 MHz smx:14 arch:20 mem:5257/5316 MB
  [2019-11-25 11:19:11.467] net use pool XXXX:1111 XXXX
  [2019-11-25 11:19:11.468] net new job from XXXX:1111 diff 80000 algo cn/r
  [2019-11-25 11:19:11.469] nv use profile cn/2 (1 thread) scratchpad 2048 KB
  | # | GPU | BUS ID | I | T | B | BF | BS | MEM | NAME
  | 0 | 0 | 42:00.0 | 448 | 4 | 112 | 6 | 25 | 896 | Tesla C2075
  [2019-11-25 11:19:11.576] nv READY threads 1/1 (106 ms)
  [2019-11-25 11:19:11.577] nv thread #0 failed with error <cryptonight_core_gpu_hash>:751 "invalid
  resource handle"
  [2019-11-25 11:19:16.637] Ctrl+C received, exiting
  [2019-11-25 11:19:16.644] nv stopped (7 ms)

But according to: "Fixed compatibility with CUDA 8, RandomX support not tested and potentially broken with this CUDA version.", I guess it won't be supported anyway with the change of POW, so it isn't worth to keep trying to use it, right?

Thank you for your effort in the development of this great tool!