Bendr0id / xmrigCC

RandomX, CryptoNight, Argon2 and GhostRider CPU/GPU miner with Command&Control (CC) Server and Monitoring
GNU General Public License v3.0
311 stars 108 forks source link

Ubuntu 20.04.2 - AMDGPU 21.50.2 - Radeon VII - HSA_STATUS_ERROR_MEMORY_APERTURE_VIOLATION #392

Closed blackmennewstyle closed 1 year ago

blackmennewstyle commented 1 year ago

I'm trying to mine CryptoNightGPU with my Radeon VII and i'm getting the following error:

sudo /home/ceedii/xmrigCC-miner_only-3.3.1-linux-dynamic-amd64/xmrigDaemon --api-worker-id=epimetheus --api-id=epimetheus --http-host=192.168.0.35 --http-port=4454 --http-access-token=epimetheus --http-no-restricted --algo=cn/gpu --tls --url=stratum+ssl://sg.conceal.herominers.com:1115 --user=ccxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx--pass=epimetheus --no-cpu --opencl --opencl-devices=0 --cc-disabled
 * ABOUT        XMRigCC/3.3.1 gcc/10.3.0 (RELEASE)
 * LIBS         libuv/1.44.1 OpenSSL/1.1.1o hwloc/2.7.1
 * HUGE PAGES   supported
 * 1GB PAGES    disabled
 * CPU          Intel(R) Celeron(R) CPU G3900 @ 2.80GHz (1) 64-bit AES
                L2:0.5 MB L3:2.0 MB 2C/2T NUMA:1
 * MEMORY       0.7/7.7 GB (9%)
                DIMM_A0: 8 GB DDR4 @ 2133 MHz HyperX              
                DIMM_A1: <empty>
                DIMM_B0: <empty>
                DIMM_B1: <empty>
 * MOTHERBOARD  Gigabyte Technology Co., Ltd. - H110-D3A-CF
 * DONATE       5%
 * ASSEMBLY     auto:intel
 * POOL #1      stratum+ssl://sg.conceal.herominers.com:1115 algo cn/gpu
 * CC Server    disabled
 * COMMANDS     hashrate, pause, resume, results, connection
 * HTTP API     192.168.0.35:4454 
[2022-09-02 14:44:55.923] CC feature is disabled.
 * ADL          press e for health report
 * OPENCL       #1 AMD Accelerated Parallel Processing/OpenCL 2.2 AMD-APP (3406.0)
 * OPENCL GPU   #0 03:00.0 AMD Radeon VII (gfx906:sramecc+:xnack-) 1801 MHz cu:60 mem:13912/16368 MB
 * CUDA         disabled
[2022-09-02 14:44:56.806]  net      use pool sg.conceal.herominers.com:1115 TLSv1.2 168.119.69.50
[2022-09-02 14:44:56.806]  net      fingerprint (SHA-256): "8f59a0e25168482b71eae0e75fb5023637a24ef7dbd51323e71412f8a5050743"
[2022-09-02 14:44:56.806]  net      new job from sg.conceal.herominers.com:1115 diff 10000 algo cn/gpu height 1107063 (1 tx)
[2022-09-02 14:44:56.806]  opencl   use profile  cn/gpu  (2 threads) scratchpad 2048 KB
|  # | GPU |  BUS ID | INTENSITY | WSIZE | MEMORY | NAME
|  0 |   0 | 03:00.0 |       960 |     8 |   1920 | AMD Radeon VII (gfx906:sramecc+:xnack-)
|  1 |   0 | 03:00.0 |       960 |     8 |   1920 | AMD Radeon VII (gfx906:sramecc+:xnack-)
[2022-09-02 14:44:57.424]  opencl   READY threads 2/2 (619 ms)
:0:rocdevice.cpp            :2603: 0052207347 us: 1902 : [tid:0x7f9062261700] Device::callbackQueue aborting with error : HSA_STATUS_ERROR_MEMORY_APERTURE_VIOLATION: The agent attempted to access memory beyond the largest legal address. code: 0x29
Aborted (core dumped)
 * ABOUT        XMRigCC/3.3.1 gcc/10.3.0 (RELEASE)
 * LIBS         libuv/1.44.1 OpenSSL/1.1.1o hwloc/2.7.1
 * HUGE PAGES   supported
 * 1GB PAGES    disabled
 * CPU          Intel(R) Celeron(R) CPU G3900 @ 2.80GHz (1) 64-bit AES
                L2:0.5 MB L3:2.0 MB 2C/2T NUMA:1
 * MEMORY       0.7/7.7 GB (9%)
                DIMM_A0: 8 GB DDR4 @ 2133 MHz HyperX              
                DIMM_A1: <empty>
                DIMM_B0: <empty>
                DIMM_B1: <empty>
 * MOTHERBOARD  Gigabyte Technology Co., Ltd. - H110-D3A-CF
 * DONATE       5%
 * ASSEMBLY     auto:intel
 * POOL #1      stratum+ssl://sg.conceal.herominers.com:1115 algo cn/gpu
 * CC Server    disabled
 * COMMANDS     hashrate, pause, resume, results, connection
 * HTTP API     192.168.0.35:4454 
[2022-09-02 14:45:10.614] CC feature is disabled.
 * ADL          press e for health report
 * OPENCL       #1 AMD Accelerated Parallel Processing/OpenCL 2.2 AMD-APP (3406.0)
 * OPENCL GPU   #0 03:00.0 AMD Radeon VII (gfx906:sramecc+:xnack-) 1801 MHz cu:60 mem:13912/16368 MB
 * CUDA         disabled
[2022-09-02 14:45:11.421]  net      use pool sg.conceal.herominers.com:1115 TLSv1.2 168.119.69.50
[2022-09-02 14:45:11.421]  net      fingerprint (SHA-256): "8f59a0e25168482b71eae0e75fb5023637a24ef7dbd51323e71412f8a5050743"
[2022-09-02 14:45:11.421]  net      new job from sg.conceal.herominers.com:1115 diff 10000 algo cn/gpu height 1107063 (1 tx)
[2022-09-02 14:45:11.421]  opencl   use profile  cn/gpu  (2 threads) scratchpad 2048 KB
|  # | GPU |  BUS ID | INTENSITY | WSIZE | MEMORY | NAME
|  0 |   0 | 03:00.0 |       960 |     8 |   1920 | AMD Radeon VII (gfx906:sramecc+:xnack-)
|  1 |   0 | 03:00.0 |       960 |     8 |   1920 | AMD Radeon VII (gfx906:sramecc+:xnack-)
[2022-09-02 14:45:22.312]  net      new job from sg.conceal.herominers.com:1115 diff 10000 algo cn/gpu height 1107064 (1 tx)
Bendr0id commented 1 year ago

I need some more info.

What version of the Rocm driver are you using? Did you try to use reduced intensity + memory settings? Is this happen sporadically or always? Was is happening with an older version?

Best, Ben

blackmennewstyle commented 1 year ago

Hi Ben,

It happened sporadically not always. How do i reduce the intensity + memory settings? I'm using this drivers from AMD: https://www.amd.com/en/support/kb/release-notes/rn-amdgpu-unified-linux-21-50-2 I believe they implemented ROCR which is a derivative of ROCM, how can get the specific version used with that drivers? It's my first time mining CryptoNightGPU, i never tried it with older drivers.

Bendr0id commented 1 year ago

Afaik there is no rocr/rocm support on CN/GPU. Please try to use an older driver like 20.40 and retest with that.

Best, Ben

Bendr0id commented 1 year ago

no response