Mining on Nvidia Jetson Tx2

[tonyfrost2020]

I currently have a Nvidia Jetson Tx2 just sitting around, I was wondering if anyone has managed to get a GPU miner to work with the 256 cuda cores for alt mining?
I installed ethminer but when I ran the benchmark the GPU was showing as 0MH/s.
I know this wouldn't even touch bitcoin, which is why I am looking into the alt coin market.
If anyone has managed this, would you be able to give me a push in the right direction?
And also post a benchmark of your results if possible?

[gt_addict]

Probably best to ask in te altcoins section ive asked mods to move it.

[tonyfrost2020]

Ok so I have built xmrig from source and get the following cuda error when trying to run xmrig:-

Code:

x@x:~/miner/xmrig/build$ ./xmrig -o xxx.xxxxx.xxx:xxxx -u x -p x -a rx/0 -k --no-huge-pages --cuda --cuda-loader=./libxmrig-cuda.so
 * ABOUT        XMRig/6.5.1 gcc/7.5.0
 * LIBS         libuv/1.18.0 OpenSSL/1.1.1 hwloc/1.11.9
 * HUGE PAGES   disabled
 * 1GB PAGES    unavailable
 * CPU          ARM Cortex-A57 (2) x64 AES
                L2:2.0 MB L3:0.0 MB 6C/6T NUMA:1
 * MEMORY       4.9/7.7 GB (63%)
 * DONATE       1%
 * POOL #1      xxx.xxxxx.xxx:xxxx algo rx/0
 * COMMANDS     hashrate, pause, resume, results, connection
 * OPENCL       disabled
 * CUDA         disabled ((null))

Can anyone think of why the CUDA is showing as disabled((null))?
I will keep having a dig around to see if I can find anything.

[tonyfrost2020]

I have updated the drivers on the Jetson Tx2 and installed the xmrig-cuda addon and can now see the following:-

Code:

 * ABOUT        XMRig/6.5.1 gcc/7.5.0
 * LIBS         libuv/1.18.0 OpenSSL/1.1.1 hwloc/1.11.9
 * HUGE PAGES   disabled
 * 1GB PAGES    unavailable
 * CPU          ARM Cortex-A57 (2) x64 AES
                L2:2.0 MB L3:0.0 MB 6C/6T NUMA:1
 * MEMORY       5.1/7.7 GB (66%)
 * DONATE       1%
 * POOL #1      xxx.xxxxxx.com:xxxx algo rx/0
 * COMMANDS     hashrate, pause, resume, results, connection
 * OPENCL       disabled
 * CUDA         10.0/10.0/6.4.1
 * NVML         disabled
 * CUDA GPU     #0 00:00.0 NVIDIA Tegra X2 1300/1300 MHz smx:2 arch:62 mem:2664/7859 MB

What is (hopefully) a massive step closer, the I am now facing an issue regarding the following:-

Code:

[2020-11-13 16:14:06.572]  nvidia   use profile  rx  (1 thread) scratchpad 2048 KB
|  # | GPU |  BUS ID | INTENSITY | THREADS | BLOCKS | BF |  BS | MEMORY | NAME
|  0 |   0 | 00:00.0 |       128 |      32 |      4 |  0 |   0 |    256 | NVIDIA Tegra X2
[2020-11-13 16:14:07.146]  nvidia   READY threads 1/1 (573 ms)
[2020-11-13 16:14:07.147]  nvidia   thread #0 failed with error <reg>:50 "operation not supported"

I will continue to dig away at it, any help would be greatly appreciated.

[tonyfrost2020]

After a reboot, I managed to get the GPU running with xmrig.

Code:

[2020-11-13 16:36:16.185]  miner    speed 10s/60s/15m 32.30 n/a n/a H/s max 32.82 H/s
|   CUDA # | AFFINITY | 10s  H/s | 60s  H/s | 15m  H/s |
|        0 |       -1 |    32.52 |      n/a |      n/a | #0 00:00.0 NVIDIA Tegra X2
|        - |        - |    32.50 |      n/a |      n/a |

As you can see it's not a desirable hash rate, I've had a RPI4 running faster.

Is there anyway I can speed this up?

[jadefalke]

After a reboot, I managed to get the GPU running with xmrig.

Code:

[2020-11-13 16:36:16.185]  miner    speed 10s/60s/15m 32.30 n/a n/a H/s max 32.82 H/s
|   CUDA # | AFFINITY | 10s  H/s | 60s  H/s | 15m  H/s |
|        0 |       -1 |    32.52 |      n/a |      n/a | #0 00:00.0 NVIDIA Tegra X2
|        - |        - |    32.50 |      n/a |      n/a |

As you can see it's not a desirable hash rate, I've had a RPI4 running faster.

Is there anyway I can speed this up?

you recognized that RandomX is not designed for GPU? so what exactly do you expect? try eth/x16 or something more favorable for your GPU.

[tonyfrost2020]

After a reboot, I managed to get the GPU running with xmrig.

Code:

[2020-11-13 16:36:16.185]  miner    speed 10s/60s/15m 32.30 n/a n/a H/s max 32.82 H/s
|   CUDA # | AFFINITY | 10s  H/s | 60s  H/s | 15m  H/s |
|        0 |       -1 |    32.52 |      n/a |      n/a | #0 00:00.0 NVIDIA Tegra X2
|        - |        - |    32.50 |      n/a |      n/a |

As you can see it's not a desirable hash rate, I've had a RPI4 running faster.

Is there anyway I can speed this up?

you recognized that RandomX is not designed for GPU? so what exactly do you expect? try eth/x16 or something more favorable for your GPU.

Fair enough, I have had a look at the kawpow algorithm and am now hitting 250k h/s.

[miner29]

250khs is very low for kawpow.  Generally done in Mhs. 

Also the gear you are using is simply not powerful enough.  This thing only have 1/8 of the cores of a 1070....   about 1/4 to 1/5 of a 1060..so not much there.  Even a 1050 (non ti) has more cores. 

You are wasting your time trying to mine on this thing...but that was clear 3 years ago.

Fun to play and learn but no good for mining,