Closed aapjeisbaas closed 2 years ago
ow just for later refference:
ssm-user@ip-172-31-23-130:/var/snap/amazon-ssm-agent/2012$ nvidia-smi
Wed Apr 14 06:49:33 2021
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 450.80.02 Driver Version: 450.80.02 CUDA Version: 11.0 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 A100-SXM4-40GB On | 00000000:10:1C.0 Off | 0 |
| N/A 48C P0 328W / 400W | 4720MiB / 40537MiB | 100% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 1 A100-SXM4-40GB On | 00000000:10:1D.0 Off | 0 |
| N/A 47C P0 327W / 400W | 4720MiB / 40537MiB | 100% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 2 A100-SXM4-40GB On | 00000000:20:1C.0 Off | 0 |
| N/A 50C P0 336W / 400W | 4720MiB / 40537MiB | 100% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 3 A100-SXM4-40GB On | 00000000:20:1D.0 Off | 0 |
| N/A 46C P0 324W / 400W | 4720MiB / 40537MiB | 100% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 4 A100-SXM4-40GB On | 00000000:90:1C.0 Off | 0 |
| N/A 47C P0 319W / 400W | 4720MiB / 40537MiB | 100% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 5 A100-SXM4-40GB On | 00000000:90:1D.0 Off | 0 |
| N/A 46C P0 327W / 400W | 4720MiB / 40537MiB | 100% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 6 A100-SXM4-40GB On | 00000000:A0:1C.0 Off | 0 |
| N/A 49C P0 325W / 400W | 4720MiB / 40537MiB | 100% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 7 A100-SXM4-40GB On | 00000000:A0:1D.0 Off | 0 |
| N/A 46C P0 322W / 400W | 4720MiB / 40537MiB | 100% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 3985 C /tmp/PhoenixMiner 4717MiB |
| 1 N/A N/A 3985 C /tmp/PhoenixMiner 4717MiB |
| 2 N/A N/A 3985 C /tmp/PhoenixMiner 4717MiB |
| 3 N/A N/A 3985 C /tmp/PhoenixMiner 4717MiB |
| 4 N/A N/A 3985 C /tmp/PhoenixMiner 4717MiB |
| 5 N/A N/A 3985 C /tmp/PhoenixMiner 4717MiB |
| 6 N/A N/A 3985 C /tmp/PhoenixMiner 4717MiB |
| 7 N/A N/A 3985 C /tmp/PhoenixMiner 4717MiB |
+-----------------------------------------------------------------------------+
I tried -gt 400
to push the cards a little harder but it made no difference in the hashrate
Phoenix Miner 5.5c Linux/gcc - Release build
--------------------------------------------
Unable to open log file /tmp/log.txt - 13
CUDA version: 11.0, CUDA runtime: 8.0
No OpenCL platforms found
Available GPUs for mining:
GPU1: A100-SXM4-40GB (pcie 16), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU2: A100-SXM4-40GB (pcie 16), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU3: A100-SXM4-40GB (pcie 32), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU4: A100-SXM4-40GB (pcie 32), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU5: A100-SXM4-40GB (pcie 144), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU6: A100-SXM4-40GB (pcie 144), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU7: A100-SXM4-40GB (pcie 160), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU8: A100-SXM4-40GB (pcie 160), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
Nvidia driver version: 450.80.02
...
...
...
*** 0:03 *** 4/14 06:57 **************************************
Eth: Mining ETH on ssl://eth-us-east1.nanopool.org:9433 for 0:03
Available GPUs for mining:
GPU1: A100-SXM4-40GB (pcie 16), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU2: A100-SXM4-40GB (pcie 16), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU3: A100-SXM4-40GB (pcie 32), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU4: A100-SXM4-40GB (pcie 32), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU5: A100-SXM4-40GB (pcie 144), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU6: A100-SXM4-40GB (pcie 144), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU7: A100-SXM4-40GB (pcie 160), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU8: A100-SXM4-40GB (pcie 160), CUDA cap. 8.0, 40 GB VRAM, 108 CUs
GPU1: 70C 370W, GPU2: 60C 343W, GPU3: 73C 381W, GPU4: 59C 339W, GPU5: 67C 347W, GPU6: 59C 343W, GPU7: 69C 354W, GPU8: 59C 345W
GPUs power: 2821.6 W
Current -gt 400,400,400,400,400,400,400,400
Eth: Accepted shares 23 (0 stales), rejected shares 0 (0 stales)
Eth: Incorrect shares 0 (0.00%), est. stales percentage 0.00%
Eth: Maximum difficulty of found share: 362.4 GH (!)
Eth: Average speed (5 min): 1383.154 MH/s
Eth: Effective speed: 1031.39 MH/s; at pool: 1031.39 MH/s
Eth: Share accepted in 193 ms
*** 0:03 *** 4/14 06:57 **************************************
Eth: Mining ETH on ssl://eth-us-east1.nanopool.org:9433 for 0:03
Eth speed: 1383.167 MH/s, shares: 24/0/0, time: 0:03
GPUs: 1: 172.871 MH/s (2) 2: 172.871 MH/s (1) 3: 172.890 MH/s (3) 4: 172.893 MH/s (5) 5: 172.927 MH/s (3) 6: 172.871 MH/s (1) 7: 172.944 MH/s (7) 8: 172.901 MH/s (2)
Eth: Accepted shares 24 (0 stales), rejected shares 0 (0 stales)
Eth: Incorrect shares 0 (0.00%), est. stales percentage 0.00%
Eth: Maximum difficulty of found share: 362.4 GH (!)
Eth: Average speed (5 min): 1383.154 MH/s
Eth: Effective speed: 1066.67 MH/s; at pool: 1066.67 MH/s
Thanks, I've added this instance.
Just out of curiosity I tried the p4d.24xlarge I had to request a limit increase to be able to start this 96 core 8GPU beast
Results: Eth: Average speed (5 min): 1383.154 MH/s
which will yield about $6 per hour, unfortunately the machine as a spot instance costs $9.81 And in your MH/$ metric it scores 140.767 MH/$
The machine has 8 x A100-SXM4-40GB cards and the run at +/- 170 HM/s with PhoenixMiner yes I changed out the user data in my tests to use a different miner...
I think AWS uses something like the dgx A100 for the p4d.24xlarge
For my own mining I'll keep using a simple home rig with nvidia-docker: https://gitlab.com/aapjeisbaas/docker-miner https://hub.docker.com/r/aapjeisbaas/docker-miner
Thanks for the template, I had fun with it!