Grin mining with cuckatoo31

Mining
1

Hello!

I have the version of Grin-Miner version 1.0.2 (git v1.0.2-14-g143db3b), built for x86_64-unknown-linux-gnu by rustc 1.33.0 (2aa4c46cf 2019-02-28). I’m running on Debian GNU/Linux 9.8 (stretch).

I’m running on a computer with 4 GPUs, each with about 15 GB of memory. I’m therefore running the Cuckatoo31 algorithm. This is giving me about 12 GPS in total. I’m using the nicehash mining pool to mine, as described here: https://www.nicehash.com/algorithm/grincuckatoo31

However, after about 12 hours, I see nothing in the log files or on the nicehash interface with either rejected or accepted shares. Why do I see no accepted or rejected shares at this point?

Here’s some parts of the mining output (bits redacted) that was logged about 10 minutes after starting the process:

Mar 06 18:24:14.100 INFO Mining: Cuck(at)oo at 12.060275270280519 gps (graphs per second)

Mar 06 18:24:15.001 DEBG Received message: {"id":"0","jsonrpc":"2.0","method":"job","params":{"difficulty":8,"height":3018,"job_id":1917665189930,"pre_pow":"00010000000000000bca000000005c80104d00121c6e0afc2287b4edae17d917add9ab83bf6ecc79f60926ee0c7212e5df9567a5636e197b833f9cf4ffd13ca4324e3080828375497bbfa23268df43c31bbdaeee65b0f46b848623379c0935537ed4fc77778b2e1fd2cda394fc45532a020d8f5f4d2bc4ff54c93f8e2513cc789b82c029b314a88153e03f626ee9e33edd7660a55cad19de9af79c9888dda0f38d26391d9a00bc69621a321e60cbb7dfda920000000000000000000000000000000000000000000000000000000000000000000000000000178e000000000000178e0000000a7c879d5600302974","xn":"0091","cleanjob":true}}

[...]

Mar 06 18:30:50.083 INFO Mining: Cuck(at)oo at 12.088301131365798 gps (graphs per second)

Mar 06 18:30:53.085 DEBG Mining: Plugin 0 - Device 0 (...) at Cuck(at)oo31 - Status: OK : Last Graph time: 0.293413449s; Graphs per second: 3.408 - Total Attempts: 2654

Mar 06 18:30:53.086 DEBG Mining: Plugin 1 - Device 1 (...) at Cuck(at)oo31 - Status: OK : Last Graph time: 0.343349107s; Graphs per second: 2.912 - Total Attempts: 2286

Mar 06 18:30:53.086 DEBG Mining: Plugin 2 - Device 2 (...) at Cuck(at)oo31 - Status: OK : Last Graph time: 0.34038395s; Graphs per second: 2.938 - Total Attempts: 2283

Mar 06 18:30:53.086 DEBG Mining: Plugin 3 - Device 3 (...) at Cuck(at)oo31 - Status: OK : Last Graph time: 0.349973934s; Graphs per second: 2.857 - Total Attempts: 2235

Here are some parts of the toml file:

[[mining.miner_plugin_config]]
plugin_name = "cuckatoo_mean_cuda_rtx_31"
[mining.miner_plugin_config.parameters]
device = 0
expand = 1
cpuload = 1
ntrims = 176
genablocks = 4096
genatpb = 128
genbtpb = 128
trimtpb = 512
tailtpb = 1024
recoverblocks = 1024
recovertpb = 1024

[[mining.miner_plugin_config]]
plugin_name = "cuckatoo_mean_cuda_rtx_31"
[mining.miner_plugin_config.parameters]
device = 1

[[mining.miner_plugin_config]]
plugin_name = "cuckatoo_mean_cuda_rtx_31"
[mining.miner_plugin_config.parameters]
device = 2

[[mining.miner_plugin_config]]
plugin_name = "cuckatoo_mean_cuda_rtx_31"
[mining.miner_plugin_config.parameters]
device = 3
2

Hello,
I had a similar problem…
One question… is it an RTX card? (are they?)
If not, the plugin WONT WORK, and will output strange results if it runs!

I had the same numbers with a P6000… but it was not sending shares… (because it was not solving anything really)

3

They’re not RTX or GTX cards, this might be the problem. They are high-end NVIDIA Tesla GPUs that use CUDA though.

Do you recommend I use cuckatoo_lean_cuda_31 or cuckatoo_mean_cuda_gtx_31 instead? How did you resolve the problem on your end @josevora?

4

use cuckatoo_mean_cuda_gtx_31
the lean miners are painfully slow…

5

Wow, that change is dissapointing. The GPS for all 4 GPUs together dropped from ~12 GPS with RTX down to ~2.5 GPS with GTX. That seems painfully low for 4 GPUs with 15+ GB memory bandwidth.

The good news is I am seeing rejected shares finally. What does the following mean in the miner TUI:
Solutions found: 5. Accepted: 0, Rejected: 3, Stale: 0, Blocks found: 0

How can there be 5 solutions found without any accepted shares? Something I’m missing here?

6

I solved it by following tromp’s advice…
I’ve had the exact same numbers tho… (V100, and P6000 outputs about the same, no miners optimized for them yet…)

You found 5 solutions, you submitted 3, and all 3 got rejected, as far as I understand… why 2 ignored solutions?.. maybe they weren’t 42 Cycles, maybe they were less… (I don’t know if that qualifies as a Solution tho, just as a Cycle…)

7

@tromp, Is there a timeline for when optimized miners will come out for these kinds of chips? The GPS from the grin-miner for these chips is lower than expected.

Also will the K80 work with the GTX or RTX plugin? It doesn’t say online that the K80 is part of either GTX or RTX.

@josevora, did you see about 2.5 GPS for 4 GPUs on your rig as well?

8

Check the architecture on your cards and compare with RTX and GTX, it should work fo same architecture!

i don’t remember exactly te GPS, but i remember V100 was ~1.5 per card and P6000 ~1.1 (not sure)