Cuckoo Cycle GPU memory requirements


#1

With the default -E 0, the GPU mean miner for 2^29 edges requires 7 GB.
With option -E 1 this drops to 5 GB along with a slight slowdown.
With option -E 2 it drops to the minimum of 3.5 GB with slightly more slowdown.


#2

If the memory is 3GB what arguments should I give? Because I am using 1060 3GB GPU’s


#3

I don’t see a way to efficiently GPU mine with only 3 GB.


#4

Thanks. So you mean I cannot mine aeternity with 3GB cards. So is there any tweak to mine with 3GB i have 140 cards


#5

With 3GB cards, you cannot mine aeternity efficiently.
There’s no tweak I can think of.


#6

Do you think it is possible to make a lean executable that works on arm? as to be running an aeternity node on pi3/tinkerboard.

I know there is a specific x86 library that is nto available for the mean/lean executable.


#7

It’s possible, but a single graph will take minutes on a pi3; so most of your efforts will be wasted…


#8

I don’t want it to mine, just to be able to sync and verify.

I understand a single graph in a raspberry pi3 would take a long long time (too little ram, not enough cpu power).


#9

±----------------------------------------------------------------------------+
| NVIDIA-SMI 410.78 Driver Version: 410.78 CUDA Version: 10.0 |
|-------------------------------±---------------------±---------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 106… Off | 00000000:01:00.0 Off | N/A |
| 32% 61C P0 33W / 200W | 18MiB / 6078MiB | 0% Default |
±------------------------------±---------------------±---------------------+
| 1 GeForce GTX 106… Off | 00000000:04:00.0 Off | N/A |
| 28% 59C P0 33W / 200W | 18MiB / 6078MiB | 0% Default |
±------------------------------±---------------------±---------------------+
| 2 GeForce GTX 106… Off | 00000000:05:00.0 Off | N/A |
| 29% 59C P0 33W / 200W | 18MiB / 6078MiB | 0% Default |
±------------------------------±---------------------±---------------------+
| 3 GeForce GTX 106… Off | 00000000:06:00.0 Off | N/A |
| 24% 60C P0 33W / 200W | 19MiB / 6078MiB | 4% Default |
±------------------------------±---------------------±---------------------+
| 4 GeForce GTX 106… Off | 00000000:07:00.0 Off | N/A |
| 24% 59C P0 35W / 200W | 23MiB / 6078MiB | 0% Default |
±------------------------------±---------------------±---------------------+
| 5 GeForce GTX 106… Off | 00000000:08:00.0 Off | N/A |
| 0% 57C P0 34W / 200W | 73MiB / 6078MiB | 0% Default |
±------------------------------±---------------------±---------------------+

±----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 2486 C ./cuda29 21MiB |
| 1 2489 C ./cuda29 21MiB |
| 2 2487 C ./cuda29 63MiB |
| 3 2485 C ./cuda29 63MiB |
| 4 2488 C ./cuda29 63MiB |
| 5 2490 C ./cuda29 63MiB


#10

Sun Dec 2 11:22:39 2018
±----------------------------------------------------------------------------+
| NVIDIA-SMI 410.78 Driver Version: 410.78 CUDA Version: 10.0 |
|-------------------------------±---------------------±---------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 106… Off | 00000000:01:00.0 Off | N/A |
| 32% 63C P2 104W / 200W | 4929MiB / 6078MiB | 100% Default |
±------------------------------±---------------------±---------------------+
| 1 GeForce GTX 106… Off | 00000000:04:00.0 Off | N/A |
| 29% 62C P2 36W / 200W | 4929MiB / 6078MiB | 100% Default |
±------------------------------±---------------------±---------------------+
| 2 GeForce GTX 106… Off | 00000000:05:00.0 Off | N/A |
| 0% 59C P2 73W / 200W | 4929MiB / 6078MiB | 100% Default |
±------------------------------±---------------------±---------------------+
| 3 GeForce GTX 106… Off | 00000000:06:00.0 Off | N/A |
| 28% 62C P2 87W / 200W | 4929MiB / 6078MiB | 100% Default |
±------------------------------±---------------------±---------------------+
| 4 GeForce GTX 106… Off | 00000000:07:00.0 Off | N/A |
| 28% 62C P2 95W / 200W | 4929MiB / 6078MiB | 100% Default |
±------------------------------±---------------------±---------------------+
| 5 GeForce GTX 106… Off | 00000000:08:00.0 Off | N/A |
| 29% 61C P0 36W / 200W | 55MiB / 6078MiB | 78% Default |
±------------------------------±---------------------±---------------------+

±----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 3225 C ./cuda29 4919MiB |
| 1 3227 C ./cuda29 53MiB |
| 2 3229 C ./cuda29 63MiB |
| 3 3224 C ./cuda29 4919MiB |
| 4 3226 C ./cuda29 4919MiB |
| 5 3228 C ./cuda29 45MiB


#11

Hello, Mr., my rig runs often with the screenshot reply. In the last second, the GPU load was very low, and the load reached 100% in the next second. Is this a normal working state? I use the -E 1 option, 6*1066. Thanks for any help!


#12

You’re bound to see inefficiencies when the solver is used in the wrong way; see https://forum.aeternity.com/t/expected-1080-ti-performance


#13

Sorry,sir.I read the content of this topic. But I can’t fully understand it. Is it necessary to compile a new miner with the latest released code? Then put it in the epoch directory?


#14

Or just need to modify some configurations?


#15

You need to wait for AE devs to release new code that interfaces to my solver the proper way (preserving the solver context from nonce to nonce).


#16

Can you pinpoint where in the grin repo you have provided the function call to the solver ?


#17

Got it, thank you.Continue to wait until the new release


#18

See https://github.com/mimblewimble/grin-miner/search?q=run_solver&unscoped_q=run_solver


#19

Where do these E options go??? In the epoch.yaml??? Like this???
cuckoo:
miner:
executable: cuda29
extra_args: “-E 0”
instances: 2
edge_bits: 29
hex_encoded_header: true
miner:
edge_bits: 29
executable: mean29-avx2
extra_args: -t 26


#20

Yes, but you can’t mix CPU mining with GPU mining at the moment. So only one set of miner: configuration options is allowed…