madMAx released GH 3.0 today. C29-C33 compression

Gigahorse 3.0 has been released: Release 2.1.4.giga30 Gigahorse 3.0 Node / Farmer · madMAx43v3r/chia-gigahorse · GitHub
New in this release:
Gigahorse 3.0 plot format, C30 to C33
Fixed an issue with remote compute where fail-over to another server did result in a lost lookup once every 100 sec. (ERROR: connect() failed with: Connection refused)

The new C30 to C33 are only supported with the CUDA plotter (cuda_plot_k32_v3, 8G RAM minimum)
Farming C30 to C33 is only supported with Nvidia GPU (starting with 900 series)
K32 only for 3.0 format

Need to use cuda_plot_k32_v3 binary for plotting the new format.
Partial difficulty is now even more important for maximum farm size. Minimum recommended partial difficulty is 10k for the new 3.0 format.
Using a too low partial difficulty will increase GPU load by a factor of 2, in the worst case.
Higher -S can help with phase 3 plotting times, like -S 8 if you have enough VRAM.
Use ProofOfSpace farm --diff 800000 to benchmark for solo farming.
chia_recompute_server is backwards compatible to old clients, but needs to be updated for new clients running giga30.
Both Node & Farmer & Harvester need to be updated at the same time to support 3.0 format.
Lower levels of the 3.0 format to replace C16 to C18 will be released later (work in progress).


So, no CHIP-22 at this time ?

1 Like

Arm race never stops.

What desktop 30xx 40xxs gpu equivalents RTX A5000? Thanks.

1 Like

No not yet, but I believe its in the works.

1 Like

Here are some initial Farm size calc’s and Plotting times.

1 Like

Waiting for C26 - C28 improvements…

What sort of farm size can an RTX A4000 (16GB) do at C31, would it be around 600GB native.

1 Like

Any idea on how to set difficulty on Space Pool?

Does difficulty need to be set now if solo mining.

They don’t equate precisely, as the Ax000 series tends to much lower clocks on higher core counts than the RTX 4xxx series (down side of their lower power consumption in general).
Probably the closest equal would be ABOUT a RTX 4070.

For those of us with older cards, the 3060ti is fairly close to the 1080ti.

1 Like

No, there is no changing difficulty when solo mining, only when using pools.

It’s been a while since I’ve been on Space Pool, and I do believe it was very limited range you could adjust on Space Pool. Some of the other pools have unlimited adjustment of difficulty. It’s pool dependent.

Need to try Foxypool…

1 Like

I got this 2 errors on C30 on Space Pool. need to set difficulty. When removing C30 everything goes back to normal

ERROR Exception fetching qualities for X:\plot-k32-c30-2024-03-07-11-16-42970532ed9fd11ac9bac7382d0ac441c9b12c7fcecac96cc3f25c5b6045ec95.plot: recv() failed with: timeout (> 20000 ms) PlotInfo(prover=<chiapos.DiskProver object at 0x0000015EE51A0AB0>, pool_public_key=None, pool_contract_puzzle_hash=<bytes32: 63753f83b962c9d40d245f5694b4bc84c91be47cd7696e23e708a56797cf8fdc>, plot_public_key=<G1Element 9081553bf55b14574ad127e10e91ed0871b327e9cba0207d00ecbbe7c22e4f6f7e51c33c469ac68c15ebac69065defb1>, file_size=46490361096, time_modified=1709828475.8327825)

ERROR Exception fetching qualities for L:\plot-k32-c18-2024-02-29-10-10-cc12ac9468220b9f398ea6d46a905337e2f08f6a46693bdfc410009c84197777.plot: recompute failed to produce a candidate (table = 3) PlotInfo(prover=<chiapos.DiskProver object at 0x0000015EE63FDEF0>, pool_public_key=None, pool_contract_puzzle_hash=<bytes32: 63753f83b962c9d40d245f5694b4bc84c91be47cd7696e23e708a56797cf8fdc>, plot_public_key=<G1Element 8e598c3003aff4048da552be4610cec6cf59cda51ff1b20b5237c9283a23624c771ff84d903142303cba2f0a06df0e19>, file_size=64002719208, time_modified=1709219744.259228)

1 Like

330 c30 plots and pooling on Foxy without issue, mind you I’d had my difficulty set to 120 for the last 10 hours, just changed it to 1000, and I’ll increase it as I add more plots, so we’ll see what happens.

Hi. How much RAM is needed for plotting without SSD, as before - 256 GB or 128 GB?
And which pool can I farm on now? they write the following:
“Currently, you can’t use this version with our community nodes. We are working hard to get the new nodes ready.”

Only with Fast Farmer.

Will be very close, maybe run a mix of C30 and C31.

Same as before, 256G.

Hi @madMAx43v3r

I am having this message when plotting C31

[P1] Table 6 took 32.356 sec, 4291538144 entries, 2.34797 GB/s up, 3.29738 GB/s down
P1 download thread failed with: P1 download thread failed with: failed to allocate 16777216 bytes of MEM_TYPE_HOST: malloc() / aligned_alloc() returned nullptr
failed to allocate 16777216 bytes of MEM_TYPE_HOST: malloc() / aligned_alloc() returned nullptr
P1 download thread failed with: failed to allocate 16777216 bytes of MEM_TYPE_HOST: malloc() / aligned_alloc() returned nullptr
PS C:\Users\HP\Desktop\CUDA PLOTTER>

My Spec’s:
Intel Xeon E5-2690 v2 @ 3.0 GHz
256 GB RAM DDR3 1600 MHz
RTX 3060 12.0 GB

At least 3 attempts, rebooting, but still the same message.

Any recommendation?

Really strange, it fails to allocate normal host memory. Did you use -M 128 ?