Question about the Madmax new pool


I ask stupidly into the group because it is not yet very clear to me how what works here.

1: Madmax has developed a new plotting method that generates the plots via the GPU (?)
2: This new method can also compress the plots. on around 70GB (-30% smaller)
3: The Chia client recognizes these plots as valid because the GPU unpacked the plot requested and thus outputs as a full plot.
4: These plots must be assigned “Nossd Pool”. (Other pools don’t go?)
5: This “Maxmax-Nossd pool”. Is that now n Forck or a full chia pool like Space pool or something?

Maybe someone can answer that with short sentences.
Many Thanks

1 Like
  1. Yes, to my understanding.
  2. They are not “Compressed”, they are basically incomplete plots. Missing data. The data is then created/calculated when the plot is called for a challenge. So it takes up less space but more processing power to use.
  3. No, the chia client does not recognize these plots. Only in Madmax’s pool (at least that is my understanding). NoSSD also has a similar format only recognized in their pool. Supposedly, Chia is working on an official plot version with these features.
  4. See #3.
  5. No, in the end, they are recognized as real plots for the official chain.

Have a great day!

1 Like

Max is having a pool party, who is bringing the beer???


to 2: ahhh… now I understand the whole thing.

But isn’t that a limitless waste of electricity?

If I start chia then all plots are read in briefly.
I have almost 5,000 plots.
They are then generated for a short time and then discarded again. later they have to be generated again. When I participated 3 times, I generated almost 1 plot (at 30% compression).

And that also takes time. So far I have about 0.2ms. at 0.5ms you’re almost out.
It can’t be that fast to finish (?)


Agreed and agreed. The arguments I have seen so far point to the space savings being worth the trade off. I shrug at this point. If you’re unsure on how to proceed you could take the cautious approach - old plots are still making XCH and it will take a couple of months for the whales to replot PiBs of storage (or at least get the files transferred, that seems to be the bottleneck now) anyway.

Also it’s kind of like the ASICS argument: look how efficient my silver little box is - it does the work 90 desktop computers can do. The problem is that everyone now has these shiny silver boxes pulling 1200 watts and Bitcoin is the scourge of the planet.

If the plots are not compressed, why do they mention it as compressed plots? Isn’t this a misleading? It is told in such a way that the new plot is advertised to everyone. Also forcing farmers to spend more money to buy a rtx graphic card. Moreover, you should also buy a new psu with at least 300 - 400 watts higher. Also the electricity cost will go up?


You are correct on all points. Why are they called compressed? Marketing I guess. Because they are not actually compressed. They are just smaller. Not from compression but by just leaving out parts of it that can be quickly created on demand. Silly, but whatever. And yes, additional hardware needed, and more electricity. People say that the savings in space needed (or the additional earnings from more plots per TB) offsets the additional cost of hardware/electricity. But I can’t confirm that. I’m sure as it becomes more and more established, there will be plenty of spreadsheets outlining the costs/benefits.

For me, I won’t do any of them that restricts me to a specific pool. Once there is an official version that makes them portable, I’ll consider it. Until then, to me, it’s just noise.


I think they are called compressed, because if he calls them condensed, people will complain that they can’t make Viet coffee with them.

I think condensed would be a better term, since you apparently have to add something (i.e. GPU juice or CPU juice) to make them whole again?

What isn’t, if done enough?

You can probably stay on the standard farmer and just eventually have earnings drop relative to the Chiasphere as a whole.

I would expect that it’s not every time a plot is considered, but every time a plot passes the filter. I’m not sure on this.

I’m not sure where these numbers come from. You have 25 seconds roughly to submit a partial. That’s 25000ms, not 0.2ms.

There are a lot of people ignoring the advice to wait for the details to come out before panicking and setting their rigs afire with plastic explosives or something in protest.

When the general Chia technology is released, there will be more detail and explanations available. Same with Max’s Chia version of his plotter.

It’s not reasonable to assert that Max’s MMX plotter and Chia’s new plotter and arcplot and nossd are the same thing.

It is reasonable to assert that when the Chia products come out from Chia and Max, they will work with existing standards-compliant pools in short order, just like pooling did back in July 2021. Some pools were faster to adopt, some took a bit longer for various reasons, a couple didn’t go to the new protocol and those probably won’t work with this development either.


It is suggested to take a look at the official description of this: GPU Plotting is Real - and Very Fast - Chia Network

No, nothing has to be done when loading the plots. Lookups require some compute, and when you find a proof it requires some more compute. It’s all on-demand, nothing is done for no reason and then discarded.

It’s more energy efficient up to a certain compression level, when comparing rewards / Watt. The exact level depends on your setup and future plot filter reductions.

1 Like

This is all very interesting to someone who actually enjoys the technical and hardware aspects of running a farm beyond the monetary. Out of curiosity, are the compressed plots just for MMX or regular Chia?

Thanks, Henry

1 Like

For now my software only works on MMX.

1 Like

didn’t you say it’s not a fork (?)

Mmx is not a fork, no.