Mad Max cryptocurrency - The Chia Plot

Every crypto is getting beaten right now, so there is no point for him to make waves, as that will just make some people go away for the lack of “progress.” If his target is around Jan, that implies it could be a bit earlier, or later. Maybe, if crypto is not going to start moving up, he will delay it even further to get better timing. If the test net will be working smoothly (no reason to think that it will not), he will get another investment to bring the project up when ready for production. Right now, there is no point to run it top-heavy. So, I would say that whatever we see right now is kind of irrelevant, but at least from the technical point looks rock solid.

On the other hand, looking how the node works, to me it looks much better than what Chia produced so far. Just one node process, but taking over all resources when needed (really multithreading, not multi-processing - as Chia does it), not choking on one ‘dispatching’ full node process. Just one process, so potentially no db corruptions from day one during shutdowns, … As you stated fast, … So, all that is really good, but again kind of irrelevant for time being.

Yeah, one can argue that Chia made a big blunder by not hiring him (and officially stating some lame excuses for it not happening).

Would seem so > k31s are ~51.5GB, k30s ~25.0GB

Have you run this in windows 10?

Yup, that zip file has MMX installer.

Yes, straight Win10 Pro install. It’s a single .exe file. That all and it’s done. Super easy, it’s on my everyday Ryzen 3600 using C: as the single plot temp and, I must say, an absolute no-brainer to use…refreshing. Even shows you the CL it’s using as a command it will run as you set the performance parameters. If only ‘someone else>’ had this orderly and logical an interface. And currently there’s 5.339 PB of NETSPACE running so peps are using the testnet. Just set 20 k31s going.

1 Like

Add to that that this is still far out test-net, but already has MMX exchange for testing. Looks like will be ready to go from day one.

I read somewhere, that the node will perform better if there is a video card in the box. I have it on Intel CPU with a built in video, and there is some usage on that GPU already, but not that much. Also, the requirement for that extra GPU card should be really, really low (like some Hercules type card).

It uses Chia’s PoS consensus but everything else is written from scratch by madMax.
It’s goals are defined in the wiki on github, as are instructions for installing and running it.
Currently testnet6, offering already some very impressive stuff, like smart contracts, exchange, staking, virtual plots and a browser based GUI.
Plotting is done with current version of madMax plotter, all options available but using a specific port number using a new -x parameter. Plots are like Chia’s in size.
You can’t use testnet plots for mainnet! Even so curent netspace is around 5 PB.
All nodes can (not must) run a timelord, the fastest one in the network at any time rules.
Timelord uses 2 CPU cores, so fastest CPU is fastest timelord.
A node performes VDF verifications using OpenCL when available, all available CPU cores otherwise.
Without GPU you need a really beefy CPU -many cores, high clockspeed) to keep up, meaning VDF verifications time below 2 sec (considered save from missing proofs).
GPU/OpenCL accelerates VDF verification and offloads the calculations from the CPU.
Using an Intel CPU with integrated iGPU you need coffee lake generation or later, NVidia K620 is highly recommended on the discord from speed vs power perspective. AMD GPU’s cause serious headaches;-)

Currently madMAX is writing his own database (!), as a few generally available ones all failed on one on the other demand his (and Chia’s) blockchain implies. Maybe Chia can improve on their db and borrow his code when available :crazy_face:

Wiki: https://github.com/madMAx43v3r/mmx-node/wiki

Discord: https://discord.com/channels/852982773963161650/925017012235817042

Some GUI screenshots:

1 Like

So this morning I installed MMX Node-0.6.12-win64.exe on a test node (Windows 10)

image
This is what it comes up with. So I went to the github site and got ver 10 and 11 same error. Is this bad Wednesday???

It may be that @drhicom is closing on a fix, so stay tuned.

I would only farm a PoST alt-coin to dump what I make immediatly, then go back to Chia when the size of the network shrinks back to unprofitability. Only worth it if you are a whale already with lots of plotting and farming capacity, even then it’s probably not going to net you anything substantial relative to Chia.

Latest supported Visual C++ Redistributable downloads | Microsoft Docs from our Guy Jacek. I installed the latest C++ modules and the App now works.

I won! I won! :cool: :boom: :100:

:baby: My first MMX - I’m so proud. All worthless 0.504343 of it!

2 Likes

I have just installed it on a Celeron J4105 / 12 GB RAM / 128 GB Gigabyte SSD Gigabyte NUC. I let it run / sync for about 5 minutes, and it download about 10% of the blockchain, so expected full sync time would be around an hour / two.

I shut it down and brought db that I synced yesterday on a bigger box. It took a couple of minutes to fully sync and is running nicely on it. It is using around 5-40% of CPU (in spikes), about 0-100% iGPU, and virtually no movement on the SSD (bc db), so looks like even on this underpowered box it still has plenty of headroom (maybe not for a dust storm, though, but this is still mostly a test code, not yet production ready, so too early to really test it).

On the other hand, when I was starting with Chia, and wanted to use this NUC as a node, I just gave up after a week of syncing as it was going nowhere (it was still relatively small bc db at that time, and yet it just couldn’t process blocks fast enough).

So, this node is written by mostly just one person in about 3-6 months, and basically it beats Chia’s node hands down. Sure, the overall node architecture was there, but all the code is written from scratch.

How long does VDF Verification take on that iGPU? Max advises to have this <2 sec to minimize the risk of losing a block-reward. With faster timelords in the network VDF verification time will go up (linear). As there will be some kind of reward for fastest timelord nodes with fast CPU’s will be tempted to turn timelord on and accept the extra Watts;-)
I have been following the discord for quite some time now. Impressive to see what this guy can do and how well informed on blockchain and applications thereoff he is. After giving his plotter to the community he took some serious time to study this from several angles and has been able to answer/explain his design choices to people who seem pretty knowledgeable on that stuff.

1 Like

Part of the testnets is to have duststorms. Next one organized will be to test the new db Max is making. Great fun to join and burn those testnet coins I farmed. Even though I have nothing to add on the constructive side of the project, I sure like demolition work :grin: :axe: :pick: :bomb:

How do I check the VDF speed?

Don’t know the Windows version but suppose it works similarly.
On the NUC run a browser and point to http://localhost:11380/gui/ to get to… the GUI. Password asked for is in the PASSWD file in the mmx-node directory.
The node tab shows the running log as in screenshot below.

VDF Speed is the fastest timelord, took… is your node’s VDF verfication.

I really would not worry at this point whether the box is fast enough or not, just would like to have it on the network, so it contributes to the overall testing. On the other hand, the fact that it runs on the box is really amazing. The fact that right now there is still plenty of headroom implies that it is close to being feasible when the production code will come.

I have also installed it on AMD Athlon 5350 APU (a bit slower than that Celeron, but not that much - also four cores). The iGPU usage is not that heavy (maybe, as it is faster) as on Celeron, but the CPU usage is tad higher. Stil, manageable, could potentially take 4-5x the traffic.

Yeah, I noticed that db is really fast, blazingly fast. It is almost too fast for this stage, I mean that a bit slower db would also do, giving it a bit more room to work on other parts. I am not that much fun of db side, so I would rather prefer to have a slower one, but potentially replaceable (maybe some sort of SQL? :slight_smile: ). At least all those are heavily tuned / optimized (yeah, for general usage, but still). It is just too often that a slow db is really the fault of the dev that doesn’t really know how to tune db, and how to layout tables (e.g., devs like me - barely knowing how to use it).

Here are my nodes: AMD APU:

Celeron:

In both cases, from about 1.8, up to past 3 secs. So, tad too slow, but it just means, more wins for you during the testing phase :slight_smile:

What is the unit for that speed here? It is M/S, nothing really that would make sense right away (like Million (in bits) per S). Was that meant to be milliseconds (ms, msec, or something like that)?

Which channel? I tried to watch it, but kind of cannot really stomach Discord.

Yeah, going back to Discord, still nauseated :frowning:

Mega Iterations per second I think. Must be what timelords do, iterating…

https://discord.com/channels/852982773963161650/925017012235817042

mmx-general mostly. Just scanning new entries most days, but agreed that it takes some persistence…