Gingeropolous continued Monerosim work and experimentation, MRC repair / improvements
Who:
Gingeropolous, long time community member, tinkerer, maintainer of MRC, wrangler of monerosim
What
The first Monerosim CCS is complete - yay! Thank you prior donors! We have a functioning monero simulation tool, and hardware to run massive simulations. Now the task is to use it and improve it. There have been various requests from the community for various experiments in the simulation (most recently "the p2p ssl PR would be interesting to run on monerosim", and lets get cuprate running in there – I actually did the cuprate one). While monerosim was built so that anyone can use it, running large simulations requires massive hardware - a 16 node simulation runs fine on a standard desktop, a 1000 node simulation brings the MRCs largest machine to a crawl.
Currently, this hardware and software is being used to investigate eclipse attacks using additional tooling added to monerosim, and I’ve been investigating selfish mining mitigations.
Tasks
Task 1 - Monerosim Continued Development and Research
I am therefore requesting some funding to support my time and resources to running monerosim experiments and improving monerosim over the next 6 months. I can reasonably expect to spend 15 hours a week directly working on these things, using $65 / hour, this comes to about $23400. This will help cover my time actually running experiments and tweaking / improving monerosim. This time will also cover hardware maintenance time.
I spoke with Rucknium, and he has agreed to assist in the statistical analysis component of these experiments. (Rucknium mentioned they have some hours available in an existing CCS for this.)
Total for time = $23,400
Task 2 - hardware maintenance
As additional items / milestones, the MRC needs some maintenance and improvements.
The main storage array on Junior is now degraded due to a drive failure (these things have been spinning since 2022). Thus, i want to get a replacement for the failed drive and a couple spares. This comes to around $1.2k. Storage will become critical for the monerosim experiments and analysis.
Junior, (the threadripper with 256 gb of ram), has been a favorite of the MRC users, and its limited ram means that it enters swap storms often. While the ram is maxed out on the board, intel optane pcie 750 GB cards can be obtained for about $350. This will take the pressure off of the nvme array (where the current swap lives). And we can put a second optane onto the Dell R7525.
Truth be told, I wrote this in July, and sat on it, and now its october. I’ve already obtained the hardware specified above, and the ZFS array is back up to fully functioning status, the optane drive is in there as well. Also discovered the SLOG mirror nvmes had reached 125% of rated wear, so they were decommissioned and SLOG now lives on the optane, which is a much better home for a SLOG. So this component of the CCS is reimbursement as opposed to planned fixes.
Total for already acquired hardware = $1900
Additional hardware – Junior has been the most popular machine, I assume because of the 17tb of nvme scratch storage. I want to get Senior (the 1tb machine) to be more utilized, and I think it needs more storage to become a proper workhorse. I have my eyes set on some 7.68TB NVME PCIE3.0 2.5in U.2 SSD for about $900 each. Large monero simulations use a LOT of storage. I was going to get 2 of these, but I think 4 is the right move. bringing a scratch space on Senior up to ~30 TB.
Total for additional hardware = $3600
Notes
(a note on timeframe – as many may have noticed, I am horrible at estimating time frames, and I am dedicated to delivering something that is worthy of receiving community funding. For example, I thought monerosim could be done in the summer of 2025, and I just posted the milestone and received payout in 2026. Thus, I expect to complete a worthy amount of work in 6-8 months, but if it takes me another year to feel confident that I have done enough to warrant payment, then thats when I will request payout).
The MRC could use some newer generation compute (IPC increase numbers are real!!), but I’m hesitant to directly request funds for compute – my model has been that I buy the stuff thats directly useful for mining, and then request CCS to make the mining machines capable of performing monero research (more RAM and storage). The problem is that these days storage and RAM, especially for current generation cpu architecture, is crazy priced. And from my time watching usage of MRC users, higher thread, ram, and storage on a single machine has been preferred over distributed compute (so an ideal new compute purchase would be a single threadripper / epyc machine with massive ram as opposed to multiple 9900x boxes). If the community wants the MRC to have more modern hardware, let me know and we can make a second CCS for that.
So the total comes to $28900 , and the 21day EMA on price today comes to $539, so that comes to about 50 xmr. Add 2 xmr for volatility (added to first hardware tranche), 53.6 xmr.