My take:

1)Your $60/month dedicated machine is either 32+ cores and/or 10 years old in technology. As I (and others) detailed in this thread about a year ago, I have many Mac Mini M2 (8 threads) machines that use about 25watts of power under full load and many Wintel machines (16 threads) from 5 years ago that use 100watts under load. None of my machines are generating $60/month. My entire electric bill goes up about $250/month for running those 20+ machines...which means on average across all my machines, each is costing me about $15 a month. I think you will find my link a very interesting read. Due to a bunch of advantages of the Macs over the Wintels that I detailed on that thread, I'm trying my best to dump the Wintels and switch 100% to Macs.

2)I'm unclear if the charity you mention is the same project/cause as what you are crunching. However, on a personal level, I feel it's better for me to donate $5000 over the course of 1-3 years via WCG (I only crunch MCM) and electricity vs. giving some Cancer organization a check for $5000. Without going off topic too much, I just don't believe the money that is given/raised for Cancer is going to good (or at least efficient) use. I think about the billions and billions and billions of dollars that have been donated over the last 15-20 years, the endless road races, the non-stop sports Public Relations, Breast Cancer Awareness Month and all its publicity, etc. and there's almost nothing to show for it other than some headway here and there on very specific cancer issues. With WCG/MCM, I can see real, concrete results of my computers crunching. Now, the same may be true with MCM that we haven't had any major breakthroughs, but at least I know where my money went.

3)Contrary to what others have mentioned here, AI is not going to solve these DC projects faster or with less computing power. Computing power is computing power. AI (a software application) simply uses large models and databases to, well, be intelligent and find an answer or suggest an answer all by itself. If the math problem 1238934571371 X 485974352981234 is presented to either AI or a traditional calculator app on your personal computer, you're still gonna get a)the same answer and b)the calculator app would be faster because there's no AI acting as a middleman trying to decipher what you are asking, then asking the CPU to do the calculation, then the AI returns the result. Therefore, AI isn't going to make WCG crunching any faster or better. I am not an expect on how WCG and MCM WUs are designed and the math required to process them, but it's math. Math is math. If someone here would like to share in somewhat laymen terms how the WUs are crunched, that would be nice to read.

True, but I think a lot of people who donated to DC weren't just thinking "hey, I'll let my unused cpu cycles donate to this worthy cause"...many were thinking "hey, I can run this thing 24x7 or as close to that as possible to really help...and maybe I have a few other machines laying around or know someone that has numerous machines that would be willing to help"



That is what I was asking...and yes, that $250/month could be substantial or even worthless to some organizations depending on their size. I say this without tooting my horn and simply want to mention it as it's part of your question: my wife and I donate a LOT of money to various organizations every year...everything from the Diaper Bank to homeless shelters to local schools to Girls Who Code to local school scholarships to our Youth And Family Services org. It means a lot for us to give monetarily and these organizations can use the money.

But, cancer research is my #1 issue I want to donate to. I don't feel that it's worth me sending any one or any set of cancer orgs $5000+ a year. I just feel it's a drop in a very large bucket and the bucket never seems to produce results. I could be wrong, but that's how I feel. Again, with the MCM project, I can see actual work I'm doing, results, and what slice of the pie I'm helping. I'm a techie at heart...so if I can get some neat new machines every few years and set them up to run 24x7 crunching for cancer, that makes me happy...and happy from a "let's find a cure" point of view. :)

What's really cool is when people come to our house and see this datacenter of 20+ machines with wires and blinking lights and some fans humming and they ask what it's all about...and after the explanation they go home and at least look up WCG as well as consider that there are lots of ways to "donate" to cancer or other projects.

In this regard: the website totally dedicated to the Africa Rainfall Project: https://africarain.readthedocs.io/en/latest/

Adri

I disagree with your post about AI. It certainly speeds up to solve DC projects and we have seen that projects that have withdrawn from DC to go for an AI solution instead.

To keep it simple: A lot of DC projects are working with brute force method. Compare everything with everything or e.g. calculating with a certain algorithm what's the least necessary energy used to fold a protein an so on.
AI recognizes patterns and can exclude a lot of those pathes to get to the solution faster.

A very famous example is Deepmind with Alphafold. In some months after launching Alphafold 2 they folded successfully all known proteins, which was seen as "won't happen during our lifetime" before. And we were exactly doing that with Folding@home and Rosetta@Home. Certainly also with a bit of an another focus (finding the single steps in foldings proteins and what could go wrong / finding new proteins). There was not such a thing before and how most of the proteins fold was unknown until Alphafold came along.
There was a very nice comparison between how proteins were folded before. It took on a an at that time very good NVIDIA gpu one day to fold like 0,0001 nanoseconds of a protein. With the alphafold model they could fold it completely on the same hardware in around 10 minutes. I once read in an article that it speeds up 10.000 - 1.000.000 with Alphafold 2.

I think the same! Thanks!