pwhidden

Sorry calendar slipped a week!

Corrected version:

Thank you AL. I had one overnight as well.

078 indicates we are at about 42.6% but for this month I am assuming 2 generations behind to allow for the stragglers so 41.5%.

The latest interval is 4.13223 days and the 10-interval average is 5.25432 days. Consequently, the end date forecast is still January 2023.

Mike

The latest news says
"At the current pace, the project will be able to simulate an entire rainy season by the end of 2021. However, our tech team has determined that we can send out work more quickly."

how does that compare to the guess of Jan 2023 for finish?

I hope the 2023 number is right as I have a long way to get the year badge.

Some random thoughts on rainy seasons and project completion...

The major rainy season is, I think, April to June. There is a shorter rainy season from October to December, and we appear to be processing late November and early December at present. If someone knows otherwise, please set me right...

I believe the simulation started at 1st July 2018, in which case the end of the main rainy season will coincide with a year's worth of runs, which we're unlikely to hit this calendar year - it would require processing more work in just over 5 months than has been processed to date! I therefore presume they're referring to the other rainy season.

If the current rate of progress is sustained, we're likely to start seeing 2019 simulations in about a month or so (if I've done the arithmetic right!) but I suspect there'll still be 2018 stragglers out there for months yet. Perhaps that's why they cited the end of 2021?

I'd be surprised if the project will be cut off at what is effectively the half-way stage, so I reckon Mike's early 2023 completion estimates are probably still right, though that will depend on how many more work units they can get out (and get back!) at once - late 2022 might be possible without too much more activity. Unfortunately, many of the crunchers who were doing a bit of MIP1 may have machines better suited to MCM1 than ARP1, so it isn't clear how much extra ARP1 "crunching capacity" there might be. It'll be interesting to see if the returns rate increases significantly over the next few weeks - we can then have fun with new completion estimates!

Cheers - Al.

[Edit:] P.S. I have assumed that their "end of 2021" wasn't a typo for "end of 2022" :-)

My current target date is January 2023. That is based on the start of the last 11 generations to give 10 intervals. It automatically adjusts to changes in workflow over the following 10 imtervals, so one tenth of the change each generation.

Apart from entity's 15,000 limit, ARP1 does not include Android or ios as far as I am aware. The long time between checkpoints would be a problem for thase machines.

Mike

Thank you, Rayburner.

This is an oddity. It is very early and the deadline is 8 days instead of 7.

079 indicates we are at about 43.2% but for this month I am assuming 2 generations behind to allow for the stragglers so 42.1%.

The latest interval is just 2.68104 days and the 10-interval average is down to 4.95187 days. Consequently, the end date forecast is now December 2022.

Ordinarily I would expect the next generation to start about 25 July but it could be later to balance up this early start. Maybe 28 July!

Mike