Project 13424 (Moonshot) very low PPD

[Burlingtonian]

I realize there are myriad reasons for user attrition and have two suggestions for keeping FAH contributors engaged. First would be for FAH to send out to registered users a monthly or quarterly layman-friendly newsletter updating the program status, results and objectives going forward. Second would be for an automated messaging program acknowledging PPD goals reached, adjusted for active users' average PPD. In my case, where I am now generating around 2.9M PPD, perhaps every 100M points. These are just suggestions and I am aware that this information is available online but in this age of limited attention spans, I think that a slightly more proactive approach would be beneficial for user retension. Personally, I am in for the long haul and will be shopping the Black Friday/Boxing Day sales to upgrade my rig.

[cine.chris]

Happy Times!!
Thinking that I sound like a U-boat captain, perhaps?
5M this morning, 8.6M currently. 9AM>2PM
This reflects the previous results I was seeing after restacking GPUs on two dedicated folders in July.
System #1: ~3M PPD, usually hovers in low 3M.
System #2: ~4M PPD, usually hovers in high 3M.
Other points are from two 1660s on monitor station & cine-edit station, which is transient.
5ea-13422, 1ea-13424
Currently no 2080Ti, but it matched system #2 @4M** on 13422 WU with 89,709 atom count
** didn't have htm installed then, most WU iirc completed ~3.6M.
===============================
testing flickr bbcode
1wk of htm. Logged 283 WU.
Slots got shuffled after power failure reinstall, resulted in an extra data sets. I had a 2060s with a slight overclock + pwr cap, which was reset... which apparently, resulted in a 4th data group, whereas the 2070s data overlayed exactly.
Edit: 4th grp was an 80% pwrcap on the 2060KO after working through the color codes. Bumped it to 85%, now running @68C.
Also noticed that the HTM benchmark data tracks the GPU info... adding that to the tabular data generation would be useful.
Useful, as if I can validate which data group is connected to a specific config. It will likely require an export to a spreadsheet.
But seeing a delta visually across multiple WU projects within a couple of days will be a fast descriminator, forcing a slot switch so data isn't averaged shouldn't be to difficult. Much better than guessing.
Edit: It looks like simply changing the htm host name will work to create a new comparison data set. No, it didn't, but deleting the client in htm & adding with a different name appears to have worked.

2020-09-06-htm by Chris Pettit, on Flickr

[JohnChodera]

@Burlingtonian: Thanks for the great suggestions! I've passed these on to our internal team!

~ John Chodera // MSKCC

[AnimalFriend]

Here's a small taste of the real-time science dashboard that's coming for each sprint:
https://fah-ws3.s3.amazonaws.com/covid- ... index.html

Wow, this is sooo crazy cool! I think this is a great way to keep ppl engaged.

Amazing that you Not only get the biochemistry and Computer science stuff done, but also the crowd communication!

Sorry for the off topic post, I know this is a support forum, but I had to get this out of my system.

Best regards
Animal

[JohnChodera]

@AnimalFriend: The Sprint 4 Science Dashboard is live now too!
https://fah-ws3.s3.amazonaws.com/covid- ... index.html

We'll cover what this is showing in a blog post coming in the next couple of days.

~ John Chodera // MSKCC