FYI, there has been no credit stats from MW since around the first of last month.
The have been no validations of pending credits since the work generator has been turned off.
Therefore no wingman validation _1 tasks have been created for over a month.
This lack of credit exports and lack of validatable work should exclude Milkyway from consideration for the Super Bowl Sprint.
Do not send me PMs. The PM system here doesn't notify like it's supposed. Contact me through Discord (link at top) or post in the forums here. I check those regularly. At least until I figure out the PM notification problem.
GPUGrid has not had "plenty" of work for the last two months.
2024-02-10 06_32_38-BOINC - Grafana.png (40.75 KiB) Viewed 9161 times
As you can see they have gone plenty of days with 0 work to send and only recently do they have a lot of work.
The other issue is whether or not their servers can handle a large load unexpectedly. They already limit communications to the project every ~30 seconds by IP (not host) which makes it extremely difficult for a user with multiple hosts to get a lot of work when they get blocked if they all ask within' a 30 second time frame.
The fact that they go through such lengths to prevent too many connections to their server leads me to believe their Internet/Network infrastructure isn't capable of handling it.
OK, I get your reasoning. I've not had any trouble getting work nor for contacting the project. That seems to have improved greatly from what it has been in the past.
The type of work they have had in the past made it not an issue when the tasks ran for 10-25 hours at a time. Current Quantum Chemistry work that you show in your graph is running from 15 minutes to 2 hours.
My teammates have had no issues keeping 2-20 gpus fed constantly all from the same IP address.
The network think seems more like DDoS prevention at GPUGrid. If I update some clients then go try and browse the forums it won't work. Waiting a couple minutes it's fine. Or switching to another IP via a VPN allows me to browse the forum.
Keith Myers wrote: ↑Sat Feb 10, 2024 5:52 pm
My teammates have had no issues keeping 2-20 gpus fed constantly all from the same IP address.
Which teammates? Because I know one who started paying for a VPN to fix the problem.
Do not send me PMs. The PM system here doesn't notify like it's supposed. Contact me through Discord (link at top) or post in the forums here. I check those regularly. At least until I figure out the PM notification problem.
