Half-day task without pause may not suitable for me...
By the way, it's also a trouble for my poor laptop that, install gbs of thing in ...\slot\1\ for every task, rather then in ...\projects\www.gpugrid.net\ for once.

:(

Tasks are always copied into and crunched in a slot in BOINC.

Never in the main BOINC directory or project directory. That is by design.

yup. the only slight difference with GPUGRID ATM (and QChem) is that the environment package is very large (2-3GB when extracted). you do need to provision for that in this case. you don't notice it nearly as much with other projects shipping tiny binaries.

this is the way the project has decided to run their work.

but BOINC is all about choice. if the applications here do not mesh well with your system, probably best to find a different project that works better for you.

---

Boinc server doesn't have correct character library.

You can use this https://www.online-decoder.com/

Heads up: A new batch of ATM (not Beta) was released this morning - just 200 tasks, all allocated now, but it might be a sign of things to come.

Two things to note:

a) because of the shared DCF at this project, they were given a very low initial runtime estimate - less than 10 minutes, but likely to take nearer 4 hours on this card.
b) the task limit has also increased - up to four per card.

Don't be too greedy fetching work until this settles down!

Two things to note:

a) because of the shared DCF at this project, they were given a very low initial runtime estimate - less than 10 minutes, but likely to take nearer 4 hours on this card.
b) the task limit has also increased - up to four per card.

Don't be too greedy fetching work until this settles down!

Thank you for your notes.
Maybe a third thing to point:
I've noticed that recently the delay for GPUGRID server to attend task requests has been decreased to 11 seconds, instead of previous 31 seconds.

| GPUGRID | update requested by user
| GPUGRID | Sending scheduler request: Requested by user.
| GPUGRID | Not requesting tasks: don't need (CPU: ; NVIDIA GPU: job cache full)
| GPUGRID | Scheduler request completed
| GPUGRID | Project requested delay of 11 seconds

This affects to every apps, not only ATM.
And for small batches, it can cause that tasks ready to send decrease triple the rapid until they run out.

bonjour y a t il des taches quantum chemistry pour windows?


hello are there quantum chemistry stains for windows?

---

Good morning, is it possible to process processes lasting 7, 10 hours with RTX 3xxx series cards or possibly resume from the point of view that was stopped without making an error?
Thanks

Good morning, is it possible to process processes lasting 7, 10 hours with RTX 3xxx series cards or possibly resume from the point of view that was stopped without making an error?
Thanks

Process tasks in less than 7 hours with RTX 3000 cards?

YES

Stop running tasks and resume them without error or losing progress?

NO

bonjour y a t il des taches quantum chemistry pour windows?


hello are there quantum chemistry stains for windows?

not yet.

---

Gonna go out on a limb here and say a very high percentage of the computers that look for work units from you are Windows based. Why on earth would you design something for Linux first?

Because that’s what they know and use in their lab. They develop on Linux, test, tweak, then port to Windows later.

---

Gonna go out on a limb here and say a very high percentage of the computers that look for work units from you are Windows based. Why on earth would you design something for Linux first?

Very, very little in the world of computational science starts in Windows. You should see the world of genomics! Basically nothing can be accomplished in Windows.

The non-beta ATMs are flowing again. Hopefully I can get up to 11 completions so the server will start to recognise the true processing speed this time.

This seems to be a second, larger batch of 0-7 tasks. Waiting to see if the runtime progress bug has been finally eliminated for this app.

Waiting to see if the runtime progress bug has been finally eliminated for this app.

I know this was fixed on ATMbeta, did it not translate to ATM on the previous batch?

---

They were all 0-7 too - and progress during that first segment was always OK. I've not seen a 1-7 or later yet.

b) the task limit has also increased - up to four per card.

I can say that running 4x on the 4090 has pushed the GPU harder than any other work in the context of power utilization but are running really smoothly. I won't probably run 4x for too long though- pulls almost 400w(!) (never goes above 58c though).

My time estimations look like they are working on the systems I have checked today.

Edit: Time estimation is not accurate on all of our systems (yet).

"No tasks available for ATM", says BOINC

700+ tasks ready to send, says gpugrid.net.

???

"No tasks available for ATM", says BOINC

700+ tasks ready to send, says gpugrid.net.

???

Same problem here.

Edit: My linux machines can get the ATM tasks. But my windows machines cannot.

Update: My win machines finally got some of the ATM tasks. Not sure what changed, but happy to see it.

---

Reno, NV
Team: SETI.USA