Climate has routine trickle credits which come in, which is also verification that it's making it to the next checkpoint. Some climate models were weaks though, not hours...

If the "excess" data is necessary; which is resulting in these failed uploads, it might be possible to provide a compression algorithm in the app depending on what the data is. If it's text, it should be very compressible, otherwise less so. This could then be decompressed server side to allow the data to go into whatever processing you have; so long as compression doesn't result in a loss of critical data due to compression errors.

---

It seems that today a new batch of ATMbeta tasks is on the field.
A subtle difference comparing to previous ones:
Name for my last task from previous batch was:
TYK2_m15_m16_3-QUICO_ATM_Sch_GAFF2_AIMNet2_RE-4-5-RND5965_0
Name for my first task from today's batch is:
CDK2_m10_m04_4-QUICO_ATM_Sch_AIMnet2_10-0-10-RND2956_0
If it is as usual, each task for a current investigation line will generate a chain of ten links (from 0 to 9), instead of previous 5 (from 0 to 4).
Verification pending...

what I notice is:
on both ATM tasks which I downloaded and started about 1 hour ago, the percentage progress bar in the BOINC manager works fine, i.e. it shows real values.
Either this was fixed, or it is coincidence.

We're still working on the first task in each sequence - 0-5, or now 0-10. This looks to be a big batch, so that may last some time.

The interesting point (the sudden jump to 100%) will come with the 1-10 tasks. But we're still working on the 1.09 application version from late March, so I'm not expecting a miracle.

what I notice is:
on both ATM tasks which I downloaded and started about 1 hour ago, the percentage progress bar in the BOINC manager works fine, i.e. it shows real values.
Either this was fixed, or it is coincidence.

Does that mean that ATM tasks are available now? I haven't received anything since.. May or June, if i remember correctly. Even now, i explicitly clicked the Update button - nothing. nVidia RTX3070 on Win 11, if that matters.

what I notice is:
on both ATM tasks which I downloaded and started about 1 hour ago, the percentage progress bar in the BOINC manager works fine, i.e. it shows real values.
Either this was fixed, or it is coincidence.

Does that mean that ATM tasks are available now? I haven't received anything since.. May or June, if i remember correctly. Even now, i explicitly clicked the Update button - nothing. nVidia RTX3070 on Win 11, if that matters.

the tasks now are "ATMbeta".

you need to go into your preferences and allow beta work.

---

the tasks now are "ATMbeta".

you need to go into your preferences and allow beta work.

That's been done long time ago. My current settings are:

ACEMD 3: yes
ACEMD 4: yes
ATM (beta): yes
Quantum Chemistry (CPU): yes
Quantum Chemistry (CPU, beta): no
Python Runtime (CPU, beta): no
Python Runtime (GPU, beta): yes

That's strange. 'Server Status' page reports 1,296 unsent units, but my BOINC client doesn't receive any. This is what i have in the logs:

31/10/2023 11:11:12 PM | GPUGRID | Sending scheduler request: To fetch work.
31/10/2023 11:11:12 PM | GPUGRID | Requesting new tasks for NVIDIA GPU and Intel GPU
31/10/2023 11:11:15 PM | GPUGRID | Scheduler request completed: got 0 new tasks
31/10/2023 11:11:15 PM | GPUGRID | No tasks sent
31/10/2023 11:11:15 PM | GPUGRID | No tasks are available for ACEMD 3: molecular dynamics simulations for GPUs
31/10/2023 11:11:15 PM | GPUGRID | No tasks are available for ACEMD 4: molecular dynamics simulations for GPUs
31/10/2023 11:11:15 PM | GPUGRID | No tasks are available for ATM: Free energy calculations of protein-ligand binding
31/10/2023 11:11:15 PM | GPUGRID | Project requested delay of 31 seconds

Got my first 1-10 task of this run - specifically,

JNK1_m06_m05_5-QUICO_ATM_Sch_GAFF2_xTB_MM_False_v3-1-10-RND0728_0

It jumped immediately to 100%, as before.

Same here for JNK1_m06_m05_2-QUICO_ATM_Sch_AIMNet2_10-1-10-RND3720 - direct to 100%, whereas TYK2_m03_m15_1-QUICO_ATM_Sch_GAFF2_xTB_MM_False_v3-0-10-RND2341_3 progress worked just fine.

Interesting thing I happened to notice though - which might or might not help the devs fixing this:
- On the 0-10 task, I noticed that progress jumped up in 1.446% increments every 5 minutes.
- On the 1-10 task, there's a 'progress' file in the slot directory with a number that jumps up in 0.01449 increments every 5 minutes. this number starts with a 1, like this: 1.391304348 => 1.405797101 => ...

Hypothesis is that this number should start with a 0 instead of a 1 and then progress would be reported OK?

<edit> would be interesting to see if I get a chunck 2-10 if this progress number starts with a 2.xxx then. Might be an 'easy' fix to subtract the chunk number from the progress number before writing?

the tasks now are "ATMbeta".

you need to go into your preferences and allow beta work.

That's been done long time ago. My current settings are:

ACEMD 3: yes
ACEMD 4: yes
ATM (beta): yes
Quantum Chemistry (CPU): yes
Quantum Chemistry (CPU, beta): no
Python Runtime (CPU, beta): no
Python Runtime (GPU, beta): yes

there's more to it. there are two options you need to select.

make sure this one is checked:
"Run test applications?"

---

Same here for JNK1_m06_m05_2-QUICO_ATM_Sch_AIMNet2_10-1-10-RND3720 - direct to 100%, whereas TYK2_m03_m15_1-QUICO_ATM_Sch_GAFF2_xTB_MM_False_v3-0-10-RND2341_3 progress worked just fine.

Interesting thing I happened to notice though - which might or might not help the devs fixing this:
- On the 0-10 task, I noticed that progress jumped up in 1.446% increments every 5 minutes.
- On the 1-10 task, there's a 'progress' file in the slot directory with a number that jumps up in 0.01449 increments every 5 minutes. this number starts with a 1, like this: 1.391304348 => 1.405797101 => ...

Hypothesis is that this number should start with a 0 instead of a 1 and then progress would be reported OK?

<edit> would be interesting to see if I get a chunck 2-10 if this progress number starts with a 2.xxx then. Might be an 'easy' fix to subtract the chunk number from the progress number before writing?

there's no need to reanalyze all this. it's been done ad nauseam many months ago and discussed over and over again. all of the behavior is known at this point.

tasks in the "0" group will all process and count the progress naturally.
tasks in "1+" groups will all jump to 100% after the extraction phase. but will complete successfully in the normal time if you leave it alone.

the root cause very likely has to do with how they have decided to configure their asyncre.cntl file. using a relative value rather than absolute based on what group it's in.

for all run types, the last line of this file is listed as just "MAX_SAMPLES = +70". it's very possible that the app is communicating to boinc the run percentage based on the ratio of current sample to max sample.

so for all "0" runs, the current sample will be less than the max sample. 1 through 70.

however, for all "1+" runs, the current sample will always be greater than max sample (71+). boinc can't report greater than 100% completion, so it jumps to 100% and stays there until the task is actually finished.

---

the root cause very likely has to do with how they have decided to configure their asyncre.cntl file. using a relative value rather than absolute based on what group it's in.

for all run types, the last line of this file is listed as just "MAX_SAMPLES = +70". it's very possible that the app is communicating to boinc the run percentage based on the ratio of current sample to max sample.

so for all "0" runs, the current sample will be less than the max sample. 1 through 70.

however, for all "1+" runs, the current sample will always be greater than max sample (71+). boinc can't report greater than 100% completion, so it jumps to 100% and stays there until the task is actually finished.

Well that would match with what I'm seeing, if chunk 0 = 1-70 and chunck 1 = 71-140, then progress would run 0-1 in the first case, and 1-2 in the second (and probably 2-3 for chunck #2 and so on).
If the app reports progress in that way, and they can just change it to subtract the chunck number, then that would bring all progress back to 0-1 range where it should be.
Sounds like a reasonably easy fix - if there isn't more to it obviously, and if they have the time to do that...

there's more to it. there are two options you need to select.

make sure this one is checked:
"Run test applications?"

Sure, i have it checked:

Run test applications?
This helps us develop applications, but may cause jobs to fail on your computer yes
If no work for selected applications is available, accept work from other applications? yes
Use Graphics Processing Unit (GPU) if available yes
Use Central Processing Unit (CPU) yes

I have both test and other applications, that's why i'm surprised to not have been receiving anything when WUs are available now.[/i]

what is your cache setting in BOINC? how much work are you asking for? I think with GPUGRID if you ask for "too much" you end up getting this response and getting nothing. if you're asking for something like 10 days of work, that might explain it. set your work cache to like 1 day or less.

---

I even removed the project from BOINC and re-added it, to no avail:

1/11/2023 9:46:37 AM | GPUGRID | New computer location: work
1/11/2023 9:46:39 AM | GPUGRID | Started download of logogpugrid.png
1/11/2023 9:46:39 AM | GPUGRID | Started download of project_1.png
1/11/2023 9:46:39 AM | GPUGRID | Started download of project_2.png
1/11/2023 9:46:39 AM | GPUGRID | Started download of project_3.png
1/11/2023 9:46:41 AM | GPUGRID | Finished download of logogpugrid.png (0 bytes)
1/11/2023 9:46:42 AM | GPUGRID | Finished download of project_1.png (0 bytes)
1/11/2023 9:46:42 AM | GPUGRID | Finished download of project_2.png (0 bytes)
1/11/2023 9:46:42 AM | GPUGRID | Finished download of project_3.png (0 bytes)
1/11/2023 9:47:11 AM | GPUGRID | Sending scheduler request: To fetch work.
1/11/2023 9:47:11 AM | GPUGRID | Requesting new tasks for NVIDIA GPU and Intel GPU
1/11/2023 9:47:15 AM | GPUGRID | Scheduler request completed: got 0 new tasks
1/11/2023 9:47:15 AM | GPUGRID | No tasks sent
1/11/2023 9:47:15 AM | GPUGRID | No tasks are available for ACEMD 3: molecular dynamics simulations for GPUs
1/11/2023 9:47:15 AM | GPUGRID | No tasks are available for ACEMD 4: molecular dynamics simulations for GPUs
1/11/2023 9:47:15 AM | GPUGRID | No tasks are available for ATM: Free energy calculations of protein-ligand binding
1/11/2023 9:47:15 AM | GPUGRID | Project requested delay of 31 seconds

My settings are:

Store at least 0.1 days
and up to an additional 0.75 days of work.

Are these settings too low?

those should be OK to get at least one task I think. as long as BOINC doesnt think they will take so long to miss the deadline.

---

My computing prefs:

1/11/2023 9:52:09 AM | | General prefs: using separate prefs for work
1/11/2023 9:52:09 AM | | Reading preferences override file
1/11/2023 9:52:09 AM | | Preferences:
1/11/2023 9:52:09 AM | | - When computer is in use
1/11/2023 9:52:09 AM | | - 'In use' means mouse/keyboard input in last 3.00 minutes
1/11/2023 9:52:09 AM | | - max CPUs used: 5
1/11/2023 9:52:09 AM | | - Use at most 85% of the CPU time
1/11/2023 9:52:09 AM | | - suspend if non-BOINC CPU load exceeds 80%
1/11/2023 9:52:09 AM | | - max memory usage: 7.84 GB
1/11/2023 9:52:09 AM | | - When computer is not in use
1/11/2023 9:52:09 AM | | - max CPUs used: 8
1/11/2023 9:52:09 AM | | - Use at most 85% of the CPU time
1/11/2023 9:52:09 AM | | - suspend if non-BOINC CPU load exceeds 80%
1/11/2023 9:52:09 AM | | - max memory usage: 14.12 GB
1/11/2023 9:52:09 AM | | - Leave apps in memory if not running
1/11/2023 9:52:09 AM | | - Store at least 1.00 days of work
1/11/2023 9:52:09 AM | | - Store up to an additional 1.00 days of work
1/11/2023 9:52:09 AM | | - max disk usage: 80.00 GB
1/11/2023 9:52:09 AM | | - (to change preferences, visit a project web site or select Preferences in the Manager)

those should be OK to get at least one task I think. as long as BOINC doesnt think they will take so long to miss the deadline.

My laptop used to crunch ATM tasks successfully in the past, before the summer break. Is there anything different now? I mean, system requirements - have they changed dramatically in terms of available memory and graphics cards?