897 and 898 are N216 type, so they like lots of L3 cache.

This is longer than the previous batch (891), which ran for 9 1/2 days under comparable conditions.
So either they are bigger work units, or else take up more cache are are slowing down the processing.

Do you know which?

Batches 891 and 895 were both faster. All previous and now later N216 batches are slower. I don't know what the special case was with 891 and 895 but they seem to be the exception.

Estimated times for 895 and 898 are the same on my box. My #898s are waiting to run. I have suspended one of my 895s to compare but it will be a while till I know what happens on my machine. Also running 8 hadcm3s tasks alongside my N216 tasks my affect the answer.

---

Please do not private message myself or other moderators for help. This limits the number of people who are able to help and deprives others who may benefit from the answer.

Sorry, my bad, the fast ones were 889, 891 and 895, basically the CDDHDD N216 tasks. All the others were slower. So, when the 889 tasks were released in mid Dec, my i7 4790K grabbed 8 of them (all 5 month models) and finished them in a bit less than 930000 CPU seconds, running 4 at a time. That averages out to about 21.35 sec/TS. After those were through, the i7 grabbed 2 from batch 883 and 2 from 863 and those averaged about 22 sec/TS. So for running them over a week and a half, a 5 month run was about 8 hours faster in those batches. Not a big deal but noticeable when looking at sec/TS since it was consistent.

On my 3600X, running 5 N216 5 month models at a time, they averaged about 658000 CPU secs to complete. The older runs when they got thrown into the mix averaged about 680000 CPU secs. So a difference of 15.1 sec/TS to 15.6 sec/TS. Not a big deal, but consistent across batches.

Now that the hadcm3s are getting thrown into the mix, it makes the comparison harder.

The one hadcm3 that I have currently running is going at about 2.1sec/ts whereas the two N216s are about 50sec/ts if that is a guide.

The one hadcm3 that I have currently running is going at about 2.1sec/ts whereas the two N216s are about 50sec/ts if that is a guide.

What cpu are you using?

Both my Ryzen 7 3700x (3 cp + 4 Rosetta + 9 WCG/tn-grid) and my Ryzen 9 3900 (4 cp + 6 Rosetta + 14 WCG/tn-grid) are running at about 20 sec/TS for the 216s and about 0.7 sec/TS for the hadcm3s.

Interesting,
On my Ryzen3700X the 216s are all going between 16 and 19S/TS and the HADCM3'S AT 1.2S/TS. That is running 4x 216 tasks and 8 of the hadcm3s ones. Some of that is I suspect due to only having 32GB of RAM.

---

Please do not private message myself or other moderators for help. This limits the number of people who are able to help and deprives others who may benefit from the answer.

Interesting,
On my Ryzen3700X the 216s are all going between 16 and 19S/TS and the HADCM3'S AT 1.2S/TS. That is running 4x 216 tasks and 8 of the hadcm3s ones. Some of that is I suspect due to only having 32GB of RAM.

Woops, I’m still getting away with 16gb on each machine. I keep meaning to upgrade but I still have free memory with my mix of tasks.

I am also impressed, which amount some people need to run CPDN.
Right now i am running 3 N216 hadam4h wus in a vm, and they need together with the os 4,9-5gb of ram. all three together!
By looking per task, the need around 1,3-1,4gb/task.

Greets from Germany
Felix

Out of 8 hadcm3 tasks one of them has failed with, "invalid theta" which usually means in impossible climate has been produced e.g. a -ve pressure is one that has occurred in the past.

This one failed shortly before the 9th zip was created.

Out of 8 hadcm3 tasks one of them has failed with, "invalid theta" which usually means in impossible climate has been produced e.g. a -ve pressure is one that has occurred in the past.

This one failed shortly before the 9th zip was created.

That did make me laugh, "an impossible climate". Or.... have you discovered a new law of physics? Who's to say we can't have negative pressures? Contact James Dyson now!

You have'n' a clue about the subject matter, have you?

I now have four trickles, for my hadam4h (three 897 and one 898), and they are all 29 sec/TS.
It took three days for the first ones though on a Ryzen 3600.

i5-4690 @ 3.5GHz running Ubuntu20.04 in a Virtualbox VM. I think that VB has a high overhead somewhere as the N216 tasks run faster in a VMware VM.

i5-4690 @ 3.5GHz running Ubuntu20.04 in a Virtualbox VM. I think that VB has a high overhead somewhere as the N216 tasks run faster in a VMware VM.

That's what I found as well. I have not idea why.

You have'n' a clue about the subject matter, have you?

I have an honours degree in Physics, so yes I do. I also have a sense of humour and an open mind.

You have'n' a clue about the subject matter, have you?

I have an honours degree in Physics, so yes I do. I also have a sense of humour and an open mind.

OK, care to describe how a negative pressure (as an absolute value rather than a relative value) would work or is that an example of your humour?

You have'n' a clue about the subject matter, have you?

I have an honours degree in Physics, so yes I do. I also have a sense of humour and an open mind.

OK, care to describe how a negative pressure (as an absolute value rather than a relative value) would work or is that an example of your humour?

Same way as electron holes in semiconductors, antimatter, etc, etc. Just because we haven't discovered it yet doesn't mean it doesn't exist.

You have'n' a clue about the subject matter, have you?

I have an honours degree in Physics, so yes I do. I also have a sense of humour and an open mind.

OK, care to describe how a negative pressure (as an absolute value rather than a relative value) would work or is that an example of your humour?

Same way as electron holes in semiconductors, antimatter, etc, etc. Just because we haven't discovered it yet doesn't mean it doesn't exist.

You cannot have a lower pressure than an absolute vacuum in the same way that you cannot have a temperature lower than absolute zero. It isn’t a case of we haven’t discovered it yet - the scale stops.

You have'n' a clue about the subject matter, have you?

I have an honours degree in Physics, so yes I do. I also have a sense of humour and an open mind.

OK, care to describe how a negative pressure (as an absolute value rather than a relative value) would work or is that an example of your humour?

Same way as electron holes in semiconductors, antimatter, etc, etc. Just because we haven't discovered it yet doesn't mean it doesn't exist.

You cannot have a lower pressure than an absolute vacuum in the same way that you cannot have a temperature lower than absolute zero. It isn’t a case of we haven’t discovered it yet - the scale stops.

The scale we currently use. If you lived back when the earth was flat, going right round would seem equally insane.