Thats a pity because every time Seti goes down my 2990 with its 64 available threads runs out of work in a couple of hours and then goes on holiday. A bit of a waste because it is solely dedicated to seti.
jsm

This really only effects us "top 1%'ers" (and maybe some of the "top 5%'ers as well) so project wise it's better than upsetting rest with more server crashes than necessary. ;-)

Cheers.

just have to create the so famous and new Underclock Tuesday , the seti's green day :D

---

Thats a pity because every time Seti goes down my 2990 with its 64 available threads runs out of work in a couple of hours and then goes on holiday. A bit of a waste because it is solely dedicated to seti.
jsm

Some projects that run cpu crunching will accept "0" as the resource share. That project will immediately download and start crunching until Seti downloads more tasks. Then it will stop downloading and probably crunching until you run out of Seti tasks again.

The downside is some of those projects download a hugh cache and basically halt Seti processing for well after Seti comes back (I am looking at you Eienstien @Home[sp]) A possible fix is to set your amount of work to 0.5 with 0.01 for additional work. This MIGHT take care of load balancing with E@H which does accept 0 resources..

My favorite biology/medicine project won't accept 0 resources so I can't use it that way. :(

Tom

---

A proud member of the OFA (Old Farts Association).

My post was an observation not a complaint. I am surprised that it is difficult to organise the project in a way that does not ensure participants run out of work during server downtime. Indeed with my perhaps naive suggestion of a benchmark calculation it would be easy to set a 'backstop' (EU please note) so that rinse and repeat did not create a runaway situation. Also a recalculation at each benchmark run would take care of reduced or increased capability cpu wise.
As the previous owner of a number of computer companies including an ISP I am horrified that a figure of 3/4 million dollars might be required to stabilise the project!
jsm

Don't be too surprised there are quite a number of ageing servers (probably the cause of a number of the recent crashes), and there are many Tb of "live" storage.
The public facing servers are:

bruno: Intel Server (2 x 2.66GHz Xeon, 8 GB RAM)
carolyn: Intel Server (2 x quad-core 2.4GHz Xeon, 96 GB RAM)
centurion: Intel Server (2 x hexa-core 3.4GHz Xeon, 512 GB RAM)
georgem: Intel Server (2 x hexa-core 3.07GHz Xeon, 96 GB RAM)
khan: Intel Server (2 x 3.0GHz Xeon, 32 GB RAM)
lando: Intel Server (2 x quad-core 2.4GHz Xeon, 12 GB RAM)
marvin: Intel Server (2 x 2.66GHz Xeon, 16 GB RAM)
oscar: Intel Server (2 x quad-core 2.4GHz Xeon, 96 GB RAM)
paddym: Intel Server (2 x hexa-core 3.07GHz Xeon, 132 GB RAM)
synergy: Intel Server (2 x hexa-core 2.53GHz Xeon, 96 GB RAM)
muarae1: Intel Server (2 x hexa-core 3.07GHz Xeon, 76 GB RAM)
muarae4: Intel Server (2 x hexa-core 3.07GHz Xeon, 76 GB RAM)
thumper: Sun Fire X4500 (2 x dual-core 2.6GHz Opteron, 16 GB RAM)
vader: Intel Server (2 x dual-core 3GHz Xeon, 32 GB RAM)

If you look on the Server Status Page (https://setiathome.berkeley.edu/show_server_status.php) you will see more or less what each of the servers does for a living.
And there another couple that lurk in the background that we never see.
What the list above doesn't show are the disk arrays - someone may know the actual figures, but last time I checked I think there were three arrays of about 32 in each, each array running some form of RAID. Plus a couple of smaller arrays and individual disks or small clusters....

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Aging indeed. But how many Epyc servers would you get for 3/4 million dollars even accepting that they have to be installed and configured?
jsm

But how many Epyc servers would you get for 3/4 million dollars even accepting that they have to be installed and configured?

What we really need is new storage, preferably an AFA (All Flash Array). Several of them, and they don't come cheap. Then several new servers to be able to take full advantage of that new storage.
A couple of hundred thousand would get some very capable hardware, but it would be nice if it were capable for a long time to come, not just the next couple of years. And while you do get warranty with such hardware, a service contract for on going support would be nice- and allow Eric to spend his time doing work other than just keeping the servers up and running. Hence my WAG (Wild Arse Guess) of around $750k to replace the existing servers and storage.

---

Grant
Darwin NT

Weekly report. Full week at 75% gives ratio of 1.16 so the trend is dropping but it is still holding up quite well considering 10% less cpus working. Now dropped to 65%.
Sorry cannot interpret any graph on BOINC stats to answer the query. Probably me.
jsm

Weekly report. Full week at 75% gives ratio of 1.16 so the trend is dropping but it is still holding up quite well considering 10% less cpus working. Now dropped to 65%.
Sorry cannot interpret any graph on BOINC stats to answer the query. Probably me.
jsm

What about the BoincManager Statistics tab?

Tom

---

A proud member of the OFA (Old Farts Association).

Tom
I have screenshot the user average. Is that what you want to look at?
If so please send a note to my private email and I will reply as I have mislaid your private email ):
I dont think I can post a doc file here which has the screenshot on it but if I can by all means tell me how.
jsm

If you've got an image you want to post then you can simply save it to one of the many, many image sharing sites, then post a link to it here, enclosing the link between [ img ] and [ / img ]

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Does this work?
img mckenzietrust.com/screenshotboincstat.pdf/img
or does it need []?

Sorry not to be up to speed with the protocol - it is new to me. However if you use Firefox or Edge on that url you should get the picture.
jsm

No you need the brackets. For how to use the BBCode tags refer to the help document that is present in any message creation/editing page at the top of the editing window on the left.
https://setiathome.berkeley.edu/bbcode.php

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

As a PDF it might not work, so try using the "url" tag pair instead.

Keith - It never fails to amaze me how many people don't see that link when trying to compose "fancy" posts.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

or perhaps it should be the url tag - but it is not an htm file

jsm
p.s. I am trying.....

As a PDF it might not work, so try using the "url" tag pair instead.

If you are trying to show or view a PDF in this forum or your browser it will NOT work.

The only option as Rob points to is to use the [ url ] tag.

This allows for someone to download and view the PDF - IF they have a PDF reader ;)

MY Image Wrapped In A PDF

[url=mckenzietrust.com/screenshotboincstat.pdf]MY Image Wrapped In A PDF[/url]

Having said that. If your browser does show PDFs in frame then once you click the link it may offer to open the file in the browser window itself.... BUT it will have to be downloaded first in the background and still won't show as part of the forum page.

Another option is to find a service that decompiles a PDF to an image file (jpg, png, bmp, gif, etc..)

Or of course don't use "Save To" PDF in the first place.

Personally I dislike saving a simple image file wrapped in a PDF :(

7 day report still holding up at 1.23. This is quite interesting. Now reduced to 55%
jsm

7 day report still holding up at 1.23. This is quite interesting. Now reduced to 55%
jsm

Ha ha . . . . . shouldn't say anything . . . . but . . . . . isn't this what we predicted and suggested way back at the beginning? I guess, comforting to know you proved it to yourself on your own hardware.

Keep going until you hit the other side of the bell curve and then can definitively know just where optimum utilization for the cpu resides on your cpu.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

7 day report still holding up at 1.23. This is quite interesting. Now reduced to 55%
jsm

Ha ha . . . . . shouldn't say anything . . . . but . . . . . isn't this what we predicted and suggested way back at the beginning? I guess, comforting to know you proved it to yourself on your own hardware.

Keep going until you hit the other side of the bell curve and then can definitively know just where optimum utilization for the cpu resides on your cpu.

But, he is generating lots more discrete data points which are good when you are trying to firm up the reliability of the results.

If he was trying for the absolute maximum RAC results per machine he would be running as many gpus as he could afford and fit on each machine. So his RAC on a cpu only machine could become some of the highest possible for the number of cores/threads being crunched.

Don't get me wrong, some of the Intel "heavy metal" Seti machines out there have a lot more active threads/cores and are not necessarily running a lot of gpu tasks. So that is why I said "some of the highest".

-edit---
I was looking at the wall clock time on the Threadripper tasks, and I am not seeing any of the 4 hour+ times. The highest I saw with a quick skim were about 2 hours and 15 minutes. -edit--

Tom

---

A proud member of the OFA (Old Farts Association).

Well one of the side benefits of a lower thread count is that the running temp is 10 degs C lower than with 95% utilisation which has to be good - and this m/c has high level liquid cooling in an EATX.
jsm