ration of valid to invalid results

Redshift
Redshift
Joined: 20 Feb 05
Posts: 13
Credit: 285565
RAC: 0
Topic 187928

From the server status page:

Results VALID : 88,040
Results INVALID: 20,828

A bit less than 25% invalid.

I wonder if that indicates some problem.

---------------------------------------------------------------------------
stlforsale.com

Bruce Allen
Bruce Allen
Moderator
Joined: 15 Oct 04
Posts: 1119
Credit: 172127663
RAC: 0

ration of valid to invalid results

> From the server status page:
>
> Results VALID : 88,040
> Results INVALID: 20,828
>
> A bit less than 25% invalid.
>
> I wonder if that indicates some problem.

No. The category 'invalid' includes all results that have 'failed' for some reason. This includes for example work which never arrived at host machines, or which was trashed for some trivial reason. The number of successful results that are invalid is extremely small.

I decided that this was misleading and so have changed the server_stats page so that the invalid results are those which were successful on the host machine but later found to be invalid when compared against other results for the same WU.

Bruce

Director, Einstein@Home

Ronald McNichol
Ronald McNichol
Joined: 28 Feb 22
Posts: 27
Credit: 99853798
RAC: 0

RE: Invalid results. I

RE: Invalid results. I started keeping track of the could of the various categories as I run BOINC on my two Raspberry Pis (24/7) (4 jobs each), Laptop (16x7)(6 jobs), and desktop (11 jobs) (when it is on). Even as more and more jobs are completed, my invalid results counted down from 40 yesterday to 32 today. Is this like Lazeruz getting raised from the dead? I was thinking that number would be monotonically increasing. :o

 

Ronald McNichol
Ronald McNichol
Joined: 28 Feb 22
Posts: 27
Credit: 99853798
RAC: 0

Raspberry Pis. I have

Raspberry Pis. I have three Raspberry Pi 4s, two have 4GB and one has 8GB. I recently upgraded the 8GB one to the (official) 64 Bit Raspberry Pi OS. At first, it was crunching the same 32-bit jobs it had before the upgrade and was gaining credits at about the same rate as the unmodified one. Without ME doing anything, I guess it finally got some 64Bit jobs and is NOW earning credits >4 times what the unmodified one is getting! I was expecting probably double the number, due to the doubling of the register and ALU sizes. Perhaps doubling the number of GPRs also came into play. The remaining Pi 4 is used as a sharing file server, so I ain't lettin' BOINC touch that one! ;) As I have the first two Pis set up the way I want with a LOT of (unremembered) tinkering, I am leaving them alone with 32 Bit Buster. The 8GB one was already the outlier running 64 bit Ubuntu, 'til the Raspberry Pi OS 64 got out of Beta.

I was also surprised the GPU in my laptop's APU was used in addition to the dedicated nVidia GPU. I was also happy that the AMD GPU in my desktop is getting used as well. I was afraid it would be left out in the cold. 

Harri Liljeroos
Harri Liljeroos
Joined: 10 Dec 05
Posts: 4289
Credit: 3175381896
RAC: 1934680

The validated results are

The validated results are removed from the website database after a few days as it would grow too big if all results were kept visible for us. The scientific results are stored elsewhere. 

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.