[digiKam-users] Performance for facees db

[KRODER Gerhard]

Hi,

some question about performance...

I resartet update  of faces db  on all pictures, not just new ones. out of ~25k images there are ~10k faces found already in previous runs.

Now update runs since about 5h, and is at about ~10%.  Is this a reasonable time scaling? Any way to support/accelerate?

No other jobs running, only SSD in system, i57300U runs at 100% (digKam process takes ~ 80-95%);

[Gilles Caulier-4]

Hi,

The performance so far is not optimum, and we plan to work on this topic while this summer with a student. See my other mail on this list posted a few minutes ago.

Best

Gilles Caulier

Le ven. 14 mai 2021 à 09:34, KRODER Gerhard <[hidden email]> a écrit :

Hi,

some question about performance...

I resartet update  of faces db  on all pictures, not just new ones. out of ~25k images there are ~10k faces found already in previous runs.

Now update runs since about 5h, and is at about ~10%.  Is this a reasonable time scaling? Any way to support/accelerate?

No other jobs running, only SSD in system, i57300U runs at 100% (digKam process takes ~ 80-95%);

Cheers, Gerhard

[Gerhard]

Hi,

tnx for insights...  My system is now at about 22%, and occupies all CPU. I'd like to interrup an restart at night... If i would interrupt db update, will intermediate results be saved/accessable, or would they be lost?

Am 14.05.2021 um 09:44 schrieb Gilles Caulier:

Hi,

The performance so far is not optimum, and we plan to work on this topic while this summer with a student. See my other mail on this list posted a few minutes ago.

Best

Gilles Caulier

Le ven. 14 mai 2021 à 09:34, KRODER Gerhard <[hidden email]> a écrit :

Hi,

some question about performance...

I resartet update  of faces db  on all pictures, not just new ones. out of ~25k images there are ~10k faces found already in previous runs.

Now update runs since about 5h, and is at about ~10%.  Is this a reasonable time scaling? Any way to support/accelerate?

No other jobs running, only SSD in system, i57300U runs at 100% (digKam process takes ~ 80-95%);

Cheers, Gerhard

[Gilles Caulier-4]

No problem to interrupt. Press the button on the progress manager (bottom right side of main gui). It will shutdown the face workflow safety and close the database.

You can close digiKam no. At the next session, restart the face workflow as the previous stage, it will continue as well...

Best

Gilles Caulier

Le ven. 14 mai 2021 à 14:43, KRODER Gerhard <[hidden email]> a écrit :

Hi,

tnx for insights...  My system is now at about 22%, and occupies all CPU. I'd like to interrup an restart at night... If i would interrupt db update, will intermediate results be saved/accessable, or would they be lost?

Am 14.05.2021 um 09:44 schrieb Gilles Caulier:

Hi,

The performance so far is not optimum, and we plan to work on this topic while this summer with a student. See my other mail on this list posted a few minutes ago.

Best

Gilles Caulier

Le ven. 14 mai 2021 à 09:34, KRODER Gerhard <[hidden email]> a écrit :

Hi,

some question about performance...

I resartet update  of faces db  on all pictures, not just new ones. out of ~25k images there are ~10k faces found already in previous runs.

Now update runs since about 5h, and is at about ~10%.  Is this a reasonable time scaling? Any way to support/accelerate?

No other jobs running, only SSD in system, i57300U runs at 100% (digKam process takes ~ 80-95%);

Cheers, Gerhard