I think the performance tab is not updating for a couple of days.

Yes🎃

Thanks for letting us know. We'll look into it!

Is it updated now? The database is showing me the same results as you see in performance tab.

Is it updated now? The database is showing me the same results as you see in performance tab.

Yes, it's up to date.
However there in an anomaly:
This workunit took 19181.034s to complete, however according to the performance tab, it took only 2.59 hours (~ 9324s).

My GTX 980 host is sooooo fast, that it took the first and the second place at the same time on the performance ranking:




EDIT: workunit 10529667 and 10571801 are not in the same batch.

Haha, I guess it's never too late to discover new bugs :D It seems to me that as you have two WU with exactly the same computing time, mySQL decides to reward you with the two first positions. KUDOS!

Haha, I guess it's never too late to discover new bugs :D It seems to me that as you have two WU with exactly the same computing time, mySQL decides to reward you with the two first positions. KUDOS!

+1 ;)
Impressive stats for your GTX 980 Retvari Zoltan :)

---

[CSF] Thomas H.V. Dupont
Founder of the team CRUNCHERS SANS FRONTIERES 2.0
www.crunchersansfrontieres

Top performers per batch- please update this section -)

I did a GERARD_FXCXCL12_LIG_1214901 on my oc 780ti in 7.71 Hrs but does not show up in THE TOP PERFORMERS IN BATCH list even though thats faster than any on the list (so far).

Also the Top average performers (last week long runs) has the same user in 8 9 and 10 spot?

The "Top performers per batch" section takes the "GPU description" from the host's properties.
This results in misleading information for hosts which have mixed GPUs, or a recently changed GPU.
Could this behavior be changed to take the GPU description form the task the "WU id" column refers to?
Even this column should be changed to show the task, not the WU id.
Also there should be a timestamp on this page.

Thanks for your feedback! I'll look into it in the following days.

(a) I have added a timestamp.

(b) I now link to task instead of WU id.

(c) Could you post an example of such description misleading information? I have checked couple of examples and host description is consistent with the log file of the task.

(a) I have added a timestamp.

Thank you.

(b) I now link to task instead of WU id.

Thank you.

(c) Could you post an example of such description misleading information? I have checked couple of examples and host description is consistent with the log file of the task.

Sure.
On the current "Top performers per batch" table on the GERARD_FXCXCL12_LIG_12411681 tab at the first place is petebe's task# 14489902 which were actually run on a GTX 980, but the GPU description says "[3] NVIDIA GeForce GTX 980 Ti (4095MB) driver: 353.30". His host has multiple but mixed GPUs (GTX 980 and GTX 980 Ti). There is the same inconsistency on the GERARD_FXCXCL12_LIG_13878001 tab's 1st place (petebe too) task# 14516484
EDIT:
On the GERARD_FXCXCL12_LIG_14701011 tab 1st place (my host) task# 14512608 the GPU description says "NVIDIA GeForce GTX 980 Ti (4095MB) driver: 355.82" but this workunit was crunched on a GTX 980, which I've replaced by a GTX 980 Ti since then, and I also updated the driver (the task's Stderr output says: "# Driver version : r353_23 : 35330").

Hmm... I will have to investigate further. Do you guys set the Computer Information manually or it gets automatically updated every time you make a change in the computer? I ignore this particular behavior or BOINC. If the update occurs automatically, I suspect it takes some time to become effective.

Do you guys set the Computer Information manually

I don't.

or it gets automatically updated every time you make a change in the computer?

Yes.
But for hosts with mixed GPUs only the best one is showed (and used by default) by the BOINC manager, so if you (apparently) take the given host's GPU name from what the BOINC manager reports, it will be always inaccurate for hosts with mixed GPUs. The only accurate source for this information is the stderr output, but it's difficult to use, as if a given workunit is switched over to a different GPU (for example by a system restart), then the stderr output will have two (or three etc) different GPU names.

I ignore this particular behavior or BOINC. If the update occurs automatically, I suspect it takes some time to become effective.

The refresh of a host's details occures right after the first time the host communicates with the project.

I'm afraid there's no easy way to set this up correctly... I guess users interested in particular systems will still have to inspect the standard output if they want to know the exact details of the configuration.

Best,

Gerard

I'm afraid there's no easy way to set this up correctly...

I thoght so.

I guess users interested in particular systems will still have to inspect the standard output if they want to know the exact details of the configuration.

It's easier than before, so I think it's good enough.

I have another suggestion:
Could you make the list of the top performers begin in the same position as the selected batch? As the list of batches in the left is three times longer than the list of the top performers, it is so time consuming to scroll up & down that it makes this list nearly unavailing.

One might say that history repeats itself, but it's simply a bug which is more hard to fix than you thought.
This time my GTX 980 Ti host is sooooo fast, that it took the first and the second place at the same time on the performance ranking:


Just as for the first time, these workunits are not in the same batch.
14511591: e3s13_e2s8p1f553-GERARD_FXCXCL12_LIG_8902112-0-1-RND1917_0 and
14528607: e5s9_e2s5p1f150-GERARD_FXCXCL12_LIG_6067831-0-1-RND5688_0
(I've found it under GERARD_FXCXCL12_LIG_606783)
To be honest, I think it's sooo incredibly fast, that it could even take the first three places at the same time ;)

Thank you Gerard for putting a scrollbar to the list of batches.