Questions, comments and problems on new Fermi LAT gamma-ray pulsar search

Richard Haselgrove
Richard Haselgrove
Joined: 10 Dec 05
Posts: 2139
Credit: 2752987905
RAC: 1373867

RE: RE: Depends on how

Quote:
Quote:
Depends on how that feature is implemented. If it's implemented by a way that requires no more than having each workunit download a file specific to your settings, app_info.xml shouldn't be needed. If it's implemented by a way that requires the OpenCL feature allowing GPUs to run more than one program at a time, it may well take until sometime after a new version of BOINC provides more support for OpenCL GPU workunits.

I think you missed Richard's point: He already *has* an app_info.xml file installed, in order to do several BRP3/4 WUs in parallel. If you are using an app_info.xml file, only those apps listed in that file will get work, all other apps that are supported by the project, but missing from the app_info.xml file, are ignored. So the question was to receive the data needed to be included in app_info.xml file to support the Fermi search.

HBE


Actually, I think that was Sid, in message 112884 - I'm running Einstein au naturel, as nature intended. But Bikeman's point is well made - it's only if you have an app_info.xml file already that you would be excluded from the LAT tests without manual updating.

And if you did have Fermi-LAT in an app_info, you might have been caught out by the silent upgrade to v0.22 last Thursday: it's good to see some new test WUs.

One question, perhaps for Bernd: I see 240202413 and 240204286, created less than ten minutes apart, in the queue for the same computer - yet the second is estimated to run for five times as long as the first (2hr 08 vs. 10hr 42). Are we trying out some test scheme for processing variable length WUs, too?

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4267
Credit: 244933143
RAC: 16332

The tests with the last app

The tests with the last app version 022 were pretty successful, and we intended to start generating WUs continuously today. However apparently there seems to be a bug in the validator (and a less important one in the workunit generator) that led us to disable both for now until this has been solved.

It could actually be that the workunits end up in different 'sizes', which in this case would mean a different number of sky points (visible in stderr). Flops estimation and credit should be adjusted automatically.

BM

BM

[AF>EDLS] Polynesia
[AF>EDLS] Polynesia
Joined: 1 Apr 09
Posts: 24
Credit: 2273003
RAC: 0

A new application GPU? great

A new application GPU? great

robertmiles
robertmiles
Joined: 8 Oct 09
Posts: 127
Credit: 20996335
RAC: 79119

RE: RE: Depends on how

Quote:
Quote:
Depends on how that feature is implemented. If it's implemented by a way that requires no more than having each workunit download a file specific to your settings, app_info.xml shouldn't be needed. If it's implemented by a way that requires the OpenCL feature allowing GPUs to run more than one program at a time, it may well take until sometime after a new version of BOINC provides more support for OpenCL GPU workunits.

I think you missed Richard's point: He already *has* an app_info.xml file installed, in order to do several BRP3/4 WUs in parallel. If you are using an app_info.xml file, only those apps listed in that file will get work, all other apps that are supported by the project, but missing from the app_info.xml file, are ignored. So the question was to receive the data needed to be included in app_info.xml file to support the Fermi search.

HBE

I think you're missing something. Have you considered the possibility of having each workunit download a user-specific file, not app_info.xml, that contains all the information needed to tell the application programs they can run more than one workunit (of the same type) at once on GPUs?

I suppose you could find it easier just to offer the missing app_info.xml section, though.

robertmiles
robertmiles
Joined: 8 Oct 09
Posts: 127
Credit: 20996335
RAC: 79119

RE: We will probably send

Quote:

We will probably send out the first few tasks (for testing) on Monday.

If you don't want to receive any, opt-out of the "Gamma-ray pulsar search #1" in your Einstein@home project preferences.

BM

How can I opt IN?

When I looked at my preferences, there was no checkpoint for this search, and it ignored all my attempts to add one.

For my laptop only, though; not my desktop at present.

Gundolf Jahn
Gundolf Jahn
Joined: 1 Mar 05
Posts: 1079
Credit: 341280
RAC: 0

Those are the applications I

Those are the applications I can choose from:[pre] Run only the selected applications Binary Radio Pulsar Search
Binary Radio Pulsar Search
Gravitational Wave S6 GC search
Gamma-ray pulsar search #1[/pre]
Gruß,
Gundolf

Computer sind nicht alles im Leben. (Kleiner Scherz)

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4267
Credit: 244933143
RAC: 16332

RE: I think you're missing

Quote:
I think you're missing something. Have you considered the possibility of having each workunit download a user-specific file, not app_info.xml, that contains all the information needed to tell the application programs they can run more than one workunit (of the same type) at once on GPUs?

This is something you need to tell the BOINC Core Client, not the (Einstein@Home) application. The Client schedules the GPU devices, not the App.

BM

BM

Richard Haselgrove
Richard Haselgrove
Joined: 10 Dec 05
Posts: 2139
Credit: 2752987905
RAC: 1373867

Well, actually where it

Well, actually where it belongs is in the section of client_state.xml (specifically, the tag within ). You, the user, can put it there yourself via an app_info.xml file, or the project can put it there via the application details on the server.

The trouble with asking the project to do it is that the server setting applies to everyone, and not every user wants to go down this route - some cards (like my 9800GTs, for example) perform well on single BRP4s, but lack the hardware context switching facilities of the later Fermi-generation GPUs which is what makes running multiple instances viable in the first place.

Anyway, all of this is out of place amongst the technical news of the gamma-ray search - this is a CPU application only, at least for the time being.

@ Bernd - now that we have three different classes of application running on the project, and Fermi-LAT tasks are still much rarer than the other two, it's getting quite difficult to pick out and monitor results for a particular application. I know you have a long-term project (perhaps stalled while the new apps roll out) to upgrade the entire website, but in the meantime would there be any possibility of importing the result-display-filtering code in newer copies of results.php? There was a further useful modification a couple of months ago, in [trac]changeset:23624[/trac], to add task counts to the filter links in the page header - you can see the combined effect at seti. I'd particularly like the ability to filter the user task list down to Fermi-LAT only, for quick monitoring.

[AF>EDLS] Polynesia
[AF>EDLS] Polynesia
Joined: 1 Apr 09
Posts: 24
Credit: 2273003
RAC: 0

Hello, I still do not have

Hello,

I still do not have units for this application?!

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 686159127
RAC: 549612

RE: Hello, I still do not

Quote:

Hello,

I still do not have units for this application?!

Hi!

Have some patience :-). I have not received any of them myself so far. The scheduler will eventually give some of those to us as well.

However, the last time your PC contacted E@H, it didn't even ask for more CPU work. Maybe it's busy with other projects. The GW search units you got earlier were all aborted by you, which will eventually limit the amounts of work units that you will be able to download.

HBE

Comment viewing options

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