App info question

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1364
Credit: 3562358667
RAC: 1580
Topic 194733

Can I delete all the legacy ABP1 and S5R5 entries from my file without affecting the S5R6 WU's it has running on 3.05?

Edit: on second thought this probably won't help me in getting some ABP2 WU's so that I can transition without having to run my cache out completely.

archae86
archae86
Joined: 6 Dec 05
Posts: 3145
Credit: 7023084931
RAC: 1831799

App info question

Quote:

Can I delete all the legacy ABP1 and S5R5 entries from my file without affecting the S5R6 WU's it has running on 3.05?

Edit: on second thought this probably won't help me in getting some ABP2 WU's so that I can transition without having to run my cache out completely.

Yes to both points. In order to get the new type of work with the new application while still using an app_info.xml file, you would somehow have to download all the required files to run the new application, and edit in an appropriate reference set of references to the app_info file. If you have multiple compatible hosts, it might not be too hard to do this after a first one running without app_info happens to get some work.

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1364
Credit: 3562358667
RAC: 1580

I have hosts already running

Message 96610 in response to message 96609

I have hosts already running ABP2. Forcibly thumping a dual core isn't that big a deal even with a large cache; aborting the 120ish WU's my i7 is stitting on would leave it out of work for a while until the error counter rolled down.

The one thing I'd need to put in my app info that I don't have are, or know where to find, are the API versions for the 3.06 ligo executable or the ABP2 executable.

archae86
archae86
Joined: 6 Dec 05
Posts: 3145
Credit: 7023084931
RAC: 1831799

RE: The one thing I'd need

Message 96611 in response to message 96610

Quote:
The one thing I'd need to put in my app info that I don't have are, or know where to find, are the API versions for the 3.06 ligo executable or the ABP2 executable.

I think you might find it to work if you just omitted the API version entry. For a few hours I ran the now deprecated ABP2 with an app_info from which I omitted that section, and it both downloaded and ran ABP2 work.

Caution, I don't know what lack of security or other undesirable effect this omission might cause. My attempt did get ABP2 work, but failed to get GW work (and got me complaint notices asserting that I lack a version of Hierarchical S5 all-sky GW search #6 in my app_info). As I never figured out what I had wrong, it could be that at that time GW required API (which was missing), but ABP2 did not (as that was also missing). Or maybe I just goofed in another way. My luck at composing app_info files which work first time has not been very good.

archae86
archae86
Joined: 6 Dec 05
Posts: 3145
Credit: 7023084931
RAC: 1831799

RE: The one thing I'd need

Message 96612 in response to message 96610

Quote:
The one thing I'd need to put in my app info that I don't have are, or know where to find, are the API versions for the 3.06 ligo executable or the ABP2 executable.


I don't yet have a copy of the ABP2 executable, but on searching the Windows ABP1 executable using the text editor Textpad, I found the string:

API_VERSION_6.7.0 deep in the file, in a section with a lot of text which appears to be the contents of user alert messages possibly generated by the program.

This matches the relevant section of the app_info I was using, which I think came straight from the project:

6.7.0 The small executable (switch chooser, I think) I'm currently running GW with is tagged. 6.7.4 right at the end of the file.

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1364
Credit: 3562358667
RAC: 1580

I appear to be doing

I appear to be doing something wrong. I haven't crashed and WU's but I haven't got any ABP2 (could be coincidence), but my attempt to get new s5r6 wu's to use the new s5r6 app instead of being routed onto the older s5r5 one are failing.

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 686042288
RAC: 599448

Well, but why are you using

Well, but why are you using an app_info.xml in the first place? There is no beta app at the momemnt. I'd just remove the app_info.xml after setting to "no new task" and the cache has been depleted.

CU
HB

DanNeely
DanNeely
Joined: 4 Sep 05
Posts: 1364
Credit: 3562358667
RAC: 1580

RE: Well, but why are you

Message 96615 in response to message 96614

Quote:

Well, but why are you using an app_info.xml in the first place? There is no beta app at the momemnt. I'd just remove the app_info.xml after setting to "no new task" and the cache has been depleted.

CU
HB

because draining the cache without a significant amount of time with no work is a major pita because the boinc scheduler always prefers to run WU's of my gap filling project instead of finishing off the last few E@H WU's. I tried bludgeoning it into submission on one of my PC's by manually adding 50M LTD but whatever the scheduler uses to decide to defer the E@H WUs still wins (I tried assigning the debt both ways so this wasn't just a sign issue). One possibility might be the deadline since I've been using Leiden as my backup recently and they've got a shorter deadline. OTOH I think the same's been true of my other backup projects in the past so I'm not sure what I could use to test it.

This is managable on a dual core since I only need to intervene at two points in time (unsuspending a project B WU when the first core runs dry, and removing the app info after the second is done), but is a major pain in my i7 with it's 8 concurrent tasks.

What I was hoping to be able to do was to have my app info assign new s5r6 WU's to the s5r6 default executable while not causing those assigned to the old s5r5 exe to crash. Then once my cache was empty in a few days I could safely delete the app info file.

Richard Haselgrove
Richard Haselgrove
Joined: 10 Dec 05
Posts: 2139
Credit: 2752648530
RAC: 1492503

If you want the Einstein apps

If you want the Einstein apps to run preferentially, it's the STD you need to bump. Plus 86,400 is the highest you can go (and you would need to set Leiden to minus 86,400 to balance it - STD still sums to zero across all projects).

Michael Karlinsky
Michael Karlinsky
Joined: 22 Jan 05
Posts: 888
Credit: 23502182
RAC: 0

RE: If you want the

Message 96617 in response to message 96616

Quote:
If you want the Einstein apps to run preferentially, it's the STD you need to bump. Plus 86,400 is the highest you can go (and you would need to set Leiden to minus 86,400 to balance it - STD still sums to zero across all projects).

Was not there a change in BOINC, so that 0 was the lowest STD possible?

Michael

Richard Haselgrove
Richard Haselgrove
Joined: 10 Dec 05
Posts: 2139
Credit: 2752648530
RAC: 1492503

RE: RE: If you want the

Message 96618 in response to message 96617

Quote:
Quote:
If you want the Einstein apps to run preferentially, it's the STD you need to bump. Plus 86,400 is the highest you can go (and you would need to set Leiden to minus 86,400 to balance it - STD still sums to zero across all projects).

Was not there a change in BOINC, so that 0 was the lowest STD possible?

Michael


No - you're thinking of LTD, where zero is the highest LTD possible: STD is still centered.

Believe me - I've been monitoring to see if the devs have got it right! This was v6.10.25:

http://img51.imageshack.us/img51/6843/200912237daystd.png
http://img38.imageshack.us/img38/579/200912237dayltd.png

Comment viewing options

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