April 2005 - jsdl-wg - lists.ogf.org

Process Topology
by Christopher Smith 21 Apr '05

21 Apr '05

Ok ... here are my thoughts on process topology and what's currently expressible in JSDL. First, I'll list some use cases (they're all parallel jobs): 1. Simple MPI job. Wants 32 processors with 1 processor per resource (in JSDL, a host is a "resource"). 2. OpenMPI job. Wants 32 processors with 8 processors per resource. 3. An OpenMP job. Wants 32 processors. Shared mem of course, so one resource. 4. A "homegrown" master/slave parallel job (say a ligand docking job). Wants 32 processors. No tiling constraints at all. * Note that I'm specifically leaving out the Naregi "coupled simulation" use case (sorry guys), since we determined at the last GGF that it was a case which could be decomposed into multiple JSDL documents. Second ... what is process topology? It provides the user a way to express how resources should be _allocated_ given the characteristics of the job (usually in terms of IO patterns ... e.g. network communication, disk IO channel contention, etc). Thus, it's used when the resource manager is _allocating_ the resources, not when the job is being started/launched. Therefore, none of the elements used to express process topology should be in the POSIXApplication section What we have in JSDL now: ResourceCount (how many "resources" i.e. hosts I want) CPUCount (how many processors _per resource_) TileSize (how many processors to allocate per resource as a unit) ProcessCount (total number of _processes_ that the job will use to execute the job) I will argue that ProcessCount is useless for the purposes of process topology, since a) it isn't about allocation, and b) there isn't enough information to tell me how to start/launch a parallel job. It isn't about allocation since it is irrelevant to the scheduler whether I'll be computing using threads or processes. It isn't useful for launching because it doesn't tell me how to spread the ProcessCount processes given a particular allocated topology. So that leaves the rest of them. TileSize and CPUCount are pretty much the same thing. At least for 80% (or more) of the uses I've seen. The only thing that might cause them to differ is that I could possibly allocate more than one tile on a host. Given that CPUCount is a range and that we could express step values in the range (we can express step values in the range, right?), we don't need TileSize any more. Note: I'm making an assumption here that CPUCount is the number of cpus that I want from the resource, rather than an expression of how many cpus the host needs to have configured. If it is the latter, then we do need TileSize, and replace CPUCount in my examples below with TileSize. So let's see how these map to the use cases. 1. ResourceCount == 32, CPUCount == 1 -> LSF : "-n 32 -R span[ptile=1]" -> PBS : "-l nodes=32:ppn=1" (ppn=1 might be the default) 2. ResourceCount == 4, CPUCount == 8 -> LSF : "-n 32 -R span[ptile=8]" -> PBS : "-l nodes=4:ppn=8" 3. ResourceCount == 1, CPUCount == 32 -> LSF : "-n 32 -R span[hosts=1]" (hosts=1 equivalent to ptile=<-n val>) -> PBS : "-l nodes=1:ppn=32" 4. ResourceCount == 32, CPUCount == 1 -> oops ... it doesn't care about tiling ResourceCount == 1, CPUCount == 32 -> hmm ... artificial constraint ... would suck on a blade cluster ResourceCount == 1-32, CPUCount == 1,32 -> oops again ... I might get a total allocation of 32*32 cpus * there seems to be a gap! If we had a term called "TotalCPUCount" for the entire job, I could do: 4. TotalCPUCount == 32 -> LSF : "-n 32" -> PBS : "not sure how to express" It basically means to grab 32 cpus, regardless of how they are spread. Basically I just need cpus. This is used a whole hell of a lot within our customer base. So ... in summary ... I propose: CPUCount (as is if it's allocated cpus per resource) TileSize (iff CPUCount is an expression of configured cpus in a host) ResourceCount (as is ... hmmm ... maybe the default value needs to change) TotalCPUCount (how many cpus this jobs needs to run in total) -- Chris

3 12

JSDL: Telecon minutes 19 Apr 2005
by Ali Anjomshoaa 21 Apr '05

21 Apr '05

Hi all, the minutes of yesterday's telecon are on GridForge: https://forge.gridforum.org/projects/jsdl-wg/document/Minutes_-_2005-04-19/… Those on the call, you can find your actions in-lined in the minutes. Please ask for help with your actions if you don't have time or need input. Target someone you know is able to help and perhaps has time. Catch you on the call next Tuesday, when we can tick off a few more issues. Cheers, Ali -- ---------------------------------------------------- |epcc| - Ali Anjomshoaa EPCC, University of Edinburgh James Clerk Maxwell Building Mayfield Road E-mail: ali(a)epcc.ed.ac.uk Edinburgh EH9 3JZ Phone: + 44 (0) 131 651 3388 United Kingdom Fax: + 44 (0) 131 650 6555 -------------------------------------------------------------

3 2

Re: [jsdl-wg] Process Topology
by Karl Czajkowski 20 Apr '05

20 Apr '05

On Apr 20, Andreas Savva loaded a tape reading: > Afaik Resource should be 0-n.... Thanks for pointing it out. I'll > confirm it and change the spec. > > When were heterogeneous resource mixes discussed and are there any notes > of that discussion? > I think it came up at GGF-13 and I thought it was 0-n too at the time. I got confused by the 0-1 section title in the draft... I thought we said the general model was a sequence of Resource elements each with its own localally scoped attributes and the resource count was just a shorthand to avoid writing redundant sequences for homogeneous scenarios. If it remains 0-n, I wonder if we should have an enclosing Resources element or something like that to hold the "total" metrics that summarize allocation across all resources? karl -- Karl Czajkowski karlcz(a)univa.com

2 3

Tracker status value proposal - take 2
by Andreas Savva 20 Apr '05

20 Apr '05

Ok, here's a slightly revised proposal for the status values. Again note that this is just for the *specification* tracker and not for the general 'Agenda/Topic' tracker. At the moment the tracker artifact status values are the gridforge default: open, pending, closed. These cannot be changed but new values can be defined. I propose to augment them with the following values: resolved, fixed "open", "pending", "resolved" and "fixed" would be of status class Open "closed" of status class Closed Status value definitions will be clearly documented on the tracker summary page. A typical workflow would be open->pending->resolved->fixed->closed (Michel's email gives a more detailed proposal for these transitions.) The meaning of these values (for the typical path) would be as follows: Open: the issue has been proposed by a member but not yet accepted by the WG. Pending: the issue has been accepted and is still pending resolution by the WG. (Includes 'waiting response from author' mentioned by Donal.) Resolved: The WG has decided and accepted a resolution, but the actions are still pending Fixed: The editor has completed the proposed actions, but the actions are awaiting verification. Both the schema and spec may have to be revised in order to reach this state. Closed: The issue has been resolved and is reflected in the work products of the WG. The completion of the changes must be reported by the primary editor and verified by the secondary editor (or other WG member). Issues that have been deferred or rejected also get closed with an appropriate comment. Issues that have been deferred are moved to the post-v1 tracker. (After checking the tracker settings again, "deferred" or "rejected" should probably be 'resolution' values rather than status values. The resolution field exists but is not enabled at the moment. I propose we first try out the new status values. We can come back and enable the 'resolution' field if we decide we really need it.) Andreas

4 5

[karlcz@univa.com: Re: [jsdl-wg] Process Topology]
by Karl Czajkowski 20 Apr '05

20 Apr '05

oops, hit reply instead of reply-all and missed the JSDL list... -- Karl Czajkowski karlcz(a)univa.com

2 1

process count fields
by Karl Czajkowski 19 Apr '05

19 Apr '05

I wasn't sure if Chris was taking on sorting out the posix element, so I wanted to add my memory regarding the process count fields because they seem misleading in the minutes document from the call. There are two mostly unrelated concepts: 1) a process or task count for the job, which should use the range value type 2) a posix limit "max user processes" that should have integer type and is a sibling to the other limits like stack size, VM size, file descriptors, etc. it is probably a mistake to make these have names which look too similar and incorrectly imply that they are very closely related... karl -- Karl Czajkowski karlcz(a)univa.com

1 0

None
by Ali Anjomshoaa 19 Apr '05

19 Apr '05

http://www.epcc.ed.ac.uk/~ali/WORK/GGF/JSDL-WG/index.txt -- ---------------------------------------------------- |epcc| - Ali Anjomshoaa EPCC, University of Edinburgh James Clerk Maxwell Building Mayfield Road E-mail: ali(a)epcc.ed.ac.uk Edinburgh EH9 3JZ Phone: + 44 (0) 131 651 3388 United Kingdom Fax: + 44 (0) 131 650 6555 -------------------------------------------------------------

1 0

JSDL: Here it is: MeetingPhone.co.uk Conference Call Confirmation
by Ali Anjomshoaa 19 Apr '05

19 Apr '05

Try this: Conference Details Date: 19 April 2005 Start Time: 3:15 p.m. British Summer Time End Time: 5:10 p.m. British Summer Time UK number: Primary Dial-in Number: 0871.711.1515 (UK 10p National Rate) International number: * +44.870.088.5700 (For Callers Outside of the U.K.) * 01805.004.101 (For Callers in Germany) Participant Access Code: 20050419 Touch-Tone Quick Reference for Conference Call 0 Exit the Conference 4 Access the Help Menu 6 Toggle Mute/Unmute 8 Time Remaining in the Conference See you there!

1 0

Re: [jsdl-wg] JSDL: Telecon...
by Ali Anjomshoaa 19 Apr '05

19 Apr '05

Hang on folks, I'm about to set up a another call... Be with you in a minute. Ali On Tue, 19 Apr 2005, Michel Drescher wrote: > "... The meeting has ended." > > WTH? > > Michel > > > On 19 Apr 2005, at 14:45, Ali Anjomshoaa wrote: > > > > > ...on the hour in 15 mins... > > > > > > Agenda (last week's minutes): > > > > https://forge.gridforum.org/projects/jsdl-wg/document/Minutes_-_2005 > > -04-13/en/1 > > > > > > Web: > > > > Subject: JSDL > > Meeting > > URL: https://www.livemeeting.com/cc/xangoit/join > > Meeting ID: JSDL > > Attendee Key: Wxx*}$Hk4 > > Role: Attendee > > > > > > Phone: > > > > --> ext 3000 pin: 6444 > > > > Belgium - 0800-7053 > > Bezeq - 1-800-940-1786 > > France - 0800-903109 > > Germany - 0800-1815765 > > Hong Kong - 800-968932 > > Italy - 800874509 > > Japan - 00531123656 > > Korea-Tele - 00798-14-800-0512 > > Netherlands - 0800-0225447 > > Singapore - 800-1200766 > > Sweden - 020-7905043 > > Taiwan - 00801-13-8546 > > UK - 0800961294 > > US - 800-933-3594 > > > > --> ext 3000 pin: 6444 > > > > > > Cheers, > > > > Ali > > > > -- > > > > ---------------------------------------------------- |epcc| - > > Ali Anjomshoaa > > EPCC, University of Edinburgh > > James Clerk Maxwell Building > > Mayfield Road E-mail: ali(a)epcc.ed.ac.uk > > Edinburgh EH9 3JZ Phone: + 44 (0) 131 651 3388 > > United Kingdom Fax: + 44 (0) 131 650 6555 > > ------------------------------------------------------------- > > -- ---------------------------------------------------- |epcc| - Ali Anjomshoaa EPCC, University of Edinburgh James Clerk Maxwell Building Mayfield Road E-mail: ali(a)epcc.ed.ac.uk Edinburgh EH9 3JZ Phone: + 44 (0) 131 651 3388 United Kingdom Fax: + 44 (0) 131 650 6555 -------------------------------------------------------------

2 1

JSDL: Telecon...
by Ali Anjomshoaa 19 Apr '05

19 Apr '05

...on the hour in 15 mins... Agenda (last week's minutes): https://forge.gridforum.org/projects/jsdl-wg/document/Minutes_-_2005-04-13/… Web: Subject: JSDL Meeting URL: https://www.livemeeting.com/cc/xangoit/join Meeting ID: JSDL Attendee Key: Wxx*}$Hk4 Role: Attendee Phone: --> ext 3000 pin: 6444 Belgium - 0800-7053 Bezeq - 1-800-940-1786 France - 0800-903109 Germany - 0800-1815765 Hong Kong - 800-968932 Italy - 800874509 Japan - 00531123656 Korea-Tele - 00798-14-800-0512 Netherlands - 0800-0225447 Singapore - 800-1200766 Sweden - 020-7905043 Taiwan - 00801-13-8546 UK - 0800961294 US - 800-933-3594 --> ext 3000 pin: 6444 Cheers, Ali -- ---------------------------------------------------- |epcc| - Ali Anjomshoaa EPCC, University of Edinburgh James Clerk Maxwell Building Mayfield Road E-mail: ali(a)epcc.ed.ac.uk Edinburgh EH9 3JZ Phone: + 44 (0) 131 651 3388 United Kingdom Fax: + 44 (0) 131 650 6555 -------------------------------------------------------------

3 2