Work Unit Page

From Unofficial BOINC Wiki

Contents 1 General 2 Work Unit Data 2.1 Application 2.2 Created 2.3 Name 2.4 Error 2.5 Canonical Result 2.6 Granted Credit 3 Results Listing - Column by Column 3.1 Result ID 3.2 Computer (ID) 3.3 Sent 3.4 Time Reported or Deadline 3.5 Server State 3.6 Server State Descriptions 3.6.1 Inactive 3.6.2 Unsent 3.6.3 In Progress 3.6.4 Over 3.7 Outcome 3.8 Possible Outcomes 3.8.1 Unknown 3.8.2 Success 3.8.3 Couldn't Send 3.8.4 Client Error 3.8.5 No Reply 3.8.6 Didn't Need 3.8.7 Validate Error 3.9 Client State 3.10 Client States 3.10.1 New 3.10.2 Done 3.10.3 Downloading 3.10.4 Computing 3.10.5 Uploading 3.11 CPU Time (sec) 3.12 Claimed Credit 3.13 Granted Credit 4 Notes, Oddities, and "Watch Out For This" 5 Web Addresses of the "Work Unit" Page

[edit] General

This is one of those pages that is common to all of the BOINC Powered Projects, yet, is also unique to each. The reason is simple. Each BOINC Powered Project has a different type, size, shape, form, fit, function, etc. to its Work Unit.

Naturally, this means that we will have a fairly wide range of actual work. However, because all of the work performed using the BOINC Client Software uses the identical process; this makes very little difference when it comes to the accumulation of information about a specific item of work.

This item of work is, of course, the Work Unit. These Work Units are issued to Participants, processed by the Participant's Computer, and returned as a Result (and one or more Result Data Files).

The returned Result is compared to other Results returned by other Participants using a Project specific mechanism called Validation, those Results that pass Validation will then be granted an amount of Credit.

Because of these factors, the Work Unit Page has two basic parts, one is the top identification and description information while below that is a #Results Listing that lists all of the Results that have been registered against this Work Unit.

Note:

Because the pages used in the active sites have to look up a Work Unit by its Work Unit ID, it is not feasible for me to create links into the sites to example Work Units.

This is especially true because most of the Projects are making the "retirement" of those Work Units and Results that have been completed from the BOINC Database a priority. This "pruning" helps to keep the database lean and fast. Unfortunately, it also makes it practically impossible for me to link into the sites for specific Work Units and Results that you can look at for a specific example.

Note:

The Rosetta@Home Project, at the time this image was taken, was not using redundant processing so only one Result was needed to make up a Quorum of Results.

[edit] Work Unit Data

The top portion of the "Work Unit" Page contains the information that is specific to the Work Unit itself. Not all of this information is relevant for all of the BOINC Powered Projects; and so, not all of these rows will be listed for all of the Projects. The information identifying the Work Unit includes:

[edit] Application

This line on the page lists the application or program that created this specific Work Unit. Because a Project may have multiple applications creating work, this is the field that will identify which of the Project's applications created this item of work.

[edit] Created

This field contains the date and time when the Work Unit was created. This time stamp is usually recorded against a clock set to UTC.

[edit] Name

This is the "Name" of the Work Unit. The Work Unit has, in fact, two items that will uniquely identify it. One is the human readable "Name" and the other is the Work Unit ID. The Work Unit ID is a number which is used by the database to track the Work Unit (for database freaks, the Primary Key value). This second, usually hidden value, is the one that is a positive integer that is used in the links connecting pages like the "Result" Page to the "Work Unit" Page.

The one we are concerned about here is a descriptive name that uses a Project specific scheme to create uniquely named object of work. For example, the SETI@Home Project identifies the tape from which this Work Unit has been extracted, along with a bunch of other information about the source of the Work Unit.

Other BOINC Powered Projects will be doing the same type of thing, but in the context of their Project. There is one constant across all Projects the last number in the Work Unit Name is the Result Identifier. For the most part this is a lot of "who cares?" other than we know that the Work Unit has a unique identifier.

The way that the Work Units are identified should be described on that Project's Web Site (if you really care that much). The identification data may be a simple, but long identifier, or may have within its seemingly random collection of numbers, letters, and punctuation encoded information that can be decoded if you send in enough box-tops and get the Genuine Captain Paul Marvell's Magic Secret Decoder Ring (the ring itself is free, but the shipping and handling is a low, low $56.95 U.S.; don't wait, rush down to the post office while supplies last …).

A Result Name is simply a modification of the Work Unit Name by adding an identifier. This identification number, as a suffix to the Work Unit Name, provides a means of establishing a unique Result identifier. So, when the Work Unit is split into Results to be sent out to the Participant's Computers, a Result Identifier is appended to the Work Unit Name. The format of this identifier is "_#". An underscore ("_") separates the Result Identifier from the Work Unit Name. The "#" is replaced with the sequence number beginning with zero (0) and continuing upwards in increasing value.

This value may, if the Project requires, be more than a single digit in length.

[edit] Error

If the Work Unit cannot be successfully processed this line will contain an error message indicating the reason that the Results are not accepted as Valid Results.

Errors include:

• Too many success results
• Checked no consensus yet

[edit] Canonical Result

This value, when present allows you to link to the "example" or "standard" Result that was selected as part of the Validation Process to represent all of the Results that are within the Quorum of Results for this Work Unit. In scientific terms, the "example" is the Canonical Result.

Clicking on the link for the Canonical Result will take you to the "Result" Page for the specific Result selected as the Canonical Result.

[edit] Granted Credit

This line lists the amount of Credit that will be granted to all of the Participants that returned a Result that was part of the Quorum of Results. Remember, you ask for credit using Claimed Credit which may allow you to receive Granted Credit that will be added to your Total Credit.

Note:

At this time, the Climateprediction.net Project does not annotate this field because for this Project, credit is granted as the Participant's Computer returns "Trickles" and they are received and recorded.

[edit] Results Listing - Column by Column

[edit] Result ID

This is the unique identifier of a specific Result that was issued to a Participant and which records the progress of that Result by that Participant. As this identifier is normally only available as a link, it allows the navigation to the "Result" Page where detailed information about this specific Result is available. This identification number is the internal tracking number that is given to this specific Result so that it has a unique identity in the BOINC Database.

This identifier, the Result ID, is a number which is used by the database to track the Result (for database freaks, the Primary Key value). This is a positive integer that is used in the links connecting pages like the "Work Unit" Page to the "Result" Page. You will find that Result ID is labeled as "Task ID" when viewing results.

There is no information coded in this number, it is a simple large integer number that will grow until it reaches a value of about 2,147,483,647 if they are using signed long integer and 4,294,967,294 if they are using unsigned numbers. And if they are using a string data type, an effectively unlimited numeric range is possible (as long as they stick to strings shorter than 2,000 characters).

Ok, so what does all that mean? It means that this is the internal tracking number that is given to this specific result so that it has a unique identification number in the BOINC Database.

Contrast this with a Result Name. The Result Name is a derived name that includes the Work Unit Name with a suffix that is an underscore followed by a number that indicates the position of this specific Result in the series of Results generated from that specific Work Unit.

The Results are numbered from 0 to "N" with the maximum number of trials being set by the Project. For example, if the LHC@Home project only wants to try 5 times to process a Result to a successful conclusion we would see a Result series from "RESULT-NAME_0" to "RESULT-NAME_4"; with these Results being issued to five different Participants.

[edit] Computer (ID)

This is a unique identifier for the computer that processed this Result. The Computer ID is a link to the descriptive record for the computer/host that processed this Result. Obviously, clicking on the link will take you to the page that will list the characteristics of the Participant's Computer. In some cases the link will not be there because the Result was not needed.

Clicking on one of the links will take you to the "Computer Summary" Page for the Participant's Computer that was assigned the Result.

[edit] Sent

This column contains the date when the Result was issued (sent) to the Participant.

[edit] Time Reported or Deadline

A point in time. As each Work Unit (or Result; depending on how you want to look at it, what the Data Server gives you are Work Unit Data Files and a Result, your computer processes the data files and you return the Result and the Result Data Files) is issued by the Project's Data Server this value will be established in the BOINC Database for that Project.

Keep in mind that Results that are returned after the Deadline has been passed may generate no Credit.

For the Projects that are currently in a production status, that is, they are actually performing the intended science even though the Science Application (or Science Applications in the case of Projects like the Predictor@Home Project) may still be in a Beta Testing status the current Deadline lengths are:

Project	Application	Deadline	Quorum
Climateprediction.net (CPDN)	hadsm3	347 Days, 5 Hours, and 20 Minutes	1
	Sulphur Cycle	150 Days, 5 Hours, and 20 Minutes	1
Einstein@Home	einstein	14 Days	3
LHC@Home	sixtrack	Variable, around 5 days	3
Predictor@Home	mfold	7 Days	3
	Charmm	1 day
Rosetta@Home	rosetta	7 or 28 Days	1
SETI@Home	setiathome enhanced	Variable, 4 to 60 days	3
SIMAP@Home	simap	10 Days	2
SZTAKI Desktop Grid	search	4 Days	4
World Community Grid	faah	unknown	unknown
	rosetta	unknown	unknown

One of the newer features of the BOINC Web Site Software is the color coding of the Deadline, where the date and time listed are shown in one of three colors:

• When the date and time is BLACK; the Deadline was not reached before the Result was returned, or
• When the date and time is GREEN (GREEN); the Deadline is in the future, or
• When the date and time is RED (RED); the Deadline is in the past

This column has an "Explain" link that will take you to a page that explains the purpose of the column and how to interpret it.

[edit] Server State

This is a display of the current state of the Result in the opinion of the Project's BOINC Database, Data Server, and Scheduling Server.

If you need a reminder of the meaning of the status states, you can click on the "explain" link to get the definition.

This is a categorization of a Result when the Result has been created for a Work Unit by the Project and is ready to be issued by the Data Server to a Participant to be processed by the Participant's Computer.

There are 4 Server States that are related to any one specific Result (for a specific Work Unit) that is processed by the Science Application under the control of the BOINC Client Software. Several of these states will not be seen by the Participants as a normal part of their processing because they are only of interest by the Project's managers.

This is a categorization of a Work Unit when the Work Unit has been created and is ready to be issued by the Data Server to a Participant to be processed by the Participant's Computer.

Normally, the only two Server States that the Participant will see are the "In Progress" and "Over" states.

Server State Descriptions

Inactive

The Result is not ready to send (for example, because its input files are unavailable).

A Result in this state can only be seen by a Participant if they look for a specific Result by Result ID number, which is not a task that most Participants will perform for any particular Result.

A Result with this state have been generated for the specific Work Unit but the Result is not ready to be issued. As stated in the "official" explanation one or more of the input files are missing. With some of the current BOINC Powered Projects you not only need the Work Unit itself, you may also need one or more "library" or reference data files before the Result can be issued for processing.

Unsent

The Result is ready to send, but hasn't been sent yet.

This is a Result that has not yet been issued to any Participant but is ready and available on the Data Server.

In Progress

The Result has been sent; waiting for completion.

This is the state of a Result that has been issued to a Participant and the Project is waiting for a return of the completed Result.

Over

The Result has been sent to a Participant's Computer and either it has timed out or the Participant's Computer has reported its completion.

In general, we have two cases:

1. The deadline is in the future and the Result was returned, or
2. The Deadline is in the past and the result was not returned

In sum, the Result has been processed on the Participant's Computer and a Result has been returned or an error occurred. Results in this state can still be in error, not processed completely, returned past the Deadline, exceeded allowable Result Data File size, etc. but the Scheduler and the Data Server are no longer waiting for this Result to be returned by this Participant.

[edit] Outcome

The end state of the analysis for the specific Result.

Possible Outcomes

Unknown

A Result in this state has not yet been processed, the Result was issued to a Participant as an "empty" Result, but that Participant has not yet returned the completed Result.

Success

A Result in this state was processed successfully and there were no errors detected by the BOINC Client Software and the Science Application on the Participant's Computer. This does not mean that the Result is scientifically valid or meaningful. It is entirely possible that the Result is "bad" and when compared to other Results will not achieve consensus and the Work Unit will be rejected because it exceeds the Max success results limit.

Couldn't Send

During the time that this Result was available on the Data Server no Participant that requested work had sufficient resources to process this Result (Work Unit).

Client Error

The Result was issued and the Participant's Computer returned an error during processing of this Result. There are a number of Client Error conditions that can occur and that can be detected by the BOINC Client Software and recorded.

No Reply

The Result was issued to a Participantt and the Result was not returned within the time specified by the Deadline. This error may occur due to problems in the data transfers, processing, or simply because the Participant's Computer was not able to complete the processing within the time allotted.

Didn't Need

The Result was not sent out at all, either due to already had enough Results returned by other Participants and a Quorum of Results was created before this Result was issued to someone, or the Work Unit was aborted. Most commonly, these aborts are due to the Work Unit reaching one of the error-limits before this Result was issued to someone.

Validate Error

The Result was reported but could not be Validated. This error, most typically, is because the Result Data Files were lost on the Data Server. Other causes include Result Data Files, or the Result itself was not correctly received usually due to data transfer errors or heavy server loads.

Note: A Result Outcome of "Validate Error" is not the same as a Valid Result. See Validate State for an explanation of the various types of Validation outcomes. As an example, an Outcome can be "Success" but the Validate State may be "Invalid".

This column has an "Explain" link that will take you to a page that explains the purpose of the column and how to interpret the specified Outcome.

[edit] Client State

Broadly, this is the stage at which the processing of a Result encountered an error.

The Client State of a Result is defined as one of three basic states:

1. Not started (New), or
2. Processing is Finished (Done), or
3. Processing interrupted with an error condition (Client Error)

Client States

New

The BOINC Client Software has not yet completed the work. Since the processing is not over, the the final Client State at the Result Outcome is not yet known.

The Result has been issued to the Participant's Computer and the Data Server is awaiting a returned Result.

Done

No error occurred.

The Result has been processed on the Participant's Computer and a Result has been returned.

Downloading

The BOINC Client Software couldn't download the Science Application or Work Unit Data Files.

A Result with this state had an error detected while it was being downloaded to the Participant's Computer and the data transfer cannot/could not be completed even after several retries have occurred.

Computing

An error occurred during computation.

A Result with this state had an error detected while it was being processed by the Participant's Computer. These errors can be errors that are detected by the BOINC Daemon or the Science Application.

Uploading

The BOINC Client Software couldn't upload the Result Data Files.

A Result with this state had an error detected while it was being uploaded from the Participant's Computer and the data transfer cannot be completed even after several retries have occurred.

This column has an "Explain" link that will take you to a page that explains the purpose of the column and how to interpret the specified Client State.

[edit] CPU Time (sec)

This is the amount of time that was spent by the Participant's Computer processing this Work Unit to obtain the indicated Result. Keep in mind that this is not the elapsed wall-clock time, but the actual processing time , or CPU Time tracked by the BOINC Daemon.

[edit] Claimed Credit

The Claimed Credit value is the amount of Credit (in Cobblestones that the BOINC Daemon calculates should be awarded for the creation of this Result.

[edit] Granted Credit

This is the amount of Credit that the Project has awarded for the generation of this Result. This value may be:

1. Zero - for Results returned past the Result Deadline.
2. Zero - for Results returned that fail Validation.
3. Equal to the Claimed Credit - if this is the Canonical Result.
4. Higher than the Claimed Credit - if this is a Valid Result and the calculated Credit for the Canonical Result is higher than this Claimed Credit.

[edit] Notes, Oddities, and "Watch Out For This"

1. None.

[edit] Web Addresses of the "Work Unit" Page

Because most BOINC Powered Projects "purge" their BOINC Databases, where the Work Unit and Result data is stored, I cannot give you direct links to the pages on the Project Web Sites.