This application is outdated – it does not support the “partxx” format for RAR archives and it does not support the PAR2 file format.

Multismart – Multipart Archive Tool

Multismart is primarily a tool that groups files. Instead of displaying long lists of related files like you see in a regular file view, Multismart groups the files and displays the groups along with statistics about the groups.

Multismart integrates the display of disk files, RAR archives, PAR files, SFV files, Descript.ion files and calculated information to show you a complete overview of the state of all your archives.

Quick Start

This application should be fairly intuitive, but there are some details you need to know before you start. I’ll try to keep this short. Read the Features in Detail section below if you would like to know more.

Installing and Starting the Application

Starting the application is as easy as unzipping the file and starting the application. No installation is necessary. The application is a single executable file. It touches only the folders you monitor and it creates one key in the registry where it stores your settings.

Monitoring Folders

After you point Multismart to a folder that contains multipart archives, the application will monitor that folder. You will see one or more lines of text in Multismart’s main window. Each line is an “item”. An item can be a real file that exists on disk, a file that Multismart learned about by reading another file or a piece of calculated information. The “Show existing and missing” check boxes turn on display of existing files and missing files. When both of these are turned off you see only archive items. Each archive item displays statistics about one distinct archive that exists in part or in whole on your hard drive. It is in this mode that Multismart is most useful. When you turn on display of existing and or missing files you clutter up the display without receiving much more information.

Interpreting the display

For this short description, I will focus on archive items. An archive item is in one of three states marked by the names “archive”, “archive pending” and “archive guess” in the Type column. When Multismart first reads the monitored folder, it only finds filenames and file sizes. Multismart is able to make a number of educated guesses about your archives from this information alone, but they are still only guesses. In particular, there may be undetected errors in the archive parts and the filenames may be misleading as to which archives the parts belong to. If checksum information exists for an archive in the form of PAR or SFV files, the archive item is in the state of “archive pending” until Multismart has verified the checksums of all the parts, after which the state changes to “archive”. If no checksum information exists, the archive item is displayed as “archive guess” and left at that.

The quickest way to describe how to interpret the information that Multismart displays is probably to use an example. Imagine that you’re downloading a multipart RAR archive called BigArchive that consists of 5 parts (rar, r00, r01, r02, r03). Here’s what happens as you get the parts in random order:

Lets say the first archive part you get is BigArchive.r02. Seeing this file alone, Multismart knows there’s an archive called BigArchive that has at least 4 parts (rar, r00, r01, r02). It will then display an archive guess item saying “size ok, have 1 of 4 or more parts”. (archive guess, size ok, 1 / 4+). It’s the best it can do at this point.

Adding BigArchive.r03: The last file in an archive is almost always shorter than the other files. When Multismart sees that this file is shorter than BigArchive.r02, it makes a guess that r03 is the last part and alters the archive guess item to say “size ok, have 2 of 5 parts”. (archive guess, size ok, 2 / 5).

Adding BigArchive.rar but it is truncated by a download error: Multismart sees that the new part is shorter than expected and adds a size error to the display. (archive guess, size error: 1, 2 / 5). As you see, the part with the size error doesn’t count towards the number of parts.

Adding a parity file for the archive, say BigArchive.p02: Multismart no longer has to make guesses on the archive, and it will display an archive pending item until it is done calculating checksums and verifying all files in the archive. It will then display an archive item saying that two files are missing or have checksum errors and that one of those errors is being covered by a parity file. (archive, md5 error: 1, 3 / 5 (1 / 1)). Read about parity archives below if you don’t undertand this. The number in parentheses shows that 1 out of 1 available parity files were used for coming up with the total of 3 / 5 parts present.

Adding another parity archive and archive part, say BigArchive.p03 and BigArchive.r00: The archive is now complete because the two parity archives can be used for fixing one truncated file (BigArchive.rar) and one missing file (BigArchive.r01). This is displayed as archive, complete, 5 / 5 (2 / 2). Multismart will then automatically fix the archive the next time you click the par button. After this, the par files are deleted, and the archive displays simply as archive, complete, 5 / 5.

Caution and Tips

Like all powerful tools, this one can do some serious damage if you don’t use it right. Tips for the beginner are:

When performing Delete operations, carefully look at the list of actual files in the confirmation dialog to see what you are deleting. If you are unsure how a specific file ended up in the delete list, try selecting items one at a time and click Delete to see which of them imply deleting the file in question.

If you ever wonder how Multismart came up with a specific piece of information, turning on display of existing and missing files may help.

If you don’t understand how a specific combination of files resulted in what is being displayed, you can point Multismart to an empty folder, and move files in and out of that folder one at a time while refreshing the display after each move. You’ll then see exactly what’s going on.

What is a Parity Archive?

Parity archives are files with an extension of .Pxx (P01, P02 etc). They protect a set of files by storing redundant information derived from the files in the set. The redundant information can be used to restore broken and missing files in the protected set of files. Think of a parity archive as a “wildcard” file – it will replace any one broken file in the set. Any two parity archives will replace any two broken files and so on.

How does it work? Consider this simple example: Say you have 4 files, each storing a single number. Let’s choose the random numbers 3, 4, 9 and 5. You now create a file containing the sum of all the numbers, which is 21. This file is the “parity archive”. If you now lose a random file of the 4 original files, say file 3, you can find the lost number by taking the number stored in the parity archive and subtracting the numbers stored in all the files you do have: 21 - 3 - 4 - 5 = 9.

Parity archives let you use any number of random parity files generated for a set to fix the same number of random broken files in the set and this makes the math very complicated, but you get the idea. Note that because of a mathematical weakness in the PAR system there are some combinations of files that can not be restored. These combinations are very rare though.


Only RAR archives are supported. The first archive part must be ”.RAR”, not ”.EXE”.

Multismart assumes that a set of PAR parents cover exactly one archive. It will not display information about archives correctly if the archives are covered only in part by a PAR set, or one PAR set covers more than one archive.

Multismart relies almost entirely on filenames for grouping files into the archives they are part of. So if you, for some weird reason, have files that have the exact same names but are parts of different archives, that will confuse Multismart.

Multismart Features in Detail

In the following, I use the terms “PAR/SFV parent” and “PAR/SFV child”. A PAR/SFV parent is a file with the extension of .SFV, PAR, Pxx or Qxx where xx is a number between 00 and 99. The files contain lists of files and corresponding checksums. Pxx and Qxx files also contain recovery data – more about that later. A PAR/SFV child is a file that is listed in one of these files. A PAR set is a collection of PAR parents that all have the same children and those children.

Monitored folder overview

Without running any additional functions you can check which archives you have, how many parts you have and how many parts are missing in each archive, in which time period the archive parts were created on disk, if any parts of an archive failed MD5/CRC32 (if the archive has a PAR/SFV parent), the size status of each archive (if the archive does not have a PAR/SFV parent), Descript.ion information for the first existing archive part and which archives have PAR/SFV parents.

By turning on “show existing files” you can also read the Descript.ion information for each existing part of the archive, check the MD5/CRC32 status for each file (if the archive has a PAR/SFV parent), delete individual archive parts (for instance those with MD5/CRC32 errors) and see other files you have in your monitored folder.

By turning on “show missing files” you will also find details about all PAR/SFV children that don’t exist in your monitored folder and all files that don’t exist in the monitored folder or in any PAR/SFV parents but still need to be present to complete an archive (gaps).


Here’s a list of the different items you may see and what the information displayed in the items mean.

Archive items

Archive items display information about archives. The information is (from left to right):

Number of days since a part of this archive was first created, number of days since a part of this archive was last created, size of all files that exist on disk, estimated size of the complete archive (if the archive is not yet complete), number of parts that exist on disk, number of parts in this archive (guestimate) and status field (described below).

Status field:


All the known parts of the archive are present. The number of known parts was found from a PAR/SFV parent, and there are no MD5/CRC32 errors.


All the known parts of the archive are present. The number of known parts were found by assuming that the part found on disk that has the highest sequence extension is the last part because it is smaller than the other parts (the last part will almost always be smaller). If the last part found has been truncated by an error, it may not be correct that this archive is complete. MD5/CRC32 has not been calculated, but all the parts have the correct size (the size of the last part has not been checked).


A single archive part is present and that archive does not have a part number AND it does not have any parents AND it does not have a size that is a round number (as almost all parts of multipart archives have).

md5/crc bad:

A PAR/SFV parent exists. One or more parts of the archive have MD5/CRC32 errors. “md5 bad” is displayed if there’s a PAR parent, and “crc bad” if there’s an SFV parent.

size bad:

No PAR/SFV parents exist. One or more parts of the archive do not have the expected size.


A PAR/SFV parent exists. MD5/CRC32 has not been calculated for one or more of the files. Use the CRC button to run MD5/CRC32 calculations.

open error:

One or more files could not be opened for MD5/CRC32 calculation. They are probably open in another application. Use the MD5/CRC32 button to run MD5/CRC32 calculations later.


One or more parts are missing, but the parts that do exist seem to be ok.

The guestimated numbers may or may not have a “+” behind them. If they don’t have a “+”, that means that the archive has a PAR/SFV parent and so Multismart can be reasonably certain that the exact number of parts in the archive is known. If there is a “+”, there were no PAR/SFV parents and a guess was made based on the same logic as described in the “complete?” section above.

Disk items

Disk items represent real files that exist on disk.

Status field:

crc ok:File exists on disk and has a PAR/SFV parent. The MD5/CRC32 for the disk file has been calculated and corresponds to the one listed in the PAR/SFV parent.
crc bad:File exists on disk and has a PAR/SFV parent. The MD5/CRC32 for the disk file has been calculated but it does not correspond to the one listed in the PAR/SFV parent
pending:File exists on disk and has a PAR/SFV parent. The MD5/CRC32 for the disk file has not yet been calculated so it is not known whether the MD5/CRC32 is correct or not.
size ok:File exists on disk but it does not have a PAR/SFV parent. The file is considered to be part of an archive and the size of the file is the same as the size of other parts of the archive.
size bad:File exists on disk but it does not have a PAR/SFV parent. The file is considered to be part of an archive but the size of the file is not the same as the size of other parts of the archive.
<blank>:File exists on disk but it does not have a PAR/SFV parent. File is not considered to be part of an archive.

PAR/SFV child items

PAR/SFV child items represent files that are listed in a PAR and/or SFV file (parent) but do not exist on disk.

Info1 field:

The name of the PAR/SFV parent the child comes from.

If the same child item exists in both a PAR parent and an SFV parent, the information stored in the PAR parent takes precedence because the checksums stored in PAR parents are more secure.

Gap items

Gap items represents gaps in the logical sequence of archive parts. For instance, if a file named BigArchive.r02 exists and no other BigArchive files exist, gap items will be created for BigArchive.RAR and BigArchive.r01.

No more information is given about gap items.

PAR Resolver

Clicking the “PAR” button starts the PAR Resolver which attempts to fix all known archives that have errors. The PAR Resolver searches each archive for broken files. If it finds any broken files, it searches for healthy PAR parents to use for fixing those files. If enough PAR parents are found, an attempt is made to fix the archive. If, for any reason at all, the attempt fails, all involved files are left exactly as they were. If on the other hand, fixing the archive is successful, as verified by thoroughly checking all involved files using MD5 checksums, the PAR parents are deleted as they no longer have any value.

This function ignores names of files. It works exclusively based on file MD5 checksums, eliminating the possibility that any error is made because of mangled filenames or filename collisions. This approach maximizes the chance that an archive is fixed because all relevant files will be used no matter what their filenames are. After an archive has been completely fixed, all files in the archive are renamed to match the names of the files as they are listed in the PAR parents. This maximizes the chance that you will be able to successfully extract the archive.

Note: Any file that has not yet been MD5/CRC32’ed is effectively invisible to this function. So, to fix as many archives as possible, use this function after all files have been MD5/CRC32’ed.

Duplicate Cleanup and Fixing Archives

Clicking the “Dup” button starts the duplicate cleanup process. This function automates fixing archives by replacing original broken files with duplicate intact files. It also automatically deletes duplicates of intact original files.

Multismart’s Dup function works on the assumption that all of the following files are the same file:

Copy of BigArchive.r03
Copy (2) of BigArchive.r03

As you see, I’m attempting to cover any automatic renaming scheme that may have been applied to your files. If, for some reason, you have archives with names like above that are actually different archives, you should not use this function.

The basic rules for renaming and deleting files

The best way to describe what Dup does is probably by examples:

If BigArchive.r03 has the correct size, or better, a correct MD5/CRC32 checksum (verified by a PAR/SFV parent), all other permutations of BigArchive.r03 will be deleted (as specified above).

If BigArchive.r03 has an incorrect size or an incorrect MD5/CRC32 checksum (again as verified by a PAR/SFV parent), all permutations of BigArchive.r03 will be searched for one that matches the required size or MD5/CRC32. If one is found, the original BigArchive.r03 file is replaced with the found file, and all other BigArchive.r03 files are deleted.

Note: To avoid that files such as newyears-1998.r03 confuse the function, it is a requirement that the two first digits of the -nnnn sequence are 00. This will only be a problem if you have more than 100 duplicates of a file.

Select, Delete, Copy and Missing

These functions work together to let you delete specific archives and create lists of archives to use for different purposes.

Creating Selections

Manual and semi-automatic selections

The Delete, Copy and Missing functions work on items that you have selected. You can select multiple items manually in the normal Microsoft Way by clicking items while pressing the shift and/or ctrl keys, or semi-automatically by double-clicking the following items:

Double-click Item Effect
Archive select all parts of the archive
archive part select all parts of the archive
.SFV file select all the children of the .SFV file
.PAR file select all the children of the .PAR file

Automatic Selections

The Select function selects archives that match your criteria. Use the For Delete tab to select items to delete (archives that have MORE than a certain number of missing parts), and use the For Copy / Missing tab to select items to request (archives that have LESS than a certain number of missing parts).

Note that the Select function only performs the selection. After the selection has been done, you can view and modify the selection by clicking items while holding the ctrl key before starting one of the functions described below.


Delete all marked items. Note that deleting a archive item will delete the entire archive. Deleting “PAR only”, “SFV only” or “gap” items is not possible since they don’t exist on disk. These items will disappear automatically when the actual files that they display information about are deleted.

After clicking Delete, a confirmation dialog will appear. The dialog shows you exactly which files will be deleted if you complete the function.


Copy a list of filenames to the Clipboard. If you have made a selection that implicitly selects more than one type of items, a dialog box will appear where you can specify what kind of items you would like to include in your list. Existing files are files that are actually present in your monitored folder and virtual files are files that are missing in your monitored folder but which Multismart thinks are needed for creating complete archives.


Copy a list of missing files to the Clipboard in a compact format. For instance, if Copy would have generated the following list:


Missing will generate the following list:

BigArchive r03-r05, r10

Missing automatically excludes files that exist in your monitored folder, even if they are in your selection.


When you first open a folder and then each time you click the Refresh button, the folder is scanned for new files, and special files are opened and scanned. This can be a lengthy process if you have a huge amount of files in the folder. After the folder is scanned, MD5/CRC32 calculations of any new files are started. This can take a very long time, but you may work with the application during this process. Be aware though that any uncalculated files are invisible to all PAR functions. So, to increase the chance that PAR functions succeed, wait until all MD5/CRC32 calculations are finished before running them.

To avoid hard drive trashing, simultaneous MD5/CRC32 calculations of different folders on the same drive are queued and executed consecutively while calculations of folders on different drives are executed concurrently.

Keyboard Shortcuts

Key Function
Ctrl-D, Delete Delete
Ctrl-C Copy
Ctrl-R Missing
A - Z Jump to item in list
F5 Refresh

Automatic Descript.ion cleaning

If you have Descript.ion files, Multismart will display the information about files that is records there. To prevent the Descript.ion database from growing too big, Multismart also keeps track of how much unused information it contains. If more than 50% of the information in the database is unused, you are given the option to remove the unused data. The unused data is descriptions of files that no longer exist.

SFV Function

Multismart integrates functionality similar to that of QuickSFV, FastSFV and WinSFV.

File MD5/CRC32 performance in Multismart should equal or exceed that of the applications mentioned above because it uses the fastest disk access method provided by Windows.

Note that because MD5/CRC32 is a lengthy process; Multismart runs the MD5/CRC32 processes on a priority that is slightly lower than default. There is a potential disadvantage in that if you run less “courteous” applications that perform lengthy operations at higher priorities, Multismart will wait until these are done before doing work itself. The advantage is that you will see very little impact on the usability of your computer while the MD5/CRC32 process runs.

Notes on implementation

Known bugs


Known minor issues

When the descript.ion cleaning dialog pops up, the progress dialog goes behind the main window.

If a file is deleted before the MD5/CRC32 process gets to it, it is displayed with “open error”. For non-programmers, this is probably not intuitive. The file should be marked with “removed” or should just disappear.

When “Show” state is changed, should jump to item that had focus.

Files with MD5/CRC32 errors

Multismart currently considers files with MD5/CRC32 errors absent even though it might be possible to recover the file if it is a RAR part that was created with the “recovery” option. The best way for Multismart to deal with such RAR parts would probably be to detect if they were created with a recovery option and then try to recover it, but for now, that is not implemented.

The file database, .Multismart

Multismart maintains a database of file MD5s and CRCs, so that it doesn’t have to MD5/CRC32 the same file several times. Multismart also stores information such as last modified date and size of all existing files in this database. This prevents Multismart from making any mistakes if you change or overwrite files. The name of the database file is .Multismart and it is stored in the monitored folder. It you delete this file while Multismart is running, Multismart will automatically restore it, but if you delete it when Multismart is not running, Multismart has to MD5/CRC32 all your files again before it can display full information about the archives the next time you start it.


Uninstalling Multismart is as easy as removing the executable file and removing the following key from the registry:


Multiple instances

It should not be necessary to run multiple instances of Multismart because you can monitor as many folders as you like with a single instance of the program. However, you can run multiple instances of Multismart as long as you point them to different monitored folders. If you try to run two or more instances of Multismart pointing to the same monitored folder you may get inconsistent results as instances bicker over the data in the database files (.Multismart and Descript.ion).

Only the settings from the last instance you close will be retained because instances will share the Registry key.

“Double” selections

If you select both a archive item and one or more corresponding archive parts that exist on disk, you have essentially selected some items multiple times because archive items are virtual items and refer directly to the real items. Deleting this selection will cause the real items to be deleted twice as hard – just kidding :)

The rule is that Multismart will always try to process as many files as possible when you select files both implicitly (through selecting an archive item or a PAR/SFV parent) and explicitly (through select corresponding archive parts or PAR/SFV children) at the same time.

Finding how many parts of which a RAR archive consists

The RAR archive format does not provide a way to find directly how many parts make up a given RAR archive. Multismart deals with this problem in the following way.

Without a PAR/SFV parent

Multismart knows that RAR archives consist of many parts. If Multismart sees the following two files:


it knows that there is a RAR archive called “BigArchive” that has at least 9 parts (RAR, r00 - r07) where two parts are present and the rest are missing. It looks at the size of the last known file (r07 in this case) and compares it to the size of the other files in the archive. If the file is smaller, chances are that it is the last file of the archive. If the file is the same size, there are probably files beyond r07 but there is no good way for Multismart to find out how many.

With an PAR/SFV parent

If there is a PAR/SFV parent, Multismart uses it to find out how many parts an archive consists of. It will also use the checksums in the SFV file to check if there are checksum errors in the corresponding files.

Deciding whether or not to delete the archive

Multismart keeps track of the following key information about an archive:

How many days since the first time and the last time a part of the archive was encountered and the number of existing and missing parts of an archive.

Multismart then compares these values with values you supply to find out if a particular archive should be selected for deletion. For instance, you might say that archives where there’s more than 5 missing parts and the last part was encountered more than 2 days ago should be marked.

If a few new parts of the now deleted archive appear later, they will automatically be marked for deletion when you run the function with the same criteria again.


  • Speed optimizations.
  • Scripting support (so that all the functions that the application supports can be performed from a script or batch file).
  • Built in uncompress functions for archives.
  • Continuous folder monitoring (find and process files as they appear in the download folder).
  • Functions to move and rename complete archives.
  • Function to merge archives that have different names but are actually the same archive.
  • Improve handling of archives that have errors but contain built-in recovery information.
  • Improve handling of duplicate archive parts (in this version. duplicate parts are seen as archives of their own).
  • Grouping of related archives (for instance. selecting BigArchive_1_of_2 might automatically select BigArchive_2_of_2).
  • Display combined size of selected files.
  • Function to search for specific archive based on various search criteria.
  • Function to select all archives that match a given string.
  • Put database files in the application folder instead of in the download folder.
  • Functions to hide special files like PAR/SFV parents and non-archive files.
  • Automatically perform Dup and PAR functions in the background.