[dencorso]

dencorso, on Nov 22 2008, 11:59 PM, said:

As for the limit for SCANDSKW.EXE (4.90.0.3000), it is between 15.2 and 30.4 million clusters, because it worked with my former 500 GB partition but crashed with Marius '95 1 TB raid single partition (link), although 98-Guy has reported it works up to 31.2 million clusters (follow the links inside this post).

I've done even more tests:

I've partitioned as a single active primary partition, using RPM (the Ranish Partition Manager), a Corsair Flash Voyager 32 GB pendrive. The partition was formated (with fat32format.exe, under Win XP) as FAT-32 with 02 sectors/cluster and, using all the space available in the pendrive, had 32,628,384,768 bytes (32,6 GB or 30.4 GiB), 31,863,657 clusters and 248,935 sectors/FAT. ScanDskW.EXE (the Win ME one, version 4.90.0.3000) was unable to check this partition and threw the following error message: "ScanDisk could not continue because your computer does not have enough available memory. If any other programs are running, quit one or more of them, and then try running ScanDisk again.", precisely as reported by Marius '95 to have happened when he tried to use ScanDskW with his own 1TB RAID.

I've then reduced the size of the partiton progressively, and eventually created a 27.2 GB (25.4 GiB; 27,226,775,552 bytes) partition, with 1024 byte clusters, which had about 26.6 million clusters (26,588,648 clusters) and 207,724 sectors per fat and then a 27.0 GB (25.2 GiB; 27,022,737,408 bytes) partition, also with 1 kiB clusters, which had about 26.4 million clusters (26,389,392 clusters) and 206,168 sectors per FAT. ScanDskW.EXE (v. 4.90.0.3000) worked OK with the latter (taking, however, 6h to finish!), but threw the same error message reproduced above with the former. So it seems that the limit for ScanDskW, at least for my own computer/OS-configuration, is somewhere between 26.4 and 26.6 million clusters, or somewhere between 206.1 and 207.7 thousand sectors per FAT. This result confirms Marius '95 report and seems to disprove the result reported by 98-Guy.

Yet, things might be more complicated than this. The tests conducted by 98-Guy were very careful and well designed, so I do believe he managed to use successfully ScanDskW with a partition of 31.2 million clusters... The facts are that ScanDskW throws an out-of-memory error and ScanDskW is a 16-bit executable windows program (that is: a NE executable), and all such programs run together with the system dlls and the 16-bit part of windows kernel in the 1-GiB-wide arena begining at the virtual address 2 GiB, that Microsoft call the "Shared Area" (see Q125691), and, hence, it is reasonable to imagine that the more cluttered this arena may be, the less space remains for ScanDskW to run... and my guess is that 98-Guy's test system had a far less cluttered Shared Arena than my day-to-day-use system, which is what I've used for my tests. But, in any case, while it may be possible to make it work with 31.2 million clusters is some special situations, the more usual limit of about 26 million clusters or even somewhat less than this should be the one to have in mind, when thinking about ScanDskW usability, in the general case.

As a side note, when working with more than 25 million clusters, ScanDskW slows the system to a crawl and prevents one from loading almost any additional program before it finishes, so that it virtually works as if the system were an one-task-at-a-time environment. On the other hand, DOS is a true one-task-at-a-time environment and both scandisk.exe or ndd.exe, running in DOS, checked my initial 32.6 GB pendrive partition in about 30 min each. So, whether or not ScanDskW runs with such 25-million-clusters-or-more partitions or not is less of an issue, because if one has to run it as if standalone, it is much more convenient and fast to reboot in DOS, scan the partition of interest with any of the mentioned above DOS programs, and reboot in windows once again. On the other hand, it is relatively fast to check just the file-system integrity (do the standard test) with ScanDskW, up to the cluster-number in which the out-of-memory errors begin to appear. What I think is definitely a no-no is the surface scan (a.k.a. "thorough test"), which, as I said, takes some 6h to complete.

This post has been edited by dencorso: 16 December 2008 - 09:45 PM

[RetroOS]

dencorso, on Nov 23 2008, 02:59 PM, said:

...
NDD32.EXE (v. 18.0.0.62, from NSW 2005), on the other hand, gives a BSOD with Fatal Exception 0E (Page Fault) with it.
...
So it seems that the limit for NDD32.EXE (v. 18.0.0.62, from NSW 2005) is somewhere between 7.6 and 10 million clusters, or, it dawned on me while I was doing these tests, somewhere between 59.6 and 77.7 thousand sectors per fat (as I now believe it crashes when reading the FATs to a buffer in memory).
...

Actually the latest update of NDD32.EXE from Norton SystemWorks 2005 is 18.0.3.11.
Maybe this version will fix some of your issues?

To update just Norton Utilities to 8.03, download:
ftp://ftp.symantec.com/public/english_us_....03/PatchNu.exe

LiveUpdate never seems to update Norton SystemWorks 2005 to 8.02 or 8.03...

To update Norton SystemWorks 2005 to 8.02, see step 3 of this page:

Updating Norton SystemWorks 2005 to version 8.02

To then update to 8.03 see... Huh? The support page in english has been purged from Symantec's knowledgebase... A common occurance with Symantec... Grrr...

Thanks to The Wayback Machine:
Updating Norton SystemWorks 2005 to version 8.03

see section To download the Norton SystemWorks 2005 updater

* Please note: To download updates on this page from Symantec, copy the download link and remove http://web.archive.o...20060220235820/ from the front of the url, then paste into your browser.

The FTP update links are for english, but the Symantec FTP site has other language updates as well.

[dencorso]

Thanks a lot, RetroOS! You rock!
I've grabbed the patches and shall patch my NSW in the next few days, after my next back-up.
But I have no way to repeat my tests just now, because the 32 GB pen drive is being heavily used in other projects.
As soon as I can I'll get back to it, though.

[Philco]

RetroOS, on Jan 10 2009, 12:37 AM, said:

...
Actually the latest update of NDD32.EXE from Norton SystemWorks 2005 is 18.0.3.11.
Maybe this version will fix some of your issues?
...

How can I detect version NDD.exe? Only by date and size?

[RetroOS]

Philco, on Jan 13 2009, 07:52 PM, said:

...
How can I detect version NDD.exe? Only by date and size?

NDD.EXE is the DOS version.
NDD32.EXE is the Windows GUI version. This one you can go to Properties, Version.
It's normally sitting under C:\Program Files\Norton SystemWorks\Norton Utilities

[MDGx]

To my knowledge newest NDD.EXE version [strictly native DOS tool] is from Norton Utilities 2002.
Newer Norton Utilities editions do not include DOS tools anymore, or if they do their versions are the same.

HTH

[dencorso]

MDGx, on Jan 13 2009, 04:59 PM, said:

To my knowledge newest NDD.EXE version [strictly native DOS tool] is from Norton Utilities 2002.
Newer Norton Utilities editions do not include DOS tools anymore, or if they do their versions are the same.

MDGx is correct. There are no newer versions than the one from 2002. It was included in many later packages, the latest being NSW 2005, if I'm not mistaken. In any case, the four DOS utilities (DISKEDIT, NDD, UNERASE and UNFORMAT) accept the /version switch. And although they refuse to do any other work under windows, you can run them with the /version switch from a DOS box and they will tell you their version, even under windows. The latest ndd.exe, run in a DOS box, gives this output:

Quote

C:\>ndd /version
NDD, Norton Utilities 2002 ..10E, Copyright © 2001 Symantec Corporation

This post has been edited by dencorso: 13 January 2009 - 09:51 PM

[Philco]

dencorso, on Jan 13 2009, 09:48 PM, said:

Quote

C:\>ndd /version
NDD, Norton Utilities 2002 ..10E, Copyright © 2001 Symantec Corporation

Thank you!

This post has been edited by Philco: 14 January 2009 - 09:22 AM

[Queue]

I have NDD32.exe from SystemWorks 2006 (it reports as version 19.0.0.48) and it still chokes and blue-screens on a 500 GB partition. It's a real shame; it's a jillion times faster at scanning my smaller drive (and my old drive that the 500 GB one replaced) than Scandisk.

Queue

[RetroOS]

Queue, on Jan 17 2009, 05:45 PM, said:

I have NDD32.exe from SystemWorks 2006 (it reports as version 19.0.0.48) and it still chokes and blue-screens on a 500 GB partition. It's a real shame; it's a jillion times faster at scanning my smaller drive (and my old drive that the 500 GB one replaced) than Scandisk.

Queue

Hi Queue, what version of Windows and service pack are you using?

Well, it looks like Symantec may not have addressed this problem...

[Queue]

98SE, all official patches. Only unofficial stuff I have is large hard drive support (BHDD31), UberSkin (for taskbar locking), IE6 SP1 (all of which I got within the past 24 hours). I have a very slighty modified (only a few bytes hex edited) explorer.exe to handle better system tray icon colors, but I got that back in 2000 or 2001, long before I knew of this site.

My primary drive is IDE 127 GB, secondary is SATA (via a PCI card) 500 GB, which is why I grabbed the large hard drive support package, so I could scandisk the 500 GB drive (since NDD32 kept choking on it).

This place is amazing by the way. I love my old 9x machine; I'm shocked so many others still love 9x as well. As for my join date for these forums... I must've registered to download a link months ago...

Queue

[RetroOS]

Well in that case, welcome, welcome!
So you are using 98SE... I would assume that SystemWorks 2006 is running on a dual boot Windows 2000 or XP?
SystemWorks 2005 was the last to install in Windows 98SE.
I use all versions of Windows, even the beta of Windows 7... But I prefer 98SE!
This forum and it's contributing members is a life saver for 9x

[Queue]

Not dual-booting; this machine is 98SE only. I have other machines for XP.

Some (if not all, I didn't really test) of the SystemWorks 2006 Norton Utilities run on 9x. I didn't install it, I extracted files manually from the CD.

Queue

This post has been edited by Queue: 17 January 2009 - 03:54 AM

[Queue]

I did a little digging and NU files with a version number of 19.0.1.8 are the last that seem to work on Win9x. Version 19.0.1.8 of NDD32 still chokes on very large hard drives.

Queue

[dencorso]

Queue, on Jan 21 2009, 11:12 PM, said:

I did a little digging and NU files with a version number of 19.0.1.8 are the last that seem to work on Win9x. Version 19.0.1.8 of NDD32 still chokes on very large hard drives.

Thanks for your good work, Queue!
We now know for sure that, up to now, no existing version of NDD32 is able to handle Big Partitions (with more than about 250 GB).
But there is a very thin chance that a better version may yet come out because, AFAIK, 19.0.1.8 is the version reported by some of the files in the latest 2008 update to NSW. It is offered for owners of NSW 2006/7/8 and effectively updates their NSW to the latest version available. We don't really know whether v. 20.x.x.x and later work with 9x or not simply because they're not out yet, or am I much misinformed?

[Queue]

dencorso, on Jan 21 2009, 06:35 PM, said:

We don't really know whether v. 20.x.x.x and later work with 9x or not simply because they're not out yet, or am I much misinformed?

I have a set of NU files with version number 22.0.0.52 (all Digitally Signed) that I can NOT get to work on Win98SE. They are (apparently) files from SystemWorks 2009. Bad news, I know. =/

Edit: 21.0.0.67 also doesn't work on 9x (and when you try to run them they specifically say they're not compatible with the current OS).

Queue

This post has been edited by Queue: 22 January 2009 - 12:08 AM

[dencorso]

So, I was misinformed. Bad news indeed. So much for my last hopes...
Well, your info settles it, Queue. There's no chance of us getting a better NDD32 for 9x/ME.
And on the basis of what I reported some posts above, for partitions above about 250 GB,
SCANDSKW can be used to check the file system, but to do a surface test the DOS utilities
SCANDISK or NDD are the only way to go. Thanks a lot, Queue! You rock.

[Wolfgang16]

Hi all,

I got an 500GB IDE drive WD5000AAKB, made in XP one extended partition including one logical drive. I formatted it using fat32format -c32, i.e. with 16kB clusters instead of 32KB, so it has 30,508,600 clusters. Then I moved it into my Win98 system. With DriveImage2002 I made an image of another drive, which consists of about 42 files of 2GB size. I copied that repetitively to the 500GB drive until 95% was full. Up to this WinME ScanDskW and Defragment worked fine. Of course there was nothing to do for Defrag. I let DriveImage check some of the copied files.

Then I started some work on the big drive, generating some smaller files and moving a few GB around. After that ScanDskW showed the out-of-memory message. Defragment still worked. When it finished, ScanDskW still showed out-of-memory, but after a reboot it worked (no surface test). At this point there were 1,253,082 clusters free. Then I unpacked an about 2GB RAR archive, after that ScanDskW showed again the out-of-memory message, after reboot it worked.

As far as I remember in the FAT there are 4 bytes per cluster. When ScanDskW has to held 2 FATs in memory it needs about 240MB. This could be the origin of the limitation.

I use a standard Win98SE modified with Maximus decims BHDD3.0, but instead of the English I use the localized German (ME) versions. I have 512MB memory.

bye
Wolfgang

EDIT: I did some further testing: upgraded memory to 786,432 kiB and added MaxFileCache=356352 to system.ini according to the recommendation of Igor Leyko, but this didn't change anything. I can run ScanDskW only after a fresh reboot, otherwise the out-of-memory message shows up.

This post has been edited by Wolfgang16: 21 March 2009 - 06:47 AM

[rloew]

DOS SCANDISK can handle a 1.5TB Partition without problems and did not complain when given a 2TB experimental Partition.

Windows ME SCANDISK and DEFRAG can handle more than 1TB, but the Partition must be Formatted with 64KB Clusters, which is non-standard.