Nikon LS4000, AI 5.5, Dual Processor is Unstable

[UrbanVoyeur]

In my experience, Silverfast is unstable on Dual PIII systems. I tried all of their fixes including a clean win2k system.

Silverfast has NOT noted this in their literature, and has not posted a fix.

My solution?

Since I have several dual proc PC's, I removed one of the processors, reset the bios, and clean installed win2k, photoshop 6.01 and silverfast. And lo and behold, wonder of wonders it works!
I just scanned in 160 slides at 24 bit. I am testing the 48 bit 5.5.2 now.

Am I happy with this solution? No. I don't like buggy expensive software and i don't like having to cripple a system and essentially dedicate it to scanning. But I do need the IT-8 calibration, and I am lucky enough to have spare machines to experiment with.

[President_LSI]

Dual Pentium Processor not properly supported

Dear User,

Obviously you are not aware that our SilverFast development is based on the Nikon Maid SDK. Problems within the SDK cannot be solved by us. We can only forward them to Nikon Japan and hope for a fix.
So far the usage of dual processors is rare with sophisticated colour software with colour management. That is why you are the first to report this problem.
You also should not call our software buggy, since it is much more stable then other software and by far not expensive except if you compare with unprofessional shareware or can you relate to any other scan software at that price with the functionality that SilverFast has?

Also I recommend you work with us, in a constructive way, and we will do our best to solve problems. None of our developers that work hard get motivated by negative comments!

regards
Karl-Heinz Zahorsky, CEO

[LSI_Magnussen]

The reason that you cannot find a note about your problem with your Dual PIII system in LaserSofts literature, and that we haven't posted a fix is simple: the problem is (or was) unknown.
Could you please specify exactly, which problems do you have? "SilverFast is unstable" is not very specific. Does SilverFast crash? If yes, what do you do to make it crash? Which operating system are you using? etc.
Please give us as many information as possible and we will try to fix the problem.

[UrbanVoyeur]

Hmmm....

Actually I was aware that Silverfast is based on the Nikon SDK. I gleaned that from the files that Silverfast installs and the fact that part of your fix involves re-installing the Nikon software.

Its just that I don't care. What you base your software on is YOUR choice. Buying your software is MY choice. At $400 (more that twice the cost of the operating system and 75% of the cost of Photoshop, 1/3 the cost of the scanner) I expect the software to work! And yes, for me, $400 for scanning software is EXPENSIVE.

In another post, I noted that v**s**n, a $40 competitor works flawlessly on dual processor systems. Whether or not it uses the Nikon SDK (it does not) is inconsequential. It works.

You may not call your software buggy - and that's understandable. Its yours. However I consider software that crashes constantly on an otherwise stable system to be unstable and by definition "buggy". We differ.

I have been very patient in the 4 months I've been trying to ge this to work. I posted my negative experience to get your attention and to alert other users. I also, by the way posted a work around, and ulitmately, an endorsement (>160 consecutive scans). How much more construcive do you want me to be?

I am also a developer, and I know that negative comments can hurt. But that's part of the job. It is necessary to know when and how your code, your pride and joy, fails.

Now, according to the forum user "schweikl", he reported this bug 6 months ago, so I am not the first!

I do hope this problem is fixed, but either way I will continue to use Silverfast, albeit on a dedicated scanning workstation.

[UrbanVoyeur]

Problems On the dual P3
( I gave all these details to cust support over the past few months)

Silverfast crashes at different points. It can crash on load, prescan, or scan. By "crash" i mean that all silverfast and photoshop windows close and i get either of these dialog boxes (roughly)"this application has completed an illegal operation" OR "photoshop.exe has completed an illegal operation"

Sometimes it crashes on the 2nd or 3rd scan. I can never complete a 3rd scan. It does not matter if set it to quit after each scan or re-start. Once a crash occurs, i can sometimes get is going again after 5-10 tries, sometimes after re-booting the scanner, and sometimes after re-booting the machine. Sometimes it can complete an IT-8 calibration, sometimes not.

It will crash regardless of which film adatper is used.

I tried it on 4 different dual processor systems - 3 tyans and 1 MBI DVVAA with exsiting and clean OS installs.

Dual System details:
System 1,2,3
Tyan Tiger 100 Dual P3 Motherbaord
2 800 MHz PIII processors
1 GB Kingston PC 100 RAM
Nikon Firewire Card (came with the scanner)
Windows 2000 Pro SP-2, fully updated
Photoshop 6.01, fully updated
Silverfast 5.5.1r03 & 5.5.2r02

System 4
MBI DVVAA Dual P3 Motherboard
2 800 MHz PIII processors
372 MB Kingston PC 100 RAM
Adaptec Firewire Card
(Otherwise the same software)

Silvefast DOES work on the MBI DVVAA Motherboard with a single celeron 600 chip.

[Rob]

UrbanVoyeur,
I'm running a Dual PIII system with SF 5.5.1r03 with no problems. The only three differences in your setup over mine is I have an ABIT VP6 board, 1.5 GB of 133Mhz ram, and I use a Adaptec 3940 UW PRO scsi card. With this configuration I have never had SF crash like you describe.

The only time I have had any problems is when I have my Micronics flat bed scanner on the same SCSI card turned on with the Nikon LS III any try to scan with the Nikon. Even the few scans with 5.5.2 work I just do not get 48 bit output.

Rob

[UrbanVoyeur]

I have no problems with Silverfast 5.5.1 for the PIE 1800 (USB) or Umax 2400 S (adaptec 2940 SCSI) with a dual processor machine.

The problems I experience only happen with Nikon LS-4000 (firewire)... which is substantially different from your LSIII (SCSI).

[schweikl]

Hello, I have been out of the forum for about 3 months.
UrbanVoyeur, you mentioned my reply 6 months ago. At that time I sent my bug reports to LSI without getting any helpfull answers. I did not even get the information that SF is depending on the (more stable) Nikon software.
For the moment I am using the v**s**n which is even able to do slidefilm batch scans from the negativ film feeder. I don't believe that v**s**n is based on the Nikon SDK. As it is a single thread program it is not depending on any race conditions in the dual processor system. The backdraw of v**s**n is it's speed. It is very slow, but faster than SF or CoolScan with several restarts after crash.

My Hardware: aOpen Dual-PIII with plenty of harddisk and ram space (40GB / 768MB)

I am professional software developer and would be glad in helping LSI (and Nikon) getting the drivers more stable. I told this offer to LSI on telephone about 6 weeks ago and they promised me to send an email answer. I did not get some! !!!

Does LSI not need any satisfied customers or are they overloaded when working on all requests?

<font size=-1>[ This Message was edited by: schweikl on 2002-02-25 14:19 ]</font>

[schweikl]

Additional hints for LSI developers:
Isn't it possible to restrict the driver to just using only one of the two processors in the system. There are system calls to do so.
This would be a workaround for any thread being too fast.
I would be glad in testing any patches.
Please contact me directly:
"richard@schweikl.de"

[LSI_Flyvbjerg]

I have just tested SilverFast with the Nikon LS IV ED on my Dual PII system (2 * 400MHz) under Win2K. No problems there. The LS IV ED is connected via USB, but else it is very similar to the LS4000, expect for a lower optical resolution. The LS IV ED and the LS4000 are using the same Nikon SDK.
Possibly the problem is a Pentium III or chipset problem.

Schweikl,
Your proposal to restrict the use of only one CPU could be a workaround. Thanks for the tip. We will make a test version, which we will send directly to your EMail adress. It can take some days, because we are very busy for preparations for the CEBIT.
After the test, we will publish the results in this group.

Eric.

[schweikl]

Eric,

>> Possibly the problem is a Pentium III or chipset problem.
May be, because I already read some other mails stating that their dual processor system is working very fine on the LS-4000. I am sure it's just a matter of some nanoseconds (which is different on various chipsets) when one thread is maybe slower than another, which is depending on the first.

Do you have any direct contact to the Nikon developers? I tried to get contact to them via german support, but, I failed. They don't seem to be interested in constructive feedback.

Richard.

[LSI_Flyvbjerg]

Richard,

Yes, we have an excellent contact to Nikon?s developer in Japan. I will ask them for help.


Eric.

[Michael Gay]

I revisited the site to see if anything had been done to resolve the problems of using the LS4000 scanner on a dual processor machine.
I was astonished to see a statement "you are the first to report it" dated in February.
I reported the problem in December last year and received a reply.
At that time the Nikon scanner interface was blamed but the problem remained with the 3.1.2 Nikon version, which works perfectly with the Nikon software. I also informed Silverfast of this.

Can the company not manage its data base, or is it simply a policy to try to deny problems?

[schweikl]

I tried the nikon coolscan 3.1.2 and my experience was not good:
It is the most unstable scanner software I ever tried.
It crashes during startup. There is no chance to press any button on the program window.
The only usable software for the LS-4000 is v**s**n.
I am still waiting for that promised test version of SF.

[LSI_Flyvbjerg]

The test version is still in the queue, but as I mentioned before, we are very busy until the CEBIT next week.

Eric.