Conference noted::ibmpc-95

    Well I took the plunge. I went with a Smart and Friendly 1002 (2x2).
    
    So far I have made 1 CD using CDRFS and 1 based on ISO standards.
    Both worked fine... Although Easy CD-Pro I wound up recording at 1x
    speed because I kept getting 'buffer under run warnings' during the
    testing phase. (Hard to belive with a system that has 72 Mb memory, 133
    Mhz Pentimum and writing from a 'created image')
    
    I do have 1 'minor problem'... When using CDRFS, and Explorer I can
    check 'properties' to see how much free space is left on the CD-R. If I
    then dump some data onto the CD-R (Like this weekend I archived over
    100 floppies!) I then go back to Explorer and click properties it crashes
    (don't recall the specific message but it drops to an all Blue Screen
    and brings up an exception message) the only recovery is a 'Hard
    Reset'.  (hitting return brings me back to W95 but all my shortcuts are
    gone and using the Start - Shutdown only cause the system to 'hang
    forever in the shutdown phase'.
    
    Anyone else seen this with CDRFS? (V1.0 came with the unit)
    
    Jim

    What's your hard disk? Sounds like it's not fast enough... Common with
    IDE disks, BTW.
    
    Go thru the EASY-CD Pro help on buffer underruns - ther are a number of
    things you may be able to do to improve things. 
    
    There is also a lot more info on the Adaptec Web site (they bought
    Incat systems)
    
    Chris
    

    I have three drives all are IDE. The newest drive is 2.1 GIG which has
    a transfer rate about double of the 'required' speed. I did the
    defrag (even tried creating an ISO image on the hard drive)... kept
    failing during the testing phase (by the way it ran greater than 60%
    through the test both times before it failed.)
    
    Personally my gut feel is the 'problem' is more likely with the test 
    program than my system's ability to keep up. Eventually I just told it 
    to write single speed and not bother with the test. Next time I just
    may tell it to write the CD at 2X and see what happens...
    
    I wish CDR-FS was the standard it works great (except for the
    Properties/Crash problem in my original note ;-)  )
    
    Jim

    re: .27 by CHOWDA::FAHEY
    
    Jim,
    
    On an IDE system without DMA, the cpu does a bit more work.  Things are
    a little better with SCSI.  Also, you're running Win95, right?  There
    may be some 16-bit code still in use that's taking up more than its
    fair share of the processing power.  Make sure you have nothing running
    that isn't absolutely necessary.  Use CTL-ALT-DEL on Win95 to see what
    else is running.  If you have Find Fast, get rid of it.
    
    I have the same drive as yours running on NT 3.51 SP5 and it works fine
    at 2X.
    
    >> ...not bother with the test...just may tell it to write the CD at 2X
    >> and see what happens...
    
    I think what will happen is that you'll waste a CD-R disk, but it's up
    to you.  The test is a pretty good simulation of whether you can supply
    data to the CD-R fast enough.
    
    Cheers,
    
    /Bill

    I'll second Bill's comment - if the test run fails, you will almost
    certainly have a failure if you try to burn a disc.
    
    The test run is done -exactly- as if you were really burning the disc,
    but with the laser power set low enough that it does not actually write
    the disc. The data is transfered to the drive, thru the drive buffers,
    to the laser, all at the proper speeds. The drive is spinning the
    media, following the track on the CD blank. If it runs out of data, you
    get an underrun. It will do the same if the laser is on at full power.
    If the media is too small, it will detect that at the end of the test,
    if not before.
    
    I'm in the process of trying to put together a CD-R package for Digital
    as I write... Getting a reliable combination of drives, SW,
    controllers, and system configs is non-trivial, as I am unfortunately
    finding.
    
    Win95 is probably the easiest of the O/S's to support - with enough CPU
    and memory, things work most of the time. NT, with its multi-tasking,
    seems to be a bit harder to make work.
    
    Chris
     

    I also have the S&F 1002, and found that I had to record at 1x as
    well.  This on a Dell Dimension XPS166s system that is just a little
    under a year old.  I also found that I had to CTRL-ALT-DEL and stop
    all active tasks (except Explorer!) to ensure a successful burn-in!
    
    Ah well.
    
    john

    re: last few
    
    I guess I was lucky to get it working on NT.  Funny, I would have
    thought NT's multitasking more reliable than Win95's.
    
    My system's a P133 with on-board ultra-wide SCSI, but the disk is just
    a good plain old 7200rpm narrow SCSI Seagate.  The CD-R is on a
    separate PIO SCSI controller.
    
    /Bill

    The multitasking problem seems to be that you can get swapped out if
    too much else is going on... It's easier to lock out the other
    processes in '95.
    
    What are you using for SW, Bill? (just curious...)
    
    Chris
    

    re: .32 by LEFTY::CWILLIAMS
    
    >> What are you using for SW, Bill? (just curious...)
    
    Just the no frills Incat software that came with the Smart and Friendly
    bundle.
    
    /Bill

    
    We've seen problems with a Sony 2x writer (not sure of read speed) when
    using an internal SCSI cdrom (2x) through an adapter to the external
    writer where doing 2x gives us buffer failures and trashes the media
    making it unusable. If we run it with the test run and then copy and 1x
    speed it works fine for regular cdrom media and takes approx. 74 mins
    to complete.
    
    

    re: .34 by SUBSYS::MSOUCY
    
    You need a 4x reader to do CD-to-CD-R writing at 2X.  I've done it
    with no problems.
    
    /Bill

    Based on all that I have read in this notes file I went home last night
    and tried a couple of things.
    
    1. I did an end task on everything except explorer.
    (two Clean Sweep programs, System Agent, and OCRware (for my scanner)
    
    I then created 270 meg of dummie data....
    
    Did a 'speed test' which said I should be able to write at 4x on the
    FLY!
    
    Started a write test at 2X which failed after about 20 minutes. (Buffer
    Under Run) Selected the option to try the test at 1X. It also failed
    after a similar time period. (with buffer under run) 
    
    I tried building the ISO image on Hard Drive (actually a couple of
    different hard drives) but the CD-R write test failed at 2X and 1X.
    
    I reduced the size of the Data Set to about 170MB and got a successful
    test of write on the fly at 2X!! Tried it again and it failed :-(.
    
    At this point I suspect that I am having a problem with the infamous
    'Thermal Recalibration'. I do hear a lot of drive clicks as the failure
    is occuring but it's hard to tell if the hard drive activity is the
    cause or the result of the failure.
    
    I will fix this problem (hopefully without having to purchase a new
    computer....;-)    )
    
    Jim
    
    P.S. Ignoring the CD-R issues for a moment I have noticed that from
    time to time my system will have lots of disk activity for no apparent
    reason... For example an application that normally takes only a couple
    of seconds to open will take 15-20 seconds while lots of disk thrashing
    is going on... Any thoughts?

>    At this point I suspect that I am having a problem with the infamous
>    'Thermal Recalibration'. 
    
    If you think that's to blame, just leave your PC on for a couple hours
    before you try to burn a CD-ROM.  The internals should be at a nice
    steady state by then.
    
    I have an old Quantum drive in one PC that used to do this every five
    or ten minutes.  Now this machine is left running 24 hours a day, I
    hardly ever hear it re-calibrating itself.
    
    John

    One thing I would do if I were burning CD-ROMs and had either Office-95
    or Office-97 installed:
    
    Find FINDFAST, and nuke it.  It'll come up about once an hour and beat
    the crap out of your hard drives.  It's certain to disrupt a CD-ROM
    burn.  While you're at it, killing the system agent probably wouldn't
    hurt, either.

    The saga continues.... but I'm getting close...
    
    I have finally been able to get to the Adaptec Web Site and there is a
    lot of information regarding 'buffer under runs' and Win95. It
    certainly does mention FASTFIND but many other disk and cache related
    parameters are also discussed. 
    
    It really appears (as I had suspected) that the problem is related to
    Win95 waking up and trying to do something while the CD burn is in
    progress.
    
    For the record I tried one of those 'don't do this at home' things last
    night. I turned off Virtual memory. For the first time I was able to
    successfully do two 'test burns' at 2X!!! Then turned VM back on and it
    failed!!! sounds good right? Turned VM back off and system 'locked'
    during burn. Hard re-set re-try... (VM still off) buffer under run  
    ^%&^%*(&)(*&...
    
    I liked computers better in the days when you could conduct an
    experiment and get the same ouput based on the same input... Nowadays
    it seems the typical WIntel system has has a random results
    generator...
    
        Jim

    Well, yeah, W95 is fond of spontaneously reorganizing its paging file,
    just to get in your way.

    re: .39 by CHOWDA::FAHEY
    
    >> Nowadays it seems the typical WIntel system has has a random results
    >> generator...
    
    Bill Gates would rather you call them Operating Systems.  ;-)
    
    While doing some video capture work on Windows for Workgroups, I
    discovered some rather counter intuitive (to me) bits of information
    relative to caching.  It turned out that I got smoother data flow with
    caching turned off.  It appeared that the steady state flow of data
    from the video capture device, plus the bursts of data when the cache
    started to fill up caused a bottleneck getting to the disk, thereby
    causing dropped frames.  I suppose if the cache were smarter, the
    bunching up of data would not have happened.  This was with Workgroup's
    32-bit file access by the way.  I did not see any way to affect cache
    policy, whether write through (preferred in this case) or write back.
    
    I'm not sure if this applies to Win95, but it may give some insite.
    
    /Bill

I've had the same experiences doing video captures.  Only problem is that I
can't figure out how to disable the disk cache on W-95, which is rather
moot anyway, 'cause W-95 has recently decided to pretend my capture board
doesn't have a proper IRQ assignment anymore.

    Well I finally found some useful technical information. it's a users
    helping users file on the adaptec web site. (heaven forbid that any of
    the manufacturers would publish something useful for buffer under runs
    on W95  ;-)   )
    
    1. Turn OFF auto insert notification on all CDs (recorder and ROM)
    2. If you have Microsoft office installed, find the 'Fast Find' utility
    and kill it. (Fast Find apparently wakes up periodically and causes
    disk activity while it re-indexes.)
    
    3. Don't let W95 control the size of your disk cache!! (I think this
    was my problem) Apparently when you are doing a lot of reading W95 may
    take all available memory as disk cache!. At some point this
    consumption of memory could cause either paging or swapping. The Theory 
    is that this won't 'normally' happen except when doing something like 
    reading hundreds of megabytes. To control the cache size edit the 
    SYSTEM.INI file.
    
    Find the VCACHE section... The section is there but there are no
    entries in it.
    
    add a min file and max file line
    
    [VCACHE]
    
    MinFileCache=512   (Do NOT use 0 this will cause WIN95 to manage your
                        cache)
    MaxFileCache=16384 (Rule of Thumb use 25% of physical memory, in my
                        case I have 72MB total - 16MB cache)
    
    (There is also a reference to making the swapfile MIN and MAX the same
    value. This apparently will prevent W95 from 're-sizing' the swap file
    in the middle of the burn.... I did not do this as I don't like to
    change too many things at once when I'm in trouble shooting mode)
    
    After editing the file and re-booting (of course) I did a test burn of
    about 300 Meg. It ran fine!! I can also tell you that the behavior of
    the hard drive during the burn is much differnt than it was. The drive
    now does more frequent reads for shorter periods of time. (this is
    based solely on observing the disk activity light). This would seem
    contrary to what you want to accomplish but it apparently prevents the
    system from paging which is a possible cause of buffer under runs.
    
    So far this works for me!! (one test doesn't necessarily prove anything
    but I have good vibes about this)
    
    Jim
                                           

    re: .43 by CHOWDA::FAHEY
    
    Those tips make a lot of sense to me.  I use the MIN=MAX pagefile
    setting on NT to prevent fragmentation, but I suppose it helps by
    not requiring the O.S. to resize as needed.
    
    Cheers,
    
    /Bill

    re .43.
    
    I've got another question that would go along with that too....
      Do you notice (just a guess would be ok.. I guess..) if the system
    speed is any different as a result of the changes you made?  One might
    think that if the system had to do less overhead you might notice a
    difference in speed too.  
    
    L8tr..
    SteveM....

    Well the short answer is yes. BUT... I finished my 'test burns' around
    10pm and didn't really do anything in the way of running
    applications...   It did 'feel' like windows were popping open a bit
    faster etc.
    
    When I ran shut down the, message "It is now safe to shut 
    down" 'seemed' to occur much faster than I have ever seen it before....
    
    I think this is one of those cases where large memory systems (ie
    greater than 32MB) haven't been heavily tested in the W95 environment.
    I can see where there can be a whole set of tuneing parameters that are
    based on a 16MB system and not optimized to take advantage of larger
    memory configurations...
    
    By the way I have not seen much comment on the Sony CDR File System
    (CDRFS) It came with my unit (at first I thought it was a driver). I am
    just curious as the 'buffer under run' problem does not seem to be an
    issue with CDRFS. I have made one CDRFS disk... I was curious as to what
    'formatting a CD' was all about. I have added  to that disk many times
    and am getting close to 'freezing it'. I realize that it's not an ISO
    disk but it would seem that what ever they are doing that doesn't
    require a 'continuous burn' should be applied to creating an ISO
    standard  disk. (I guess it has something to do with creating 'tracks'
    versus a helical type write..?)
    
    Jim                     

    The Sony CDRFS is a "packet write" file system for CD-R's. It writes
    data in smaller packets, which are always smaller than the drive's
    buffer size, so the data can be fully transferred to the buffer before
    writing. Buffer underruns cannot happen if the data is written this
    way, but there is a significant capacity overhead.
    
    CDRFS is Sony proprietary, and incompatible with anyone else.
    
    The standard filesystem for packet write, CD-E, and DVD is CD-UFS
    (universal file system), which is being implemented by Microsoft, all
    the CD-R SW vendors, the DVD consortium, etc... It will probably be
    available in the next few months.
    
    The standard is online at www.osta.com, available for download.
    
    Also look at www.cd-info.com for more refernce info on CD and DVD
    technology.
    
    Chris
    

    My last (I hope) update on CD-R technology.  As a point of reference my
    system is a Starion 942 (133 Mhz) 72MB memory. I use the Smart and
    Friendly 1002 (2X2) CD-Recorder with the included Adpaptec SCSI
    controller (ISA version).
    
    I have burned 3 CDs at 2X, on-the-fly, without any difficulties. 
    (Coasters=0) I made extensive use of SYSMON and the 'TEST WRITE'
    feature of Easy-CD Pro. When I do a 'Test Write' SYSMON is all 'Flat
    Lines'. Disk Reads, Writes, Swapfile size, etc are all stable during
    the test. (I do not run SYSMON during an actual burn.)
    
    For anyone considering CD-R technology I would strongly urge you to use
    the adaptec web page technical information and review it carefully. If
    you are not comfortable altering the system parameters discussed there
    then stay away from CD-R technology and WIN95.
    
    The highlights once again were.
    
    1. Stop all background processes (CTRL-ALT-DEL end all tasks but
    Explorer)
    
    2. Modify the minimum Virtual Memory to be greater than what is needed
    when burning a CD. (you don't want the swap file to re-size up or down
    during the burn)
    
    3. Defrag Defrag Defrag (even if it says drive is 0% fragmented)
    
    4. Source Disk should have a cluster size of 16K or less. (ie partition
    your 1Gig and larger drives) I created at 650 MB partition as my source
    for CD-R input. 
    
    5. Esatblish a minimum and maximum read file cache size. 
    
    Jim
     

    All good pointers, and well appreciated.
    
    If you ever try this under NT, I'd be interested in a similar list of
    pointers.

    We're doing evals on NT CDR SW also... Adaptec's Easy-CD Pro seems to
    be the best implementation we've seen so far, for both W95 and NT.
    
    Heavy network traffic will likely cause problems during a burn. Non-DMA
    SCSI controllers cause problems. ATAPI source CD's, or IDE disks of any
    sort degrade the margins also...
    
    But it all works. No frisbees yet, and we've been trying to make it
    fail.
    
    Chris
    

	We have a 486 running Windows NT and Gear by Elektroson. It
	works well. The system disk is an IDE disk and we also
	have a SCSI controller. On the SCSI controller is the burner
	(a Phillips 522. BIG thing), a CD drive and 2 SCSI hard disks.
	These are used for staging the files to be burned and for
	creating the physical image file which the CD burns from.

	This cuts down on all the problems of network delays. All the
	traffic stays on the SCSI bus. It's worked flawlessly for 
	well over a year.

							mike

    The network issues I mentioned in .50 are due to network traffic eating
    CPU and system resources, not transferring data to the CD-R over the
    network. I was assuming that you were doing local mastering.
    
    Mastering from the network directly -will- cause production of
    frisbees.
    
    Chris
    

	Which is why the configuration I have includes the staging
	disk. Move your files there first, then burn.

For those who are interested in another package, written for a former Digit:

See www.mainstream.net/goldenhawk

I know there is some free software there.  This is one of the few packages that
will copy almost ANY format music or otherwise...

	- mark

3 of us have tried a JVC2010. It came with v4.03 of their Personal
Archiver. This didn't work with NT 4 at all. We called JVC and they
said we would get a free upgrade to v4.1 (NT4 support) if the drive
was manufactured after Aug 96. We send the serial #, but they replied
saying they couldn't find it & it was possibly OEM'd by HIVAL. We
contacted th eplace where we bought it & they gave us a beta copy of
Adaptec's Easy CD Writer.

#1: P166 w/64MB Symbios SCSI
The attempt to use v4.03 completely screwed his system &
he rebuilt from scratch. With Adaptec software he got through the
test mode, and it failed about 10 minutes in - scratch 1 blank! He did
get it to write 1 CD.

#2: 486/133 Adaptec 1540
Win95 - worked fine using JVC software (v4.03)
NT4 - worked fine too (CD->CD)!

#3: P133 w/64MB Symbios SCSI (always CD->HD->CD)
NT4: The Adaptec software continually failed with insufficient memory
during the test phase.
Win95: JVC software failed during the test phase
Adaptec software actually started the test phase but failed

Got JVC V4.01 from HIVAL
NT4: Testing phase got to 99% before reporting an error - same the 2nd
time, after that it died immediately test writing the lead-in track
Win95: - same lead-in error
tried another blank with the same results

#1: 2nd attempt
NT4: Test phase worked fine - wrote CD - no errors - can't read CD!

We have had enough. We are going to return the drive tomorrow are
looking for a drive/software combination that will work reliably.

 mark

PS 2x, 1x didn't make any difference

    re: .55 by FSCORE::KAYE
    
    There must be a lesson in there somewhere.
    
    e.g. Stick with Adaptec hardware if using Adaptec software  ;-)
    
    Cheers,
    
    /Bill (still happy with Smart and
    Friendly 2x/2x [Sony mechanism] with
    Adaptec Incat software and Adaptec
    host adapter; works reliably at 2x
    on NT 3.51 Pentium 133 with all SCSI
    disks)

    Since making the tuning adjustments I referred to in a previous note I
    have successfully created 12 CDs. 10 were standard burns, 1 was a music
    CD and 1 was a 'mixed mode'. I have had two failures. One failed halfway
    through at 2X. The second wrote the CD at 1X but I couldn't read it in
    any drive. Both failures were on the same media type. The media
    is currently being evaluated by 3M.
    
    Media I have used successfully is TDK and Fuji plus the Smart &
    Friendly supplied disk (unknown manufacture)
    
    Windows 95, Digital Starion 942 (133 Mhz), 72 Mb Ram, EIDE hard Drives,
    Smart and Friendly 2x4 connected with supplied Adaptec Controller.
    
    Overall very pleased with the drive.
    
    Jim