Conference giadev::decstation

    re: NETRIX::"[email protected]"
    
    Steve,
    
    Please escalate via official channels.
    
    Thanks,
    
    /Bill

    Steve,
    
    I'm very interested in this topic.  We are getting ready to order 180
    of these systems for my US Army customer...  Please post your findings!
    
    Thanks,
    Tom

    It is unclear from your note what the old rev's are and what the new
    rev's are. Are you saying that all the replacement MLB's are rev H01?
    
    If you feel that this is a serious problem and not just a high DOA
    rate, then you should open an IPMT case. If you do open a case, you
    should be perpared to provide as much information as possible on the
    failing boards:
    
    1. MLB part # and rev
    2. data from the SIS chipset (there are 3 SIS chips on the MLB: 5511,
       5512, 5513 - provide the complete chip data - all the markings on
       each chip)
    3. Any other add-on boards in the systems
    4. Any other applications other than the standard factory installed
       software.
    5. Operating system (WFWG or WIN95)
    6. Exact error messages
    7. Whether or not each system fails in the same manner (if different
       failures on different system, tie the MLB part # to the failure)
    8. Any other information that may have a bearing on the problem
    9. It would also speed up an anaylsis of the failing MLB's to have a
       complete system (in the failing state) with complete step by step
       instructions on reproducing the failure sent to the PCBU Engineering
       Support Group.
    
     
    For your inforation:
    
    54-24544-02 is a FX MLB without multimedia - this was the first etch
    for the FX.
    54-24588-02 is a replacement etch for the 24544-02
    54-24588-03 is a FX with multimedia.
    
    The 54-24544-02 only had a rev H06 (released with BIOS rev 1.07).
    Higher rev's were because of new BIOS releases (rev J06 for 
    example was released with BIOS 1.08).
    
    The 54-24588-02/03 went from rev F01 to H01 with BIOS 1.07.
    
    regards,
    
    glen kelley
    pcbu engineering support

    Are these Win95 systems or DOS/WinV3.11?
    

Glen,

Thanks for your reply on the different MLB's...
I realise that a lot more info is required to raise an IPMT case, of which I
am
compiling now. I was just wondering if anybody else has had something similar.

The engineers reported that MLB's stamped H02 have this problem. Even units
that
have no client software, just OS installed either Win95 or WFW 3.11 can
"corrupt"
disks. No other hardware is installed in the machines.

I will post more info as I find out.
Thanks and regards,

SteveK
[Posted by WWW Notes gateway]

Hi everyone,

A little more info.........

IT seems now that this problem may only be with WFW 3.11 and the SIS chipset 
drivers. After editing Config.sys and system.ini .. ie : adding software
switches to drivers, seems to have "stabilized" the system.

Does anybody know if the FIS version has changed for the FX's ?
 
Glen,

I am getting a lot more calls now on this and would greatly appreciate if you 
could let me know anything that you find out. I have another 3 sites with
multiple
machines of this problem.

Thanks and regards
Steve Kaniowski

PS.. Upgrading the BIOS to 1.09 did not resolve problem. Also, the SIS chipset
is version 1.18.

[Posted by WWW Notes gateway]

    Steve,
    
    BIOS v1.09 did not really "fix" any problem, it just added support for
    the AMD K5 chip (the name that flashes on the boot screen).
    
    The WFWG FIS has been consistent since release. Also, the SIS driver,
    SISIDE, should only be used for WFWG. The shipping and tested version
    is 1.18.
    
    We've seen some problems with 3rd party memory managers such as QEMM
    causing this problem while running WFWG and using the SISIDE driver. If
    possible take all systems back to a clean install state without any
    additional software or drivers loaded, then run system tests. If all
    the tests pass, add in the other SW one app at a time.
    
    For Win95, there is a patch from MS that addresses problems with
    systems using large HD that are partitioned into two or more logical
    drivers. Without this patch, you will see data corruption and other
    strange problems. If this matches the customer's configuration, get the
    patch and see if that fixes the problem (look on MS home page at
    HTTP://WWW.MICROSOFT.COM  and find the "Free Software" section that
    talks about updates).
    
    Again, if this is a major problem, you really should open an IPMT case.
    
    regards,
    
    glen
    

G'day Glenn !!!

Some more info.....

For now it seems that it is an SIS problem, with WFW 3.11 only.....

It sounds unfortunate but, in the readme files of the SIS drivers V1.18 there
is 
important information about running PIO mode etc.... We have put these in
place
in New Zealand a few days ago, and the systems seem to have "stabilized".

We are trying another site for TAFE here in Sydney to see if in fact it does
resolve the issue.

There is also a recommendation by SIS that if there are problems such as 
intermittent hanging with mode 4 drives, to edit the device driver in 
Config.sys with a couple of software switches... 

            ie; Device=c:\siside\sisdr.sys /A3 /R3 .

There was also talk of that currently the drivers are not shipped with the
FIS,
and the fact that they used to be... means somebody may know about this ???
I know you mentioned that the FIS has'nt changed from what you know but, 
apparently in the MakeMaster diskettes it isn't there anymore....or maybe the 
Win95 FIs has changed ?  Who knows ?

Anyway I will post an update of the results of the tests....

Thanks again
Steve Kaniowski
Region Support 
Australia




[Posted by WWW Notes gateway]

Oh.... By the way....

Apparently replacing a rev H02 system board with a J03/J04 "fixes" the problem
immediately.... I'm not too sure about this as of yet, because I suspect
that the engineers may be getting a bit "lost"....(Not their fault, just one
of those things).

I was'nt able to find a J04 but I did find a J03 and compared the SIS 5513
markings with a H02. 

The markings are as follows :

     H02                 J03
  
   PGD2749              PGD1633
   9639IT B1            9626IT A8

I wonder what the difference is ?

SteveK
[Posted by WWW Notes gateway]

    I went back to some of the notes that we had collected on this problem
    and what we found had to do with the rev of the 5511 chip. If you used
    the SIS 1.18 driver with the E step of the 5511, you had problems in
    WFWG. If you use the F step of the 5511 and the 1.18 SIS drivers you're
    OK.
    
    As to the FIS, the WFWG FIS has not changed since first customer ship.
    The SIS drivers were ONLY shipped with WFWG, not Win95 or NT.
    
    Also, I don't believe that you can use the DMA mode with the SIS
    drivers, only the PIO mode.
    
    Let me know what you find.
    
    thanks,
    
    glen 

For info only .....

Markings on 5511 on H02 : PGD3053
                          9642DT E3

                    J03 : PGD1562 
                          9625DT E3

            5512 on H02 : PGD2776
                          9639BT B1

                    J03 : PGD1577
                          9625BT B1

Steve Kaniowski
[Posted by WWW Notes gateway]

Hi Glen,

Sorry to confuse the matters worse but .....

I have 2 working units ... 1 is board rev H02 and the other is J04. The 5511 
                           has a stepping F2 on both.

Also 2 none working units .. Both rev H02 and both have stepping F2.

Now I don't think that this problem may have anything to do with SIS chipset
revs.

I have recieved a couple more calls on this ..... again ..... disk
corruptions,
but we still have'nt got anything conclusive as of yet. We are still waiting
on
New Zealand, who have said just today their machines are still running OK so
far.

I have currently 2 "faulty" machines with me of which I myself am trying
various
things and will be loading software from scratch.

I'll keep you posted.

SteveK

[Posted by WWW Notes gateway]

    Can you send me a list of the 55xx chips on the good MLB's and the
    failing MLB's?
    
    Also, can you sent me the chip info on chip E21 (located next to the
    SIMMs, CPU, and cache). Also include the vendor of the chip.
    
    thanks,
    glen

OK Glen .... ( Venturis GOD ) !!!!



3 x failing PC's E21 chip : Toshiba 74F245D     Motorola F245         Texas
F245
                                    2253440              XAB9542           
6AAEEHK
                                    9518hL



3 x failing PC's 5511 chip : PGD3053      PGD2863     PGD2861
                             9642DT E3    9640CT F2   9640CT F2

3 x failing PC's 5512 chip : PGD2776      PGD2775     PGD2973   
                             9639BT B1    9639BT B1   9641BT B1

3 x failing PC's 5513 chip : PGD2749      PGD2517     PGD2743
                             9639IT A8    9637IT A8   9639IT A8


All above are board revs H02 ......




                                Board rev
                            H02         J04
                            ---------------------                           
2 x "good" PC's 5511 chip : PGD2861     PGD3686
                            9640CT F2   9648CT F2

2 x "good" Pc's 5512 chip : PGD2775     PGD3762
                            9639BT B1   9649BT B1

2 x "good" PC's 5513 chip : PGD2743     PGD3467
                            9639IT A8   9646NT A8


This is all the info I have at this point in time. The customer obviously
would
not hand over the working units for inspection, so these markings are relayed
information.

I have also at this stage blown away the software for the 2 failing units and
re-
loaded DOS/ Windows. I have then loaded the SIS driver V1.18 and followed the
instructions from README's. At this stage I have not been able to re-produce
the
fault.

I am now trying to chase up staging / PCI to find out what their procedure is
in
loading this software or if in fact we are at all.

Glen, once again I appreciate your prompt replies and suggestions, and I will
add
more notes as I find more detail.

Thanks
Steve Kaniowski
[Posted by WWW Notes gateway]

I don't know what happened to the last post but here are the markings again :

3 x failing PC's E21 chip : Toshiba 74F245D   Motorola F245    Texas F245
                                    2253440          XAB9542      6AAEEHK
                                    9518hL









[Posted by WWW Notes gateway]

Glen.....  some progress....

I've just been to staging / PCI and got one of the suspect machines re-fised.
I have found a problem where the FIS software adds an entry SIS_MASTER=ON in 
system.ini.

This is very critical as it changes the drive to run in DMA mode and obviously
we don't want this.

This problem does not seem to appear when 32bit disk access is not enabled, 
which would explain why some machines have this problem and some don't, and
why some customers have this problem and others don't.

As a test I copied all *.ini files to floppy. I then changed the system to run
32bit disk access in control panel. The system would start loading Windows,
and drop back to c:\. It would also "hang" and do other strange things like 
advise of corrupt swap file when loading.

These same symptoms was described by most engineers.

I then either rem'ed out SIS_MASTER=ON or changed it to SIS_MASTER=OFF, and
the 
problem disappeared. Files seem to be intact for now. I believe if this was
continually done then the files would definately be corrupt.

So now the question is ..... 

I believe the FIS for WFW is standard world wide. So are we the only one's 
seeing this problem ?

Keep you posted.
Steve Kaniowski
[Posted by WWW Notes gateway]

G'Day,

	Just to add a little data on the version of the SIS FIS image that we
are installing (which would be adding the seemingly offending line to
system.ini).

	The version of the driver is 1.18 but the FIS image module is known as
0C0SIDEA.004 (this will make sense to guys like Andy Fandel down on the ground
floor of AKO1).

	If the setting needs to be changed it will have to be co-ordinated
through the FIS engineering group so that it gets properly released to all of
the plants.

Regards,

David.

    Steve,
    
    Great news!
    
    I'll check with the FIS engineering group about that switch along with
    Taiwan enigeering (I'm sending you a separate note about this also).
    
    thanks,
    glen

I have been having a similar problem in the uk with a customer getting
corruption and GPF's in WFWG, we have have applied the fix and we are now
working all ok !

Cheers


Jason Lane
NBD Technical Specialist UK
[Posted by WWW Notes gateway]

    Did you check NOT to use EDO RAM  54-23623-JA  8MB SIMM's with
    4x Hyundai chips. See IPMT CFS.46870. They cause (timing?-) problems in
    Venturis FX, Celebris FX and Prioris systems and WIN95 or WIMNT.
    I also heard about problems with these chips in Compaq systems.
    
    There are no known problems with version 54-23623-JA with 4x NEC or SEC
    memory chips.
    
    Charles le Febre   hlfs00::febre_c