🚀 go-pugleaf

JR>Subject: What's holding up PKZIP v. 2.0 (message from PKWARE)
JR>Date:  7 AUG 92 12:10:54

JR>During the development of PKZIP v. 2.0 for MS-DOS, PKWARE has
JR>discovered hardware faults in newer 80386 & 80486 computers.
JR>These faults can cause anomalies in the operation of PKWARE
JR>software as well as other software.

JR>PKWARE has been endeavoring to locate an anomaly that has been
JR>observed by a number of customers and associated vendors.  The
JR>intial symptoms of the problem were:

JR>  - Particular files extracting with errors on one machine, but not
JR>    on another.

JR>  - Installs of some products failing.

JR>These problems have been observed primarily on newer, fast (33Mhz
JR>or faster) 80386 and 80486 machines.  Investigation of the problem
JR>revealed no difference in software or hardware between machines
JR>that failed and machines which did not.  The files which fail do
JR>not fail every time.

<RAM copy test program message deleted>

JR>If the EXTERNAL CACHE is turned off through the BIOS
JR>configuration, the problem does not occur.

JR>This is indeed a hardware flaw.

I would like to take this chance and chip in my $.02

I had had this problem EXACTLY as described above. I was trying to
upgrade my external cache from 64K to 256K. My original cache is 15ns
8K*8 SRAM, however my 32K*8 SRAM is 20ns. After I finished the
installation, I ran some diag. programs overnight and everything was
checked out OK.

Starting from that day, occassionally ROBOCOMM and SLMR (BBS automated
off-line mail programs) reported error when unpacking (most of the
time) mail packets. At first, I thought it was cause by transmission
errors.

One day I needed to test some of my zipped files and found a lots of
them failed 'PKUNZIP -t' test (CRC error, I think). Again, I guessed it
was cause by my HD and I went ahead and deleted all those files. At the
mean time, I restored original copies and ran '-t' test again. BOOM !
Error ! I started to question if something had gone wrong. So I reran
the test again and found some of them passed ! Reran again, and they
would fail !

After trying all software configurations (boot clean, load DOS low, use
HIMEM instead of QEMM, etc), problem still exists. Logically, I started
to change my hardware settings. First thing was to disabled EXT. CACHE.
Ran some tests and they ALL passed !!

Knowing the ext cache was the cause, I switched back to 64K. The problem
went away and did not happened again. Later I called Micronics, their
tech. told me I need to use 15ns SRAM with my motherboard.

So I think memory/cache access timing has something to do with the
problem. I've seen a lots of motherboards with wrong speed SRAM on them,
particularly TAG RAM. Moreover, my opinion on it usually happened on
newer 386/486 MB is because they are the boards which usually come with
ext. cache.

Just a though.

Wilfred Wong
The Cutting Edge BBS
Houston, Texas, USA.
InterNet: wilfred.wong@cutting.hou.tx.us
---
 . SLMR 2.1 . It's only a hobby ... only a hobby ... only a