AminetAminet
Search:
84491 packages online
About
Recent
Browse
Search
Upload
Setup
Services

util/misc/ZIPTest-1.1.lha

Mirror:Random
Showing:m68k-amigaosppc-amigaosppc-morphosi386-arosi386-amithlonppc-warpupppc-powerupgeneric
No screenshot available
Short:test Amiga 3000 ZIP DRAM memory chips
Author:amiga at cdh.eebugs.com (Chris Hooper)
Uploader:amiga cdh eebugs com (Chris Hooper)
Type:util/misc
Version:1.1
Architecture:m68k-amigaos
Distribution:Aminet
Date:2024-01-17
Download:http://aminet.net/util/misc/ZIPTest-1.1.lha - View contents
Readme:http://aminet.net/util/misc/ZIPTest-1.1.readme
Downloads:207

ZIPTest is a command line utility to test ZIP fast memory installed in an
Amiga 3000 or Amiga 3000T.  This test requires AmigaOS 2.0+ in ROM.
It might run on other Amiga computers having Ramsey (Amiga 4000 and 4000T
systems), but will probably not provide more useful information than the
DIMM (bank) which has failed.

The utility is designed to detect and diagnose ZIP memory regardless of
whether there are defective chips which would prevent AmigaOS from
recognizing and using that memory.  This is achieved by accessing the
physical addresses where the memory is mapped by Ramsey in the Amiga.
The program should be useful for diagnosing bad connections to individual
ZIP ICs or individual bad cells within those ZIP ICs. In order to operate
on memory where some or all may have been detected and put in the freelist
by AmigaOS at startup, the program is designed to run from chip memory.
It does all accesses with interrupts disabled and restores memory contents
after testing.  If you are running Enforcer or equivalent (MuForce), you
may need to disable it before running this utility.  This is because ZIPTest
potentially accesses CPU physical addresses which may not have been added
to the memory list.


DISCLAIMER: THE SOFTWARE IS PROVIDED "AS-IS", WITHOUT ANY WARRANTY.
THE AUTHOR ASSUMES NO LIABILITY FOR ANY DAMAGE ARISING OUT OF THE USE
OR MISUSE OF THIS UTILITY OR INFORMATION REPORTED BY THIS UTILITY.


The motivation for writing the utility started with me receiving an Amiga
which had been stored in a basement for the past 21 years. After re-capping
and debugging other board issues, I saw that none of the fast memory
was being recognized. Not having much luck with swapping around ZIP ICs,
I started poking at the CPU's physical addresses where the memory should
appear and found nibbles in each bank which were not functional. The
process was not very difficult, but did require a significant amount of
time to research how the ZIP ICs were mapped into memory.  I decided to
share my experience by writing a utility which others could use to diagnose
problems with their Amiga ZIP memory.

The utility implements three main tests:
1) Data line test
2) Address line test
3) Memory cell test

If started without arguments, ZIPTest will run all three of these tests
in sequence. If the data line test fails a particular ZIP IC, then you
should not rely on results of the address line test for that IC. If either
the data line test or the address line test fails a particular ZIP IC, then
the memory cell test will also likely fail for that chip, and maybe for
other chips.

When run, ZIPTest will first display some information about your CPU and
memory controller.  For example:
    CPU: 68060 without Burst, MMU Active
    Memory controller: Ramsey-04 $d $3a (25.05 MHz)
    Memory config: 1Mx4 (4MB per bank) Burst (SCRAM required)
    Memory refresh: 240 clocks (9.60 usec)

In the above, the indication of "without Burst" means that ZIPTest has
determined that enabling Ramsey burst does not make a performance
difference for fast memory access. The likely reason is that the CPU's
bus interface does not support translation of the CPU's burst cycle to
the Amiga 3000's onboard bus, which includes Ramsey. This unfortunately
means that ZIPTest will not be able to probe and report whether the
individual ZIP ICs are Static Column (SC) or Fast Page Mode (FPM).
ZIPTest knows the difference between Ramsey-04 and Ramsey-07 registers
and can, for example, report when Skip mode is active with Ramsey-07.
The clock speed reported for Ramsey is calculated based on how fast
Ramsey DRAM refresh cycles are occurring relative to the ECLOCK
provided to the 8520 (715909 Hz or 709379 Hz) for high resolution timing.
This benchmarked number may not be accurate in all configurations,
especially if your memory or CPU is not fast enough to accurately track
Ramsey refresh cycles.


HARDWARE COMPATIBILITY
----------------------
Early Amiga 3000 computers have Kickstart 1.4 in ROM and will load 1.3 or
2.0+ from the hard disk and soft kick. ZIPTest may not correctly test the
fast memory mapped for Kickstart, or testing of that memory might crash the
computer. With the version 1.1 change where the MMU is disabled during
testing, ZIPTest might work correctly on this machine. If it does not work,
a work-around (untested) would be to hold down both mouse buttons at cold
power-on.  This will take you to the OS 1.3 / OS 2.0 selection menu. Click
in the upper left corner of the screen and it should attempt to boot with
the ROM-based Kickstart 1.4. You may need to boot from floppy.

ZIPTest has been tested with an Amiga 3000 and onboard 25 MHz 68030, as well
as multiple Amiga 3000 computers with A3640 and A3660 accelerators installed.
It also runs on the ReA3000, but test failure output is tuned to an A3000
with ZIP memory. On the ReA3000, bank 0 and 1 are the right SIMM. Bank 2 and
3 are the left SIMM. ZIPTest has also been run on the Amiga 4000 and 4000T,
but usefulness for reporting failing components is marginal there.

A feature of the address line test is that it can potentially also identify
whether each ZIP IC supports SC (Static Column) or just FPM (Fast Page Mode).
This identification can only happen if the CPU can burst accesses to Ramsey,
which is probably only the A3000 onboard 68030 CPU. The A3640 and A3660
do not support CPU burst accesses.


=============================================================================
Example test output from a fully working Amiga 3000 with 16MB Fast Memory
and A3640.

1> ziptest
ZIPTest 1.1 (2024-01-17) by Chris Hooper
CPU: 68030 with Burst, MMU Active
Memory controller: Ramsey-07 $f $28 (24.99 MHz)
Memory config: 1Mx4 (4MB per bank)
Memory refresh: 240 clocks (9.60 usec)

Data line test
  Socket   IO1  IO2  IO3  IO4   Socket   IO1  IO2  IO3  IO4
  -------- ---- ---- ---- ----  -------- ---- ---- ---- ----
  U881 3.7 Good Good Good Good  U879 3.5 Good Good Good Good
  U873 2.7 Good Good Good Good  U871 2.5 Good Good Good Good
  U865 1.7 Good Good Good Good  U863 1.5 Good Good Good Good
  U857 0.7 Good Good Good Good  U855 0.5 Good Good Good Good
  U880 3.6 Good Good Good Good  U878 3.4 Good Good Good Good
  U872 2.6 Good Good Good Good  U870 2.4 Good Good Good Good
  U864 1.6 Good Good Good Good  U862 1.4 Good Good Good Good
  U856 0.6 Good Good Good Good  U854 0.4 Good Good Good Good
  U877 3.3 Good Good Good Good  U875 3.1 Good Good Good Good
  U869 2.3 Good Good Good Good  U867 2.1 Good Good Good Good
  U861 1.3 Good Good Good Good  U859 1.1 Good Good Good Good
  U853 0.3 Good Good Good Good  U851 0.1 Good Good Good Good
  U876 3.2 Good Good Good Good  U874 3.0 Good Good Good Good
  U868 2.2 Good Good Good Good  U866 2.0 Good Good Good Good
  U860 1.2 Good Good Good Good  U858 1.0 Good Good Good Good
  U852 0.2 Good Good Good Good  U850 0.0 Good Good Good Good

Address line test
           A A A A A A A A A A Mem            A A A A A A A A A A Mem
  Socket   9 8 7 6 5 4 3 2 1 0 Type  Socket   9 8 7 6 5 4 3 2 1 0 Type
  -------- - - - - - - - - - - ----  -------- - - - - - - - - - - ----
  U881 3.7 G G G G G G G G G G FPM   U879 3.5 G G G G G G G G G G FPM
  U873 2.7 G G G G G G G G G G FPM   U871 2.5 G G G G G G G G G G FPM
  U865 1.7 G G G G G G G G G G FPM   U863 1.5 G G G G G G G G G G FPM
  U857 0.7 G G G G G G G G G G FPM   U855 0.5 G G G G G G G G G G FPM
  U880 3.6 G G G G G G G G G G FPM   U878 3.4 G G G G G G G G G G FPM
  U872 2.6 G G G G G G G G G G FPM   U870 2.4 G G G G G G G G G G FPM
  U864 1.6 G G G G G G G G G G FPM   U862 1.4 G G G G G G G G G G FPM
  U856 0.6 G G G G G G G G G G FPM   U854 0.4 G G G G G G G G G G FPM
  U877 3.3 G G G G G G G G G G FPM   U875 3.1 G G G G G G G G G G FPM
  U869 2.3 G G G G G G G G G G FPM   U867 2.1 G G G G G G G G G G FPM
  U861 1.3 G G G G G G G G G G FPM   U859 1.1 G G G G G G G G G G FPM
  U853 0.3 G G G G G G G G G G FPM   U851 0.1 G G G G G G G G G G FPM
  U876 3.2 G G G G G G G G G G FPM   U874 3.0 G G G G G G G G G G FPM
  U868 2.2 G G G G G G G G G G FPM   U866 2.0 G G G G G G G G G G FPM
  U860 1.2 G G G G G G G G G G FPM   U858 1.0 G G G G G G G G G G FPM
  U852 0.2 G G G G G G G G G G FPM   U850 0.0 G G G G G G G G G G FPM

Memory cell test
  Bank 0 [................................]
  Bank 1 [................................]
  Bank 2 [................................]
  Bank 3 [................................]

  Socket   Result   Socket   Result
  -------- ------   -------- ------
  U881 3.7 Good     U879 3.5 Good
  U873 2.7 Good     U871 2.5 Good
  U865 1.7 Good     U863 1.5 Good
  U857 0.7 Good     U855 0.5 Good
  U880 3.6 Good     U878 3.4 Good
  U872 2.6 Good     U870 2.4 Good
  U864 1.6 Good     U862 1.4 Good
  U856 0.6 Good     U854 0.4 Good
  U877 3.3 Good     U875 3.1 Good
  U869 2.3 Good     U867 2.1 Good
  U861 1.3 Good     U859 1.1 Good
  U853 0.3 Good     U851 0.1 Good
  U876 3.2 Good     U874 3.0 Good
  U868 2.2 Good     U866 2.0 Good
  U860 1.2 Good     U858 1.0 Good
  U852 0.2 Good     U850 0.0 Good

=============================================================================

Data line test
--------------
The Data line test walks all four data lines (IO1, IO2, IO3, and IO4)
of each ZIP, and verifies that each line is connected and that writes to
the chip can retain contents. Output from this test is a table in the
same order as the ZIP ICs are populated on the Amiga 3000 motherboard -
the top is toward the connectors and bottom toward the CPU expansion socket.

A result of "Good" means that the data line is functioning correctly.
A result of "!" means that am indeterminate value is always read from that
data line (sometimes 0 and sometimes 1). This could be due to a defective
or missing part. A "1" means that a 1 value is consistently read, and a
"0" means that a 0 value is consistently read. Unless you are diagnosing
at the board level, anything other than "Good" means that you need to either
re-seat or replace the entire ZIP IC. If the ZIP IC is not installed, you
might receive a variety of "1", "!", or "0" values.

Example data line test output:

1> ziptest data debug quiet

Data line test
  Socket   ADDR    IO1  IO2  IO3  IO4   Socket   ADDR    IO1  IO2  IO3  IO4
  -------- ------- ---- ---- ---- ----  -------- ------- ---- ---- ---- ----
  U881 3.7 71fffff Good Good Good Good  U879 3.5 71ffffe !    !    !    !
  U873 2.7 75fffff !    1    !    !     U871 2.5 75ffffe !    !    !    !
  U865 1.7 79fffff !    1    !    !     U863 1.5 79ffffe !    !    !    !
  U857 0.7 7dfffff Good Good Good Good  U855 0.5 7dffffe Good Good Good Good
  U880 3.6 71fffff Good Good Good Good  U878 3.4 71ffffe Good Good Good Good
  U872 2.6 75fffff 1    !    !    !     U870 2.4 75ffffe 1    !    1    !
  U864 1.6 79fffff !    !    !    !     U862 1.4 79ffffe 1    !    1    !
  U856 0.6 7dfffff Good Good Good Good  U854 0.4 7dffffe Good Good Good Good
  U877 3.3 71ffffd !    !    1    !     U875 3.1 71ffffc 1    1    1    1
  U869 2.3 75ffffd 1    1    1    1     U867 2.1 75ffffc 1    1    1    1
  U861 1.3 79ffffd 1    1    1    !     U859 1.1 79ffffc 1    1    1    1
  U853 0.3 7dffffd Good Good Good Good  U851 0.1 7dffffc Good Good Good Good
  U876 3.2 71ffffd 1    1    1    1     U874 3.0 71ffffc 1    1    1    1
  U868 2.2 75ffffd 1    1    1    1     U866 2.0 75ffffc Good Good Good Good
  U860 1.2 79ffffd !    Good !    Good  U858 1.0 79ffffc 1    1    1    1
  U852 0.2 7dffffd Good Good Good Good  U850 0.0 7dffffc Good Good Good Good

Comparing the above table against readily available Amiga 3000 ZIP memory
installation hints on the internet, one can quickly decide this configuration
is not going to work so well. Only "Bank 0" of ZIP memory might be usable
by the operating system. Note that for the example above, not all ZIP
ICs were installed. One ZIP is defective, and another has an address pin
purposely stuffed in the wrong pin socket (to inject a fault).

The Amiga memory addresses above may look unusual. These addresses are
chosen because there isn't a simple mapping between the Amiga memory address
and the Ramsey assertion of the RAS and CAS address lines. There is a table
in the comments of ziptest.c which documents the mapping that I observed
when probing with a logic analyzer against power-of-two addresses. Simply
stated, the address lines A1 through A9 are inverted from the CPU's physical
memory address. In 1Mx4 mode, the RAS lines A0 through A9 are also rolled
one to the right from the CPU's physical memory address.


=============================================================================

Address line test
-----------------
The address line test performs a modified walking zero's and walking one's
test on the address lines connected to the ZIP memory. The test is designed
to detect shorted or floating address lines by walking adjacent address lines,
checking if writes land at the wrong memory address. The test can't detect
all possible address line-to-line shorts, but does a good job in the general
case of a line shorted to an adjacent line or one which is not connected.

At each address line being tested, address lines immediately below and
above are also used to pattern and test. For example, if A4 is being
tested, then A5, A4, and A3 will be walked with 000, 001, 010, 011, 100,
101, 110, and 111 values, while the other address lines (A0-A2 and A6-A9)
will all be held either at 1 or at 0.

Note that due to the nature of the address test, some address lines may be
indicated as questionable around others which are bad. For example, view the
report for U880 below, which has its A8 pin stuffed in the same socket as A9:

1> ziptest addr quiet

Address line test
           A A A A A A A A A A           A A A A A A A A A A
  Socket   9 8 7 6 5 4 3 2 1 0  Socket   9 8 7 6 5 4 3 2 1 0
  -------- - - - - - - - - - -  -------- - - - - - - - - - -
  U881 3.7 G G G G G G G G G G  U879 3.5 ! ! ! ! ! ! ! ! ! !
  U873 2.7 ! ! ! ! ! ! ! ! ! !  U871 2.5 ! ! ! ! ! ! ! ! ! !
  U865 1.7 ! ! ! ! ! ! ! ! ! !  U863 1.5 ! ! ! ! ! ! ! ! ! !
  U857 0.7 G G G G G G G G G G  U855 0.5 G G G G G G G G G G
  U880 3.6 ? ! ? ? G G G G G ?  U878 3.4 G G G G G G G G G G
  U872 2.6 ! ! ! ! ! ! ! ! ! !  U870 2.4 ! ! ! ! ! ! ! ! ! !
  U864 1.6 ! ! ! ! ! ! ! ! ! !  U862 1.4 ! ! ! ! ! ! ! ! ! !
  U856 0.6 G G G G G G G G G G  U854 0.4 G G G G G G G G G G
  U877 3.3 ! ! ! ! ! ! ! ! ! !  U875 3.1 ! ! ! ! ! ! ! ! ! !
  U869 2.3 ! ! ! ! ! ! ! ! ! !  U867 2.1 ! ! ! ! ! ! ! ! ! !
  U861 1.3 ! ! ! ! ! ! ! ! ! !  U859 1.1 ! ! ! ! ! ! ! ! ! !
  U853 0.3 G G G G G G G G G G  U851 0.1 G G G G G G G G G G
  U876 3.2 ! ! ! ! ! ! ! ! ! !  U874 3.0 ! ! ! ! ! ! ! ! ! !
  U868 2.2 ! ! ! ! ! ! ! ! ! !  U866 2.0 G G G G G G G G G G
  U860 1.2 ! ! ! ! ! ! ! ! ! !  U858 1.0 ! ! ! ! ! ! ! ! ! !
  U852 0.2 G G G G G G G G G G  U850 0.0 G G G G G G G G G G

The question marks in the address test results above indicate that the address
line had a low number of errors, which could be caused by side-effects of
bad adjacent address lines. If you add the "debug" flag to the addr test,
it will give you an exact count of failures per address line (up to 99).
Here, we can see more clearly the low failure counts on all address lines
except A8 (on U880):

1> ziptest addr debug quiet

Address line test
  Socket   A9 A8 A7 A6 A5 A4 A3 A2 A1 A0  Socket   A9 A8 A7 A6 A5 A4 A3 A2 A1 A0
  -------- -- -- -- -- -- -- -- -- -- --  -------- -- -- -- -- -- -- -- -- -- --
  U881 3.7  0  0  0  0  0  0  0  0  0  0  U879 3.5 45 45 44 45 46 47 48 48 48 48
  U873 2.7 40 39 42 39 41 43 44 42 40 40  U871 2.5 35 41 39 38 39 36 38 39 32 36
  U865 1.7 44 43 42 43 46 45 46 43 42 44  U863 1.5 45 47 48 48 48 48 48 48 47 48
  U857 0.7  0  0  0  0  0  0  0  0  0  0  U855 0.5  0  0  0  0  0  0  0  0  0  0
  U880 3.6  6 33  4  4  2  0  0  0  0  2  U878 3.4  0  0  0  0  0  0  0  0  0  0
  U872 2.6 31 38 33 36 33 32 29 34 28 34  U870 2.4 41 43 47 44 41 44 45 42 44 43
  U864 1.6 30 32 31 31 32 33 31 31 29 33  U862 1.4 47 45 47 44 40 43 44 44 48 45
  U856 0.6  0  0  0  0  0  0  0  0  0  0  U854 0.4  0  0  0  0  0  0  0  0  0  0
  U877 3.3 45 45 43 42 44 44 44 41 42 44  U875 3.1 45 44 44 44 44 45 45 46 46 45
  U869 2.3 47 45 46 45 47 43 44 42 42 44  U867 2.1 48 48 48 48 48 48 48 48 48 48
  U861 1.3 44 45 46 48 48 48 46 43 44 45  U859 1.1 45 47 44 47 45 45 45 41 44 45
  U853 0.3  0  0  0  0  0  0  0  0  0  0  U851 0.1  0  0  0  0  0  0  0  0  0  0
  U876 3.2 47 47 43 43 44 44 48 45 43 44  U874 3.0 41 40 40 40 36 37 39 41 44 39
  U868 2.2 43 46 46 46 43 45 45 44 43 44  U866 2.0  0  0  0  0  0  0  0  0  0  0
  U860 1.2 32 34 32 33 30 27 28 33 30 30  U858 1.0 42 40 40 41 41 41 38 40 41 39
  U852 0.2  0  0  0  0  0  0  0  0  0  0  U850 0.0  0  0  0  0  0  0  0  0  0  0


You might wonder about U860. There is a failed ZIP which is installed in
this socket. A couple I/O lines (A2 and A4) still work, but this is not
sufficient to confirm any of the address lines are good. There is not a
ZIP IC installed in U864, U871, or U872, so it's odd why these also present
lower error counts (one must assume a bit of randomness since there is no
IC driving the data lines when addressing unpopulated sockets).

If the ZIP memory is configured in 256Kx4 mode (4MB max fast memory),
then the A9 address line test can not be tested. Output will reflect that
difference (notice '-' in that column):

Address line test
           A A A A A A A A A A           A A A A A A A A A A
  Socket   9 8 7 6 5 4 3 2 1 0  Socket   9 8 7 6 5 4 3 2 1 0
  -------- - - - - - - - - - -  -------- - - - - - - - - - -
  U881 3.7 - G G G G G G G G G  U879 3.5 - ! ! ! ! ! ! ! ! !
  U873 2.7 - ! ! ! ! ! ! ! ! !  U871 2.5 - ! ! ! ! ! ! ! ! !
  U865 1.7 - ! ! ! ! ! ! ! ! !  U863 1.5 - ! ! ! ! ! ! ! ! !
  U857 0.7 - G G G G G G G G G  U855 0.5 - G G G G G G G G G
  U880 3.6 - ! ? ? G G G G ? ?  U878 3.4 - G G G G G G G G G
  U872 2.6 - ! ! ! ! ! ! ! ! !  U870 2.4 - ! ! ! ! ! ! ! ! !
  U864 1.6 - ! ! ! ! ! ! ! ! !  U862 1.4 - ! ! ! ! ! ! ! ! !
  U856 0.6 - G G G G G G G G G  U854 0.4 - G G G G G G G G G
  U877 3.3 - ! ! ! ! ! ! ! ! !  U875 3.1 - ! ! ! ! ! ! ! ! !
  U869 2.3 - ! ! ! ! ! ! ! ! !  U867 2.1 - ! ! ! ! ! ! ! ! !
  U861 1.3 - ! ! ! ! ! ! ! ! !  U859 1.1 - ! ! ! ! ! ! ! ! !
  U853 0.3 - G G G G G G G G G  U851 0.1 - G G G G G G G G G
  U876 3.2 - ! ! ! ! ! ! ! ! !  U874 3.0 - ! ! ! ! ! ! ! ! !
  U868 2.2 - ! ! ! ! ! ! ! ! !  U866 2.0 - G G G G G G G G G
  U860 1.2 - ! ! ! ! ! ! ! ! !  U858 1.0 - ! ! ! ! ! ! ! ! !
  U852 0.2 - G G G G G G G G G  U850 0.0 - G G G G G G G G G


=============================================================================

Memory cell test
----------------
The last test is a comprehensive data pattern test which walks each memory
location in each ZIP package. This test takes considerably longer to run
than the data line test or the address line test.

In default mode, this test will only use alternating 0x5 and 0xa values at
the four bits of each ZIP IC to verify all memory cells. In this mode,
each cell gets tested for retention of both a 0 value and a 1 value.
When the "LONG" parameter is specified, then each ZIP IC is tested with the
following additional patterns: 0xc, 0x3, 0x1, 0x2, 0x4, 0x8, 0x7, 0xe, 0xd,
0xb, and 0x0.

At the end of the test, a summary will be displayed.  "Good" means that
all cells of the specific chip passed the test. An "!" means failures
were detected.

Example output:

1> ziptest cell quiet

Memory cell test
  Bank 0 [................................]
  Bank 1 [X                               ]
  Bank 2 [XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX]
  Bank 3 [XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX]

  Socket   Result   Socket   Result
  -------- ------   -------- ------
  U881 3.7 Good     U879 3.5 !   
  U873 2.7 !        U871 2.5 !   
  U865 1.7 !        U863 1.5 !   
  U857 0.7 Good     U855 0.5 Good
  U880 3.6 !        U878 3.4 Good
  U872 2.6 !        U870 2.4 !   
  U864 1.6 !        U862 1.4 !   
  U856 0.6 Good     U854 0.4 Good
  U877 3.3 !        U875 3.1 !   
  U869 2.3 !        U867 2.1 !   
  U861 1.3 !        U859 1.1 !   
  U853 0.3 Good     U851 0.1 Good
  U876 3.2 !        U874 3.0 !   
  U868 2.2 !        U866 2.0 Good
  U860 1.2 !        U858 1.0 !   
  U852 0.2 Good     U850 0.0 Good

=============================================================================
Example ziptest output where a good bank 0 of 256Kx4 ZIP ICs are installed
in an Amiga while Ramsey is configured for 1Mx4 mode.  As you can tell from
the patterns, the ICs are installed in U850, U851, U852, U853, U854, U855,
U856, and U857.

1> ziptest
ZIPTest 1.1 (2024-01-17) by Chris Hooper
CPU: 68040 without Burst, MMU Active
Memory controller: Ramsey-07 $f $28 (25.08 MHz)
Memory config: 1Mx4 (4MB per bank)
Memory refresh: 240 clocks (9.60 usec)

Data line test
  Socket   IO1  IO2  IO3  IO4   Socket   IO1  IO2  IO3  IO4
  -------- ---- ---- ---- ----  -------- ---- ---- ---- ----
  U881 3.7 1    1    1    1     U879 3.5 1    1    1    1
  U873 2.7 1    1    1    1     U871 2.5 1    1    1    1
  U865 1.7 1    1    ?    ?     U863 1.5 ?    1    1    1
  U857 0.7 Good Good Good Good  U855 0.5 Good Good Good Good
  U880 3.6 1    1    1    1     U878 3.4 1    1    1    1
  U872 2.6 1    ?    1    1     U870 2.4 1    1    1    1
  U864 1.6 1    1    1    1     U862 1.4 ?    ?    1    ?
  U856 0.6 Good Good Good Good  U854 0.4 Good Good Good Good
  U877 3.3 1    1    1    1     U875 3.1 1    1    1    1
  U869 2.3 1    1    1    1     U867 2.1 1    1    1    1
  U861 1.3 1    1    1    1     U859 1.1 1    1    1    ?
  U853 0.3 Good Good Good Good  U851 0.1 Good Good Good Good
  U876 3.2 1    1    1    1     U874 3.0 1    1    1    1
  U868 2.2 1    1    ?    1     U866 2.0 1    1    1    1
  U860 1.2 1    1    1    1     U858 1.0 1    1    ?    1
  U852 0.2 Good Good Good Good  U850 0.0 Good Good Good Good

Address line test
           A A A A A A A A A A Mem            A A A A A A A A A A Mem
  Socket   9 8 7 6 5 4 3 2 1 0 Type  Socket   9 8 7 6 5 4 3 2 1 0 Type
  -------- - - - - - - - - - - ----  -------- - - - - - - - - - - ----
  U881 3.7 ! ! ! ! ! ! ! ! ! ! ?     U879 2.5 ! ! ! ! ! ! ! ! ! ! ?
  U873 2.7 ! ! ! ! ! ! ! ! ! ! ?     U871 3.5 ! ! ! ! ! ! ! ! ! ! ?
  U865 1.7 ! ! ! ! ! ! ! ! ! ! ?     U863 1.5 ! ! ! ! ! ! ! ! ! ! ?
  U857 0.7 ! ? ? G G G G G ? ? ?     U855 0.5 ! ? ? G G G G G ? ? ?
  U880 3.6 ! ! ! ! ! ! ! ! ! ! ?     U878 3.4 ! ! ! ! ! ! ! ! ! ! ?
  U872 2.6 ! ! ! ! ! ! ! ! ! ! ?     U870 2.4 ! ! ! ! ! ! ! ! ! ! ?
  U864 1.6 ! ! ! ! ! ! ! ! ! ! ?     U862 1.4 ! ! ! ! ! ! ! ! ! ! ?
  U856 0.6 ! ? ? G G G G G ? ? ?     U854 0.4 ! ? ? G G G G G ? ? ?
  U877 3.3 ! ! ! ! ! ! ! ! ! ! ?     U875 3.1 ! ! ! ! ! ! ! ! ! ! ?
  U869 2.3 ! ! ! ! ! ! ! ! ! ! ?     U867 2.1 ! ! ! ! ! ! ! ! ! ! ?
  U861 1.3 ! ! ! ! ! ! ! ! ! ! ?     U859 1.1 ! ! ! ! ! ! ! ! ! ! ?
  U853 0.3 ! ? ? G G G G G ? ? ?     U851 0.1 ! ? ? G G G G G ? ? ?
  U876 3.2 ! ! ! ! ! ! ! ! ! ! ?     U874 3.0 ! ! ! ! ! ! ! ! ! ! ?
  U868 2.2 ! ! ! ! ! ! ! ! ! ! ?     U866 2.0 ! ! ! ! ! ! ! ! ! ! ?
  U860 1.2 ! ! ! ! ! ! ! ! ! ! ?     U858 1.0 ! ! ! ! ! ! ! ! ! ! ?
  U852 0.2 ! ? ? G G G G G ? ? ?     U850 0.0 ! ? ? G G G G G ? ? ?

Memory cell test
  Bank 0 [XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX]
  Bank 1 [XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX]
  Bank 2 [XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX]
  Bank 3 [XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX]

  Socket   Result   Socket   Result
  -------- ------   -------- ------
  U881 3.7 !        U879 3.5 !
  U873 2.7 !        U871 2.5 !
  U865 1.7 !        U863 1.5 !
  U857 0.7 !        U855 0.5 !
  U880 3.6 !        U878 3.4 !
  U872 2.6 !        U870 2.4 !
  U864 1.6 !        U862 1.4 !
  U856 0.6 !        U854 0.4 !
  U877 3.3 !        U875 3.1 !
  U869 2.3 !        U867 2.1 !
  U861 1.3 !        U859 1.1 !
  U853 0.3 !        U851 0.1 !
  U876 3.2 !        U874 3.0 !
  U868 2.2 !        U866 2.0 !
  U860 1.2 !        U858 1.0 !
  U852 0.2 !        U850 0.0 !

Although this memory (or a portion of it) might pass testing by Amiga OS,
the ziptest utility will still detect a problem.  In this particular case,
adjusting J852 is all that is necessary for the installed memory to be
reported as Good.

=============================================================================

Other options
-------------
The ziptest utility has a few additional command line options which may
be listed by supplying a "?" argument.  All supported arguments:

    ADDR   - perform address line test
    ASCII  - show ASCII ART of chip positions and pins
    CELL   - perform memory cell test (verify every bit)
    DATA   - perform data line test
    DIP    - show DIP RAM positions
    DEBUG  - enable debug output
    INFO   - only show system information
    FORCE  - ignore fact enforcer is present
    LONG   - perform more thorough (slower) line test
    MAP    - just show map of corresponding bits (no test)
    QUIET  - do not display banner
    SPROBE - probe for static-column memory (68030 only)
    STROBE - generate power-of-two address strobes for a probe

ADDR
----
Perform only the address line test.  No other tests will be executed unless
they are also specified.  The LONG option may be specified to run more
iterations of the address line test, which slightly increases the chances
of finding a floating line.

ASCII
-----
Display ASCII art showing the placement and pinout of the ZIP and DIP ICs
on the board.

CELL
----
Perform only the memory cell test.  No other tests will be executed unless
they are also specified.  The set of patterns for each ZIP IC will be 0x5
and 0xa, which tests each cell set to 0 value and set to 1 value.  The LONG
option may be specified to run each ZIP through more patterns.  The complete
set of patterns for each ZIP IC in this mode will be: 0x5, 0xa, 0xc, 0x3,
0x1, 0x2, 0x4, 0x8, 0x7, 0xe, 0xd, 0xb, and 0x0.

DATA
----
Perform only the data line test.  No other tests will be executed unless
they are also specified.  The LONG option may be specified to run a more
comprehensive test by checking data lines to each ZIP IC separately from
each other, rather than in parallel with data lines of the other ZIP ICs.

DIP
---
Also display fast memory DIP chip positions (these mirror the first bank
of ZIP memory, so the information is likely redundant).

DEBUG
-----
Display test debug output.  This is most useful for the Address line test
as it changes the good/bad result to be a count of failures.  Adding the
flag a second time will generate more output for the address line test.

FORCE
-----
Ignore the fact that Enforcer or MuForce is running.  This will likely
lead to a hang when the test runs as the address exceptions are handled by
software.

INFO
----
Just display Amiga system information, including CPU and Ramsey memory
controller configuration.

LONG
----
Run a more comprehensive version of the data, address, or cell tests.
See the individual test descriptions above for more details.

MAP
---
Display a map of how Amiga CPU physical memory addresses and bit positions
correspond to data bits and address bits of the ZIP ICs.  Note that the
maps displayed will be different depending on whether the memory is jumpered
for 1Mx4 or 256Kx4 mode.

QUIET
-----
Do not display the banner showing the name, version, and date of ziptest.
The CPU and Ramsey configuration are also not displayed unless the INFO
command is also specified.

SPROBE
------
Not only will this code probe for static column memory. It will also do
some benchmarking of read.l with cache and burst on and off. Note that
the speed reported will not be as high as bustest because the code executes
from chip memory and is also not using movem.

STROBE
------
This option is for board level debug.  It will cause read strobes on the
address lines in order to assist with debug using an oscilloscope.  You can
use any of the parallel port data lines as a trigger for the oscilloscope
or logic analyzer to capture the address lines during the RAS and CAS cycles.

=============================================================================

Source code notes
-----------------
Source may be compiled with either DICE 3.15 (use dmake) or VBCC 0.906
13.08.2017 (use compile_vbcc script).  I use DICE for the distribution build
as it compiles faster and the final executable is a bit smaller.  I didn't
notice a performance difference between the two for this particular
application.  Although gcc can also compile the utility, I wasn't able to
figure out how to set the hunk attributes to MEMF_CHIP, which is required.

Since the code must run from chip memory, there is a small startup function
in stack.asm which relocates the program's stack to chip memory.  It then
calls c_main() in ziptest.c.

There are a few defines at the top of ziptest.c -- you shouldn't need to
adjust any of the settings.


Contents of util/misc/ZIPTest-1.1.lha
PERMISSION  UID  GID    PACKED    SIZE  RATIO METHOD CRC     STAMP     NAME
---------- ----------- ------- ------- ------ ---------- ------------ ----------
[unknown]                  163     226  72.1% -lh5- e372 Nov 28  2017 ZIPTest/compile_vbcc
[unknown]                  782    1717  45.5% -lh5- dd10 Jan 14 20:11 ZIPTest/DMakefile
[unknown]                  769    1893  40.6% -lh5- 0727 Nov 27  2017 ZIPTest/stack.asm
[unknown]                 1502    6313  23.8% -lh5- 1ffa Jan 17 01:52 ZIPTest/util.asm
[unknown]                  273     852  32.0% -lh5- 351f Nov 27  2017 ZIPTest/vbcc_config
[unknown]                14248   24572  58.0% -lh5- 77f0 Jan 17 01:56 ZIPTest/ziptest
[unknown]                23146   84865  27.3% -lh5- 147b Jan 17 01:55 ZIPTest/ziptest.c
[unknown]                 9264   29173  31.8% -lh5- 995c Jan 17 01:53 ZIPTest/ziptest.readme
---------- ----------- ------- ------- ------ ---------- ------------ ----------
 Total         8 files   50147  149611  33.5%            Jan 18 04:05

Aminet © 1992-2024 Urban Müller and the Aminet team. Aminet contact address: <aminetaminet net>