home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
DP Tool Club 25
/
CD_ASCQ_25_1095.iso
/
dos
/
prg
/
4p_v311
/
1st_read.me
next >
Wrap
Text File
|
1995-08-27
|
13KB
|
241 lines
Dear user! v3.1.1 from 01.09.1995
==========
This file contains some information about the new PROGRAMMER's PROCESSOR POWER
PACKAGE called 4P_Vxxx.ZIP from Christian Ludloff, Germany. This software pack
includes a text file with knowledge about the 80x86 processors - even with the
undocumented and secret details added in - and some very useful utilities. The
included INFO program can be used to show almost all details about the proces-
sor and with STAT you can do the performance monitoring on iPentium PCs.
Please, read this long text until the last line before you try to use the pack
on your PC, because there are some details, you should know, and because there
are some risks too (i.e. crashing the actual DOS session)!
1. this is shareware!
======================
Before describing the details I wrote about above - don't worry! - I will tell
you some words about the shareware concept I used for this software package. I
designed all the software as shareware, which means, that you may and you also
should test the programs during a 14 day trial period.
If you hate the programs after this time, then you may erase them from your HD
or give them to your friends. But if you think, that the software is good and
you plan to use it in the future (more than once a year or so...), then you're
fair when sending me the amount of $10 or 15,-DM as a registration fee. I will
send you the newest version without the shareware message screen and with your
name added in - that's the so called 'full-version'.
The details for the money transfer methods, my address and the possible eMail-
addresses to contact me you may read at the nice shareware message screen when
running the INFO or the STAT program. This screen stays stable for 60 seconds
or as long, as you paused the program with the [PAUSE] key. Please, press any
key, if you want to continue with the programs.
The registered versions of my programs will show you the registered user, when
you run them with 'REG' as the first command line parameter.
2. hard- and software
======================
To run the programs included in the 4P software package you need an IBM compa-
tible PC with at least an i8088 processor. To use the STAT program you must be
able to run it on an iPentium based PC - on older processors this software can
not be used due to the lack of some iPentium processor hardware details.
I checked, that 512KB of RAM will be enough to run all the programs I included
into the 4P package. Of course you can run the programs from a floppy disk, if
there's no HD available. But using the HD is much faster...
If you run the STAT program in TSR mode, then STAT needs about 85KB to stay in
your memory. Do not cry about so much RAM for such a small TSR - it needs only
16KB for saving the actual screen during the pop-ups, about 8KB for the stack,
then about 45KB for program code (not written with an Assembler) and some more
KBs for other parts.
The programs will run on all graphic boards from Hercules up to the newest PCI
accelerator cards with many MBs of VRAM. Nevertheless only the text modes will
be used. Someone asked me for a WINDOWS version? Just run a DOS box - if there
will be PCs without any DOS emulation in the future (next century?), then I do
begin thinking about a non-DOS version of my software.
The onliest software which is required by the programs from the 4P package is
the good old DOS. You will need a version 3.3 or newer. I checked the programs
on MS-DOS versions 5.x/6.x and on some DR-/Novell-DOS systems. If you get into
trouble with any system software, then contact me, please!
I do not recommend running my software in a WINDOWS or OS/2 DOS box! I tested
these two system and found them to be incompatible to my software. These newer
systems have their own memory managers included in their kernels, not allowing
some special processor instructions due to the fact, that these systems mostly
run in the virtual mode of an 80x86 processor.
If you nevertheless try to run INFO or STAT in a DOS box, then you will get an
error message and the DOS box will be closed (without any other problems). You
can try this without getting into trouble or loosing data.
3. the INFO program
====================
There are two different versions of the INFO program: one before march, 20th,
1995 and another one after this date. The older versions have revision numbers
like 2.x.x and the newer versions have 3.x.x or later.
Nevertheless all new versions will include the latest INFO v2.x.x program, due
to the fact, that it is useful even today. The older versions will only have a
summary of single blue screens, showing the results of the tests. The newer v3
INFO has a nice Turbo-Vision desktop and an integrated on-line help.
Both versions include a built-in self-test which is started after you runned a
INFO program. This self-test protects you against getting a virus-infected co-
py of my software and it protects my software against debugging too. Due to an
Intel-NDA (non-disclosure agreement) for some details supported by their newer
processors I used in my software I have to protect my software.
The older INFO program versions support some command line parameters which are
shown after the program start, but before doing any tests. You may read them,
stop INFO by pressing the [ESC] key an run it again with parameters. It is not
useful to describe the parameters here, because INFO tells you, what they will
do (or not) - don't worry - it's very easy to use them!
If you're running one of the following systems, then you should use the listed
parameters to protect the actual DOS session against crashing, if you test one
of the older INFO versions (means: execute INFO v2.x.x from INFO v3.x.x):
DOS without any EMM - INFO2.EXE [/R] [/C] - this is true real mode
DOS with its EMM386 - INFO2.EXE /P - use this on iPentium
DOS with QEMM v7.04 - INFO2.EXE - supports the iPentium
WINDOWS 3.1 DOS box - INFO2.EXE /P /M - no iPentium support
OS/2 DOS box - INFO2.EXE /D /T /P /M - not a good DOS box...
The triple fault reset tests and the cache switch will be performed only, if a
parameter /R and/or /C is specified, due to the fact, that they are not compa-
tible with all mainboard or BIOS versions and with all processors. The CPU-re-
set will sometimes enable the internal caches - I do not correct this! Another
little problem is getting the EFLAGS register value after the processor reset,
because the first instruction must disable all interrupts. So the 'IF' bit may
bet set to zero or one - it seems to be impossible to get the correct value.
If the RDTSC instruction for reading the value of the Time Stamp Counter on an
iPentium processor is disabled, then INFO v2.x.x trys to read the Model Speci-
fic Register #10h instead, which contains the TSC on all iPentiums.
4. the STAT program
====================
The STAT program did not changed its character since the first version. It's a
very hard task to make this program better. If you run STAT and you use an EMM
memory manager, which makes your 80x86 processor switches to the virtual mode,
then you must use a newer EMM, supporting the iPentium instruction use even on
the CPL=3 - otherwise the DOS session will be crashed when running STAT. Only,
if the VCPI API is supported and the processor is running in standard (but not
enhanced!) virtual mode, then the STAT program is able to by-pass the standard
virtual mode. (The CPL=0 is entered via the VCPI functions and there all these
new iPentium instructions can be executed without crashing the machine.)
After starting the two internal counters by pressing the shown keys you leave
the STAT program with the [ESC] key - but it stays resident in your memory. To
pop it up, just press the [F12] key. If you want to unload STAT from your me-
mory, just run it again - if it's possible, then STAT will remove the resident
copy from RAM. This will not work if you started other TSRs after STAT. If so,
then remove them earlier or STAT will stay resident, but inactivated. To acti-
vate it again, just run it again. It's very easy to handle!
The pop-up will work at all text modes, except 132x44, 80x60 or so, except all
self-made graphic modes like 320x480x256 or so and except all High- and True-
Color modes. For all these modes no simple standards exists, so that I can not
save the screen data correctly on all graphic boards - sorry!
If you're using a mouse, then remember, that the mouse pointer state will not
be saved or restored, if you interrupt a mode, where the pointer is visible. I
did not included a routine for saving the mouse pointer data, because the dif-
ferent size of the data block (there's no heap inside the STAT program).
The pop-up logic is very intelligent: The program will look for HD accesses or
activities on the serial and parallel ports, before interrupting the system on
a pressed [F12] key. I think, that loosing data should be impossible.
You can also use the STAT program for the non-resident performance monitoring,
if you use the possible command line parameters. If you run STAT, then a list
of these parameters will be shown.
If you want to use STAT on a iPentium-based PC with an iPentium processor with
the FPU-bug - just run INFO to test for this bug! - then you should use PSTAT.
This little patch will make, that STAT doesn't uses the FPU for its mathemati-
cal tasks. Nevertheless the FPU will not be disabled for any other software!
5. any questions?
==================
If you have any questions, then you may contact me in several different ways:
Christian Ludloff TEL +49-371-242091 eMail cl@box.in-chemnitz.de
Ludwig-Kühn-Straße 15 FAX +49-371-242091 eMail cl@vgasoft.com
D-09123 CHEMNITZ, Germany Fido 2:2426/2240.14
Please, write or talk English or German and (at phone) ask for Christian 1st!
6. known bugs
==============
The INFO program v3.0.1 uses the interrupt vector for the INT B0h to check, if
it is still resident in memory. This will avoid starting a 2nd copy from a DOS
shell - just type 'EXIT' and press [RETURN] to go back to the INFO program. If
you have problems with the used INT B0h vector, then let me know, please!
The INFO program v2.x.x and also the STAT program check the clock speed of the
iPentium processor, using a short timer loop. If you press the [PAUSE] key and
stop the speed test for a while, then you will get incorrect 'MHz-results' for
the clock speed. This will cause incorrect results for STAT too.
The clock speed test will work correctly inside an IBM OS/2 DOS box only, when
you set the HW_TIMER option to ON, because the routine uses the timer chip.
Sometimes the INFO program 2.x.x will detect an 486SX with an 487SX, even when
there really is an 486 processor. This is a timing problem, which will be sol-
ved, when I created a new (non-timing) test routine.
The triple fault reset test from the INFO program v2.x.x will not disable the
internal L1-cache of the processor during the reset and it will also not cor-
rect the cache state after the reset. Please, use the /C parameter, to correct
the cache state, when using the /R parameter for the reset test.
The FPU location test from the INFO program v3.x.x will not be able to show a
correct result, if MS-WINDOWS is running. Rely on the results from the CPUID
instruction tests only! This test crashs the IBM OS/2 DOS box too.
The 'A20 gate' switch crashs the DOS box of WINDOWS and therefore it will be
disabled, if WINDOWS is running. The 'A20 gate' switching also crashs the ses-
sion, if you use the QEMM memory manager - so it will be disabled too. Also an
IBM OS/2 DOS box would be crashed - so disabled here too.
The Borland TD386 debugger crashs, when you try to debug some of the programs.
Also the programs crash, when you set DR7.GD=1, while the processor is in vir-
tual mode, because this bit (which disables any access to the Debug Registers)
can only be cleared in real mode. So you have to use the /D parameter for the
INFO2 program, which disables the Debug Register access, when the processor is
in virtual mode and when DR7.GD is set to 1 - otherwise the session crashs.
The test for an IBM 386/486SLC or 486BLC processor needs to check for possible
access to some Model Specific Register. Therefore it is only performed, if the
processor is in real mode (no EMM drivers loaded and/or Windows not running).
The Cyrix M5, M6 and M7 processors can only be detected via their triple fault
reset value in real mode (no EMM drivers loaded and/or Windows not running).
The screen grab feature will crash the programs, when you try to grab to write
protected disks. Setting the SCR_SHOT.TXT file to read only, system and hidden
is no problem, because these attributes are removed first. If the file was de-
leted, then a new one will be created. Otherwise the data will be appended.
*** END OF FILE ***