Numerological
Unal Z
and remove them), EZoto's inquiry made me look closer at numbers and
letters. I found the enumeration pattern and mention here only the three
character submodel code, e.g. 9577-XYZ. Very simple: X is release code, Y
processor code and Z hard disk code. The coding holds for the second MCA
wave, 95xx, the first wave 85xx (mod. 60, 70, 80) indicated the HD capacity,
like 121 for 120MB, 071 for 70MB etc.
95xx-XYZ ::
X = Release number, 0 (zero) for first releases, if letter then special (?)
model or a variant.
Y = Processor type:
------------------
5 = 386SLC20
B = 486SLC2-50
G = 486SX-20
H = 486SX-25
U = 486SX-33
X = 486SX-33 (X is weird)
J = 486DX-25
K = 486DX-33
L = 486DX-50
M = 486DX-50 w. 256kb L2
N = 486DX2-66
P = Pentium 60
Q = Pentium 66
T = ?
Z = Hard disk capacity in MB
-------------------------
5 = 80
6 = 104
9 = 160
A = 212
B = ?
C = ?
D = 320
E = ?
F = 400
G = 540
T = 1024 (1 GB)
V = 2048 (2 GB)
The list may not be complete. If all that has been already posted, sorry for
the tautology.
--
UZ
Tim Clarke
>
> X = Release number, 0 (zero) for first releases, if letter then special
(?)
> model or a variant.
>
> Y = Processor type:
> ------------------
> 5 = 386SLC20
> B = 486SLC2-50
>
> G = 486SX-20
> H = 486SX-25
> U = 486SX-33
> X = 486SX-33 (X is weird)
>
> J = 486DX-25
> K = 486DX-33
> M = 486DX-50 w. 256kb L2
> N = 486DX2-66
>
> P = Pentium 60
> Q = Pentium 66
Carroll Bloyd
news:3db5e867$0$20510$91ce...@newsreader01.highway.telekom.at...
> 95xx-XYZ ::
>
> X = Release number, 0 (zero) for first releases, if letter then special
(?)
> model or a variant.
>
> Y = Processor type:
> ------------------
> 5 = 386SLC20
> B = 486SLC2-50
>
> G = 486SX-20
> H = 486SX-25
> U = 486SX-33
> X = 486SX-33 (X is weird)
>
> J = 486DX-25
> K = 486DX-33
> L = 486DX-50
> M = 486DX-50 w. 256kb L2
> N = 486DX2-66
>
> P = Pentium 60
> Q = Pentium 66
> T = ?
T = 486DX4-100
Excerpted from the IBM Sales Manual 9577 model matrix:
Speed Memory/70ns Memory
Model Processor (Mhz) Min-Max Type Cache Hard Disk
ATB IntelDX4 100 8MB-64MB Pty 256Kb L2 270MB IDE
ATG IntelDX4 100 8MB-64MB Pty 256Kb L2 527MB IDE
BTG IntelDX4 100 8MB-64MB Pty 256Kb L2 540MB SCSI-2
VTG IntelDX4 100 16MB-64MB ECC 256Kb L2 540MB SCSI-2
6TG IntelDX4 100 8MB-64MB Pty 256Kb L2 540MB SCSI-2
7TG IntelDX4 100 8MB-64MB Pty 256Kb L2 540MB SCSI-2
BTB IntelDX4 100 8MB-64MB Pty 256Kb L2 270MB SCSI-2
My 9577-VTG came with a 486DX4-100. (Now replaced with a POD83.)
Carroll
rd...@spamcop.net
Unal Z
Thanks, Carroll, your list contains B and G entries for the hard disks as
well.
Below is the updated list, 4 and E for processor type added:
95xx-XYZ ::
X = Release number, 0 (zero) for first releases, if letter then special (?)
model or variant.
Mod. 9577:
----------
A = IDE hd
0, 1 = SCSI hd (as in the announcement)
6, 7, B, V = SCSI-2 hd
Mod. 9595:
----------
3 = (3 disks) Array models (95A), SCSI-2 F/W RAID
P = MGA Ultima, CD-ROM II (Server 95 only)
Y = Processor type:
------------------
4 = 80386SX-20 (mod. 8557, new addition)
5 = 386SLC-20
A = ?
B = 486SLC2-50
C = ?
D = ?
E = 486SLC3-75 (mod. 9557, new addition)
F = ?
G = 486SX-20
H = 486SX-25
I = ?
U = 486SX-33
X = 486SX-33 (X is weird)
J = 486DX-25
K = 486DX-33
L = 486DX-50
M = 486DX-50 w. 256kb L2
N = 486DX2-66
T = 486DX4-100 w. 256Kb L2 (new addition, courtesy of Carroll Bloyd)
P = Pentium 60
Q = Pentium 66
Z = Hard disk capacity in MB
-------------------------
5 = 80
6 = 104
9 = 160
A = 212
B = 270 (new addition, courtesy of Carroll Bloyd)
C = ?
D = 320
E = ?
F = 400
G = 540, 527 (527 new addition, courtesy of Carroll Bloyd)
T = 1024 (1 GB)
V = 2048 (2 GB)
Ed: It seems as if all HDs <= 540MB are SCSI and not SCSI-2, if the
descriptions in the announcements and the options lists exactly
differentiate between SCSI and SCSI-2 (there is an Options 540MB hd noted as
SCSI-2).
--
UZ
Unal Z
I would have expected it to be in the PS/2 files.
Everything is of course somewhere documented.
--
UZ
Helmut P. Einfalt
>Ed: It seems as if all HDs <= 540MB are SCSI and not SCSI-2, if the
>descriptions in the announcements and the options lists exactly
>differentiate between SCSI and SCSI-2 (there is an Options 540MB hd
noted as
>SCSI-2).
I wouldn't trust the announcement letters too much in that respect --
not even IBM is faultless...
Did you get the compilation I mailed you?
Helmut
rd...@spamcop.net
Unal Z
> not even IBM is faultless...
Well, they are also very precise in describing components (they have very
good legal advisors). Spock is denoted always as SCSI and not SCSI-2, it is
indeed SCSI which understands SCSI-2 but works SCSI. The disks in the
Options lists are then accordingly denoted as SCSI. Later they differentiate
between SCSI and SCSI-2 disks. The difference is by now only historical, and
all that is just for the records.
> Did you get the compilation I mailed you?
Big thanks, Helmut. Your magnificient list confirmed the "coding algorithm",
brought new processor codes and showed some deviations:
New:
0 = 386DX-20 (mod. 8590-402, Helmut Einfalt)
S = 486SX2-50 (mod. 76i, Helmut Einfalt)
Deviations:
B = 486SLC2-50 (486SX-33, 486SX2-66 in 9577-QB6/QBA Helmut Einfalt)
D = 486DX2-66, (486DX4-100 in 9576i, Helmut Einfalt)
H = 486SX-25 (code 4 in 9576-I49, Helmut Einfalt)
U = 486SX-33 (code B in 9577-QB6/QBA Helmut Einfalt)
Now it should be possible to write a very small program which reads the
submodel code and tells you what CPU and HD this model has. This can be done
pretty quickly in Javascript.
Herzlichen Dank,
--
UZ
David L. Beem
> ...I found the enumeration pattern and mention here only the three
> character submodel code, e.g. 9577-XYZ. Very simple: X is release
> code, Y processor code and Z hard disk code. The coding holds for
> the second MCA wave, 95xx, the first wave 85xx (mod. 60, 70, 80)
> indicated the HD capacity, like 121 for 120MB, 071 for 70MB etc.
have already been contributed to your post, but I will list it in full for
completeness). This excludes what you define as "first wave": Models 25, 30,
50, 60, 70, & 80. It does seem to cover the EduQuest models in my
experience, but ValuePoint & later have their own charts.
First character (your "X"):
0 = Regular Systems
1 = Special Systems or Ethernet Systems
2 = Multimedia or Token Ring Systems
3 = Application oriented
4 = Special Bid model
6 = MM w/ DOS, Win
7 = MM w/ OS/2 2.1 Installed
D = Orderable w/ choice of operating system
K = OS/2 2.1 installed
Q = DOS/Windows 3.1 installed
Second character (your "Y"):
1 = 386SX-16
2 = 386SLC-16
3 = 386SX-16
4 = 386SX-20
5 = 386SLC-20
6 = 486SLC2-40
8 = 386DX-20 (Tim also says this is "0" as well)
B = 486SLC2-50
C = 386DX-25
D = 386DX-33
E = 486SLC3-75
G = 486SX-20
H = 486SX-25
J = 486DX-25
K = 486DX-33
L = 486DX2-50
M = 486DX-50
N = 486DX2-66
P = Pentium 60
Q = Pentium 66
T = 486DX4-100
U = 486SX-33
X = 486SX-33
Third character (your "Z"):
0 = No hard drive
3 = 40Mb
4 = 60Mb
5 = 80Mb
6 = 100Mb/104Mb
7 = 120Mb
9 = 160Mb/170Mb
A = 200Mb/212Mb
B = 245Mb
C = 170Mb (mobile)
D = 340Mb/400Mb
E = 340Mb (mobile)
F = 400Mb
G = 540Mb
T = 1Gb
David
Da...@gilanet.com
rd...@spamcop.net
Unal Z
Thanks indeed for the list. Having compared quite a many numbers from the
vintage site and Helmut Einfalt's vast compilation, here are the updates:
> First character (your "X"):
Have no definite opinion about that. There are quite a many new codes in
this category, e.g. 8, A, B, I, P, V, W, Z. Most of them are country codes,
e.g. P = Portugal. I was more interested in the last two character codes as
a handy quick reference for opportunities. Some Canadian French have a
four-char code, the 4th char is "F".
> Second character (your "Y"):
> B = 486SLC2-50 :: also 486SX-33, 486SX2-66, mod. 9577-QB6/QBA (Helmut
Einfalt)
> D = 386DX-33 (typo ?) :: also D = 486DX2-66, also 486DX4-100 mod.
9576i (Helmut Einfalt)
> Q = Pentium 66 :: new S = 486SX2-50 (mod. 76i, Helmut Einfalt)
> Third character (your "Z"):
> B = 245Mb :: also 270MB
> D = 340Mb/400Mb :: Also 320MB
> G = 540Mb :: also 527MB
> T = 1Gb :: new V= 2GB
We have a good list by now. There are few deviations from the rule but not
that serious.
--
UZ
David L. Beem
> Thanks indeed for the list. Having compared quite a many numbers from
> the vintage site and Helmut Einfalt's vast compilation, here are the
updates:
> > B = 486SLC2-50
> :: also 486SX-33, 486SX2-66, mod. 9577-QB6/QBA (Helmut Einfalt)
waiting to hear about the 486SX2-66, which is a fairly rare chip. Got a
CPUID on it?
> > D = 386DX-33 (typo ?)
> :: also D = 486DX2-66, also 486DX4-100 mod. 9576i (Helmut Einfalt)
For the 76i/77i it is a Lacuna planar. A true factory upgrade these
should have been marked correctly. I know that there doesn't seem to be any
PS/2 model that had a 386DX-33, but I have copied it exactly from the
manual.
> > Q = Pentium 66
> :: new S = 486SX2-50 (mod. 76i, Helmut Einfalt)
Again a Lacuna & I don't dispute the new "S" information. However on the
others I wonder where the CPU upgrades were performed. IBM is known to
heavily stick with convention: The true 486SLC2/486SLC3 daughtercard upgrade
kits for the Model 56/57 had submodel stickers for the customer to put in
place on the case.
The first example almost seems like a Bermuda planar in a 9557
(486SLC2-50) case. In particular, the "A" 212Mb drive is an IDE model that
would not be on a planar SCSI system. Also a 100Mb drive ("6") seems a bit
small for a 486 of that timeframe (but was common on the 56/57 systems).
> > Third character (your "Z"):
> > B = 245Mb :: also 270MB
> > D = 340Mb/400Mb :: Also 320MB
> > G = 540Mb :: also 527MB
> > T = 1Gb
> :: new V= 2GB
> We have a good list by now. There are few deviations
> from the rule but not that serious.
IBM is known to "fudge" the hard drive numbers a little to cover the
variety of sizes out there that they sourced. In little else though; It
would surprise me that they weren't consistant with the CPU notations. I
depend on IBM to be better than other manufacturers. Part of that is
consistency in product labeling.
David
Da...@gilanet.com
Helmut P. Einfalt
Disclaimer:
The model data I sent to Unal were compiled over the years from
various IBM sources, both online and off-line, and I won't exclude the
odd typo on my part...
>> > Second character (your "Y"):
>> > B = 486SLC2-50
>> :: also 486SX-33, 486SX2-66, mod. 9577-QB6/QBA (Helmut Einfalt)
> This is a Bermuda planar? Is the 486SX-33 a PLCC mounting? I am
also
>waiting to hear about the 486SX2-66, which is a fairly rare chip. Got
a
>CPUID on it?
Not the least idea. All 76/77 data were retrieved online and straight
from IBM... so no typo on *my* part at least...
> IBM is known to "fudge" the hard drive numbers a little to cover
the
>variety of sizes out there that they sourced. In little else though;
It
>would surprise me that they weren't consistant with the CPU
notations.
I won't exclude that IBM would deliver a given model with a different
("better", read "mass produced" and hence cheaper) CPU and that the
list (cant' find the link though right now) reflected a later change.
>I depend on IBM to be better than other manufacturers.
>Part of that is consistency in product labeling.
I wish that would hold true, but I do have some doubts about that...
Helmut
Tim Clarke
> > the variety of sizes out there that they sourced. In little else though;
> > It would surprise me that they weren't consistant with the CPU
> > notations.
Apart from the noted differences between the earlier (Models 50, 60, 70 and
80... plus a few others) and later systems, the "Model-SubModel" convention
is relatively accurate, but obviously evolved over time and developed some
"looseness" and non-sequential "holes" as a result.
> I won't exclude that IBM would deliver a given model with a different
> ("better", read "mass produced" and hence cheaper) CPU and that the
> list (cant' find the link though right now) reflected a later change.
My version, compiled from various IBM documentation and direct observation,
lurks here:
http://homepage.ntlworld.com/tim_n_clarke/Identfy.htm
> >I depend on IBM to be better than other manufacturers.
> >Part of that is consistency in product labeling.
>
> I wish that would hold true, but I do have some doubts about that...
As I said above, the consistency dropped-off as time went on and more
variations were introduced.
--
Cheers,
WBS Tim
Unal Z
> > > Second character (your "Y"):
> > > B = 486SLC2-50
> > :: also 486SX-33, 486SX2-66, mod. 9577-QB6/QBA (Helmut Einfalt)
> This is a Bermuda planar? Is the 486SX-33 a PLCC mounting? I am also
> waiting to hear about the 486SX2-66, which is a fairly rare chip. Got a
> CPUID on it?
The data is from configuration list, where only the most important
parameters are listed. Regarding also the disclaimer of Helmut Einfalt,
typos from all sources should be not excluded, but in any case this has been
the coding and I just process and output the result. I have no additional
[technical] information about the quoted models. I am too quite interested
to know more about them.
This B code for 486SX-33, 486SX2-66, a deviation in this case, appears only
at mod. 9577-QB6 and 9577-QBA. These should have at least different disk
capacities (6=100/104, A=200/212), but they are both listed as with 202MB
HDs. Let us put a question mark on this.
> > > D = 386DX-33 (typo ?)
> > :: also D = 486DX2-66, also 486DX4-100 mod. 9576i (Helmut Einfalt)
> For the 76i/77i it is a Lacuna planar. A true factory upgrade these
> should have been marked correctly. I know that there doesn't seem to be
any
> PS/2 model that had a 386DX-33, but I have copied it exactly from the
> manual.
I know already how careful you are, my suggestion was rather if you think it
were an IBM typo (I think so). This D code appears at 9576-4D6/6D6/8D6 (DX2)
9576-4D9/6D9/8D9 (DX4), all 128K cache and mod. 76i, no any HD capacities in
the list.
> > :: new S = 486SX2-50 (mod. 76i, Helmut Einfalt)
> Again a Lacuna & I don't dispute the new "S" information.
This is 9576-8S5, no HD, no cache, no video entry, according to the list.
This S code appear *only* at this model. Customized series? See also below.
> IBM is known to heavily stick with convention: The true 486SLC2/486SLC3
> daughtercard upgrade kits for the Model 56/57 had submodel stickers for
the
> customer to put in place on the case.
Two questions: When does a configuration gets a new submodel number
assigned? Would it be possible that a customer order for a few thousand
units with a specific, custom configuration eventually gets a new submodel
code? If that is so, that could explain the deviations, visible only within
a local market.
> IBM is known to "fudge" the hard drive numbers a little to cover the
> variety of sizes out there that they sourced. In little else though; It
> would surprise me that they weren't consistant with the CPU notations. I
> depend on IBM to be better than other manufacturers. Part of that is
> consistency in product labeling.
My first thoughtless assumption is that some submodel codes may have been
assigned by country branches and are "valid" only within those countries. As
long as they are intended for internal use and hidden from ordinary
customers, this could work for them. However, once the customer parts with
them and they enter the used computer market, the labelling could help us to
better identify the boxes. Personally, most often sellers do not allow to
open the box and look inside.
Perhaps we should set aside an *official* list and an *exception* list. The
official list are the retail models, the exception list the custom
configurations. I know how an organization placed an order for thousands of
preloaded PS/2s with strictly defined configs. They had once maybe the
largest number of OS/2 licenses in this country but they have switched now
to NT. I hope, I get a chance to inspect their submodel numbers. I'll have
to ask them for that.
--
UZ
Unal Z
Ooops, "Personally, in my experience ...."
Few more exceptions (9585 and 9595):
Mod. 9585
----------
9585-x01, where x is a country code in [C, D, F, G, H, I, N, P, V, W]:
This is 486SX-33, 424MB HD, SVGA.
9585-Enn, where nn in [01, 03, 13, 23, 33]:
This is 486DX2-66, SCSI, no HD capacity, no video entry.
Mod. 9595, PS/2 Server 95:
--------------------------
9595-P6T/Q6T, Pentium 66, Fast SCSI, no video entry
9595-P92/Q92, Pentium 90 with 2GB (code 2?) HD, Fast SCSI, no video entry
9595-P9T/Q9T, Pentium 90, Fast SCSI, no video entry
9595-S0n, n in [3, 6, C] :: 540MB HD, no SCSI & video entries, "LAN"
9595-S03, 486DX2
9595-S06, Pentium 66
9595-S0C, Pentium 60/90
9595-Sn0/Sn3/Sn6 where n in [1, 2, 3] but no S16 :: no HD & SCSI & video
entries, "disk base":
9595-S10/S20/S30, Pentium 60, "PS/2 Server 560"
9595-S13/S23/S33, 486DX2, "PS/2 Server 466"
9595-S26/S36, Pentium 66, "PS/2 Server 560"
9595-B0C, Pentium 66, no HD entry, RAID, SVGA
* * *
I see in the list a particular mod. 8590-402 ("PS/2 90 XP 386") with
386DX-20, XGA, no HD entry, which is a curiosity.
* * *
New disk code X = no disk.
"IBM Personal Systems", Jim Hoskins, 7th ed: the X code for the disk means
"no disk" (diskless workstation), e.g. mod. 56LS 486SLC3 = 9556-1EX/2EX.
--
UZ
rd...@spamcop.net
Peterwendt
>Don't know if a '0' got used for PS/2 as well.
Yep.
As in 8555-LT0 and -LE0 - these TR or E(i)thernet medialess workstations.
Very friendly greetings from Peter in Germany
http://members.aol.com/mcapage0/mcaindex.htm