11 Jun
2019
11 Jun
'19
5:50 p.m.
One of the projects I've been working on recently is adding floating point accelerator
emulation to the SIMH 3B2/400 emulator. I _think_ I've done reasonably well, in that
the simulator passes all of the accelerator diagnostics that AT&T wrote for their own
product, but frankly these tests are rather cursory and don't validate much.
I'd like to compile a set of IEEE-754 tests on the 3B2. Unfortunately, the only
compiler I have ready access to on the 3B2 is AT&T's pre-ANSI C compiler, so not a
lot of modern C is going to work.
Does anyone know of a period-appropriate set of IEEE-754 tests that could be compiled on
SVR3?
-Seth
--
Seth Morabito
Poulsbo, WA
web at loomcom.com
12 Jun
12 Jun
12:55 a.m.
On 2019-06-11 20:50, Seth Morabito via cctalk wrote:
One of the projects I've been working on recently
is adding floating point accelerator emulation to the SIMH 3B2/400 emulator. I _think_
I've done reasonably well, in that the simulator passes all of the accelerator
diagnostics that AT&T wrote for their own product, but frankly these tests are rather
cursory and don't validate much.
I'd like to compile a set of IEEE-754 tests on the 3B2. Unfortunately, the only
compiler I have ready access to on the 3B2 is AT&T's pre-ANSI C compiler, so not a
lot of modern C is going to work.
Does anyone know of a period-appropriate set of IEEE-754 tests that could be compiled on
SVR3?
http://www.jhauser.us/arithmetic/SoftFloat.html
?
1:06 a.m.
On Wed, 12 Jun 2019 at 09:55, emanuel stiebler via cctalk
<cctalk at classiccmp.org> wrote:
I took a quick look - the source is written in ANSI C (including the
oldest versions there), so that looks like it won't work for Seth.
On 2019-06-11 20:50, Seth Morabito via cctalk wrote:
[..]
I'd like
to compile a set of IEEE-754 tests on the 3B2. Unfortunately, the only compiler I have
ready access to on the 3B2 is AT&T's pre-ANSI C compiler, so not a lot of modern C
is going to work.
Does anyone know of a period-appropriate set of IEEE-754 tests that could be compiled on
SVR3?
http://www.jhauser.us/arithmetic/SoftFloat.html 
9:15 a.m.
On 6/11/19 5:50 PM, Seth Morabito via cctalk wrote:
One of the projects I've been working on recently
is adding floating point accelerator emulation to the SIMH 3B2/400 emulator. I _think_
I've done reasonably well, in that the simulator passes all of the accelerator
diagnostics that AT&T wrote for their own product, but frankly these tests are rather
cursory and don't validate much.
I'd like to compile a set of IEEE-754 tests on the 3B2. Unfortunately, the only
compiler I have ready access to on the 3B2 is AT&T's pre-ANSI C compiler, so not a
lot of modern C is going to work.
Does anyone know of a period-appropriate set of IEEE-754 tests that could be compiled on
SVR3?
-Seth
is there anything in
http://bitsavers.org/bits/ATT/SysV_r3.0_SV_Ver_Suite_Source_Tape_1987 ?
9:28 a.m.
On Wed, Jun 12, 2019, at 9:15 AM, Al Kossow via cctalk wrote:
is there anything in
http://bitsavers.org/bits/ATT/SysV_r3.0_SV_Ver_Suite_Source_Tape_1987 ?
Wow wow wow, thanks for pointing this out (and for dumping the tape) Al. That's a
great find, I wasn't aware it was on BitSavers!
I will definitely dig around in here. Moreover, I'll try to get it compiled on the 3B2
emulator in its entirety so I can attempt to pass verification. I say this a bit naively
-- I don't know how much of SVVS is supposed to verify hardware directly, if at all.
-Seth
--
Seth Morabito
Poulsbo, WA
web at loomcom.com
10:21 a.m.
On Wed, Jun 12, 2019, at 9:28 AM, Seth Morabito via cctalk wrote:
I will definitely dig around in here. Moreover,
I'll try to get it
compiled on the 3B2 emulator in its entirety so I can attempt to pass
verification.
Unfortunately there is no obvious way to build the SVVS sources from the tape. No
makefiles or build scripts that I could find after searching a bit. There clearly is
support for build and install system of some kind, because there are lots of definition
files scattered around that would instruct a build system how to compile, but the build
system itself is missing and/or undocumented, and the interpolated variable values (such
as %COMP% and %OPTS% and so forth) are not obvious. I suspect the official SVVS
documentation would reveal all, but I can't find it. Drat!
Nevertheless, there ARE useful bits of code in there that I can extract and build by hand,
so I'm still very happy to have it.
-Seth
--
Seth Morabito
Poulsbo, WA
web at loomcom.com
9:49 a.m.
Have you looked at http://www.math.utah.edu/~beebe/software/ieee/ ?
Which says:
"All C programs can also be compiled with old-style Kernighan and
Ritchie C. However, inasmuch as I now have access to only one such
old compiler, with more than 35 others supporting 1989 Standard C,
I no longer attempt to retain K&R compatibility in new code.
If you run into trouble, perhaps Nelson has older versions stashed away?
10:17 a.m.
On Wed, Jun 12, 2019, at 9:49 AM, Phil Budne wrote:
Have you looked at
http://www.math.utah.edu/~beebe/software/ieee/ ?
Which says:
"All C programs can also be compiled with old-style Kernighan and
Ritchie C. However, inasmuch as I now have access to only one such
old compiler, with more than 35 others supporting 1989 Standard C,
I no longer attempt to retain K&R compatibility in new code.
If you run into trouble, perhaps Nelson has older versions stashed away?
Great find, somehow I had missed this page in my searches. I'll add this to the
growing arsenal, thank you!
-Seth
--
Seth Morabito
Poulsbo, WA
web at loomcom.com
3:09 p.m.
On Wed, Jun 12, 2019 at 10:49 AM Phil Budne via cctalk <
cctalk at classiccmp.org> wrote:
Have you looked at
http://www.math.utah.edu/~beebe/software/ieee/ ?
Which says:
"All C programs can also be compiled with old-style Kernighan and
Ritchie C. However, inasmuch as I now have access to only one such
old compiler, with more than 35 others supporting 1989 Standard C,
I no longer attempt to retain K&R compatibility in new code.
If you run into trouble, perhaps Nelson has older versions stashed away?
Can't you just run unprotoize on it?
Warner
1801
days inactive
1801
days old
8 comments
6 participants
participants (6)
-
aek＠bitsavers.org -
emu＠e-bbes.com -
imp＠bsdimp.com -
kspt.tor＠gmail.com -
phil＠ultimate.com -
web＠loomcom.com