On Thu, Aug 9, 2012 at 10:22 PM, Dave McGuire <mcguire at neurotica.com> wrote:
On 08/09/2012 08:50 PM, Paul_Koning at Dell.com wrote:
Why is SIMH so slow though?
Because it's extremely portable with no architecture-specific
assembler assists. As I won't use anything that's NOT portable, I
consider it to be screaming fast.
I don't know about Hercules, but dtcyber is all C code, and it seems
to be substantially faster than simh.
...and it also only emulates one architecture. Simh has an underlying
emulation framework upon which emulations for many disparate
architectures can be constructed. There is (as I'm sure you know)
always a cost to this sort of flexibility.
At least that's my take on it.
VAX is a "big" architecture and instruction set to emulate.
That's true, but that doesn't mean any one instruction is hard to
emulate, it just means there's a lot of code. Well, the packed
instructions, and edit, and stuff like that, sure, but when you're
running NetBSD makefiles and compiles you're basically in the world
of integer instructions, and those aren't particularly hard.
Maybe it's time to do some code analysis and profiling.
I think that's a fine idea.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
Hello!
Dave did the Alpha family come in a 1U size? If so could said system
be considered to be a quiet one? For example I have here a SUN
UltraSPARC 5 machine who happens to be about that size and is
currently working headless, and serving as my website manager (hosting
entity and server for same.) it happens to be so quiet the only way
I'd know that the fellow is working is that I promptly logon to the
unit every few hours via SSH or bring up the site on the browser.
I'm trying to find a keyboard for the fellow so I can also use it as a
regular workstation, so far my regular source has not responded.
What I'd really like to do is to find another one, plus a keyboard,
and a terminal server that SUN made once. Having obtained them I would
simply configure Number Two as a webserver with its console out line
connected to that TS unit. Number One would become a regular
workstation.
Yeah I know gang, mildly off topic, but here goes. Any clews regarding
that Alpha? Oh and if you can track down a keyboard or even that Alpha
or the other two please contact off list for that address and other
things.
And Sampsa? Right now your surrounded by a collection of dangerous
individuals straight out of Doctor Who plus seven Daleks.....
-----
Gregg C Levine gregg.drwho8 at gmail.com
"This signature fought the Time Wars, time and again."
On 08/09/2012 08:50 PM, Paul_Koning at Dell.com wrote:
Why is SIMH so slow though?
Because it's extremely portable with no architecture-specific
assembler assists. As I won't use anything that's NOT portable, I
consider it to be screaming fast.
I don't know about Hercules, but dtcyber is all C code, and it seems
to be substantially faster than simh.
...and it also only emulates one architecture. Simh has an underlying
emulation framework upon which emulations for many disparate
architectures can be constructed. There is (as I'm sure you know)
always a cost to this sort of flexibility.
At least that's my take on it.
VAX is a "big" architecture and instruction set to emulate.
That's true, but that doesn't mean any one instruction is hard to
emulate, it just means there's a lot of code. Well, the packed
instructions, and edit, and stuff like that, sure, but when you're
running NetBSD makefiles and compiles you're basically in the world
of integer instructions, and those aren't particularly hard.
Maybe it's time to do some code analysis and profiling.
I think that's a fine idea.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
On Aug 9, 2012, at 6:10 PM, Dave McGuire wrote:
On 08/09/2012 04:07 PM, Sampsa Laine wrote:
Why is SIMH so slow though?
Because it's extremely portable with no architecture-specific
assembler assists. As I won't use anything that's NOT portable, I
consider it to be screaming fast.
I don't know about Hercules, but dtcyber is all C code, and it seems to be substantially faster than simh.
As far as I can understand, one MIPS is roughly one VUPS, correct?
More or less...depends on who you ask.
So how come I get 14 VUPS on the same host running SIMH whilst my Hercules install peaks at 180?
Ummm...because one is an IBM mainframe architecture and the other is a
VAX? MIPS is a very, very different thing between different architectures.
True, but a 360 is also a CISC machine, somewhat simpler than VAX but not massively so.
On Aug 9, 2012, at 6:12 PM, Dave McGuire wrote:
On 08/09/2012 05:21 PM, Paul_Koning at Dell.com wrote:
...
I did notice that SIMH vax seems surprisingly slow, at least when running NetBSD.
VAX is a "big" architecture and instruction set to emulate.
That's true, but that doesn't mean any one instruction is hard to emulate, it just means there's a lot of code. Well, the packed instructions, and edit, and stuff like that, sure, but when you're running NetBSD makefiles and compiles you're basically in the world of integer instructions, and those aren't particularly hard.
Maybe it's time to do some code analysis and profiling.
paul
On 08/09/2012 05:21 PM, Paul_Koning at Dell.com wrote:
For example, how does the SIMH pdp11 compare to E-11? I can't make a direct comparison, unfortunately.
I'm told E-11 is faster than simh. That's the trade-off for
portability. I usually run simh on the fastest (in terms of integer
performance) computer here, which is a large UltraSPARC system. I can't
run E-11 on that. The UltraSPARC machine, even while busy, runs rings
around my Core i7 quad desktop box running simh.
I did notice that SIMH vax seems surprisingly slow, at least when running NetBSD.
VAX is a "big" architecture and instruction set to emulate.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
On 08/09/2012 04:07 PM, Sampsa Laine wrote:
Why is SIMH so slow though?
Because it's extremely portable with no architecture-specific
assembler assists. As I won't use anything that's NOT portable, I
consider it to be screaming fast.
As far as I can understand, one MIPS is roughly one VUPS, correct?
More or less...depends on who you ask.
So how come I get 14 VUPS on the same host running SIMH whilst my Hercules install peaks at 180?
Ummm...because one is an IBM mainframe architecture and the other is a
VAX? MIPS is a very, very different thing between different architectures.
Are the architectures that different (obviously they are) or what is it?
They very much are, yes.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
That's a good question.
For example, how does the SIMH pdp11 compare to E-11? I can't make a direct comparison, unfortunately.
I did notice that SIMH vax seems surprisingly slow, at least when running NetBSD.
Another data point: dtcyber (a Cyber mainframe emulation) runs quite fast. I haven't tried to express it in MIPS, or VUPS (or CUPS?) but it's much faster than the original hardware, which was in the 3-10 MIPS range. So it seems to be closer to what you saw for Hercules. And a Cyber is not that easy to emulate, due to its oddball floating point.
paul
On Aug 9, 2012, at 4:07 PM, Sampsa Laine wrote:
Why is SIMH so slow though?
As far as I can understand, one MIPS is roughly one VUPS, correct?
So how come I get 14 VUPS on the same host running SIMH whilst my Hercules install peaks at 180?
Are the architectures that different (obviously they are) or what is it?
Sampsa
On 9 Aug 2012, at 23:04, Dave McGuire wrote:
On 08/09/2012 02:33 PM, Paul_Koning at Dell.com wrote:
This seems like a natural for SIMH, since it already has a large
amount of necessary prerequisites (a lot of device emulation,
framework pieces, etc.).
There is Alpha support in the development track version of simh.
Is the architecture well enough defined in
publicly available documents? I would guess yes but I don't know for
sure. Are there secret bits that are critical and hard to obtain?
The architecture manual pretty fully documents it.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
Why is SIMH so slow though?
As far as I can understand, one MIPS is roughly one VUPS, correct?
So how come I get 14 VUPS on the same host running SIMH whilst my Hercules install peaks at 180?
Are the architectures that different (obviously they are) or what is it?
Sampsa
On 9 Aug 2012, at 23:04, Dave McGuire wrote:
On 08/09/2012 02:33 PM, Paul_Koning at Dell.com wrote:
This seems like a natural for SIMH, since it already has a large
amount of necessary prerequisites (a lot of device emulation,
framework pieces, etc.).
There is Alpha support in the development track version of simh.
Is the architecture well enough defined in
publicly available documents? I would guess yes but I don't know for
sure. Are there secret bits that are critical and hard to obtain?
The architecture manual pretty fully documents it.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
On 08/09/2012 02:33 PM, Paul_Koning at Dell.com wrote:
This seems like a natural for SIMH, since it already has a large
amount of necessary prerequisites (a lot of device emulation,
framework pieces, etc.).
There is Alpha support in the development track version of simh.
Is the architecture well enough defined in
publicly available documents? I would guess yes but I don't know for
sure. Are there secret bits that are critical and hard to obtain?
The architecture manual pretty fully documents it.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
I'm trying to get the bridge built on OpenBSD 5.1 64-bit but it keep segfaulting. Anyone ever get this working?
# cc -O2 -Wall -o bridge bridge.c -lpcap
bridge.c:65:1: warning: "MAX" redefined
In file included from /usr/include/netdb.h:90,
from bridge.c:32:
/usr/include/sys/param.h:191:1: warning: this is the location of the previous definition
bridge.c: In function 'add_bridge':
bridge.c:241: warning: implicit declaration of function 'inet_aton'
bridge.c:263: warning: format '%d' expects type 'int', but argument 2 has type 'char *'
bridge.c:263: warning: format '%d' expects type 'int', but argument 3 has type 'char *'
bridge.c: In function 'add_service':
bridge.c:276: warning: format '%s' expects type 'char *', but argument 3 has type 'struct BRIDGE *'
bridge.c: In function 'dump_data':
bridge.c:524: warning: implicit declaration of function 'inet_ntoa'
bridge.c:531: warning: format '%s' expects type 'char *', but argument 4 has type 'int'
bridge.c: In function 'main':
bridge.c:564: warning: unused variable 'port'
bridge.c:561: warning: unused variable 'len'
/tmp//cc69aRp2.o(.text+0x934): In function `add_bridge':
: warning: strcpy() is almost always misused, please use strlcpy()
# cp bridge /var/bridge/
# cd /var/bridge/
# ./bridge 4711
Adding router ''local''. 00000000:0
Adding router ''sampsa''. 0afc2a0a:4711
Adding DECnet bridge local.
Trying to match local
Matching against: local
Found match: local == local
Adding DECnet bridge sampsa.
Trying to match sampsa
Matching against: local
Matching against: sampsa
Found match: sampsa == sampsa
Adding LAT bridge local.
Trying to match local
Matching against: local
Found match: local == local
Adding LAT bridge sampsa.
Trying to match sampsa
Matching against: local
Matching against: sampsa
Found match: sampsa == sampsa
Host table:
Segmentation fault (core dumped)
-brian
This seems like a natural for SIMH, since it already has a large amount of necessary prerequisites (a lot of device emulation, framework pieces, etc.). Is the architecture well enough defined in publicly available documents? I would guess yes but I don't know for sure. Are there secret bits that are critical and hard to obtain?
paul
On Aug 9, 2012, at 1:51 PM, Gregg Levine wrote:
On Thu, Aug 9, 2012 at 1:41 PM, Kari Uusim ki <uusimaki at exdecfinland.org> wrote:
There is a Non-commercial version running on Linux:
http://www.stromasys.ch/axpnce/
Kari
On 9.8.2012 19:56, Sampsa Laine wrote:
CHARON-AXP I think runs on Linux.
But not the free versions (windows only).
Sampsa
On 9 Aug 2012, at 19:50, Gregg Levine wrote:
Hello!
Okay here's a completely crazy and off four walls and probably two
dimensions, but here goes:
Does an AXP emulator exist that will run OpenVMS for Alpha and can be
built on (or run on) 32 bit Linux?
Incidentally Dave the group really left three days ago to try to find
Sampsa. They left behind watchers.
-----
Gregg C Levine gregg.drwho8 at gmail.com
"This signature fought the Time Wars, time and again."
.
Hello!
The key word in my question was 32 bit Linux. That site contains this
line " This implementation will not work on single-core,
hyper-threading, or 32-bit systems. ", which disqualifies my current
Linux system as a target. And I can't qualify the laptop as a target
for the Windows example that Sampsa mentioned, which was also
mentioned on this list a while ago.
-----
Gregg C Levine gregg.drwho8 at gmail.com
"This signature fought the Time Wars, time and again."
