How was simh compiled? I used Visual C and the difference between running the compiler with optimization on or off made quite a difference for the generated code. And gcc might compile even faster codefiles.
Hans
-----Original Message-----
From: Sampsa Laine <sampsa at mac.com>
Sender: owner-hecnet at Update.UU.SE
Date: Thu, 09 Aug 2012 23:07:24
To: <hecnet at Update.UU.SE>
Reply-To: hecnet at Update.UU.SESubject: Re: [HECnet] AXP Emulation
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
If the "fiddling" is similar to what is necessary to run VMS on the white box alphas then have a look at
http://home.zonnet.nl/hvlems
The white box alphas were tergeted at the WNT market. Adding two srm parameters works wonders for VMS. It does not involve hardware mods and AFAIK the mods are completely reversible.
Hans
PS
Sent from my cell phone which will undoubtedly upset some systems. Sorry, I'm 600 kms from home.
-----Original Message-----
From: "John H. Reinhardt" <johnhreinhardt at yahoo.com>
Sender: owner-hecnet at Update.UU.SE
Date: Fri, 10 Aug 2012 01:30:11
To: <hecnet at Update.UU.SE>
Reply-To: hecnet at Update.UU.SESubject: Re: [HECnet] AXP Emulation - 1U Alpha systems.
On 8/9/12 11:00 PM, Dave McGuire wrote:
On 08/09/2012 10:54 PM, Gregg Levine wrote:
Dave did the Alpha family come in a 1U size?
Yes, the DS10L comes to mind. My HECnet node AXPEE:: is one of those.
They are EV6-based (nice and quick!) and are available clocked at
466MHz and 600MHz.
There are other 1U models, but that's the one I'm most familiar with.
If so could said system be considered to be a quiet one?
Eh, so-so.
On 08/10/2012 01:30 AM, John H. Reinhardt wrote:
All were fairly noisy for their size. Maybe not noticeable in a
computer center, but definitely in a home setting.
Well, this is definitely a size/noise tradeoff, at least in many
cases. For low noise, you generally want large-diameter, low-speed
fans. For fans small enough to fit into a 1U enclosure, neither of
those are attainable, so you end up with the little screamers.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
On 08/09/2012 11:55 PM, Gregg Levine wrote:
Maybe I need to ship you a care package. Send me your address.
Hello!
Okay coming at via separate travel, actually private message.
Sounds good.
Thanks! One of these days I need to visit your space.....
You're always welcome!
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
On Thu, Aug 9, 2012 at 11:00 PM, Dave McGuire <mcguire at neurotica.com> wrote:
On 08/09/2012 10:54 PM, Gregg Levine wrote:
Dave did the Alpha family come in a 1U size?
Yes, the DS10L comes to mind. My HECnet node AXPEE:: is one of those.
They are EV6-based (nice and quick!) and are available clocked at
466MHz and 600MHz.
There are other 1U models, but that's the one I'm most familiar with.
If so could said system be considered to be a quiet one?
Eh, so-so.
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.
Certainly reasonable even if that's just about the worst machine Sun
ever built. ;) I'll bring some good (small) Sun hardware to the next
VCFe for you.
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.
A Sun keyboard? I'm swimming in them. Send me your shipping address.
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.
I don't recall Sun ever having made a terminal server. But, I do have
probably a hundred Sun systems here, from 1U rackmount to 1800lb
machines with 64 CPUs.
I don't have many Alphas nowadays, and the one small one I do have
(the aforementioned DS10L) is in use, so I can't help you there, at
least not right now.
Maybe I need to ship you a care package. Send me your address.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
Hello!
Okay coming at via separate travel, actually private message.
Thanks! One of these days I need to visit your space.....
-----
Gregg C Levine gregg.drwho8 at gmail.com
"This signature fought the Time Wars, time and again."
On 08/09/2012 10:54 PM, Gregg Levine wrote:
Dave did the Alpha family come in a 1U size?
Yes, the DS10L comes to mind. My HECnet node AXPEE:: is one of those.
They are EV6-based (nice and quick!) and are available clocked at
466MHz and 600MHz.
There are other 1U models, but that's the one I'm most familiar with.
If so could said system be considered to be a quiet one?
Eh, so-so.
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.
Certainly reasonable even if that's just about the worst machine Sun
ever built. ;) I'll bring some good (small) Sun hardware to the next
VCFe for you.
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.
A Sun keyboard? I'm swimming in them. Send me your shipping address.
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.
I don't recall Sun ever having made a terminal server. But, I do have
probably a hundred Sun systems here, from 1U rackmount to 1800lb
machines with 64 CPUs.
I don't have many Alphas nowadays, and the one small one I do have
(the aforementioned DS10L) is in use, so I can't help you there, at
least not right now.
Maybe I need to ship you a care package. Send me your address.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
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
