| Author |
Message |
Vasanth Venkatachalam
Guest
|
 Posted:
Mon Aug 29, 2005 10:50 pm Post subject:
History Of Hardware Perfomance Counters |
|
|
For a research project that I am doing, I am looking for any articles or
websites dealing with the history
of hardware performance counters on processors (i.e., what were they
originally intended for, etc).
Can anyone provide some pointers? |
|
| Back to top |
|
 |
Terje Mathisen
Guest
|
| Posted:
Tue Aug 30, 2005 12:15 am Post subject:
Re: History Of Hardware Perfomance Counters |
|
|
Vasanth Venkatachalam wrote:
| Quote: | For a research project that I am doing, I am looking for any articles or
websites dealing with the history
of hardware performance counters on processors (i.e., what were they
originally intended for, etc).
Can anyone provide some pointers?
|
You should probably look for the BYTE article I wrote in 1994, about
reverse engineering the Pentium EMON counters.
.... Yes, they still have it:
http://www.byte.com/art/9407/sec12/art3.htm
Terje
--
- <Terje.Mathisen@hda.hydro.com>
"almost all programming can be viewed as an exercise in caching" |
|
| Back to top |
|
|
Henry Newman
Guest
|
| Posted:
Fri Sep 02, 2005 12:15 am Post subject:
Re: History Of Hardware Perfomance Counters |
|
|
I believe that the Cray-XMP or it might have been the YMP had hardware
performance counter. XMP is circa 1982 and YMP is circa 1988
On Mon, 29 Aug 2005 21:36:00 +0200, Terje Mathisen
<terje.mathisen@hda.hydro.com> wrote:
| Quote: | Vasanth Venkatachalam wrote:
For a research project that I am doing, I am looking for any articles or
websites dealing with the history
of hardware performance counters on processors (i.e., what were they
originally intended for, etc).
Can anyone provide some pointers?
You should probably look for the BYTE article I wrote in 1994, about
reverse engineering the Pentium EMON counters.
... Yes, they still have it:
http://www.byte.com/art/9407/sec12/art3.htm
Terje |
|
|
| Back to top |
|
|
Brian Utterback
Guest
|
| Posted:
Fri Sep 02, 2005 4:15 pm Post subject:
Re: History Of Hardware Perfomance Counters |
|
|
Henry Newman wrote:
| Quote: | I believe that the Cray-XMP or it might have been the YMP had hardware
performance counter. XMP is circa 1982 and YMP is circa 1988
|
The XMP had hardware performance counters. I can only confirm this
for 1984 and after, although I believe that they are implemented
from the beginning of (at least) the XMP line.
--
blu
Remember when SOX compliant meant they were both the same color?
----------------------------------------------------------------------
Brian Utterback - OP/N1 RPE, Sun Microsystems, Inc.
Ph:877-259-7345, Em:brian.utterback-at-ess-you-enn-dot-kom |
|
| Back to top |
|
|
Sarr J. Blumson
Guest
|
| Posted:
Fri Sep 02, 2005 4:15 pm Post subject:
Re: History Of Hardware Perfomance Counters |
|
|
In article <devhst$fbl$1@news.service.uci.edu>,
Vasanth Venkatachalam <vvenkata@uci.edu> wrote:
| Quote: | For a research project that I am doing, I am looking for any articles or
websites dealing with the history
of hardware performance counters on processors (i.e., what were they
originally intended for, etc).
Can anyone provide some pointers?
|
This isn't really pointers, but gives some idea of how far back you
might need to look.
In 1969 or 1970 we had an external box (no hope of coming up with a
vendor I'm afraid) that we could attach to signals on the back plane
and count pulses, or to address lines and count references in a region.
The was on a DEC KA10. I believe that IBM gear of similar vintage had
counters that you could get at with the mysterious DIAGNOSE instruction,
but this is beyond things I actually know about.
--
--------
Sarr Blumson sarr.blumson@alum.dartmouth.org
http://www-personal.umich.edu/~sarr/ |
|
| Back to top |
|
|
Nick Maclaren
Guest
|
| Posted:
Fri Sep 02, 2005 4:15 pm Post subject:
Re: History Of Hardware Perfomance Counters |
|
|
In article <i_ZRe.438$De1.366@news.itd.umich.edu>,
sarr@umich.edu (Sarr J. Blumson) writes:
|> In article <devhst$fbl$1@news.service.uci.edu>,
|> Vasanth Venkatachalam <vvenkata@uci.edu> wrote:
|>
|> >For a research project that I am doing, I am looking for any articles or
|> >websites dealing with the history
|> >of hardware performance counters on processors (i.e., what were they
|> >originally intended for, etc).
|> >Can anyone provide some pointers?
|>
|> This isn't really pointers, but gives some idea of how far back you
|> might need to look.
|>
|> In 1969 or 1970 we had an external box (no hope of coming up with a
|> vendor I'm afraid) that we could attach to signals on the back plane
|> and count pulses, or to address lines and count references in a region.
|>
|> The was on a DEC KA10. I believe that IBM gear of similar vintage had
|> counters that you could get at with the mysterious DIAGNOSE instruction,
|> but this is beyond things I actually know about.
You could also attach wires to IBM kit, though it was not a released
facility.
Many of the first generation systems had critical lines attached to
lights, and you could observe the hardware behaviour (for both tuning
and debugging) by watching the lights. Hardware counters derived
from those.
Regards,
Nick Maclaren. |
|
| Back to top |
|
|
Stephen Fuld
Guest
|
| Posted:
Sat Sep 03, 2005 12:15 am Post subject:
Re: History Of Hardware Perfomance Counters |
|
|
"Nick Maclaren" <nmm1@cus.cam.ac.uk> wrote in message
news:df9qrm$ai$1@gemini.csx.cam.ac.uk...
| Quote: |
In article <i_ZRe.438$De1.366@news.itd.umich.edu>,
sarr@umich.edu (Sarr J. Blumson) writes:
|> In article <devhst$fbl$1@news.service.uci.edu>,
|> Vasanth Venkatachalam <vvenkata@uci.edu> wrote:
|
|> >For a research project that I am doing, I am looking for any articles
or
|> >websites dealing with the history
|> >of hardware performance counters on processors (i.e., what were they
|> >originally intended for, etc).
|> >Can anyone provide some pointers?
|
|> This isn't really pointers, but gives some idea of how far back you
|> might need to look.
|
|> In 1969 or 1970 we had an external box (no hope of coming up with a
|> vendor I'm afraid) that we could attach to signals on the back plane
|> and count pulses, or to address lines and count references in a region.
|
IIRC, there were at least two vendors. One was Comten (????), later bought
by NCR, and another called something like Tesdata. You could usually find
signals (the vendors provided the schematics for their systems!) for lots of
interesting things.
There were also sometimes instructions provided whose sole function was to
pulse a particular spot that was easily accessable to such equipment, and
perhaps to output something to some external register. By putting such
instructions in various well chosen places in the code, you could instrument
the software directly.
| Quote: | |
|> The was on a DEC KA10. I believe that IBM gear of similar vintage had
|> counters that you could get at with the mysterious DIAGNOSE
instruction,
|> but this is beyond things I actually know about.
You could also attach wires to IBM kit, though it was not a released
facility.
Many of the first generation systems had critical lines attached to
lights, and you could observe the hardware behaviour (for both tuning
and debugging) by watching the lights. Hardware counters derived
from those.
|
The S/360 had a light on the console that indicated problem state (i.e.
executing instructions in user mode). I heard that someone attached a
photocell to some simple integration electronics and a meter and placed the
device such that the light illuminated the photocell and the resulting meter
read the percentage of time the system spent in the problem state.
Inventive folks!
--
- Stephen Fuld
e-mail address disguised to prevent spam |
|
| Back to top |
|
|
Tom Gardner
Guest
|
| Posted:
Sat Sep 03, 2005 5:36 am Post subject:
Re: History Of Hardware Perfomance Counters |
|
|
On Fri, 02 Sep 2005 15:24:38 +0000, Nick Maclaren wrote:
| Quote: | Many of the first generation systems had critical lines attached to
lights, and you could observe the hardware behaviour (for both tuning
and debugging) by watching the lights. Hardware counters derived
from those.
|
Or loudspeakers instead of lights.
I've still got an audio tape of a 2kips machine that blipped the
loudpeaker every time it executed a shift instruction, IIRC. While
"fetching algol" from the magnetic film it sounded like a broody hen.
The usual music renditons went distinctly off-key when playing higher
notes. (Elliott 803)
And then there was the machine in whic some heroes generated four part
harmony derived from blipping the loudspeaker whenever bit 5 of an
instruction was set. That was my first introduction to a screen editor,
and it sure as hell beat the pants off editing 5 or 8 channel paper tape.
But the loudspeaker positively screamed when scrolling a page :)
(Ferranti Myriad) |
|
| Back to top |
|
|
|
|
|
|
FAQ
Memberlist
Register
Profile
Log in to check your private messages
Log in