my 1st quadcore

  • Thread starter Thread starter Flasherly
  • Start date Start date
F

Flasherly

Intel 8200/2.33Ghz x 4 cores, found on Ebay for $25, I used to replace
an Intel 3Ghz dualcore;- Same family of technological production
means, btw.

Surprised ** XP ** picked it up -- (shows as a Xeon in a dated Everest
HW Diag.), evidently will use it, too - "the OS _can_ [XP -fl] assign
other tasks to idle cores." First quote I noticed on Tom's HW. Seems,
perhaps, that the name of the game, then, is efficiency - e.g., W7 is
therefore more efficiently designed to, hm, about "handling threads"]
with a greater preponderance of single-core coded programs still run
on computers;- Caveats, exclusions and limitations to XP I'd think
well might as well apply. Whatever the chainmail of an eating order
is to XP, vrs NT/Vista/W7 when mightily gobbling up cores in a whoosh.

Sounds somewhat a difficult, no doubt a more complicated proposal, I
had imagined, to update this MB (and its quad) into a entertainment
system for more intensive sound processing apps I'm running, (than
most anything I can demand here at this station);- the ESys, as it is,
believe is a 2.2Ghz under AMD's typical moniker for listing it - a X2
4000 (dual core ostensibly "Intel-equal" to 4Ghz/3Ghz or some such AMD
marketing crap).

Dunno squat, though. Other than it's a monkey-barrel of a load of work
to swap these two systems out, one for the other. At a residual, as I
suspect, perhaps considerably lower in units of computing processing
power. Negligible, IOW, at what I might propose, say, to bring all of
Oxygen6's streaming sound-processing modules into play -- for still
getting gaps and pauses in playback, whilst I throttle its neck and
choke this I8200 to death. Murphy's optimism, as it were, in case I'm
totally wrong.

As it is, the ES/sound system is optimized, and well, for my having
A-B'd most of Oxygen6's modules, additionally, to a couple of select
and notably independent VST plugins. I've narrowed in to where, even
if I had the most powerful I7, 'in the world, punk' - more, such as
additional harmonics and other cake-icing from Oxygen6 isn't
necessarily going to improve inherently top-notch pro-studio session
mixes.

That brick wall, a decade ago, technology hit when it was universally
acclaimed that electrons will move only so fast through a pipeline
conduit, I think, has possibly just hit me. I feel, somehow,
splatted. Too bad I haven't anything more useful to run, really, in
this residual void of a quad-environ, unlike all the rest, the other
50% of home PC users that predominately game with them. Blam. Oh,
well, I guess it can be said that I'm replete, a tad spicier now with
a quad.

-
'My dad liked his food like his women.' -Lenny Bruce's daughter.
 
I feel, somehow,
splatted. Too bad I haven't anything more useful to run, really, in
this residual void of a quad-environ, unlike all the rest,
Click to expand...

Disregard all the above - just ran a normalization program over a ton
of sound encodes and it hit twice the performance matrix readouts I
was getting, reported by Process Lasso for red-tagging CPU utilization
over all quadcores. Sweet.

It just got more interesting.
 
Flasherly said:
Disregard all the above - just ran a normalization program over a ton
of sound encodes and it hit twice the performance matrix readouts I
was getting, reported by Process Lasso for red-tagging CPU utilization
over all quadcores. Sweet.

It just got more interesting.
Click to expand...

Regarding your quad core:

1) Win98 will run on one core only.
2) Win2K will report two cores and ignore the other two.
The Win2K desktop license, is based on cores, and
you only get two. It was designed for an era of
desktop motherboards with two processor slots
(two AthlonMP single core or two Xeon single core).
3) WinXP Home supports one socket. WinXP Pro supports two sockets.
The sockets can have as many cores as you can find.
This means either version of WinXP will report all four cores
of your processor, since only one CPU socket is involved.
WinXP marks a change from core licensing to socket licensing.

WinXP schedules task just as well as any later OS.
Many promises were made in the later OSes, but it's
hard to see their effects. For example, your quad core 8200
is actually two dual core processors sharing an FSB.

http://www.cpu-world.com/CPUs/Core_...J0534MN - AT80580PJ0534MN (BX80580Q8200).html

"(2) 2MB 8-way set associative caches
(each L2 cache is shared between 2 cores)

Later OSes may be NUMA aware, or aware of core
organization. If a process is running on Core1,
it can migrate to Core2, then back to Core1, at
zero cost. That's because those cores share an L2,
and no cache flushing is caused by moving to another
core. If Core1 and Core2 happened to be fully utilized,
an OS later than WinXP will still move the process
to Core3 or Core4, but be aware there is a cost
associated with doing so. WinXP is more or less
oblivious to the cost (a little bit of cache coherency
traffic after the move). But this fine tuning is more
or less noise in the bigger picture. WinXP is still
"sufficient" as an OS. You can even run that $4000 processor
with 18 cores if you want, because the license is by socket
and not by cores.

Windows 8/8.1 is a little less generous, because it
"reserves" some cycles. This can be seen as a frame rate
reduction, running the same game on a WinXP/Win8 dual
boot computer. One where the game was barely smooth
on WinXP, will stutter a bit on Win8. The trick to beating
the reservation system (for programs with infinite scaleup),
is to fork more threads. For example, 7ZIP on Win8 will run
slightly faster if you fork twice as many threads as there
are virtual or physical cores. If you have a quad core, and
play a game that was smooth as butter on a single core,
then you will not notice the reservation issue.

Paul
 
Regarding your quad core:

1) Win98 will run on one core only.
2) Win2K will report two cores and ignore the other two.
The Win2K desktop license, is based on cores, and
you only get two. It was designed for an era of
desktop motherboards with two processor slots
(two AthlonMP single core or two Xeon single core).
3) WinXP Home supports one socket. WinXP Pro supports two sockets.
The sockets can have as many cores as you can find.
This means either version of WinXP will report all four cores
of your processor, since only one CPU socket is involved.
WinXP marks a change from core licensing to socket licensing.

WinXP schedules task just as well as any later OS.
Many promises were made in the later OSes, but it's
hard to see their effects. For example, your quad core 8200
is actually two dual core processors sharing an FSB.

http://www.cpu-world.com/CPUs/Core_...J0534MN - AT80580PJ0534MN (BX80580Q8200).html

"(2) 2MB 8-way set associative caches
(each L2 cache is shared between 2 cores)

Later OSes may be NUMA aware, or aware of core
organization. If a process is running on Core1,
it can migrate to Core2, then back to Core1, at
zero cost. That's because those cores share an L2,
and no cache flushing is caused by moving to another
core. If Core1 and Core2 happened to be fully utilized,
an OS later than WinXP will still move the process
to Core3 or Core4, but be aware there is a cost
associated with doing so. WinXP is more or less
oblivious to the cost (a little bit of cache coherency
traffic after the move). But this fine tuning is more
or less noise in the bigger picture. WinXP is still
"sufficient" as an OS. You can even run that $4000 processor
with 18 cores if you want, because the license is by socket
and not by cores.

Windows 8/8.1 is a little less generous, because it
"reserves" some cycles. This can be seen as a frame rate
reduction, running the same game on a WinXP/Win8 dual
boot computer. One where the game was barely smooth
on WinXP, will stutter a bit on Win8. The trick to beating
the reservation system (for programs with infinite scaleup),
is to fork more threads. For example, 7ZIP on Win8 will run
slightly faster if you fork twice as many threads as there
are virtual or physical cores. If you have a quad core, and
play a game that was smooth as butter on a single core,
then you will not notice the reservation issue.

Paul
Click to expand...

That's some excellent coverage to all bases, and no reason why there
shouldn't be at least modest (very) gain improvements on sound
processing for playback. Taltube/FerricTDS - "smooth dynamic shaping
capabilities of some high-end reel-to-reel tape recorders, this
plug-in simulates three of the most distinctive and much appreciated
sonic effects generated by these devices: DYNAMICS, SATURATION,
LIMITING performance" - (two VSR DLLs off Oxygen6's sound
leveling/low bandpass filtering), for StereoTool's 9band
compressor/limiter, AGC. StereoTool, as it is, 'coded, per se,
multi-core aware,' can't be maxed without artifacts, even if it is
already close enough (via a slider for allowing it utilize more CPU
resources) to maxed, to be then nominally discernable.

So, that's about as hard as I can hit it, the 8200, in any continued
sense for efficiency. Haven't enough "tinsel," utilities I'm as yet
aware of, to make it glitter any brighter than it is (I've a fair
amount, probably half or more, of sound material at LossLess quality).
Top-notch hardware, of course - preamps/processors, mixed amps and
such over a 15-driver (quad-speaker/monitor) array.

That's also a set-environ computer, not much else going on with it
other than sound. Perhaps, at least for the time being, better off
here (dual-XP/W7 boot machine), where it's better setup to try various
apps for a testbed platform.

Really -- came out well past expectations I had for XP/SP2, which I
just didn't see overall as conducive to multi-core processing above
duals. A false presumption I had held where W7 had all those eggs in
its basket.
 
Back
Top