Chris,
Never had a problem with it. Only ran TS8 on a hyperthreading machine.
Ran versions 3159, 3247 and 3258. I am unclear what you mean by full
hyperthreading. Also unclear what you mean by
hyperthreading/multi-processors. I used a dual processor for years with
TS2ki. Only way to go. Now use a hyperthreading machine for TS2ki and
TS8. Works fine even with both TS programs running. Have not run a new
processor with two processors on one chip. Should make no negative
effects. My guess is you did something fancy and messed it up. I'm not
even sure TS8 can utilize a dual or hyperthreading processor properly.
TS2ki can.
Give us more detail as to what you did or just undo it.
Jimmy
Hi gang,
Anyone out there know much about hyperthreading/multi-processors? I
finally
gave up my trusty old Win 2000 for XP, and set up my P IV to do full HT.
I'm
having some really bizarre problems with TS 8.2. It might just be
something
going on with TS - seems quite a few folks have issues now - but given the
magnitude of what I am seeing and the new install I still wonder if
something else might contribute.
What is happening is the are amazingly large context switchs by the orplat
and orchart services, and mainly when both charts and matrix/market depth
windows are open at the same time. By large I mean stable 10-20,000/sec,
and
with volume can easily climb to 70-80,000. The same processor with w2000/
no
HT would maybe hit 2000 on very busy times. Along with the context
switches, both orchart and orplat are running way more cpu than they
should.
For example, a single matrix used to run maybe 1-2 pct on orplat. Now it
15-30%. And orchart cpu use increases considerably due to the switching as
well. The more matrix windows, the more switching.
If you close the matrix window and just run charts, everything is fine.
There will be an occasional blip really high, but the devastating cpu
effect is gone . OK, now here's the kicker... if you set the affinity of
orplat and orchart to the same cpu, the switching goes away. So that makes
me thing that it's a cache miss or common memory thing. Affinities are not
a
good solution long term though, as it is much easier to peg half the
processor.
I have tried everything I can think of and can't find anything wrong on
the
pc. Intel Inf files were run...then I ran all the Intel HT diagnostics and
they are perfect. Bios is current. I tried running with nothing else at
all
on the pc. Nothing changes. Other than inability to use matrix, HT is much
smoother than without. FED day today was super smooth.
I know there are some PC wizards out there. If anyone has any ideas, I'd
be
most appreciative.
Best,
Chris