|
PureBytes Links
Trading Reference Links
|
Ton,
When I went to bed last night, I started thinking about upgrading to a dual
processor 1GHz system to replace this "old" box.
I'm not worried about performance yet. I need to get the system running
first. Right now, speed is the least of my worries. Accuracy is much more
important since we're doing a conversion and the output from MS MUST equal
the output from our trading system.
Our basic trading system runs quite well in the latest rendition of MS. We
ran under MS for DOS until v.6.5.2 (the first MS for Windows version that
fixed the disk thrashing problem we had). It used to take us approximately
3 to 4 minutes of crunching to do our daily processing (for the S&P) and
then we had to do the other half in TAS. MS now does it all in less than 30
seconds. Great improvement.
Guy
" If at first you don't succeed, skydiving is not for you."
-----Original Message-----
From: owner-metastock@xxxxxxxxxxxxx [mailto:owner-metastock@xxxxxxxxxxxxx]On
Behalf Of A.J. Maas
Sent: Sunday, October 01, 2000 2:54 PM
To: Metastock-List
Subject: Re: calculations
Guy,
It seems that you are pushing the peddle to the iron, to me indicating that
all the calculating by Processor and back and forth I/O burstpiping and
writing
to MEM, is the "wrong doer". Perhaps in disguise.................see below.
You must have created a scientific monster ready for the Guiness book...
The Uni here and more uni's elsewhere are also masters in calculations
"until
the PC will give up", thus there is still some hope for you left since that
still
greater and faster computers are being developped all the time!!.
Seriously, see if the {at the top} mentioned ways for timing factor in
performing
the calculations can be more optimized:
-software { formulas, see hereafter }
-hardware { i840 or Server Works III-WorkStation Dual Proc. MoBo,
2x +1.13 GHZ processors, +1Gb SDRAM PC133/PC800 RIMM,
UDMA100, etc. etc.} Esp. the multi-thread capacities of
the Dual MoBo
can be of great advantage here!!.
So there is thus nothing wrong with program or within the gramma of your
formula's,
apart from the referenced, eg that what we cannot see {the referenced
formula's}.
Perhaps that, in that "referenced area" {your other formulas}, is where you
can
explore to see if you can gain more of the above mentioned optimized speed
{eg by altering their compositions}.
Good luck!!
Regards,
Ton Maas
ms-irb@xxxxxxxxxxxxxxxx
Dismiss the ".nospam" bit (including the dot) when replying.
Homepage http://home.planet.nl/~anthmaas
----- Oorspronkelijk bericht -----
Van: "Guy Tann"
Aan: <metastock@xxxxxxxxxxxxx>
Verzonden: zondag 1 oktober 2000 18:03
Onderwerp: RE: calculations
> J., Don, Ton et al who gave me suggestions about this problem.
>
> Thanks for all of the suggestions. I have tried them all, but
unfortunately
> they didn't work.
>
> This problem has been resolved by cutting and pasting each variable into
its
> own indicator, where, albeit slowly, it runs. This is the way I
originally
> coded these variables, but decided to combine them into 2 indicators.
Wrong
> choice as getting this to run has taken me a couple of weeks. Granted I
> spent 80% of my time trying to get MS to work as advertised. Like the
> subtraction problem, I just gave up and programmed around it.
>
> Now don't ask me why it was necessary to split my two indicators with 9
> variables each into 18 separate indicators and why I've encountered this
> very same problem with the next couple of indicators I checked. It's
beyond
> me. It just must be another quirk of MS (among the several I've
encountered
> during this conversion).
>
> I have other indicators that include several variables and they work
without
> a problem. Again, I seem to be learning my way around MS and those areas
> that don't function as advertised, but it's taking a long time. This
> conversion should have been done a month or two ago, but at least we're
> getting close.
>
> Thanks again to everyone. I have high hopes that I can complete the final
> 10 variables today.
>
> Guy
>
> " If at first you don't succeed, skydiving is not for you."
>
> -----Original Message-----
> From: owner-metastock@xxxxxxxxxxxxx
[mailto:owner-metastock@xxxxxxxxxxxxx]On
> Behalf Of j seed
> Sent: Sunday, October 01, 2000 7:41 AM
> To: metastock@xxxxxxxxxxxxx
> Subject: Re: calculations
>
> Guy,
> Ver.7.0 is a real "Jethro" when it comes to spelling. It doesn't like to
> have variable names split onto more than one line. It may help to change
> your variable statements to:
> COMHX03:= ( FmlVar("COMH-MXX","COMHMXX") -
> Ref(FmlVar("COMH-MXX","COMHMXX") , -3 )) / 3;
>
> This may seem insignificant but I've had formulas that would only work
this
> way.
>
> J.
>
> >From: "Guy Tann" <grt@xxxxxxxxxxxx>
> >Reply-To: metastock@xxxxxxxxxxxxx
> >To: "Metastock User Group" <metastock-list@xxxxxxxxxxxxx>
> >Subject: calculations
> >Date: Sun, 1 Oct 2000 00:41:51 -0700
> >
> >Don,
> >
> >Here's part of what I had in one indicator that I called COMHX01X09.
> >
> >COMHX03:= ( FmlVar("COMH-MXX","COMHMXX") - Ref(
> >FmlVar("COMH-MXX","COMHMXX") , -3 )) / 3;
> >
> >COMHX04:= ( FmlVar("COMH-MXX","COMHMXX") - Ref(
> >FmlVar("COMH-MXX","COMHMXX") , -4 )) / 4;
> >
> >COMHX05:= ( FmlVar("COMH-MXX","COMHMXX") - Ref(
> >FmlVar("COMH-MXX","COMHMXX") , -5 )) / 5;
> >
> >COMHX06:= ( FmlVar("COMH-MXX","COMHMXX") - Ref(
> >FmlVar("COMH-MXX","COMHMXX") , -6 ) ) / 6;
> >
> >COMHX07:= ( FmlVar("COMH-MXX","COMHMXX") - Ref(
> >FmlVar("COMH-MXX","COMHMXX") , -7 ) ) / 7;
> >
> >COMHX08:= ( FmlVar("COMH-MXX","COMHMXX") - Ref(
> >FmlVar("COMH-MXX","COMHMXX") , -8 ) ) / 8;
> >
> >COMHX09:= ( FmlVar("COMH-MXX","COMHMXX") - Ref(
> >FmlVar("COMH-MXX","COMHMXX") , -9 ) ) / 9;
> >COMHX09;
> >
> >The only thing missing is COMHX01 and COMHX02. When running, it works
> >sporadically or not at all. I cut and pasted COMHX01 and COMHX02 into
> >their
> >own indicators (which is how I tested them in the first place) and they
> >work
> >fine. I'm now moving all of these back into individual indicators. I
have
> >no idea why this didn't work since I only had 9 variables. The COMX09;
> >entry was strictly for plotting purposes so I could check out the
detailed
> >results (remember, I'm doing a conversion here and need to verify
> >calculations to 4 decimal places).
> >
> >After verifying my calculation results, I delete the final COMHX09
plotting
> >entry. Anyway, why they don't work when contained in one indicator and
do
> >work when cut and pasted into their own indicator, I have absolutely no
> >idea.
> >
> >Regards,
> >
> >Guy
> >
> >" If at first you don't succeed, skydiving is not for you."
> >
> >
>
> _________________________________________________________________________
> Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.
>
> Share information about yourself, create your own public profile at
> http://profiles.msn.com.
>
>
>
|