Vidcard Pegging my CPU?

Discussion in 'Hardware' started by Nereus, May 17, 2005.

  1. Nereus

    Nereus

    I am running a 2.4 ghz p4 with 1 gig ram.
    I have 4 crt's running on a nVidia Quadro 400 nvs and 1 Lcd running on a nVidia GeForce4 Ti 4200

    I recently (3months ago) created a custom index on esignal and i started having problems with the charts freezing up during hot moments in market action ie; economic reports, the Whitehouse Evac Scare, Ford & GM Junk news, etc.. I started investigating what is causing this... When I close the custom index charts.. the problem goes away... (these index charts are crucial to my successful trading).

    I have followed my cpu usage closely for over a year and realized that during this freeze problem, my cpu gets pegged to 100% usage... (BTW... my memory rarely reaches a usage rate over 380mb and my pagefile is set to 3gig.)

    The interesting part is... when my four monitors freeze up... the lcd (with the broker, frontend, mirc, excel) keeps on trucking... no stalls or freezes.

    This got me thinking... the vid card on the 1 Lcd is 128 mb card.
    The quad card on the 4 crt's is 64mb card.

    My questions.are...

    1) Is the 64mb card being divvied up to the 4 monitors (16 mb each) thus creating a overtaxing of my cpu because the 16mb cant handle the demand?

    2) Would replacing the quad vidcard with a more powerful quad card say... 128 or 256mb card... solve this problem?

    or

    3) Would replacing the quad vidcard with two (2) 128mb or 256mb dual vidcards solve the problem?

    4) Is it more efficient to run several vid cards ie; 3 dual vidcards or less vidcards ie; 1 dual & 1 quad vidcards?

    5) Or is it the cpu.. should i upgrade to a new machine with more powerful P4 or dual Xeon's?


    I do realize that I could try another charting program (which I am currently testing), but I have become very comfortable with my current setup on eSignal and would prefer to continue using eSignal.

    Hopefully these questions and info that might follow will prove useful to others as well.

    Thanks very much.
     
  2. Just split the load between 2 machines. Usually, charting on 1 and broker platform on another.
     
  3. gnome

    gnome

    1. No.
    2. No.
    3. No.
    4. No.
    5. No.

    It's almost certainly an application pegging your CPU. Splitting the load over 2 PCs may or may not solve the problem, but worth a try.
     
  4. Nereus

    Nereus


    1. No.
    2. No.
    3. No.
    4. No.
    5. No.

    "It's almost certainly an application pegging your CPU. Splitting the load over 2 PCs may or may not solve the problem, but worth a try."


    Yes.. I am planning someday to split the load with another puter, but would like to wait a bit and try to indentify the cause of this issue first.

    Gnome, would you have any suggestions as to indentifying the culprit and/or working this problem out?

    Thanks again.
     
  5. Your problem is eSignal. The way you have your custom studies setup is causing eSignal to compute on multiple charts on every tick. I use CQG and in it there is a feature that you can select to only recalculate at the close of every bar or on specific time intervals (every minute, etc) rather than on every tick. Use this option or reduce the number of charts/studies you have going. This almost certainly has nothing to do with your Vid Card.
     
  6. Nereus

    Nereus

    FuturesTrader71


    Thanks very much for your suggestion.. I will look into this option.


    :)
     
  7. Check your processes and see which program is hogging your cpu.

    John
     
  8. gnome

    gnome

    jficquette's suggestion is it. Or, you can load programs one at a time and check Task Manager after each until you identify the culprit. It will be obvious.
     
  9. Nereus

    Nereus

    Yes I have already done this in the task manager and identified eSignal to be the culprit.... and it only happens while I have my custom index charts running.

    When all other standard symbol charts are running... there is no problem what so ever.

    Thanks again.