The problem with multi-processor machines is programming them

Geek Speak » in reply to I think the next big push will be multi-processor chips.

The problem with multi-processor machines is programming them

Gary — 8/10/2005 7:49:14 AM (archived)

Title says it all. Writing efficient multi-threaded code that drives separate processors is really hard to do. People have been horsing with this for years and made very little progress. The machine that I'm using now is an SMP dual processor. However the only time I can get both processors running at 100% is to launch several separate CPU intensive programs. I've never succeeded in getting a single multi-threaded code to make full use of both processors.

Gary

The whole thread (40 posts)

Moore's Law has finally failed ? - SDF - 8/1/2005 10:51:33 AM
- Most computers today are limited by DISK SPEED - David F Mayer - 8/16/2005 4:15:12 PM
- Maybe not! - R.L. - 8/16/2005 7:43:37 AM
- Intel press release on failure to increase processor speed. - SDF - 8/12/2005 1:26:49 PM
  - The next question is: What is the standard for processing speed? - SDF - 8/12/2005 1:34:39 PM
    - Spend extra money building an air conditioned case for your computer. (n/t) - Scott Evil - 8/12/2005 11:28:05 PM
    - Use a believable benchmark - Gary - 8/12/2005 8:35:45 PM
- I think the next big push will be multi-processor chips. - Tom Brosz - 8/9/2005 7:21:05 PM
  - The problem with multi-processor machines is programming them - Gary - 8/10/2005 7:49:14 AM
  - re: I think the next big push will be multi-processor chips. - SDG - 8/9/2005 9:44:44 PM
    - Yech. - Tom Brosz - 8/11/2005 10:26:27 AM
- Why any faster? - Robert - 8/5/2005 5:09:46 PM
  - re: Windows Vista - Scott Evil - 8/11/2005 2:29:17 PM
    - re: Windows Vista - Robert - 8/11/2005 5:13:24 PM
      - Makes Sense - Scott Evil - 8/11/2005 9:59:13 PM
        
        re: Makes Sense - Robert - 8/12/2005 2:46:03 AM
  - re: Why any faster? - SDG - 8/6/2005 12:05:14 AM
    - Keep it simple stupid - Gary - 8/6/2005 11:48:13 PM
      - can't avoid it - SDG - 8/8/2005 5:39:46 PM
- Diminishing returns - SDAI-Tech1 - 8/2/2005 10:19:38 PM
  - von Neuman bottleneck - Gary - 8/3/2005 3:15:40 AM
    - "Couple that with CPU heat flux and it becomes clear that Moore's Law has hit the wall." - R.L. - 8/3/2005 8:02:14 AM
      - Precise point made - RNE - 8/4/2005 5:24:09 PM
  - re: Diminishing returns - SDG - 8/2/2005 11:34:01 PM
- re: Moore's Law is not a continuous curve - R.L. - 8/2/2005 8:41:02 AM
  - punctuated equilibrium, heat, and bear markets - Robert - 8/2/2005 8:29:49 PM
- It's more of a guideline... (n/t) - alcaray - 8/1/2005 11:24:55 PM
- Not the physics, but the markets - Robert - 8/1/2005 10:25:52 PM
  - I'd like to upgrade to a dual core Opteron - Gary - 8/2/2005 5:28:02 AM
    - An interesting reference on heat failure in chips - Tom Brosz - 8/11/2005 10:00:30 PM
    - Fan spring? - SDF - 8/2/2005 6:29:04 AM
      - Fan Spring and CPU cooling? - RNE - 8/2/2005 12:27:55 PM
  - I haven't seen a single ad for a 6 GHz processor (n/t) - SDF - 8/2/2005 4:23:40 AM
    - You will - R.L. - 8/2/2005 8:48:15 AM
      - That would not be a wise bet - Gary - 8/3/2005 3:26:44 AM
        
        :) I was involved in the project - R.L. - 8/3/2005 7:55:00 AM
        
        Fab plants and Rock's Law - Gary - 8/4/2005 4:06:33 AM
        
        A fab plant based on EUVL... - R.L. - 8/7/2005 4:17:04 AM
        
        I should also mention - Gary - 8/3/2005 3:31:37 AM
    - at any price (n/t) - SDF - 8/2/2005 4:25:19 AM

Recent posts

On this day in Geek Speak

The whole thread (40 posts)