Processor Use w/MCX?

[Bruce Kanzelmeyer]

All right--here goes. I ran a little experiment this morning and I'm sharing the results, as it raises a question for me regarding Mastercam's use of multiprocessors. I am running a BOXX computer with dual opteron 285 processors, 4 Gigs of RAM, XP Pro and I just setup a Ram disk of 1500 MB and set my Mastercam Temp and Regen files on the RAM disk.

 

Open up Windows Task Manager and look at Performance tab.

 

When I open Instance 1 of Mastercam and calculate a restmill operation, for example, the CPU usage jumps to 25% and about 2/3 of the activity shows up on Processor 3 and 1/3 on processor 4, with less than 10% on processor 1

 

When I open up Instance 2 of MCX and have it regenerate an large multisurface 5-axis toolpath, the CPU usage jumps to 50%. P1 runs about 50%, P3 and 4 are about 75%.

 

If I open a window of Solidworks and rotate a model around, the CPU Usage jumps to 75-77%, P1 goes to about 75%, P2 jumps to 50%, P3and P4 are 80-90%.

 

So it appears that XP automatically dispurses processor demand over the array by some predetermined algorithm and that setting affinity (as James suggested) may actually limit that. Also then, I am wondering why Mastercam limits itself to 25% of processing capability when it appears that XP actually spreads it around to all available processors?

[Bruce Kanzelmeyer]

Interesting that with 1 instance of MCX2 open, toolpath calculations (25% CPU Usage)are allocated about 60% to Processor 3 and 30% to P4 and about 10% shared between P1 and P2. However, when verifying same toolpath almost 70-80% of processing goes to P4 with 10-20% on P3 with less than 10% on P1 andP2.

 

Then when a 2nd instance of Mastercam is opened and same kind of toolpath(waterline) is calculating, we now have 50% CPU usage with P1 at 60%, P2 at 40%, P3 at 70% and P4 at 30%--so the 2 threads are spread about equally between the 4 processors.

 

Then while still calculating waterline toolpath on Instance 2, if we verify the toolpath created on Instance 1 the processing shifts to almost 90% on P3 and P4 with P1 and 2 carrying less than 10% of the load. Its just very interesting.

 

I'm also still waiting for someone to suggest why each instance of Mastercam utilizes only 25% of this computers processing power.

[3rdstring]

The software has to support hyperthreading for one instance of an app to use multiple processors. So if you have 1 instance running you use 1 of the 4 available processors or 25%.

[Bruce Kanzelmeyer]

Chuck, but in reality, one instance is operating on all 4 processors now, about half on 3 and 4 each, and just a little on 1 and 2--if I am to believe the performance tab of the Task Manager.

[Troy Gossett Indepak Inc.]

It appears that that programs are loaded in P1 and Math intensive calculations are pused to the back or P3-P4 processors to ensure the P1-P2 processors will be available for any new apps that are opened. So apps fill from the front processors to back and Math is handled from back to front. This would seem to be comming from Windows as MCAM just doesn't support/control processor managment.

[William Grizwald]

Bruce, I found on an NT box a few years back (for Vericut anyways) that 1&2 were pegged, 3 was about 3/4 and 4 was about 50%.

 

--

Bill

[MIL-TFP-41]

quote:

---

So it appears that XP automatically dispurses processor demand over the array by some predetermined algorithm and that setting affinity (as James suggested) may actually limit that.

---

It is much better to let the OS handle affinity.

 

From Wikipedia...

 

quote:

---

 

Programs running on SMP (multi-core) systems do, however, experience a performance increase even when they have been written for uniprocessor (single core)systems. This is because hardware interrupts that usually suspend program execution while the kernel handles them can run on an idle processor instead. The effect in most applications is not so much a performance increase as the appearance that the program is running much more smoothly.

---

This explains the load going across your processors. Ideally what I would like to see is some kind of benchmarking standard for mastercam, so users can compare configurations & see what is best.

[jctjct]

quote:

---

Ideally what I would like to see is some kind of benchmarking standard for mastercam, so users can compare configurations & see what is best.

---

This sounds like an excellent idea.

[Jaz]

I run dual proccesors and gained a little performance by turning off hyperthreading. Now my computer looks like it has two processors instead of splitting them and looking like four. I don't know exacly how much I gained. We have discussed this stuff before, do a search for 'hyperthread' or '64 bit' or 'dual processor" to get more info.

[Takashi Sato from Ai Solutions.]

hi

 

I made a chook for switching cpus used by mastercam process.

I dont sure if there is an adavantage from using it though..

 

http://ai-sols.sytes.net/download/ProcessPriority.zip

 

red icon is to set the current process priority above normal, and set the other mastercam processs idle.

green icon is to set all mastercam processes normal priority.

 

0 is CPU0,1 is CPU1 ,2 is CPU2 and 3 is CPU3,, if you have 4 cores.

 

currently I only have 2 core PC, so I have not checked 3rd and 4th cores..