Benchmarking a VM - excel & ycruncher

I

infinity4life

n00b
Joined
Dec 23, 2005
Messages
61
I'm a noob to a VM environment, but I'm trying to figure out if this is normal or not.

On my desktop computer if I run ycruncher it calculates to 50M places in 40seconds. (yes, slooooooooooooooooooow, lol)

If I run a financial Excel VBA macro, it takes 4minutes to execute.


If I log onto a virtual machine on the server and run ycruncher, it calculates 50M places in 18 seconds (yes, still sloooooooooooooooooooooooooow, lol).

So ycruncher runs faster on the VM. However, if I run the same Excel VBA macro, it now takes 5minutes to run.

From my limited knowledge i believe VMs will run slower, but shouldn't the faster hardware results from ycruncher offset the slowness?

basically I'm trying to see, if y-cruncher runs 2x as quickly, but Excel runs 25% slower, if this is normal, or pointing to perhaps it not being setup properly?
 
You've given no information about the systems you're using. What are the specs for both your desktop and the server?

How many cores and how much RAM is on the server and how much is dedicated to the VM?
 
Sorry! (I was actually just thinking the same thing)

Desktop is a i5 650 @ 3.2ghz; 8gig ram, 4 cores

I'm unfamiliar with the VM setup; however, there are apparently 6 virtual machines, all the same. According to y-cruncher the vm details for the one I tested on are a x5650, 2.7ghz, 2gig ram, 4core.
 
So the VM lives on a server somewhere. That explains the difference. As to why one test is faster and the other is slower that would depend on exactly what the test is doing and what resources the VM has available. The first test may be storage intensive and the VM lives on a fast SAN...several reasons why you'd see that. Just depends on what the test is doing, but it's not surprising.
 
Would reducing the number of VMs (E.g., going from 6 to 3) produce faster execution times of programs (e.g., Excel macros) then?

I realize this is a broad question, I've just been trying to figure out if perhaps this means there's too many VMs, or if perhaps they're not configured properly (I know it was a major bitch to get them setup and it took something stupid like 3months before they were even usable)
 
infinity4life said:
Would reducing the number of VMs (E.g., going from 6 to 3) produce faster execution times of programs (e.g., Excel macros) then?

I realize this is a broad question, I've just been trying to figure out if perhaps this means there's too many VMs, or if perhaps they're not configured properly (I know it was a major bitch to get them setup and it took something stupid like 3months before they were even usable)
Click to expand...

I'd need a lot more info to even begin diagnosing performance symptoms.
 
Drat, and as you can tell I won't have the info you need, lol!

Perhaps this is too general of a question, but if you had a monte carlo application that is essentially one big loop, for example (in very very simple terms) :

for (i = 0; i<= 100,000,000; i++)
{
for (j = 0; j<= 100,000,000, j++)
}
}

Would a virtual environment ever be appropraite to do that? Or, because that requires intensive horsepower to run quickly, would it always better to have a "phsyical" environment?
 
Depends on the app. Is it multithreaded? If the majority of tasks are single threaded then a VM will be fine. If the app can pound 16 cores at one time then it may be advantageous to keep it physical or limit the number of VMs on the host.
 
You must log in or register to reply here.
Back
Top