Processor utilization question

[PurdueTech]

I am working on a project for my school and I was wondering if anyone knew the answer to the
following situation:

When you start a program and it uses, for example, 6% of you cpu (like in the cpu usage monitor)
if you open a second program that uses another 6% of your cpu(again like in the cpu usage monitor)
will this slow down the operations of the first program?

If I understand this right, it would not because you would only be using 12% of your cpu.
Therefore you can run one program and it won't slow down other programs unless the cpu usage
is 100%.

Is that guess correct at all?

Please let me know if you don't understand the question and I will try to rephrase it.
Thanks a bunch for any input.

 

[Jeff7181]

I'm no expert, but since no one else responded I will...
I don't believe it would slow the first one down. It's my understanding that because of AMD's Quantispeed architechture, and Intel's Hyper Threading, that it's able to do more than one thing at a single time. However, if both programs require the same part of the CPU, then one will have to wait.
I may be wrong, but that's how I understand it.

 

[Boyne7]

yes actually it will slow down the first one unless it is a p4 3.06 w/ ht enabled or an amd athlon xp as they allow the processor to do more tasks at once

 

[zephyrprime]

AMD's Quantispeed architechture and XP have nothing to do with allowing the processor to run multiple processes at the same time. Only Intel's hyperthreading can allow more than one process to run simultaneously.

The answers provided so far are just wrong. If processor usage is not maxed out, all processes will not run slower in theory. There are exceptions to this though. If your programs become IO limited, then your programs can run slower even if they don't max out the processor. What do I mean when I say io limited? Simple, try downloading two things from the net at the same time. This will obviously make your programs take longer since they have to share your net connection and net connections are easy to max out. This general scenario applies to lots of other IO things to like HD, network, USB, video card, just about anything. But your question makes no mention of IO limitations so I will assume that the question assumes there are no IO limitations.

 

[lifeguard1999]

If two programs compete for the same item at the same time, it will slow down either or both of the programs. The "item" can be the modem, disk drive, or CPU.

Now the slowdown may not be noticeable to you. In one example, a program has 10% CPU utilization. The CPU utilization is really an average use of the CPU over a time period (for eaxample, 1 second). It could be that it fully utilizes the CPU for 0.1 second, which averages out to 10% CPU utilization over 1 second. It could be that it fully utilizes the CPU for 0.05 seconds twice within the 1 second time period the average is taken.

Add in another program (again 10% CPU utilization) and things get a bit more complicated. Do they try to run concurrently during that 0.1 second? Then they slow down. Does program 1 run its 0.1 second first, and program 2 runs in 0.1 second but half a second later? Then they are not in contention and do not slow down.

Throwing in hyperthreading makes it even more complicated. Now two programs may both run concurrently with little or no slowdown if they do not compete for the same parts of the CPU at the same time.

Most programs today cannot use all of the functional units on a CPU all the time. That means that parts of the CPU are unused. These unused parts could be used by a second program. In theory the potential speedup is 100%; two programs can run at the same time without conflict. Practically, the speedup ranges from -10% (slowdown) to 20% (speedup).

Confused yet?

 

[Jeff7181]

If I read it right... the review of HT here on the anandtech site showed that HT reduces performance on things other than multi-threaded, highly CPU intensive tasks, which is shown here in the first benchmark performed by anandtech, and still came up short of the XP2800 in the 2nd one. Also the benchmark shown here shows me that Hyper Threading doesn't do anything for gaming... look at the difference in the score between the 2.53 Ghz and the 2.8 Ghz... now look at the difference between the 2.8 and the 3.06. I'd like to see another test added... a 3.06 with HT disabled. I bet it scores even higher. My take on HT is the same as it was on SSE2... absolutely useless unless the application was designed to take advantage of it. Hopefully Intel will make some new revisions to the technology so that it only uses hyper threading when a program is capable of taking advantage of it. One might say... all the hype about hyperthreading is unfounded.