Atom Atom feed

Question: Concurrent Programming On Dual-Core Machines

I just have a simple question for all you dual-core machine owners: Has owning and developing on a dual-core machine helped with your concurrent programming?

Let me be specific: What would you miss the most, as a programmer who has to deal with concurrent programming issues, if, through some magic, I took away one of your cores and double the speed of your remaining core?

I know the theoretical answer to the question. But since I don't have a dual-core machine yet, I don't know how everything translates in practice.

Tags :
Responses[6] RSS feed for responses to this blog entry
Posted by Weiqi Gao on April 10, 2008 8:08:54 AM CDT #


Reply

Re: Question: Concurrent Programming On Dual-Core Machines

Comment from Kevin Heifner on April 10, 2008 9:11:01 AM CDT #
Remember back when we worked on the financial application together. We tested on both dual-core and multi-processor machines. We saw a number of issues on the multi-processor machines that we never saw on the dual-core machines. So in terms of testing concurrent programs, it is important to have a multi-processor machine (not just a dual-core machine).
Reply

Re: Question: Concurrent Programming On Dual-Core Machines

Comment from Weiqi Gao on April 10, 2008 9:35:41 AM CDT #

Testing to reveal bugs that will never manifest on a single core is the one thing that I thought of.

Anything else? Anything that people do actively during development?

Reply

Re: Question: Concurrent Programming On Dual-Core Machines

Comment from Guillaume Laforge on April 10, 2008 12:18:06 PM CDT #
Under heavy load, check that all your processors and cores are used at the max. Otherwise, it may mean you've got some big synchronization or lock issues which makes your threads wait for something. Careful synchronization and lock handling is required and under utilization of the CPU(s)/core(s) may show up on those kind of machines.
Reply

Re: Question: Concurrent Programming On Dual-Core Machines

Comment from Eric Burke on April 10, 2008 1:34:03 PM CDT #
I know I'm repeating what Kevin said, but when we were having our Java Concurrency lunches at OCI, we did write some example Java code that worked perfectly on my old single core CPU, but failed consistently on a dual-core processor.
Reply

Re: Question: Concurrent Programming On Dual-Core Machines

Comment from Asgeir S. Nilsen on April 10, 2008 3:03:59 PM CDT #
We develop a suite of applications which are highly concurrent. We develop on a mixture of Windows and Solaris, both single-processor, multiprocessor and dual-core.

On many occasions we have experienced test failures which are bound to number of processing cores, often related to timing between threads. On machines with several cores or processors the execution order indeterminism surprise us regularly.

Reply

Re: Question: Concurrent Programming On Dual-Core Machines

Comment from Paul Brown on April 15, 2008 1:34:02 AM CDT #
My experience with an 8-core box is that it becomes much more difficult (i.e., so far not possible) to saturate the CPU with a single operating system process than with two or four cores. Multicore computing will asymptotically approach distributed computing as the number of cores increases.

Add a comment Send a TrackBack