Please do not filter the internet. As many others have pointed out this is a very bad idea.

Previously we described a potential transformation that we believe will increase the amount of parallelism available at runtime, and use a constant number of stacks rather than a linear number of stacks with respect to the number of parallel conjunctions in a singly-recursive predicate. Read more to see the underwhelming results and conclusion.

We are already aware from previous tests that Mercury's current runtime parallelism system is not ideal. So the outcome if this test is not very useful and it should be re-evaluated later once the runtime system has been improved. That said the test is still interesting and the outcome was not what I was expecting.

This is part one, in a (probably) three part series. In this post I introduce the problem, the background information and the hypothesis

For the past week (or maybe two) I'd been learning and using CUDA, NVidia's GPGPU programming language. This is used to run arbitrary code on the graphics processor, which is excellent for large independent data-parallel problems.

There have been some performance issues with parallel Mercury programs. We understand the worst of these, but there are those that we don't. Sometimes it's also useful to measure a behavior and be sure that you understand it rather that guessing.

Modern computers execute so quickly that using a system call such as gettimeofday(2) is not satisfactory for two reasons. Firstly gettimeofday(2) doesn't return very high-precision information so it's hard to profile events that are very quick. Secondly because this is a system call it involves a transition into kernelspace and back, which can often be to expensive compared with the event you're trying to measure, it can distort the profiling data.

I discuss and define the purpose and intended audience of future blog entries.

We went skiing last weekend, it was awesome. The weather was excellent on the Friday but it was very foggy and quite cold on the Saturday and Sunday. I was too tired to go out on the Sunday, I probably would have made more of an effort if the weather was fine.

I've been trying to improve the performance of parallel programs because by default it appears to be terrible...

First Post! Announcing the re-awakening of www.bone.id.au. This site should feature blog entries about life, code/computer science, musings and family interests.

Stay tuned.