Post Reply 
WP-34S N-Queens Benchmark Based on Patrice Torchet's BITMAP Solution
10-17-2014, 08:02 PM
Post: #9
RE: WP-34S N-Queens Benchmark Based on Patrice Torchet's BITMAP Solution
(10-16-2014 06:59 PM)Claudio L. Wrote:  Is it still the same algorithm?
I thought the idea was to have the same algorithm ported in all platforms, so they can be directly compared.
I'm sure other calculators will also get a boost if you change the algorithm. I didn't look at the new code in detail, but looks to me like a completely different algorithm.

Is this sufficiently close to the original algorithm so it can be included in the original list? Otherwise will introduce noise into an otherwise very neat collection of results.
There is a problem with the benchmark, the way it is done. It is biaised in more than one way from the beginning.

1) Algorithm chosen: is it not the most powerful. It prevent some calculators from showing their full potential. Ex: hp16C wp34s

2)THe reference source code (BASIC) is poorly written. It use a double array access in the inner loop when a single array access is enough.
- Problem, C code is allowed and C compilers are smart enough to detect that one of the array access is a constant inside the loop and optimize it. Is it fair ?
- Problem, C/RegVars code is directly using a single array access, it is an advantage. Is it fair ?
- Problem, calculators are features poor. hp15c, hp16C and wp34S and many others show poor performance because they stick to the BASIC code and spend their time swaping array pointers in the innier loop. Is it fair ?

You have seen that the wp34s can be 3 times faster, that is huge. Is it fair not to show it in the benchmark?

Every program using the double array access should be rewriten to single array access, just to be fair with C/RegVars code.

“Everything should be made as simple as possible, but no simpler.” Albert Einstein
Find all posts by this user
Quote this message in a reply
Post Reply 

Messages In This Thread
IMPROVED 30% over previous post - iceman - 10-15-2014, 06:39 PM
RE: WP-34S N-Queens Benchmark Based on Patrice Torchet's BITMAP Solution - patrice - 10-17-2014 08:02 PM

User(s) browsing this thread: 1 Guest(s)