Bright 0.5c released

[beachknight]

Bright 05a (with 1cpu) got a slightly higher score than
Spark 03a (with 2cpu).

Very interesting comparison, indeed.

Best,

Bright 0.5c used 2 CPU's in STS tests, so does Spark. (Read the conditions)

Perhaps I should now test gcc compile for Spark to see if it gets significantly better scores.

Oooops.

I thought you used the recent 05a, but that is not the case.
You used 05c version, a private one for some months I suppose.

B05a < S03a < B05c could still be the outcome. Hmmm.

best,

PS: On my Q6600, I am getting better results with S03 gcc compile
than vc one.

[Mike S.]

Thanks. - Do you have a Bright 0.4a result for comparison, from the same conditions?

[beachknight]

Bright 05a (with 1cpu) got a slightly higher score than
Spark 03a (with 2cpu).

Very interesting comparison, indeed.

Best,

Bright 0.5c used 2 CPU's in STS tests, so does Spark. (Read the conditions)

Perhaps I should now test gcc compile for Spark to see if it gets significantly better scores.

Oooops.

I thought you used the recent 05a, but that is not the case.
You used 05c version, a private one for some months I suppose.

B05a < S03a < B05c could still be the outcome. Hmmm.

best,

PS: On my Q6600, I am getting better results with S03 gcc compile
than vc one.

Ooops again. Correction: Latest release is not 05a. it is 05c.

Confused again

Let me have a cigar break.

Best,

[Allard Siemelink]

Now works with firefox too.

Another Question: why is the difference (in bytes) so big ?
spark-0.3a-vc64.exe ........... 366.592
spark-0.3a-vc64-mp.exe .. 3.251.200

spark-0.3a-gcc64.exe .........221.184
spark-0.3a-gcc64-mp.exe .. 229.376 is similar

It seems to be an issue with the Microsoft VC++ 64-bit compiler.
Somehow it generates a large exe for the MP version.
If you'd compress both vc64 files (e.g. using 7zip), you will see
similar small sizes, proving that the additional bytes contain no useful information. Shouldn't hurt though.

GCC is much more consistent, and generates only a slightly larger exe for the additional SMP code, as expected.

[Allard Siemelink]

Bright 05a (with 1cpu) got a slightly higher score than
Spark 03a (with 2cpu).

Very interesting comparison, indeed.

Best,

Bright 0.5c used 2 CPU's in STS tests, so does Spark. (Read the conditions)

Perhaps I should now test gcc compile for Spark to see if it gets significantly better scores.

Probably not, the only difference is 0-10% in speed. Which is faster depends on your CPU.
When running on 2 or more cpu's, some luck is involved too, of course.

[swami]

Thanks. - Do you have a Bright 0.4a result for comparison, from the same conditions?

I didn't test it in the past because it was released way before first version of STS I believe

[fern]

Hi Allard:
How much stronger this 0.5 is?
I know that for me this is moot: the current bright I played yesterday did not give me a chance. No matter if you does not commit mistakes, just a microscopic imprecisition and you are done.
what a program allard!

Fern