an opencl binding - zcl/panama

[Michael Zucchi]

Yep sorry I hadn't checked it in at the time and was messing about with 
branches.

I check multiple different arguments as a general rule to avoid 
over-optimisation that can cloud benchmarks if for example you only ever 
call a method with the same arguments every time (stuff like unrolling 
knowing exact bounds can be quite advantageous).  I haven't looked for a 
while (like years) but previous hotspot versions would for example 
de-optimise lambdas if you used the same one 3 (or >3 i can't recall) 
times in different places which gives very different performance results 
for things like numerical calculations.  I should probably track down 
the dis-assembler plugin and poke around with it again if i'm going to 
do more performance evaluation.


On 28/1/20 7:54 pm, Maurizio Cimadamore wrote:
> Found it:
>
> https://code.zedzone.space/cvs?p=zcl;a=blob;f=src/notzed.zcl.demo/classes/au/notzed/zcl/test/TestMemoryLong.java;h=55e1c5051c7bfdfabfb62c3bbbe21d1e3f90e7aa;hb=refs/heads/foreign-abi 
>
>
> Interesting - you are testing multiple segments with the same 
> benchmark method. I'll try to replicate something similar with our JMH 
> infra.
>
> Maurizio
>
> On 28/01/2020 08:39, Maurizio Cimadamore wrote:
>> If you have the test somewhere I'd love to take a look and maybe port 
>> it on top of JMH. We are looking at these performance potholes now, 
>> so it is a great time to report such issues. 
>