Dlang on 4chan

http://boards.4chan.org/g/thread/44196390/dlang

On 09/16/2014 07:32 PM, Andrei Alexandrescu wrote:
>
> I agree that C and D should be enough. Perhaps C++ and one more near the
> top (Ada, Fortran) would be good for context.
>
> Who wants to do this? Isaac made his setup publicly available.

There was actually someone working on this a year ago or so, but I can't find the github project and don't remember who it was.

On Tuesday, 16 September 2014 at 17:32:39 UTC, Andrei Alexandrescu wrote:
> On 9/16/14, 9:44 AM, Kagamin wrote:
>> I'd say, run the damned benchmark for C and D. C would setup performance
>> scale. What would be interesting is to see, how compiler switches affect
>> performance, especially assert vs release mode and bounds checking on/off.
>
> I agree that C and D should be enough. Perhaps C++ and one more near the top (Ada, Fortran) would be good for context.
>
> Who wants to do this? Isaac made his setup publicly available.

I'll take a stab at it. Will give me something to do on my commute :-)  (assuming his scripts work, or can be made to work on OS X).

On 16.9.2014 20:07, Anonymous via Digitalmars-d wrote:
> Dlang on 4chan
> 
> http://boards.4chan.org/g/thread/44196390/dlang

Yeah, and the discussion is just in line with typical 4chan discussions :-)

A1) Andrei is fucking hot and he's not russian

A2) @A1: >Andrei will never be your husbando
Why bother living?

On Tuesday, 16 September 2014 at 21:21:08 UTC, Martin Drasar via
Digitalmars-d wrote:
> On 16.9.2014 20:07, Anonymous via Digitalmars-d wrote:
>> Dlang on 4chan
>> 
>> http://boards.4chan.org/g/thread/44196390/dlang
>
> Yeah, and the discussion is just in line with typical 4chan discussions :-)
>
> A1) Andrei is fucking hot and he's not russian
>
> A2) @A1: >Andrei will never be your husbando
> Why bother living?

Also:

A) GC bad!  I can manage memory myself, and multithreading is
child's-play - people who use D must be slow and stupid...

*snort*  Ok, you and your delusions of competence are excused
from the conversation now...

On Tuesday, 16 September 2014 at 21:21:08 UTC, Martin Drasar via Digitalmars-d wrote:
> On 16.9.2014 20:07, Anonymous via Digitalmars-d wrote:
>> Dlang on 4chan
>> 
>> http://boards.4chan.org/g/thread/44196390/dlang
>
> Yeah, and the discussion is just in line with typical 4chan discussions :-)
>
> A1) Andrei is fucking hot and he's not russian
>
> A2) @A1: >Andrei will never be your husbando
> Why bother living?

Those ones gave me a laugh.

On Tuesday, 16 September 2014 at 21:04:59 UTC, Peter Alexander wrote:
-snip-
> I'll take a stab at it. Will give me something to do on my commute :-)  (assuming his scripts work, or can be made to work on OS X).

It'll be interesting to see which linux stuff is missing:

-- without libgtop2 you could still get cpu and elapsed times (but not resident memory or CPU load)

-- without highlight you could still get gzip source code size (but the source would include comments and whitespace)


When you have questions, please ask in the benchmarks game discussion forum -- http://benchmarksgame.alioth.debian.org/play.html#misc

On Tuesday, 16 September 2014 at 22:26:48 UTC, Isaac Gouy wrote:
> On Tuesday, 16 September 2014 at 21:04:59 UTC, Peter Alexander wrote:
> -snip-
>> I'll take a stab at it. Will give me something to do on my commute :-)  (assuming his scripts work, or can be made to work on OS X).
>
> It'll be interesting to see which linux stuff is missing:
>
> -- without libgtop2 you could still get cpu and elapsed times (but not resident memory or CPU load)
>
> -- without highlight you could still get gzip source code size (but the source would include comments and whitespace)
>
>
> When you have questions, please ask in the benchmarks game discussion forum -- http://benchmarksgame.alioth.debian.org/play.html#misc

Thanks Isaac.

I think we can live without the resident memory, CPU load, and source size for now. I'll focus on getting some CPU time benchmarks first.

On 9/16/14, 2:04 PM, Peter Alexander wrote:
> On Tuesday, 16 September 2014 at 17:32:39 UTC, Andrei Alexandrescu wrote:
>> On 9/16/14, 9:44 AM, Kagamin wrote:
>>> I'd say, run the damned benchmark for C and D. C would setup performance
>>> scale. What would be interesting is to see, how compiler switches affect
>>> performance, especially assert vs release mode and bounds checking
>>> on/off.
>>
>> I agree that C and D should be enough. Perhaps C++ and one more near
>> the top (Ada, Fortran) would be good for context.
>>
>> Who wants to do this? Isaac made his setup publicly available.
>
> I'll take a stab at it. Will give me something to do on my commute :-)
> (assuming his scripts work, or can be made to work on OS X).

Fantastic. Thanks! Created this just for you:

https://issues.dlang.org/show_bug.cgi?id=13487


Andrei

On Tue, 2014-09-16 at 21:04 +0000, Peter Alexander via Digitalmars-d
wrote:
[…]
> I'll take a stab at it. Will give me something to do on my commute :-)  (assuming his scripts work, or can be made to work on OS X).

I started doing something similar for Groovy a while back (with @CompileStatic, Groovy should now be as fast as Java for all the benchmarks) but ran out of cycles and had to put it on the back burner. If I could track and chip in with a D variant, mayhap this will help get the Groovy/Java version back on track.

I would run the codes on my ancient dual Xeon workstation, so slow for sequential but excellent for dealing with scaling issues. I could easily run D codes as well as the C, C++, Fortran, Java and Groovy ones.

For D I would suggest C, C++ and Fortran as the comparison languages. Using the codes Isaac uses would seem entirely appropriate for this, then D is the only variable (other than the machine used for the tests). For the Groovy version my intention had been to use Isaac's C++ and Java as the comparison languages.

I had always been intending to publish the results via my own server by adding an extra virtual address to my Apache instance, however if there is a possibility of joining forces with a D activity "in the Cloud", I'd be happy to do that.

-- 
Russel. ============================================================================= Dr Russel Winder      t: +44 20 7585 2200   voip: sip:russel.winder@ekiga.net 41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel@winder.org.uk London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder