www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Congrats to D / Vibe-D -- techempower Round 15

reply Benny <benny.luypaert rhysoft.com> writes:
https://www.techempower.com/benchmarks/#section=data-r15&hw=ph&test=plaintext
https://www.techempower.com/benchmarks/#section=data-r14&hw=ph&test=json

D is finally showing up in the upper and middle parts of the 
tests. In some tests Vibe-D ( with LDC ) is showing a 10 to 80 
times improvement compared to Round 14 ( where D was stuck at the 
bottom of the tests ).

Plain Text:

Round 15: vibed-ldc..1,200,034
Round 15: vibed........987,534
Round 14: vibed-ldc ....15,342
Round 14: vibed.........10,657

Json:

Round 15: vibed.......206,136
Round 15: vibed-ldc...126,668
Round 14: vibed-ldc....11,813
Round 14: vibed.........8,415

It does not need to be negative postings all the time. Now a 
smart person might use this data and analyse it into a blog, hint 
hint ...
Feb 15 2018
next sibling parent reply Daniel Kozak <kozzi11 gmail.com> writes:
Unfortunately they have use wrong results.  They have selected run from
31.1.2018. But If they use the other one from february, it would be better.
But still is nice

Dne 15. 2. 2018 8:05 odp. napsal u=C5=BEivatel "Benny via Digitalmars-d" <
digitalmars-d puremagic.com>:


 https://www.techempower.com/benchmarks/#section=3Ddata-r15&hw=3D
 ph&test=3Dplaintext
 https://www.techempower.com/benchmarks/#section=3Ddata-r14&hw=3Dph&test=


=3Djson

 D is finally showing up in the upper and middle parts of the tests. In
 some tests Vibe-D ( with LDC ) is showing a 10 to 80 times improvement
 compared to Round 14 ( where D was stuck at the bottom of the tests ).

 Plain Text:

 Round 15: vibed-ldc..1,200,034
 Round 15: vibed........987,534
 Round 14: vibed-ldc ....15,342
 Round 14: vibed.........10,657

 Json:

 Round 15: vibed.......206,136
 Round 15: vibed-ldc...126,668
 Round 14: vibed-ldc....11,813
 Round 14: vibed.........8,415

 It does not need to be negative postings all the time. Now a smart person
 might use this data and analyse it into a blog, hint hint ...
Feb 15 2018
parent reply Ali <fakeemail example.com> writes:
On Thursday, 15 February 2018 at 20:43:41 UTC, Daniel Kozak wrote:

 Unfortunately they have use wrong results.  They have selected 
 run from 31.1.2018. But If they use the other one from 
 february, it would be better. But still is nice



Can you post the numbers from Feb, for us here? :)
Feb 15 2018
parent reply Petar Kirov [ZombineDev] <petar.p.kirov gmail.com> writes:
On Thursday, 15 February 2018 at 20:59:53 UTC, Ali wrote:

 On Thursday, 15 February 2018 at 20:43:41 UTC, Daniel Kozak 
 wrote:

 Unfortunately they have use wrong results.  They have selected 
 run from 31.1.2018. But If they use the other one from 
 february, it would be better. But still is nice



 Can you post the numbers from Feb, for us here? :)


My guess is that newer results can be found here: 
https://tfb-status.techempower.com

(Sorry for not posting the exact results, but I'm on the phone. I 
think we need to run some of their scripts to generate a nice 
graph from the json.)
Feb 15 2018
parent Daniel Kozak <kozzi11 gmail.com> writes:
https://tfb-status.techempower.com/raw/results.2018-02-09-10-22-34-441.json

On Fri, Feb 16, 2018 at 1:01 AM, Petar via Digitalmars-d <
digitalmars-d puremagic.com> wrote:


 On Thursday, 15 February 2018 at 20:59:53 UTC, Ali wrote:


 On Thursday, 15 February 2018 at 20:43:41 UTC, Daniel Kozak wrote:


 Unfortunately they have use wrong results.  They have selected run from
 31.1.2018. But If they use the other one from february, it would be better.
 But still is nice


 Can you post the numbers from Feb, for us here? :)


 My guess is that newer results can be found here:
 https://tfb-status.techempower.com

 (Sorry for not posting the exact results, but I'm on the phone. I think we
 need to run some of their scripts to generate a nice graph from the json.)
Feb 16 2018
prev sibling parent reply Johan Engelen <j j.nl> writes:
On Thursday, 15 February 2018 at 19:04:53 UTC, Benny wrote:

 https://www.techempower.com/benchmarks/#section=data-r15&hw=ph&test=plaintext
 https://www.techempower.com/benchmarks/#section=data-r14&hw=ph&test=json

 D is finally showing up in the upper and middle parts of the 
 tests. In some tests Vibe-D ( with LDC ) is showing a 10 to 80 
 times improvement compared to Round 14 ( where D was stuck at 
 the bottom of the tests ).


Does anybody know what exactly has changed to explain this large 
performance difference?
Feb 16 2018
parent Daniel Kozak <kozzi11 gmail.com> writes:
Hard to say, but I believe one of reasons is VibeManualMemoryManagement,
another one would be same fixes in vibe-core and eventcore
Feb 16 2018