Benchmark types

Benchmark types

Can you write here which benchark is 1st, 2nd, 3rd, 4th etc... ?Which one is :

AE12CB-16325737234926730915

and this one:

AE12CB-10353053912364647132

?

When i send my code, i get 5 results instead of 3... Is that good or bad? :D I hope they didnt invented some worse 
benchmark for us with slow programs :D
4 posts / 0 nouveau(x)
Dernière contribution
Reportez-vous à notre Notice d'optimisation pour plus d'informations sur les choix et l'optimisation des performances dans les produits logiciels Intel.
Best Reply

Here is the order of the bench I have passed (= reverse order than that of the mail):

on a 12-cores machine, using 12 worker threads, running benchmark AE12CB-7570939485803595530:
(no timing result on this bench)

on a 12-cores machine, using x worker threads, running benchmark AE12CB-16325737234926730915:
6 and 12 workers

on a 12-cores machine, using x worker threads, running benchmark AE12CB-10353053912364647132:
6 and 12 workers
on a 12-cores HT machine, using x worker threads, running benchmark AE12CB-14929312560780125584:
6 12 and 24 workers

on a 40-cores HT machine, using x worker threads, running benchmark AE12CB-14929312560780125584:
20 and 40 workers

on a 40-cores HT machine, using x worker threads, running benchmark AE12CB-9510636300373457187:

20 and 40 workers

Last tested yesterday. I don't know if there is anyting after the last bench (I guess not).

Same here,

- 1st "benchmark" more or less only checks if the submitted file is valid (Makefile OK, all sources included, etc.).
- 2nd benchmark is on a 12-cores machine using 6/12 worker threads
- 3rd benchmark is on a 12-cores machine using 6/12 worker threads, but with larger inputs
- 4th benchmark is on a 12-cores machine using 6/12/24 worker threads, with larger inputs than benchmark 3
- 5th benchmark is on a 40-cores machine using 20/40 worker threads,
- 6th benchmark is on a 40-cores machine using 20/40 worker threads, with larger inputs than benchmark 5

Hope that clarifies things.

Regards,
Jakob / Team Zubrowka (TU Munich)

I think I have some precision :

...
- 3rd benchmark is on a 12-cores machine using 6/12 worker threads, but with larger reference file
- 4th benchmark is on a 12-cores machine using 6/12/24 worker threads, with larger reference file than benchmark 3
...
- 6th benchmark is on a 40-cores machine using 20/40 worker threads, with larger inputs than benchmark 5 (I suppose it's the benchmark post on the forum)

I tell that because we are stuck on the benchmark 3 and 4 with the index building that is not parallelized on our code yet (CPU stuck at 120%).

Laisser un commentaire

Veuillez ouvrir une session pour ajouter un commentaire. Pas encore membre ? Rejoignez-nous dès aujourd’hui