Jose M Perez

[Jose M Perez @Chema_V3]

@Landertxu0 For the 14th problem, you needed to add at least one more node to the cluster (starting another connection and adding it with an ACCEPT from the first connection). Once you had at least 2, you could reduce the cluster size to 3. With 2 of 3, you have the required majority.

[Jose M Perez @Chema_V3]

@friker19 @TuentiEng Tenías una conexión persistente, por lo que no ibas a poder abrir otra sesión en el daemon. No obstante, como el hilo estaba bloqueado por el problema anterior, no detectaba la inactividad/devolvía la contraseña "-.-

[Jose M Perez @Chema_V3]

@very_good_boye @TuentiEng When a user passes the problem, the daemon logs it to persistent storage before sending the actual pass to the user. Due to some external circumstances, the persistent storage filled up during the afternoon, causing threads to block when writing to persistent storage.

[Tuenti Engineering @TuentiEng]

Ready?!

#TuentiChallenge7 is right around the corner! https://t.co/M4EnRnavds

GO! GO! GO!

[Jose M Perez @Chema_V3]

@Landertxu0 @xJakub @TuentiEng I used CUDA for matrix * some time ago and it was a blast if you used it properly (with doubles, though)

[Jose M Perez @Chema_V3]

@Landertxu0 @xJakub @TuentiEng If by reasonable time you mean less than ~10s, I don't think so (obv different/specific HW will be faster)

[Jose M Perez @Chema_V3]

@xJakub @Landertxu0 @TuentiEng I tried to make it difficult (impossible) to find a stationary distribution,so the other approach was easier

[Jose M Perez @Chema_V3]

@xJakub @Landertxu0 @TuentiEng Key parts are using a smart matrix ^ (I think numpy has it OOB) and splitting in its connected components

[Jose M Perez @Chema_V3]

@arasthel92 @Jsevillamol It compiles nicely with g++, you only need a small main that reads from stdin (given code is slow as hell)

[Jose M Perez @Chema_V3]

@arasthel92 @Disruption FFT no era necesario (La complejidad de xcorr se reduce con FFT, pero en este problema no salía rentable)

[Ignacio Martínez @imr58]

Un pequeño snack para el #TuentiChallenge5

[Jose M Perez @Chema_V3]

@Philae2014 @esa Congratulations on this amazing display of engineering :)

[Jose M Perez @Chema_V3]

@movistar_es Parece ser que ya está arreglado, ¡Gracias!

[Jose M Perez @Chema_V3]

@movistar_es Hola, ¿Hay algún problema con la red? Hay IPs inaccesibles si no uso proxy.

[Tuenti Engineering @TuentiEng]

Get to know the winners of the #TuentiChallenge4 on our blog http://t.co/voPdIedk1X Congrats guys, you're the best!;)

[Jose M Perez @Chema_V3]

#TuentiChallenge4 Ya toca dejarlo, me ha enganchado más que los dos anteriores. Cada día me gusta más LUA, facilita mucho algunos problemas.

[Jose M Perez @Chema_V3]

Trust me, I'm a graduate in engineering

[Jose M Perez @Chema_V3]

@TuentiEng The program is required to calculate the results on test time, or can precalculate them? (I'm thinking about challenge 4)