Pages

Subscribe:

Ads 468x60px

Labels

miércoles, 22 de febrero de 2012

GPU CLUSTERS

A GPU cluster is a computer cluster in which each node is equipped with a Graphics Processing Unit (GPU). By harnessing the computational power of modern GPUs via General-Purpose Computing on Graphics Processing Units (GPGPU), very fast calculations can be performed with a GPU cluster.

The hardware classification of GPU clusters fall into two categories: Heterogeneous and Homogeneous.

Heterogeneous
Hardware from both of the major IHV's can be used (ATi and nVidia). Even if different models of the same GPU are used (e.g. 8800GT mixed with 8800GTX) the gpu cluster is considered heterogeneous.
Homogeneous
Every single GPU is of the same hardware class, make, and model. (i.e. a homogeneous cluster comprising 100 8800GTs, all with the same amount of VRAM)Classifying a GPU cluster according to the above semantics largely directs software development on the cluster, as different GPUs have different capabilities that can be utilized.


The Tesla graphics processing unit (GPU) is nVidia's third brand of GPUs. It is based on high-end GPUs from the G80 (and on), as well as the Quadro lineup. Tesla is nVidia's first dedicated General Purpose GPU. The Tesla series takes its name from pioneering Serbian electrical engineer Nikola Tesla.

NVIDIA® Tesla™ 20-series GPU computing processors turn standard PCs and workstations into personal supercomputers. Based on the NVIDIA CUDA™ GPU architecture codenamed "Fermi", Tesla 20-series GPUs feature more than 500 gigaflops of double precision performance, 1 teraflop of single precision performance, ECC memory error protection, and L1 and L2 caches. Tesla 20-series GPGPU processors for workstations deliver cluster level performance right at your desk.



Geforce is the line of GPUs (graphics processors) traditional mass market. Cards are gamers and desktop PC users in general.


Quadro is Nvidia's professional line, oriented to the use of CAD and design programs. Cards are designed for architects, engineers, designers, professionals who need precision, not speed when moving graphics.



Tesla is the line of high-performance GPUs, designed to load graphics, mainly used in supercomputers. In fact there are video cards, have no outlet for connecting a monitor.



Contributions

For this week, I was researching with my partner Juan Carlos  about Beowulf cluster and MPI systems. So we made two blog entries where we share the gathered information.
Also, we implement the John The Ripper application, MD5 passwords are attacked with brute force and dictionary attacks

For the next week, we are beggining to research about Parallel CUDA, so Juan Carlos and me we are working in the construction of a GPU cluster for the next week, some examples and maybe a live execution.

NOMINATIONS:




lunes, 20 de febrero de 2012

Cluster Beowulf

Presentación de componentes principales:

Libro muy interesante te menciona la definición mas correcta de un Cluster Beowulf y las características que debe de tener



// Enlace se puede descargar el libro
http://www.phy.duke.edu/~rgb/Beowulf/beowulf_book.php

Presentación de Componentes principales


Reporte de Actividades [2]

Esta semana realice unas pruebas con el cluster se corrieron algunos Códigos  para verificar como respondía el Cluster y para ver el rendimiento al ejecutar varios procesos y cuanto tiempo tardaba en realizar cálculos y terminar procesos

Se descifro contraseñas por fuerza bruta y Ataques por diccionario de Algunas contraseñas Encriptadas por MD5 usando Jhon The Ripper MPI , un Programa de Criptografia paralelizable.

Escuche algunos compañeros que decían que no se podía hacer un Cluster Con una Live USB deUbuntu y empece a realizar pruebas y es verdad no se pueden instalar los paquetes y busque una solución solo hay que modificar los orígenes del Software y listo se pueden descargar los paquetes necesarios y otro problema es que hay que cambiarle el hostname cada vez que se inicia ya que cuando se reinicio se cambia al nombre del equipo  default y se borran los hostnames de los nodos en /etc/hosts la próxima semana buscare una solución a esto.

Añado capturas de pantallas de los resultados y del Cluster.






Resultados de algunos Códigos:

Números Primos Con 6 procesos


Hola mundo


Calculo de Pi


Calculo de Pi con 30 procesos




Decifrando MD5






Para modificar la Live usb tiene que estar asi como en la imagen