GPU Meeting (04 Oct 2011)

Lionel London (CRA) – GPU Computing in Matlab

  • Matlab Parallel Computing Toolbox (PCT) vs. Jacket
  • PCT allows for multi-cpu and GPU computing. Limited to 12 cores on the local machine. Very high level.
  • GPU computing with PCT requires an nVidia card with v1.3 compute capabilities. GPGPU with Jacket has relaxed requirements but is very expensive ($4k for 5 licenses!)
  • PCT GPU can run external .cu files.
  • Jacket has 10x more CUDA-enabled functions than Matlab.  It’s cluster capable.

Talk: Magnetar Dynamics & Grav. Waves

Kostas Kokkotas (Tübingen)

  • Magnetars exhibit regular gamma-ray flares that are preceded by star quakes or glitches in the pulsations.  The largest flares have L ~ 10^47 erg/s (3 since 1979).
  • No natural explanation for (1) sudden stop of activity P ~ 12 sec and (2) AXP-SNR association but no SGR-SNR association
  • Thompson & Duncan (1990’s) – developed a fireball model that erupts from the NS surface to explain these largest bursts.
  • Can QPOs be explained by global Alfvèn waves?  (see KK publications)
  • NS models, using perturbation theory, with crust and B-fields to predict QPO frequencies.  No crust oscillations at B > 4 x 10^14 G.
  • Gravitational wave amplitudes are unknown and must be calculated with simulations (3D GRMHD).
  • Simulation of B-field instability (Lasky et al. 2011, ApJL) and prediction for GW (Zink et al. 2011)

GPU Meeting (20 Sept 2011)

Matt Kinsey: Porting the 2D Wave Equation to the GPU
  • Optimal number of threads per block is 32*n-1, where n is an integer.  The best performance in the example shown was 63 threads per block.
  • Minimum number of blocks per grid is 32, according to the user’s guide.
  • Every time a kernel is called, the memory needs to be pushed from the CPU to the GPU.  Thus it is optimal to minimize the kernel calls.
  • In the 2D wave equation problem, Matt utilized texture memory to reduce the number of kernels to one.  The memory is indexed in a space-filling curve.  This results in better cache locality.
  • With texture memory, one can take advantage of built-in linear interpolation and boundary conditions.  Texture memory can be addressed in 1D, 2D, and 3D.

New opening for a postdoctoral researcher

Applications are invited for a postdoctoral position in the field of computational and theoretical cosmology at the Center for Relativistic Astrophysics (CRA) in the Georgia Institute of Technology. The successful candidate will work in conjunction with Dr. John Wise. The position is for 2 years with a possible extension to 3 years, depending on funding availability. The appointment can begin at anytime before September 1, 2012. Continue reading New opening for a postdoctoral researcher

AMD 16-core CPUs now shipping

AMD just announced that their 16-core Opteron “Interlagos” processors are now in production.  First they will be shipped to retailers for testing in their platform, and then later in Q4 they will be available to the public.  This is great news for the group because our new 512-core cluster will be based on the Interlagos chips.   We will be placing our order soon and will be waiting for the arrival of the dedicated cluster sometime in December!

Cover image on Science News

Let’s start out the news feed with something big.  A fantastic rendering of one of our simulations made the July 30th cover of Science News!  The associated article does an excellent job in describing the current state of the field.  This beautiful visualization was created by Ralf Kaehler at SLAC.  The featured simulation shows a Population III star at the end of its 3 Myr life.  Most of the surrounding gas is heated to 30,000 K, shown in red, whereas the dense gas associated with nearby halos shield against the stellar radiation and survive this blast of radiation.  The HII region that is created is over-pressurized and drives a 30 km/s shock from the star.  Because the host dark matter halo is only 1 million solar masses, its escape velocity is 3 km/s.  Over 90% of the gas is expelled from the halo, delaying further star formation and having possible consequences on the first galaxy.  This is clearly seen in the expanding, clumpy shell, centered on the star.