BLIS Retreat 2015
Program
Contributed talksMonday Sept. 28
| Morning | POB 2.402 | ||
| 8:30-9:00 | Breakfast (POB 2.402) | ||
| 9:00-10:00 | BLIS/libflame: past, present, and future | SLIDES | Robert van de Geijn |
| 10:00-10:30 | The Current State of BLIS | SLIDES | Field Van Zee |
| 10:30-11:00 | Coffee (POB 2.402) | ||
| 11:00-11:30 | Implementation of Linear Algebra Libraries for Embedded Architectures Using BLIS | SLIDES | Devangi Parikh |
| 11:30-12:00 | BLIS and libflame on the Myriad2 MA2100 |
SLIDES | Cormac Brick |
| 12:00-12:30 | Solving the Small Matrix Linear Algebra Problem | (Contact Greg for slides and/or video.) | Greg Henry |
12:30 - 2 Lunch (GDC 6.302 - Computer Science Faculty Lounge)
| Afternoon | POB 2.402 | ||
| 2:00-2:30 | Generation of SIMD Dense Linear Algebra Kernels with Analytical Models |
SLIDES | Richard Veras |
| 2:30-3:00 | Adding Algorithm-Based Fault Tolerance (ABFT) to BLIS |
SLIDES | Tyler Smith |
| 3:00-3:30 | BLIS-based High-Performance Compute Kernels in N-body Problems |
SLIDES | Chenhan Yu |
| 3:30-4:00 | Coffee (POB 2.402) | ||
| 4:00-4:30 | Exploiting BLIS to Optimize LU with Pivoting |
SLIDES | Xianyi Zhang |
| 4:30-5:00 | Adding Efficient Scheduling Policy to SuperMatrix on Heterogeneous Platforms |
SLIDES
| Jianyu Huang |
| 5:00-5:30 | Linkage Disequilibrium is BLIS | For video and slides contact the speaker. | Tze Meng Low |
Tuesday Sept. 29
| Morning | POB 2.402 | ||
| 8:30-9:00 | Breakfast (POB 2.402) | ||
| 9:00-9:30 | Porting the BLIS micro-kernel to PeachPy | SLIDES | Marat Dukhan |
| 9:30-10:00 | A Systematic Approach for Blocking of Convolutional Neural Networks for Architectures with Hierarchical Memories |
SLIDES | Ardavan Pedram |
| 10:00-10:30 | BLAS for Tensors: What, Why, and How | SLIDES | Devin Matthews |
| 10:30-11:00 | Coffee (POB 2.402) | ||
| 11:00-11:30 | Software implementation of correlated quantum chemistry methods that exploits advanced programming tools and new computer architectures | SLIDES | Evgeny Epifanovsky |
| 11:30-12:00 | Optimizing Tensor Contraction Sequences via Data Movement Lower Bounds | Saday Sadayappan | |
| 12:00-12:30 | ROTE + DxTer = Bigger, Faster CCSD | Martin Schatz | |