BLIS Retreat 2018 Program
Contributed talksMonday Sept. 17
| 08:30-09:00 | Breakfast (breakfast tacos, pastries, coffee) | |
| 09:00-09:10 | Welcome | Robert van de Geijn, UT-Austin |
| 09:10-09:40 | Looking back on another year of progress
[ slides ] [ video ] | Field Van Zee, UT-Austin |
| 09:40-09:50 | BLIS Performs [ slides ] [ video ] [ poster ] | Devangi Parikh, UT-Austin |
| 09:50-10:00 | Investig
ations into GEMM on the Cavium Thunder X2 [ video ] | Joe Dobson, ARM |
| 10:00-11:00 | Discussion/Coffee | |
| 11:00-11:10 |
Analysis of BLIS Multithreaded GEMM and HPC workloads
[ slides ] [ video ] |
Pradeep Rao, AMD |
| 11:10-11:20 | Intel MKL Vectorized Compact Routines [ slides ] [ video ] |
Mesut Meterelliyoz, Intel |
| 11:20-11:30 | Intel MKL small matrix multiplication optimizations using JIT compilation
[ slides ] [ video ] |
Arthur Araujo Mitrano, Intel |
| 11:30-11:40 |
CPU INFOrmation library
(Slides and video will not be made available) |
Marat Dukhan, Facebook |
| 11:30-01:20 | Discussion/Lunch | |
| 01:20-01:30 | Other BLAS Precisions | Greg Henry, Intel |
| 01:30-01:40 |
Flyte-MM: A Software Based Sub-Floating Point Precision GEMM
[ slides ] [ video ] | Richard Veras, LSU |
| 01:40-01:50 |
Low-Precision Inference Friendly GEMM and Convolution Library: Interface and Implementation | Daya Khudia, Facebook |
| 01:50-02:00 | Supporting Mixed Domain/Precision in BLIS [ slides ] [ video ] | Field Van Zee, UT-Austin |
| 02:00-03:00 | Discussion/Coffee | |
| 03:00-03:10 |
Sharing a BLISful State
[ slides ] | Maggie Myers, UT-Austin | 03:10-03:20 | Julia, BLIS, and Pedagogy [ slides ] [ video ] Examples on Github: | Sacha Verweij, Stanford and Jane Herriman, Caltech and Julia Computing |
| 03:20-03:30 |
Making Julia more inclusive and accessible
[ slides ] [ video ] | Jane Herriman, Caltech and Julia Computing |
| 03:30-04:15 | Discussion/Coffee | |
| 04:15-04:30 | Outcomes Day 1 | Robert van de Geijn, UT-Austin |
We will need to vacate the room since there is another function at 5:30 for which the room needs to be prepared. Feel free to move to the 5th floor of the south building for further discussions. (Near my office: GDC 5.704.)
5:30 - ??? Beer and food at Crown & Anchor Pub.
Tuesday Sept. 18
| 08:30-09:00 | Breakfast (breakfast tacos, pastries, coffee) | |
| 09:00-09:10 | Lots of things look like matrices [ video ] | Devin Matthews, SMU |
| 09:10-09:20 | Linear algebra in
ab initio computational chemistry
[ slides ] [ video ] | Evgeny Epifanovsky, Q-Chem |
| 09:20-09:30 |
Parallel Nonnegative CP Decomposition of Dense Tensors [ slides ] [ video ] | Koby Hayashi |
| 09:30-09:40 | Notation for distributed-memory parallel tensor computations
(Slides and video will not be made available) | Martin Schatz, Facebook |
| 09:40-10:40 | Discussion/Coffee | |
| 10:40-10:50 | Extending the BLIS analytical model to GPUs [ slides ] (video available by request to author) | Tze Meng Low, CMU |
| 10:50-11:00 | CUTLASS [ video ] | Cris Cecka, NVIDIA |
| 11:00-11:10 | Strassen's Algorithm Reloaded for GPUs [ video ] | Chenhan Yu, UT-Austin/NVIDIA |
| 11:10-11:20 | Multi-GPU GEMM: A cache-based approach
(Slides and video will not be made available) | Paul Springer, NVIDIA |
| 11:20-12:20 | Discussion/Coffee | |
| 12:20-12:30 | Outcomes Day 2 | Robert van de Geijn, UT-Austin |