High Performance Computing

We help optimize software to make the most of the HPC infrastructure equipped with CPUs and various types of accelerators (FPGA, GPU):

  • design, develop, port and optimize software to reduce time to results
  • reduce energy consumption of complex algorithms (Green Computing)
  • improve scalability (optimize for many accelerators within node and for many nodes)

byteLAKE brings:

  • massive experience in building software for supercomputers
  • own software assets to make the most of the hardware
  • various optimization techniques to boost performance
    and create energy efficient solutions
  • expertise in optimizing complex simulations for various architectures
  • vast experience with Intel, NVIDIA and Xilinx (Alveo) technologies

Projects we delivered:

Technical highlights

  • optimized HPC simulation with better data align, blocking, overlapping, streaming, regs. queue etc.
  • tweaked DVFS (dynamic voltage and frequency scaling) and CT (concurrency throttling)
  • MPI – Message Passing Interface
  • SDAccel for Xilinx Alveo
  • CUDA – Compute Unified Device Architecture
  • OpenCL, OpenMP, C, C++, C++-11, Pthreads, and others

byteLAKE's expertise in HPC

We understand heterogeneous computing not only as a type of computations performed using GPU/Intel Xeon Phi/FPGA architecture and having CPU as a management layer. Our company focuses on utilizing all the computing resources to unleash the power of the accelerators (GPU/Intel Xeon Phi/FPGA) and host processors (CPUs). In this approach we optimize the following aspects:

  • selecting the right programming model to a given problem (task parallelism, data parallelism, mixture of these two),
  • providing the right balance between CPUs and GPUs/FPGAs which are processing the algorithm with different performance,
  • optimizing the data transfers between the host memory and accelerators,
  • code adaptation to a variety of computing platforms (i.e. server/desktop GPUs, FPGAs, multicore CPUs),
  • optimization based on three criterion: performance, energy efficiency (Green Computing), accuracy of calculations
  • profiling and analysis:
    • selecting the most time consuming part of the application,
    • diagnosing bottlenecks, load unbalancing, idle resources,
    • rebuilding code by using different methods, techniques, algorithms.

Together with our partners, we deliver complete, end-to-end HPC solutions:

  • hardware components,
  • software stack,
  • fully configured hardware and a highly optimized custom software applications.