The American NVIDIA Corporation is a multinational company (has branches in many countries), specialized in production of graphics processors technologies for computers and small mobile devices (for example: smartphones). The company, based in Santa Clara, California, is main company supplying electronic chips for motherboard chipsets, smart phones' graphic controller, graphics processing units (GPUs), and game consoles.
- Next-Generation NVIDIA® CUDA™ Architecture
Breakthrough NVIDIA CUDA parallel computing architecture, code named Fermi, tightly integrates advanced visualization and compute features delivering application performance that greatly accelerates professional workflows.
Large Framebuffers with Ultra-Fast Bandwidth
Large GPU memory with fast bandwidth for display of complex models and scenes, as well as computation of large datasets. Quadro 5010M (4 GB GDDR5), Quadro 4000M, 3000M (2 GB GDDR5), Quadro 2000M, 1000M (2 GB DDR3).
NVIDIA Parallel DataCache™
Supports a true cache hierarchy combined with on-chip shared memory. L1 and L2 caches drive exceptional throughput, accelerating features such as real-time ray tracing, physics and texture filtering.
VIDIA® Scalable Geometry Engine™
Dramatically improves geometry performance across a broad range of CAD, DCC and medical applications, enabling you to work interactively with models and scenes that are more complex than ever before.
NVIDIA® Scalable Geometry Engine™
Dramatically improves geometry performance across a broad range of CAD, DCC and medical applications, enabling you to work interactively with models and scenes that are more complex than ever before
Error Correction Codes (ECC) Memory
Meets strict requirements for mission critical applications with uncompromised computing accuracy and reliability. Offers protection of data in memory to enhance integrity of results. Available only on Quadro 5010M.
NVIDIA® GigaThread™ Engine
Provides up to 10x faster context switching compared to previous generation architectures, concurrent kernel execution and improved thread block scheduling.
Dual Copy Engines
Enables the highest rates of parallel data processing and concurrent throughput between the GPU and host, accelerating techniques such as ray tracing, color grading and physical simulation. Available only on Quadro 5010M.
Fast 3D Texture Transfer
Fast transfer and manipulation of 3D textures resulting in more interactive visualization of large volumetric datasets.
- Introduction o Why is parallel computing important? o What is CUDA? o Why is it important? · Basic Terminology o Tasks o Sequential, parallel, and concurrent computing · Hardware o Flynn’s classification o CPU, GPU, cell processor, field programmable gate array, cluster, grid o Multi‐core and distributed computing o Instruction level parallelism (ILP) o Thread level parallelism (TLP)
- Concepts in synchronization o Race conditions, synchronization, atomicity, mutual exclusion o Lock, lock contention o Deadlock · Concepts in efficiency o Speed up, Amdahl’s law o Scalability o Granularity.
Hidden parallelism o Latency o Bandwidth
Throughput o Cache, false sharing o Reasoning about parallel algorithms o PRAM o PRAM pseudocode o Algorithm complexity o Data vs. Task parallelism · Parallel programming APIs o OpenMP, OpenCL, MPI, pthreads, CUDA
Activity Conducted under Nvedia
|Sr.No||Name of activity||Duration||No.of Students|
|1||Introduction to CUDA programming||2 days||45|
|2||CUDA programming||2 days||50|
|3||Two Days Workshop On NVIDIA CUDA & OpenACC||1 day||49|