Develop on your Notebook with GeForce, Deploy on Tesla

razerblade_cudaThere’s a new post over on the NVIDIA Corporate Blog by my colleague Mark Ebersole about the latest line laptops powered by new GeForce 700-series GPUs. As Mark explains, the GeForce 700 series (GT 730M, GT 735M, and GT 740M), powered by the low-power GK208 GPU has the latest compute features of the Tesla K20 (powered by the GK110 GPU), including:

  • CUDA Dynamic Parallelism, which enables the CUDA runtime API inside device code, so that threads running on the GPU can launch other kernels, call cudaMemcpy, create streams and events, and synchronize the device;
  • Hyper-Q for CUDA Streams, which improves the efficiency and performance of concurrent kernels running on a single GPU;
  • the SHFL warp shuffle instruction, which enables threads in the same warp to communicate directly; and
  • up to 255 registers per thread (increased from 63 in the Fermi architecture), which can reduce bottlenecks caused by spilling registers to off-chip memory.

The availability of the latest GPU architecture on low-cost, highly portable laptops makes it possible to develop CUDA code that uses the latest performance features for deployment on high-end Tesla GPUs. Check out Mark’s blog post for more information.

∥∀

About Mark Harris

Mark is Chief Technologist for GPU Computing Software at NVIDIA. Mark has fifteen years of experience developing software for GPUs, ranging from graphics and games, to physically-based simulation, to parallel algorithms and high-performance computing. Mark has been using GPUs for general-purpose computing since before they even supported floating point arithmetic. While a Ph.D. student at UNC he recognized this nascent trend and coined a name for it: GPGPU (General-Purpose computing on Graphics Processing Units), and started GPGPU.org to provide a forum for those working in the field to share and discuss their work. Follow @harrism on Twitter
  • Jed Brewer

    This sounds fantastic! Though, until I started trying to find out when cc3.5 notebook cards were coming out, I was completely unaware of this, as the compute capabilities for this family of cards is listed at 3.0: https://developer.nvidia.com/cuda-gpus.
    I’m guessing this is just a typo?

    • Tim Harris

      No its not a typo on the compute capability page. This article is wrong unfortunately. There are no 3.5 devices for notebooks. I have a GTX 780M and it is definitely only a 3.0 device.

      • http://www.markmark.net/ Mark Harris

        I checked on this. I’m told that any non-ASUS GeForce GT 730M is a GK208, which is SM 3.5. So the article is not wrong.