Rumors are that NVIDIA is gearing up for the release of their next-generation family of GPUs, codenamed “Pascal.” Several media outlets have reported that the upcoming Pascal GPUs will be arriving in the next few months.
Pascal GPUs may get introduced by NVIDIA during the 2016 GPU Technology Conference, which will take place in Silicon Valley from April 4 to 7. Following its debut, the Pascal GPU may also begin shipping as early as June.
Pascal GPUs will be the successor to the Maxwell GPUs, according to a post from NVIDIA’s official blog, three key design features will come along with the Pascal GPUs – 3D Memory, Unified Memory and NVLink.
Pascal GPU will have 3D Memory. By stacking memory chips and placing them adjacent from the GPU will lead to faster communication between the two networks.
With this new design, Pascal GPU will provide a greater bandwidth, double memory capacity and quadruple energy efficiency. The denser layout also means that the Pascal GPU will be able to power smaller devices.
With Unified Memory, the CPU will be able to access the GPU’s memory quicker and easily. Likewise, the GPU will also gain access to the CPU’s memory. Developers won’t have to allocate resources separately.
And finally, with NVLink, NVIDIA has designed a greater link between the CPU and the GPU. This will allow more data to pass through at a given time. To be more specific, the data will be able to flow up to 80GB per second. This is a large improvement over the current GPU’s 16 GB per second.
NVIDIA has also developed the Pascal Module to house the new Pascal GPUs with the NVLink. Compared to today’s GPUs, the Pascal Module will be one-thirds smaller. With a more compact size, the GPU will be able to provide more power even to compact devices.
No official release date for Pascal GPUs has been given out by NVIDIA yet. For more news and updates on this matter, stay tuned.