Renders Faceworks Ira demo at 2 to 3W
During SIGGRAPH 2013 conference and exhibition that is all about computer graphics and interactive techniques, Nvidia decided to show off its new Project Logan mobile chip that will be based on Cortex A15 design paired up with Kepler-architecture GPU. It showed a rather neat Faceworks Ira demo previously seen running on Nvidia GTX Titan graphics card and now running on 2-3W reference Logan design platform.
The same Faceworks Ira demo was previously described by Nvidia as big leap forward in capturing and rendering human facial expression in real time. Although it was not that much impressive on a 250W+ 28nm Titan GPU due to its sheer computing power, it is certainly impressive on a Project Logan reference platform that draws 2 to 3W of power and includes a single Kepler SMX as its GPU part.
Since Nvidia got the working sample just three weeks ago, according to reports around the net, it is most likely that we are looking at a 28nm manufacturing process and not 20nm one that Logan will be based on but nevertheless, it does look impressive considering that it will be based on Cortex A15 design and most likely feature the same 4+1 design seen on Tegra SoCs. The GPU will most likely feature a single Kepler SMX which adds up to a total of 192 CUDA cores and while the precise specs and clocks are still not finalized it still packs quite a GPU punch.
According to Nvidia's own slides, the Project Logan with Kepler-based GPU will feature performance that will exceed the one on Playstation 3 and even Nvidia's own desktop 8800 GTX graphics card. When compared to PowerVR SGX 554MP4 GPU that is currently powering the iPad 4, Nvidia expects up to 5.2 times more floating point performance.
Nvidia compared its newest "mobile Kepler" with Geforce GTX Titan and "closest mobile competitor" that reveal that, unlike Tegra GPUs, we will see a fully unified architecture with OpenGL ES 3.0, OpenGL 4.4, DirectX 11, Tessellation and CUDA 5.0 support, all thanks to Kepler.
The Faceworks Ira demo was simplified with lower texture quality but still looks quite impressive at 1080p and 2-3W power usage. Nvidia also showcased the new Logan Island Demo that showed some OpenGL 4.3 features as well as GPUs tessellation capabilities.
Source: Nvidia.com.
The same Faceworks Ira demo was previously described by Nvidia as big leap forward in capturing and rendering human facial expression in real time. Although it was not that much impressive on a 250W+ 28nm Titan GPU due to its sheer computing power, it is certainly impressive on a Project Logan reference platform that draws 2 to 3W of power and includes a single Kepler SMX as its GPU part.
Since Nvidia got the working sample just three weeks ago, according to reports around the net, it is most likely that we are looking at a 28nm manufacturing process and not 20nm one that Logan will be based on but nevertheless, it does look impressive considering that it will be based on Cortex A15 design and most likely feature the same 4+1 design seen on Tegra SoCs. The GPU will most likely feature a single Kepler SMX which adds up to a total of 192 CUDA cores and while the precise specs and clocks are still not finalized it still packs quite a GPU punch.
According to Nvidia's own slides, the Project Logan with Kepler-based GPU will feature performance that will exceed the one on Playstation 3 and even Nvidia's own desktop 8800 GTX graphics card. When compared to PowerVR SGX 554MP4 GPU that is currently powering the iPad 4, Nvidia expects up to 5.2 times more floating point performance.
Nvidia compared its newest "mobile Kepler" with Geforce GTX Titan and "closest mobile competitor" that reveal that, unlike Tegra GPUs, we will see a fully unified architecture with OpenGL ES 3.0, OpenGL 4.4, DirectX 11, Tessellation and CUDA 5.0 support, all thanks to Kepler.
The Faceworks Ira demo was simplified with lower texture quality but still looks quite impressive at 1080p and 2-3W power usage. Nvidia also showcased the new Logan Island Demo that showed some OpenGL 4.3 features as well as GPUs tessellation capabilities.
Source: Nvidia.com.
News by Luca Rocchi and Marc Büchel - German Translation by Paul Görnhardt - Italian Translation by Francesco Daghini
Previous article - Next article
comments powered by Disqus
comments powered by Disqus