Well, I continue telling you about the latest and greatest innovations demonstrated today at the IDF showfloor. And to tell the truth there were quite a few things worth telling you about.
I would like to begin with the news from one of the leading graphics chip developers: NVIDIA Corporation.
During the presentation NVIDIA made special emphasis on the new technologies implemented in the mobile field. We have already witnessed new announcements in the mobile graphics market, however, so far there have been multiple solutions requiring multiple designs, which prevented the solutions from getting really widely spread.
Now the unified design for the PCI Express mobile graphics solution is coming, which definitely has a number of indisputable benefits compared to the individual designs they introduced for every new mobile graphics chip. The snapshot from the presentation specifies this point:
As I have already said, the unified design now supported by NVIDIA will find its way in three major mobile systems segments: thin and light notebooks, mainstream notebooks, and high-performance notebooks (close to desktop replacements). For all the solutions NVIDIA introduces single 230pin chip supporting PCI Express data transfer interface:
As for the desktop and workstation markets NVIDIA also revealed a little more details about the progress they made here. Namely the company showcased the implementation of the SLI technology (Scalable Link Interface), which basically implies that the PCI Express system works with two GPUs (two graphics cards based on the NVIDIA GPUs), which are dynamically loaded with work. If the load is not even then the task in progress and the GPU status is analyzed and the load for each particular GPU is adjusted accordingly using various methods including inter and intra-frame scalability algorythms. The photos below show a running system with the SLI technology from NVIDIA:
You should have noticed a hardware bridge chip connecting the two cards together. According to NVIDIA’s partners this chip should be of a certain fixed width so that the distance between the cards didn’t change, because the bandwidth of the bus implemented in this bridge chip and connecting the two cards into a solid solution is very sensitive to the physical width changes.