What’s new? AMD’s market shares & prices, USB 4 & Intel’s AI CPUs

USB 4 and its compatibility with Thunderbolt 3

The USB Promoter Group announced USB 4 back in March 2019. A good six months later, on September 3rd, the final specification was finally published. According to the USB-IF press release, this update complements the existing USB 3.2 and USB 2.0 to create a "next generation" USB architecture.

As USB 4 is based on Intel's Thunderbolt protocol specification, the bandwidth is doubled, enabling 40 Gb/s, and multiple data and display protocols can be processed simultaneously. The USB Type-C socket, which has established itself as an external display output for host devices since its market launch, also remains with USB 4. So far, so good. But now the big disadvantage: manufacturers are not obliged to implement the functionality of Thunderbolt 3 in their USB 4 specification. In short: USB 4-capable devices are not necessarily compatible with Thunderbolt 3. As not all USB devices can achieve the promised 40 Gb/s, USB 4 devices may have to reduce their speed to adapt to the hardware. USB 4 will be available in speeds of 10, 20 and 40 Gb/s - although you can expect to find cheap and small devices only with the lower speeds.

heise.de also explains the underlying hub of USB 4 in their article. You can read more about intelligent bandwidth sharing and the power supply of USB 4 in this article.

The published specifications can already be downloaded from the USB-IF website. It will probably be the end of 2020 before products with USB 4 are available on the market.

Nervana NNP-T and NNP-I: Intel's AI processors

The topic of artificial intelligence is getting bigger and bigger, that's no secret. In order to meet the demand for special chips in this segment, Intel presented the Nervana processors at the Hot Chips HC31 conference at the end of August. With these, the chip company is competing with Google's Tensor processors, Nvidia's NVDLA and Amazon's AWL Inferentia chips.

The processor combination of the NNP-T and NNP-I chips is primarily intended for the field of machine learning. NNP stands for Nervana Neural Processor, but the chips are also known under the codenames "Spring Crest" and "Spring Hill". In a closed application environment, they are used in complementary ways: the NNP-T chip is intended for training an AI model with big data, while the NNP-I chip is responsible for inferencing, i.e. implementing the training results in the AI workflow. The SoCs are manufactured in Intel's in-house 10 nm process, NNP-T in the 16 nm process.

With the introduction of Nervana, the Xeon CPUs that were previously responsible for AI tasks will be replaced. Although these still offer sufficient power, they are less effective and efficient than Nervana.

Sources: t3n, heise

AMD's market share grows, Ryzen prices fall

We already reported on AMD's latest business figures in our last news. In the second quarter of 2019, both AMD and Nvidia recorded lower sales of graphics cards for desktop PCs, workstations and servers - but Nvidia significantly more than AMD. AMD succeeded in increasing its market share from 22.7 to 32.1 percent.

Looking at the entire GPU market, which includes graphics units integrated into processors, Intel remains the market leader with a 66.9 percent market share. But AMD also scores here: In contrast to Intel and Nvidia, the group was the only one to increase its shipments and thus raise its market share to 17.2 percent.

Source: heise, winfuture

AMD's success is due not least to its successful Ryzen APUs. The Ryzen 3000 processors are already seeing significant price reductions just two months after their launch. The largest ones measure the Ryzen 7 3800X and Ryzen 5 3600X models, whose price-performance ratio was previously considered poor. You can find a price history of the Ryzen 7 3800X on giga.de.

Source: heise

Discover our spo-comm Mini-PCs