We wrote about new AMD processors : the company unveiled a 14nm desktop APU (which said it was going to sell on 12th February), sdÄ > Lila date of issue 2000 and has reduced the generation 1000. At CES 2018, however, AMD had more than that, so now we get what happened last: GPU, roadmap aa detail
Radeon Vega for notebooks
The CES has probably the highest GPU publicity that is nowhere to be labeled “Radeon in of Intel's Kaby Lake- G which is the basis of Vega's architecture with HBM2. According to the official line, it should be a special bundle, but it looks like it could eventually be available separately. AMD has a separate label called Radeon Vega Mobile for thin notebooks.
However, the picture shows that it is a GPU mounted on a shuffle interposer and equipped with a HBM2 chip, the same ¡Configurations like the Intel. We may have the suspicion that it is the same twist, otherwise encapsulated. This variant has its own case attached to an external PCI Express interface; memory HBM2 is not an integrated EMIB technology, but a classic silicon interposer.
The advantage of the mobile Vega with HBM2 could be very compact against the competition. It takes the PCBs substantially longer than the separate GPUs surrounded by GDDR5 (which also have a higher consumption). The motherboard of many ultrabooks is so small that the battery can be as high as possible (or as much as possible), so Radeon Vega Mobile could be attractive for these purposes. AMD also says the case is low, it needs only 1.7 mm (which is the same as Intel's Kaby Lake-G). It was reportedly achieved with a larger interspacer on the substrate. More detailed specifications are not known unless they could be similar to those for Intel, if not the same.
The Roadmap is not yet listed on the market today, nor is it already listed with Radeon's notebooks. aÂ 7nm Vega 20
AMD also made another presentation as a roadmap of graphic architecture. Slajd can be seen next. It is believed that after the current Vega GPUs on the 14nm process, AMD is still planning Vegas for a 7nm process. At the same 7nm process, they can also take over the Navi GPU. They are followed by the unnamed architecture of the Generation Generation, which will be built on the improved 7th process (â € ž7nm + â € œ). This could be for example EUV or other evolutionary improvement. This development should come in the range of 2018 – 2020, and the exact generation of individual generations is not known
But AMD has revealed what is 7nm Vega. According to him, Vega 20 is coming to an end, with the unofficial information already appearing . But this GPU is obviously focused on calculating tips, not on gaming, nor can Radeon work. The Vega 20 can also be powered by FP64 for the HPC / supercomputer, a 4096 bit PCI Express 4.0 memory card slot. At the CES 2018, AMD has called it a machine processor, the GPU belongs to Instinct . Again, unfortunately, we do not know any other parameters. These cards will arrive on the market in the second half of 2018, or they will still be samples.
Zen 2 is ready
Similar to the GPU GPU roadmap, AMD has also been recognized for its processor architecture. Again, this is the plan for the years 2018 – 2020. As you can see, â € žZen + â € œ which is the 12nm processor tag, is currently in the Sample Spectrum before the market, which will be in April. It is not clear whether plus means only a better process or some architectural improvement.
For ProSoft Boost 2 It is interesting that the design of the next Zen 2-star architecture is, according to AMD, complete and will therefore only increase its output. This is a good sign that could open the path of processor availability in the first quarter of 2019 or April of that year (12 months for Ryzene 2000). But the key factor here is probably the 7nm state of GlobalFoundries. AMD also says that Zen 2 has an improved architecture against the current architecture. Also, the development of the next Zen 3 Architecture is based on a plan that does not seem to have a delay. Be careful to use the wording very well. They are tracked, for example, Intel has repeatedly claimed for its 10nm chip, which was to be launched on the market in 2016.
AMD to its various labels, CES has added yet another detail, including the Wraith cooler. The company has brought a quite capable boxing solution that has taken quite a toll. Now a new type of Wraith Prism cooler is going on. TEn will probably be a replacement for today's type of Wraith Max, which is not delivered in boxed packages but only OEM partners (it can also be purchased separately). The name refers to the fact that this evolution of Wraith Max has more light parodies. The ring around the fan now can illuminate the rainbow new is illuminated rotor blades; (19659027) Chladiæ Wraith Prism ” width=”1920″ height=”1080″ >
The innovation is not only paradise. Improvements should also be made to its own cooling function. The place of the conventional conventional basis is now using the concept of direct contact. Basically, there is an exponential flatter and smoothed heatpipe that directly removes heat from the processor heat exchanger. This solution should be considered. The chillers can also have compliant profiles (obviously white and green). The noise should remain at 39 dB. In the slaughterhouse, too, the compatibility with the base plates has been improved. The original Wraith from 2016 collided with some Mini-ITX desktops with graphics cards. So it is possible that the heatpipe was modified in the Wraith Prism so it does not move so far from the socket
The driver adds an adaptive recovery from HDMI 2.1 to today's Radeon
Between details one more favorable message. As you can remember, the HDMI 2.1 for screen output was officially completed last year . And one of its great news is the adaptive restoration of the monitor and television, which is an alternative to FreeSync and G-sync (for Nvidia). The HDMI 2.1 interface itself will require new hardware, specifically access blocks in the GPU. I'm sure the adaptive recovery itself is not. AMD has said that this part of the HDMI 2.1 standard may even add today's graphics cards with driver updates.
This should make it possible for today's Radeon to connect some of the adaptive-to-HDMI-based display adapters to HDMI 2.1 when it comes to the market. Therefore, of course, its resolution and frequency will not exceed the transmission speed limits of HDMI 2.0 or 1.4 that today's GPU delivers. If this will work, there will be a pretty smart add-on for FreeSync / VESA AdaptiveSync supported on DisplayPort in the non-standardized FreeSync format that AMD and many monitors now support for HDMI