from the now-it-looks-like-the-20st-century dept.
Future generations of virtual reality headsets for PCs could use a single USB Type-C cable for both power and data. That's thanks to a new standardized spec from the VirtualLink Consortium, a group made up of GPU vendors AMD and Nvidia and virtual reality rivals Valve, Microsoft, and Facebook-owned Oculus.
The spec uses the USB Type-C connector's "Alternate Mode" capability to implement different data protocols—such as Thunderbolt 3 data or DisplayPort and HDMI video—over the increasingly common cables, combined with Type-C's support for power delivery. The new headset spec combines four lanes of HBR3 ("high bitrate 3") DisplayPort video (for a total of 32.4 gigabits per second of video data), along with a USB 3.1 generation 2 (10 gigabit per second) data channel for sensors and on-headset cameras, along with 27W of electrical power.
That much video data is sufficient for two 3840×2160 streams at 60 frames per second, or even higher frame rates if Display Stream Compression is also used. Drop the resolution to 2560×1440, and two uncompressed 120 frame per second streams would be possible.
Framerate is too low, and it's not wireless. Lame.
The blog post in particular references a report from Digital Trends which talks about VR sales figures from Amazon, and proceeds to point out a number of ways which the data presented could be misleading.
Several points made by HTC Vive are ones that have also been addressed by VRFocus, as seen in an article about the modern VR cycle, and some comments in the weekly VR vs. article. HTC Vive were not pulling punches right from the very start, evening saying in the introduction: "Analyst reports are in and apparently, it's curtains for Virtual Reality (VR). Pardon us if we're not heeding the alarms. News of the so-called death of VR comes once a year and is greatly exaggerated."
From there, the blog post proceeds in a point-by-point fashion, discussing how early consumer VR was largely driven by smartphone-based devices such as the Samsung Gear VR and Google Cardboard. Not only have these devices been superseded by standalone units like the Oculus Go, which offer a better visual experience, but the promotional offers which were available for phone launches have now long since passed. HTC Vive also point out that PC-based VR companies are yet to release any solid sales figures, and that much of the growth of premium VR has been centered around location-based VR centres, something which the Digital Trends report did not address.
The big change here is that NVIDIA is going to be including even more ray tracing hardware with Turing in order to offer faster and more efficient hardware ray tracing acceleration. New to the Turing architecture is what NVIDIA is calling an RT core, the underpinnings of which we aren't fully informed on at this time, but serve as dedicated ray tracing processors. These processor blocks accelerate both ray-triangle intersection checks and bounding volume hierarchy (BVH) manipulation, the latter being a very popular data structure for storing objects for ray tracing.
NVIDIA is stating that the fastest Turing parts can cast 10 Billion (Giga) rays per second, which compared to the unaccelerated Pascal is a 25x improvement in ray tracing performance.
The Turing architecture also carries over the tensor cores from Volta, and indeed these have even been enhanced over Volta. The tensor cores are an important aspect of multiple NVIDIA initiatives. Along with speeding up ray tracing itself, NVIDIA's other tool in their bag of tricks is to reduce the amount of rays required in a scene by using AI denoising to clean up an image, which is something the tensor cores excel at. Of course that's not the only feature tensor cores are for – NVIDIA's entire AI/neural networking empire is all but built on them – so while not a primary focus for the SIGGRAPH crowd, this also confirms that NVIDIA's most powerful neural networking hardware will be coming to a wider range of GPUs.
New to Turing is support for a wider range of precisions, and as such the potential for significant speedups in workloads that don't require high precisions. On top of Volta's FP16 precision mode, Turing's tensor cores also support INT8 and even INT4 precisions. These are 2x and 4x faster than FP16 respectively, and while NVIDIA's presentation doesn't dive too deep here, I would imagine they're doing something similar to the data packing they use for low-precision operations on the CUDA cores. And without going too deep ourselves here, while reducing the precision of a neural network has diminishing returns – by INT4 we're down to a total of just 16(!) values – there are certain models that really can get away with this very low level of precision. And as a result the lower precision modes, while not always useful, will undoubtedly make some users quite happy at the throughput, especially in inferencing tasks.
Also of note is the introduction of GDDR6 into some GPUs. The NVIDIA Quadro RTX 8000 will come with 24 GB of GDDR6 memory and a total memory bandwidth of 672 GB/s, which compares favorably to previous-generation GPUs featuring High Bandwidth Memory. Turing supports the recently announced VirtualLink. The video encoder block has been updated to include support for 8K H.265/HEVC encoding.
Qualcomm is launching a family of chips that can add incredibly high-speed Wi-Fi — at speeds up to 10 gigabits per second — to phones, laptops, routers, and so on. It's the start of a new generation of this super-fast Wi-Fi standard, but it isn't going to be used to speed up your typical web browsing. And whether it catches on at all remains an open question.
[...] WiGig relies on a connection standard known as 802.11ad, which can hit speeds up to 5 gigabits per second over close to 10 meters, according to Dino Bekis, the head of Qualcomm's mobile and compute connectivity group. Qualcomm's latest chips move WiGig up to a new generation of that wireless standard, called 802.11ay, which Bekis says can reach speeds twice as fast, and can do so up to 100 meter away. The Wi-Fi Alliance says the new standard "increases the peak data rates of WiGig and improves spectrum efficiency and reduces latency."
So why not just use this as normal Wi-Fi, given how fast it gets? Because that range is only line-of-sight — when there's literally nothing in the way between the transmitter and the receiver. This high-speed Wi-Fi is based on millimeter wave radio waves in the 60GHz range. That means it's really fast, but also that it has a very difficult time penetrating obstacles, like a wall. That's a problem if you want a general purpose wireless technology.
[...] It's not clear if this will really catch on, though. While there's definitely room for adoption from VR gamers, the earlier version of this tech has found minimal pickup in its couple years on the market. Asus recently made interesting use of it with the ROG Phone, which is designed for gamers. And Qualcomm says it's working with Facebook to use this tech for its Terragraph project, which wirelessly delivers home internet connections.
Also at Engadget.
Related: AMD Acquires Nitero, a Maker of Wireless Chips for VR Headsets
Intel to Cease Shipments of Current WiGig Products, Focus on WiGig for VR
VirtualLink Consortium Announces USB Type-C Specification for VR Headsets
Wi-Fi Alliance Rebrands Wi-Fi Standards
While display interface standards are slow to move, at the same time their movement is inexorable: monitor resolutions continue to increase, as do refresh rates and color depths, requiring more and more bandwidth to carry signals for the next generation of monitors. Keeping pace with the demand for bandwidth, the DisplayPort standard, the cornerstone of PC display standards, has now been through several revisions since it was first launched over a decade ago. And now this morning the standard is taking its biggest leap yet with the release of the DisplayPort 2.0 specification. Set to offer nearly triple the available bandwidth of DisplayPort 1.4, the new revision of DisplayPort is almost moving a number of previously optional features into the core standard, creating what's in many ways a new baseline for the interface.
The big news here, of course, is raw bandwidth. The current versions of DisplayPort – 1.3 & 1.4 – offer up to 32.4 Gbps of bandwidth – or 25.9 Gbps after overhead – which is enough for a standard 16.7 million color (24-bit) monitor at up to 120Hz, or up to 98Hz for 1 billion+ (30-bit) monitors. This is a lot of bandwidth, but it still isn't enough for the coming generation of monitors, including the likes of Apple's new 6K Pro Display XDR monitor, and of course, 8K monitors. As a result, the need for more display interface bandwidth continues to grow, with these next-generation monitors set to be the tipping point. And all of this is something that the rival HDMI Forum has already prepared for with their own HDMI 2.1 standard.
DisplayPort 2.0, in turn, is shooting for 8K and above. Introducing not just one but a few different bitrate modes, the fastest mode in DisplayPort 2.0 will top out at 80 Gbps of raw bandwidth, about 2.5 times that of DisplayPort 1.3/1.4. Layered on that, DisplayPort 2.0 also introduces a more efficient coding scheme, resulting in much less coding overhead. As a result, the effective bandwidth of the new standard will peak at 77.4 Gbps, with at 2.98x the bandwidth of the previous standard is just a hair under a full trebling of available bandwidth.