AMD's Ryzen 5000G APUs now have a release date for the DIY market: August 5th. The 8-core Ryzen 7 5700G has a suggested price of $359, while the 6-core Ryzen 5 5600G will be $259.
AMD announced the Radeon RX 6800M, 6700M, and 6600M discrete GPUs for laptops, promising better performance, efficiency, and battery-constrained performance. The Radeon RX 6800M is a 40 compute unit design (equivalent to the Radeon RX 6700 XT on desktop) with 12 GB of VRAM.
AMD biggest announcements were the introduction of FidelityFX Super Resolution (FSR) and the demonstration of a 3D chiplet design. FSR uses a spatial scaling algorithm to upscale game graphics for higher frame rates at a given resolution. The algorithm competes with Nvidia's Deep Learning Super Sampling (DLSS), but will be released as open source and work with some older AMD GPUs, integrated graphics, as well as competing products from Nvidia and Intel (it was shown running on an Nvidia GTX 1060).
AMD CEO Lisa Su also showed off a modified, delidded Ryzen 9 5900X CPU prototype, with "3D V-Cache technology". It was identical to the standard 5900X with the exception of through-silicon via (TSV) stacked L3 cache. This allowed the 5900X prototype to have 192 MB of total L3 cache instead of 64 MB (96 MB per 8-core chiplet). AMD claims it can run games with an average of +15% performance (simply due to the larger cache size), and some version of this will appear in products that are starting production at the end of 2021.
Related: TSMC "5nm", "3nm", Stacked Silicon, and More
AMD confirms Ray-Tracing and VRS in Samsung Exynos RDNA GPU IP [anandtech.com]
The next place you’ll find RDNA 2 will be the high-performance mobile phone market. AMD has partnered with industry leader Samsung to accelerate graphics innovation in the mobile market, and we’re happy to announce we will bring custom graphics IP to Samsung’s next flagship SoC, with ray-tracing and variable rate shading capabilities. We’re really looking forward to Samsung providing more details later this year.
We need a mod for "+1 This is the Coolest Thing I have Ever Seen."
The implications of the stacked cache for the server market is going to be interesting as well. Even if they keep the max chiplet count at 8 (12 is being rumored for 5nm EPYC) it will mean up to 1.5GB of L3 cache for the CPU. This is simply a staggering amount.
AMD Confirms Zen 3 Ryzen CPUs With 3D V-Cache Stack Chiplet Design, Coming Early Next Year Before Zen 4 [wccftech.com]
Something related to 3D stacking (Milan-X) [tomshardware.com] has been rumored to be coming to the server market, but given the 5900X demo and new info, this seems to be consumer focused.
It looks like at least a couple (e.g. 12-16 core) of Zen 3-based halo products will come out in Q1 2022, with the 3D cache. The 15% average gaming performance uplift from extra cache and other changes would be used to counter Intel's Alder Lake. It could be the last gasp for the AM4 socket. Zen 4 Ryzen has likely been delayed to Q3 2022.
I'd love to see what a small machine with the 5700G in it is like :) Something like InWin's Chopin mini-itx case would be perfect for it. Ryzen still feels like alien technology.
It got down to $550 [slickdeals.net] in recent HP Pavillion deals (just $190 over this August retail price for a complete desktop), so a bunch of people have them now.
Rumor has it [videocardz.com] that Rembrandt will top out at 12 RDNA 2 compute units instead of 8 Vega. So a 5700G successor next year could have double the graphics performance (perhaps more), RDNA 2 raytracing support (made more feasible with the addition of FidelityFX Super Resolution), AV1 hardware decode support, and whatever CVML is (it's assumed to be "Computer Vision and Machine Learning" acceleration using the iGPU). Maybe even some additional 3D L3 cache, if that's what the rumored Zen 3+ is. It will also use DDR5 memory which is likely a boon for integrated graphics, and it will support PCIe 4.0 instead of Cezanne's PCIe 3.0. Oh, and USB4 (the lack of which is a sticking point for some people, apparently).
Can you imagine a Beowulf cluster of these?
Ban home supercomputing.