SoylentNews is people
SoylentNews
The SemiAccurate web site reports that AMD is putting large solid-state drives directly onto workstation-class graphics processing cards, and calling the result "SSG". Charlie Demerjian wrote in the story that this will be a revolutionary new technology:
The first demo AMD is said to be showing is a use case for movie editing and cleanup on the GPU. What is the issue here you may ask, this is old hat and has been done on the CPU for years. Some GPUs can even assist it without slowing things down in the process, so what does SSG add? How about 8K movie streaming and cleanup in realtime. At 96FPS. Sure you can do this with traditional methods but the best of them will run the same task at 17FPS.
AMD is happy to point out this is a 5.6x speedup or so for the cost of two consumer SSDs. Before SSG, possible but slow. After SSG, fast enough for most users. The impossible, realtime 8K cleanup, is now possible.
[...] It really is the technology of the year and the impossible tasks made possible already are just the tip of the iceberg. SemiAccurate is not joking when we say this is a fundamental game changer for graphics, nothing like this has happened in years.
takyon: It is a proof of concept and is being pitched for Hollywood editing/rendering, the oil & gas industry, and medical imaging.
(Score: 2) by opinionated_science on Monday August 01 2016, @02:30AM
The latency of PCIe is vastly out classed by modern GPU's , so a local memory tier would be an improvement.
Video editing sure. But molecular calculations...;-)
(Score: 1) by Decker-Mage on Monday August 01 2016, @06:42AM
Well, if we're already at the toss a bunch of money stage already, then yeah, put SSD tech on-board. But I'm not so sure that this is a proper approach. We already have sub-micro-second latency with NVMe. Heck, even with NVMe flash arrays off the servers. Having one to play with is literally mind-blowing. Where I have a problem with SSG is that it has a limited life-time. All flash-memory does unless you are using SLC. Is that something we can live with?
-- "To know the world one must construct it." - Pavese
(Score: 4, Insightful) by Immerman on Monday August 01 2016, @02:08PM
For a 5x performance increase? I'd say absolutely. Especially if it's using standard-interface SSDs that can be replaced when they die.
I mean we all know SSDs suck when it comes to long-term durability, and yet we use them anyway because even the relatively modest boost in real-world performance makes life far less annoying, and us more productive. Whereas this sounds like it would radically streamline at least some aspects of video editing. How many hours of a professional video cleanup experts time do you need to save before it pays for an SSD?
(Score: 4, Interesting) by Scruffy Beard 2 on Monday August 01 2016, @02:32PM
Some of the commentors in one of TFAs pointed out that the SSD may be a stand-in for new memory technologies that may be available around the corner.
If those memory technologies turn out to be vapourware, AMD may still have a viable niche product. If they are not vapourware, AMD gets a head-start on their competitors for designing drivers and APIs to handle it.
(Score: 2) by bob_super on Monday August 01 2016, @09:44PM
First, move all the RAM to HBM, then use the HBM to cache the SSD.
At that point, engineers will start agreeing with marketing guys about the Substrate being a real system-on-chip.