04-25-2017, 08:17 AM
https://www.extremetech.com/computing/24...-reception
Quote:One issue no one’s happy with is Intel’s decision to limit Optane to Kaby Lake CPUs in 200-series chipsets. SSD caches (and Optane caches) would be most effective if deployed as acceleration in lower-cost systems or older hardware. Intel’s decision to sandbox their platform to only the latest motherboards and CPUs means the computers that could benefit the most from Optane acceleration aren’t eligible to use it.
It’s also telling the platforms Intel shipped for Optane testing literally preclude comparing it with its most logical competitor. As we’ve previously stated, Optane should be compared against SSD drive caching, but the B250 motherboards that Intel provided literally only support Optane caching. If you try to configure the software to cache via SSD instead, it refuses to do so. For that, you need a Z270 motherboard, and that’s one reason we don’t have Optane figures ourselves today. The appropriate configuration to test against is the configuration that should serve as Intel’s primary competitor.
But any comparison of SSD-versus-Optane pricing drives home that this isn’t 2012. Back then, it made sense to buy into SSD cache drives, particularly if you wanted to add SSD-like performance to an existing rig without giving up the much larger capacities of a hard drive. Today, SSD prices have fallen so much that you can buy a 256GB SSD for the same cost as a 32GB Optane cache (~$77). While we aren’t claiming that the performance benefits of an SSD cache are equivalent or exactly matched to those of an Optane cache, look through the slideshow from our 2012 review against any of the reviews above, and you’ll see many of the same performance improvements — with the aforementioned exception of Optane’s low queue depth performance, which really is exceptional compared with SSDs.
SSD cache drives have found limited adoption in low-end desktops and laptops, but the higher power consumption of Optane when in standby (it consumes roughly 1W compared to ~.1W for SSDs) may make it a non-starter in the mobile market. SSD prices may rise this year, so Optane caches may become more compelling, but it’s unclear there’s a great consumer performance argument here just yet.
In and of itself, this isn’t really a problem. It’s easy to forget now, but when the first consumer SSDs began to hit the market, they didn’t exactly cover themselves in glory, either. The earliest SSDs were ludicrously expensive, tiny, and couldn’t even match spinning magnetic media on certain workloads due to poor random read/write performance. Optane clearly has data center capabilities and superior consumer performance may arrive with future generations. But as of today, we agree with Ars Technica’s Peter Bright, who writes:
Quote:Rather than showcasing the new capabilities that 3D XPoint brings to the table, it simply highlights how wretched Intel’s product segmentation is. It’s at best an incremental improvement over SRT, and for the money, most people are probably going to be better off with a plain flash SSD than a hybrid disk anyway. 3D XPoint may yet turn out to be something good, perhaps even something world-changing. But this ain’t it.
Not everyone calls Optane out quite that harshly — Hot Hardware allows that the caching solution “may make sense” in certain configurations, but virtually all of the review coverage agrees that limited storage options and Intel’s segmentation decisions sharply limit Optane’s addressable market. Intel’s decision to ship a test platform that prohibited the most logical comparison point to make against its own hardware also doesn’t say much for the company’s confidence in its own performance.