SoylentNews

upstart [soylentnews.org] writes:

████ # This file was generated bot-o-matically! Edit at your own risk. ████

Silent changes to Western Digital’s budget SSD may lower speeds by up to 50% [arstechnica.com]:

Western Digital's WD Blue SN550 budget SSD is a well-reviewed popular NVMe device that has regularly shown up on various sites' "best SSD" lists since it was released at the tail-end of 2019. The drive uses a four-lane PCI Express 3.0 interface and was novel for being able to perform better than SATA SSDs for about the same amount of money.

But that may be changing, thanks to quiet behind-the-scenes component changes: Chinese site Expreview [expreview.com] (via Tom's Hardware [tomshardware.com] and ExtremeTech [extremetech.com]) says that a newer version of the drive manufactured in July 2021 was writing data at speeds of about 390MB per second after the drive's cache had filled up. According to Expreview, that's about half the speed of older versions of the SN550; Tom's Hardware measured speeds of about 610MB per second during a sustained write test on the original SN550, so the exact amount of performance degradation may vary. Because both the old and new versions of the SN550 use the same SSD controller, it seems likely that the slowdown is being caused by inferior NAND flash.

Modern SSDs typically pair a large amount of slower NAND flash (for capacity) and a smaller cache of faster flash memory (for the peak speeds advertised on the box). Depending on the SSD, this cache memory is designed to sustain anywhere between a few gigabytes' and a few dozen gigabytes' worth of writes before it has to fall back on the slower flash. Most of the time, you'll never notice the drive slowing down, because you're not going to fill the cache up all the way by using your computer for basic browsing, office work, or even photo editing.

The people who will notice are professional video editors who are regularly exporting, copying, and moving huge 4K video files around all day. It's normal not to get an SSD's advertised peak performance 100 percent of the time, but for a drive to perform substantially worse than it performed in thorough, professional reviews is misleading at best.

Sourcing components from multiple suppliers is common practice when manufacturing computers, phones, tablets, and the parts that go in them; if you can get a thing from multiple places, then slowdowns or interruptions in manufacturing at any one supplier is less likely to disrupt the flow of finished products, and the competition between suppliers can keep your costs down. But just as important is making sure that the components you're sourcing all perform more or less the same, lest you create a "lottery" system where some buyers (and all reviewers) get the "good" version of your product while others get stuck with poorer-than-expected performance.

SSD-makers changing components without updating their model numbers or spec sheets is unfortunately common—Tom's Hardware has a long rundown of examples [tomshardware.com]. Sometimes, end users do benefit from component changes, like when SSD manufacturers change out older flash memory for improved flash with better specs. But in other cases, as with the SN550 and Adata's XPG SG8200 Pro [tomshardware.com], changes made to cut costs or simplify production end up hurting performance instead.

This isn't the first time in recent history that Western Digital has been caught playing fast and loose with specs. Its hard drive division has both fudged the rotational speed [arstechnica.com] of some of its drives and has tried to sell slower-performing Shingled Magnetic Recording (or SMR) drives [arstechnica.com] in its WD Red NAS drive lineup, only to backtrack later and rebrand them [arstechnica.com].

We've reached out to Western Digital to follow up on the original reporting; we'll update this story if we receive any new details.

← Previous story [arstechnica.com]Next story → [arstechnica.com]

from the this-again? dept.

Original Submission