IO Disk operations blocking the CPU is the possible cause of micro stuttering for many.

Everything about latency. Tips, testing methods, mouse lag, display lag, game engine lag, network lag, whole input lag chain, VSYNC OFF vs VSYNC ON, and more! Input Lag Articles on Blur Busters.
deama
Posts: 370
Joined: 07 Aug 2019, 12:00

Re: IO Disk operations blocking the CPU is the possible cause of micro stuttering for many.

Post by deama » 22 Sep 2020, 00:46

Chief Blur Buster wrote:
21 Sep 2020, 15:37
Windows 10 Disk Optimize already uses an SSD-aware defragmentation algorithm which runs once a month or weekly.
I'm on windows 8.1, does the internal defrag algorithm work fine there?

User avatar
Chief Blur Buster
Site Admin
Posts: 11653
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: IO Disk operations blocking the CPU is the possible cause of micro stuttering for many.

Post by Chief Blur Buster » 22 Sep 2020, 16:24

deama wrote:
22 Sep 2020, 00:46
Chief Blur Buster wrote:
21 Sep 2020, 15:37
Windows 10 Disk Optimize already uses an SSD-aware defragmentation algorithm which runs once a month or weekly.
I'm on windows 8.1, does the internal defrag algorithm work fine there?
No idea. Microsoft Optimize is quite opaque.

You need 3rd party utilities to view the fragmentation state of SSDs.

Most SSD analysis tools no longer visually since even a messy-looking fragmentation (to human eye) has no real performance degradation on SSDs, because there's no seek differences for adjacent and distant blocks, just wait-state-style latencies of beginning a new block read. Interpreting SSD fragmentation visually is much more difficult than it was for HDD fragmentation. A 1 gigabyte game file fragmented into thousands of 4 to 64 kilobyte blocks, can degrade performance much like doing many 4K or 64K reads instead of one continuous 1 gigabyte read -- so massive amounts of fragmentation is usually needed to begin to see visible game stutter on SSDs.

You see all those benchmark scores of SSDs, like a continuous gigabyte and megabyte reads, versus 4K to 64K reads, from queue depth 1 to 32. Reading two halves of a gigabyte is imperceptible difference, but reading tens of thousands of 4K-to-64K fragments definitely is slower.

The read of a super-fragmented gigabyte file on an SSD can perform similarly in benchmarks to 64K QD32 or 4K QD32 -- or occasionally worse (especially if your specific OS/drivers isn't doing a QD32-like optimized-read of all the file fragments). The kind of fragmentation that occurs when you've downloaded lots of source code from many projects from github (like millions of small files similar to 4 kilobytes) to fill your SSD disk space, and/or multiple backup trees of multiple large source-code projects. Then deleted half of the old/unwanted source code later to create lots of non-contiguous free space on your SSD. Then installed a 100 gigabyte game in Steam. Then the game may end up being peppered into a massively-fragmented SSD that fragments your game gigabytes many thousands of times. Then you'll see fragmentation-related performance degradations on your SSD when launching and/or playing big game, especially games with levels getting bigger than your unused RAM (after with background memory usage).

Think along those lines. If you're churning lots of tiny files on mostly-full SSDs, it's healthy performance-wise to do preemptive Optimize/defrag runs before AND after your flagship game install. One would do it for say, a install of Fortnite for professional use or flagship releases such as Cyberpunk 2077 or Half Life Alyx, etc.

Conversely, if you only browse and use Microsoft Word and produce a few files everyday, and often have a fair bit of space, manual SSD optimize usually isn't worthwhile -- Windows already does the weekly or monthly automatically.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

andrelip
Posts: 162
Joined: 21 Mar 2014, 17:50

Re: IO Disk operations blocking the CPU is the possible cause of micro stuttering for many.

Post by andrelip » 23 Sep 2020, 07:53

Just install Auslogic Defrag using the traditional method once and check if it makes your system better. One time will not hurt that much.

Currently, I'm doing that on a regular basis (using the paid version with SSD optimizer to do less writing operations) on my old SSD and I'm fine if it stop working in a few months. Totally worth it

Futuretech
Posts: 35
Joined: 11 Oct 2020, 23:52

Re: IO Disk operations blocking the CPU is the possible cause of micro stuttering for many.

Post by Futuretech » 12 Oct 2020, 01:22

I know this post is a bit old dating back to September. But a lot of you guys seem to have a lot of ignorance when it comes to SSD care and don't understand that this was described back in 2009 when even official people like certain Microsoft bloggers and others mentioned the same thing.

SSD COULD, CAN, AND SHOULD be defragmented by the standard defragmenter found among say Windows or some 3rd party tool.

The primary reason dating back to 2009 this was around the time of Jmicron SSD controllers, beginnings of Sandforce, and a few others before the SSD revolution a few years later coming out with the good SSD. The reason why we defragment SSDs is not because they NEED defragmentation. Their fragmented state offers benefits I'm sure someone can mention the whole parallel performance of fragmented data and the way the SSD and OS interact with said data.

But the primary reason why SSD should be defragmented and it's not because of the older controllers requiring it to improve performance. Was because of "SOFTWARE FRAGMENTATION". Hardware fragmentation isn't a problem SSDs work with fragmented data and it's their primary mode of operation hence why random performance is very important despite the fact that random performance hasn't improved in a number of years.

The problem is not the data method by which the SSD works but the data itself. Software fragmentation is a real thing kinda like bit flipping or bit rot. The fact of the matter is that the Operating System gets old and worn down and so the data needs to be defragmented in order to ensure the data integrity and maximum performance.

As far back as 2009 defragmentation of an SSD. Should be performed twice a year or once every 6 months to maintain maximum performance BECAUSE of software fragmentation. Not the hardware ignore that, that is the SSD main operating way. It's the software i.e. the data you have embedded in the SSD and the Operating System itself.

Post Reply