SSD Optimization improves LR speed

Started 3 months ago | Discussions thread
OP soloryb Senior Member • Posts: 2,351
Re: SSD Optimization improves LR speed

kelpdiver wrote:

soloryb wrote:

. But, you seem to have felt a need to interject yourself so I'll try to clarify what the issue is.

that's how forums work, bub. But don't pretend you're not here to argue - it's patently clear in your reply to any remote challenge to authority.

If I've offended you by saying you interjected, then I'm sorry. I should know how these forums work by now. Thanks for taking the time to help me with your insights anyway.

I'm not here to argue with you or anyone else, but I certainly don't agree with everything I've read on this thread. In fact neither do you as evidenced with your comment on how silly it is to defrag a SSD - something I agree with (at least for now under Win 7.)

Adobe certainly cares about the performance (speed) of their products. Just look up some of the millions of complaints regarding PS and LR performance issues, and the multitude of remedies that have been prescribed for improvement.

If they did, they wouldn't have millions, no BILLIONS of complaints. Poor use of multi core. Poor use of the GPU. Feature bloat. And then of course, their subscription model that removed any real incentive to be responsive.

Maybe you're right about Adobe's lack of concern for out welfare. I include myself among those billions you speak of. I don't love Adobe and really detest their subscription model. If and when I get an alternative to PS and LR that I can live with, I'll drop Adobe pronto.

You supply no quantitative information here, which makes it difficult to accept your conclusions. It feels faster or it got slow doesn't do the job. My random 4k read was down 40% on IOPS in crystal mark until I took this action - that's useful data.

I thought I provided all the quantitative info needed. If not, what else do you want to know?

As for my conclusions, I really don't have any definite ones yet - which is why I posted in the first place. I don't know why Defraggler SSD optimizations so significantly increased LR and system performance, other than to suspect that something was preventing the trim command to actually occur - even though trim was definitely enabled for my system. I think that Defraggler's Optimization overrided this and performed an independent trim on each SSD (or got the OS to do it.)

Defragging a desktop SSD is silly because with 10-100k in read iops and a read latency less than .1ms, you're not paying the same penalty for reading a fragmented file as a spinner that can do 100 iops and has a read latency in the 10-20ms range.

I agree.

I never do this but sygnus21 posted a link article that indicates Windows is doing this in some partial way during SSD optimization. I don't think Win 7 can optimize an SSD at all in this way, even if Win 8 and above can. As far as I can tell, Win 7 can just defrag.

https://www.hanselman.com/blog/TheRealAndCompleteStoryDoesWindowsDefragmentYourSSD.aspx

Garbage collection (trim) can be a factor on drives near space exhaustion. This is most easily alleviated with a small level of overprovisioning, and not filling the filesystem to 95% capacity. Are you doing this on your c: partition? OP ensures there are always free blocks to write to and smooths out performance variation on transactions. Anandtech has showed some pretty compelling perf graphs to describe the problem and the efficacy of the remediation.

Trim is on for C: and for my data SSD. Can't be turned on my oldest SSD according to Magician, but I only use it as a scratch disk. I do know what the trim command does and how to ensure it is enabled.

That your firmware was out of date is also rather telling and certainly could have been a contributor. Windows can be a bandaid, but its the samsung firmware and drivers that control the behavior around GC.

That's something that I hadn't checked recently and just fixed but the firmware was only a year out of date. This same problem (and the Defraggler solution) occurred a few months ago. It might be a factor.

Samsung Magician was indicating that trim was on and fsutil behavior query disabledeletenotify at command prompt indicated trim was enabled ('0' value) but my setting of VSS to Manual instead of Auto may have prevented any trim command from being performed by the OS in the first place. At least that's what I now think is possible. I'm going to have to wait and see if it occurs again within a few months time. If it doesn't, then the VSS setting is the culprit. If it occurs again in spit of changing that setting to Auto, then I won't have solved the problem and will revert to periodic Defraggler optimizations as a patch solution.

Thanks again for your help,

soloryb

 soloryb's gear list:soloryb's gear list
Fujifilm FinePix X100 Nikon D800E Nikon AF-S Nikkor 24-70mm f/2.8G ED Nikon AF-S Nikkor 70-200mm f/2.8G ED VR II Nikon AF Fisheye-Nikkor 16mm f/2.8D +9 more
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow