Skip to main content

⚠️ Please note that this topic or post has been archived. The information contained here may no longer be accurate or up-to-date. ⚠️

Found a way to increase screen preview performance

WPNL
Hi guys,

I don't know if it's covered/mentioned before but I came across this today.
My PC is fitted with a 6700K@4,4Ghz, 64Gb RAM, RAID0 SSD, Windows 10Pro x64, connected to a UHD display and I still found it 'slow'
I got it up to speed by completely turning of the paging file.

Previews are now rendered realtime!

I hope you'll experience the same (and of course be careful testing this)!

[color=#FF0000:1rvgvai8]
EDIT:
Further investigation made me find the real problem I had.
Previews were being generated too slow (several seconds).
I monitored what was going on with disk I/O because I wanted to see the SSD's reach 1GB/s
(which never happened)

What I did see was activity on another (spinning) disk (Disk 2). So I decided to eliminate that.
It turned out Virtual Memory was configured as "Automatically manage paging file size for all drives".
This is probably why Disk 2 was also active.
I then found articles of users (with 16GB+ RAM) having great experiences without a paging file at all.
So I tried that,with success, and noticeable faster rendering speed!

But...
Since Grant was 'stimulating' me to investigate further and not to give bad advice I took a 'step back' and I've now set my (System managed) paging file on the SSD's only. The rendering speed seems the same as without paging file but still a lot faster than it was before! So here we are, two options, one 'safer' than the other but both certainly worth looking into if you're experiencing long loading times for preview generation (and more?)

ps: if you're not familiar with system administration, don't mess with it yourself, instead ask someone with the proper knowledge to have a look![/color:1rvgvai8]
0

Comments

11 comments

Date Votes
  • John Doe
    What's the paging file?
    0
  • SFA
    I've never really looked at RAID (well, many years back before SSDs in a theoretical way) so I am no way expert but I seem to remember that it offers huge speed potential for some types of activity and not so much for others. This and similar disk and memory based requirements are, basically, why manufacturers provide so many variations on all types of disk technology especially ni the commercial server market. No doubt it also help with price differentiation!

    So I ran a search an dthis popped up.

    http://benchmarkreviews.com/35462/samsu ... e-tests/9/

    The conclusions are interesting. Seems like a great idea if you are shifting a lot of very large files and want speed or simple a "disk" that can deal with huge files. The video production world probably loves it.

    C1, however, mostly deals with lots of small files and then tries to work with those predicting where you might be headed next and pre-preparing the way. The may now work quite as effectively perhaps?

    JD,

    The page file is a way of using disk as a memory store going back to e the days when memory cards were small and expensive and swapping data to and from disk made sense in many application with "erratic" use. Not typed anything in there for a while, ok shove it out to disk and free up space for what you are working with then bring the other stuff back later (there is, perhaps, a reduced case for doing so in a system with a lot of memory running, say, MS Word. But for Video editing ... not so sure. For an SSD equipped system .... hmm. It may have some dependency on how the SSD management system is being run? Probably all a dark art.)

    Based on WPNL's post I wonder what exactly goes on when Windows (10,8 or 7? And I assume 64bit?) is managing the Page file(s?) on a multi SSD drive system?

    Hopefully there is a guru somewhere who knows.

    Grant
    0
  • GrahamB3
    One would think the pagefile setting would depend on the version of Windows, as well as the hardware characteristics of each user's system. While one experiencing slow performance may experiment, advising all to turn off (or adjust) pagefile doesn't seem prudent.
    0
  • WPNL
    The essence of my post is Not about RAID.
    Forget about that.
    It's about the performance gain when disabling paging on a system with 8GB + RAM.

    But...
    Small files btw, you mean 40MB raw?
    I know I see the difference when working on the SSD's or another one in the same computer so for me it's not "theoretical". 😉
    With the price of SSD these days I'd recommend this for everyone that's looking for a time saving option...
    0
  • WPNL
    [quote="NNN635465616451391497" wrote:
    One would think the pagefile setting would depend on the version of Windows, as well as the hardware characteristics of each user's system. While one experiencing slow performance may experiment, advising all to turn off (or adjust) pagefile doesn't seem prudent.

    Did you try?
    0
  • SFA
    [quote="WPNL" wrote:
    The essence of my post is Not about RAID.
    Forget about that.
    It's about the performance gain when disabling paging on a system with 8GB + RAM.

    But...
    Small files btw, you mean 40MB raw?
    I know I see the difference when working on the SSD's or another one in the same computer so for me it's not "theoretical". 😉
    With the price of SSD these days I'd recommend this for everyone that's looking for a time saving option...


    For the file size comment - read the linked piece to see whether the conclusion they make agrees with yours.

    Surely it all depends on how the Raid management deals with a paging file? Could be a factor. I'm still digging on that one.

    However in most cases one can think of, except perhaps Video work or dealing with manipulating very large databases or running a lot of memory intensive and retentive applications, a system with 64GB Ram should not really need to be using a page file in "active" mode. It may be set up to do so as some sort of recovery setting I guess but that seems unlikely.

    When I was investigating SSD performance specs one of the interesting points that came up quite a lot is that SSD throughput is relatively high per file but tends to drop off after a while(a few seconds) if processing a lot of "small" files. Small is relative in this context.

    You also have to differentiate between reading and writing to the SSD. Even though the numbers quoted for for recent drives are often similar as max throughput, sustained figures can be very different. An writing is usually much slower than reading. There's a lot of writing in a page file activity and, indeed, in C1 in certain situations.

    So the "theoretical" speeds are certainly available but not necessarily all the time nor as we may think of them from the raw numbers in the specs. Based on what you seem to have experienced with the paging system, perhaps not in all process managed circumstances.

    Perhaps the real question though, irrespective of using SSD or HDD, is what was the paging system doing that interfered so much with system performance? And why?


    Grant
    0
  • WPNL
    You know what, delete this topic.
    I'll keep things like this to myself from now on.
    0
  • John Doe
    🤓
    0
  • SFA
    [quote="WPNL" wrote:
    You know what, delete this topic.
    I'll keep things like this to myself from now on.


    Well, if you can be absolutely certain that the suggestion will have no adverse effects for anyone else no matter what configuration they are running I would encourage you to make your discovery public.

    If you have any doubt on that score your caveat to the suggestion might include some indication of what to look out for as a potential problem indicator.

    Alternatively, as you say, probably best to keep it to yourself just in case a whole load of people who don't currently have problems think they might have, follow your example and cause themselves problems which they may well then load up on someone else.
    0
  • WPNL
    I must be mistaken but I'm pretty sure I saw several "performance related" topics lately.
    That's why I thought it could be helpful for some.

    But you're right, every system is different and different problems / "features".

    Consider the first post as a tip if you will.
    0
  • SFA
    [quote="WPNL" wrote:
    I must be mistaken but I'm pretty sure I saw several "performance related" topics lately. That's why I thought it could be helpful for some.
    But you're right, every system is different and different problems / "features". Consider the first post as a tip if you will.


    That's a great edit to the first post. Thanks for listening to my concerns.

    You're right, there have been some performance related posts although not so many for Windows and what might be a disproportionately bigger number for Macs.

    However that might just be because in C1 world the proportion of Mac users might be higher than Windows users compared to user of other applications for which I have seen some attempts to obtain numbers.


    Grant
    0

Post is closed for comments.