SSD Tweaks and Tips

Discussion in 'VAIO / Sony' started by Oscar2, Nov 16, 2010.

Thread Status:
Not open for further replies.
  1. anseio

    anseio All ways are my ways.

    Reputations:
    1,940
    Messages:
    2,418
    Likes Received:
    0
    Trophy Points:
    55
    I understand that there will always be debate and discussion about SSD optimizations. Look at the SSD Thread (Benchmarks, Brands, News, and Advice). It's got over 7500 posts and is neverending.

    I think where the problem began, and I can assure you that this is not just my impression, is that the wording of your original post came across quite strongly that Les' recommendations were lunacy and were all disproved by someone with passionately different opinions on a different forum. It only seemed natural to rush to Les' defense. It wasn't until your later posts, that you came across as being less critical. By then, you had already been attacked back.

    It's a shame this discussion has turned out so poorly, on both sides.

    Edited to remove rebuke.
     
  2. Les

    Les Not associated with NotebookReview in any way

    Reputations:
    4,706
    Messages:
    5,391
    Likes Received:
    1
    Trophy Points:
    0
    Love reading conversations such as this and enjoi responding, as I did on Anand. Its unfortunate that one never read through the entire thread, otherwise his viewpoint would be different. My points, in the end, resulted in no logical rebuttal other than my friends creation of the thread as I was a new member and, well, we all don't like people jumping in to our turf now do we?

    I would more than welcome ANY thoughts you may have other than to say "Someone called him out and he backed down." The truth is that I never backed down and my points were taken...again if you read through the entire thread.

    To make it very clear, I dont think I am any more knowledgeable than any of them nor do I than any of you. I believe we all have a piece of the puzzle and at the end of the day we get it put together through intelligent conversation.

    Having said that, I invite you to bring forward any optimization that you might not agree with for discussion. Rather than stating anothers point of view lets stick to our own knowledge or that of the experts like Intel and MS Engineers if thats ok. I wouldnt want to bring forward that the person on Anand that disagreed with me also disagreed with all of the experts in the field. As well, I wouldn't want to mention that that specific Guide is probably the most popular piece of literature on the internet right now. Its traffic is worldwide and the appreciation that is received is amazing....in several languages and thank God for internet auto translation now.

    So...lets get her going. Im game!

    Oh and one other thing...good to see you got the renice. Where did you read about such an amazing card?
     
  3. Oscar2

    Oscar2 Notebook Deity

    Reputations:
    209
    Messages:
    1,030
    Likes Received:
    0
    Trophy Points:
    55
    I'm guessing that maybe three posts were removed there, and it seems like we are the better for it. ;)

    So back to the technical talk...
    I noticed that pagefile was taking up 8GB on the drive. I thought about turning it down to 1GB. But, I decided to just disable it entirely for know and see how that goes.

    Between that and turning off hibernate, I figure I freed up about 15GB. :)
     
  4. anseio

    anseio All ways are my ways.

    Reputations:
    1,940
    Messages:
    2,418
    Likes Received:
    0
    Trophy Points:
    55
    I just turned hibernation on this week to try it out. It's only 3GB.

    AND... I checked SMART data and saw that hibernating only caused .5GB in writes to my SSD. I never keep that many things open that are important. I'll time a few wake up's from hibernation and see if it's a worthy improvement over a shutdown.

    I was completely fine without the page file. The only things that wanted it were a few video games, so I have it on my 2nd HDD. Since it's there, I think a few other programs may be using it.
     
  5. Oscar2

    Oscar2 Notebook Deity

    Reputations:
    209
    Messages:
    1,030
    Likes Received:
    0
    Trophy Points:
    55
    I find that I am always using sleep/standby, so that I always set the power button to put the laptop to sleep.
    I just never really hibernate. It feels like starting up from hibernate is as slow as a fresh bootup.
     
  6. anseio

    anseio All ways are my ways.

    Reputations:
    1,940
    Messages:
    2,418
    Likes Received:
    0
    Trophy Points:
    55
    I've got BSOD issues with sleep. Found a program yesterday that can help me figure out the dmp file, but forgot about it already. Thanks for reminding me. Will try it tonight and see if I can fix the problem.
     
  7. tailwagger

    tailwagger Notebook Enthusiast

    Reputations:
    0
    Messages:
    22
    Likes Received:
    0
    Trophy Points:
    5
    In my experience it is extraordinarily rare to design an optimization that wins across the board. Optimizations by their very nature tend to rely on a set of assumptions which when violated can wind up causing more harm than good.

    As a hypothetical, before paging out memory it can make a lot of sense to compress it first. By doing so we can save both time and space as crunching the memory actually requires less time than doing i/o to a conventional disk. Given the boost in performance, we turn this behavior on by default. But is this still the case if you happen for whatever reason to be paging to a ram disk? We still save space, but at a cost of performance as minimizing the xfer size buys us far less than the cost of compression. For some, as ram is precious, that may still be the correct trade off, for others it may well not be. To pick just one thing on the list, while I havent researched prefetch deeply, I've little doubt that it is heuristic, not algorithm. If the MS engineers assumptions about the environment and work loads hold, you'll love it, if not...

    So while I think that while many of us love to cling to the notion that there is that magic combination of switches that will make everything behave optimally, IMO, that is an overly simplistic view. Rather, behind all these optimizations and tweaks there is a set of assumptions that may or may not hold for a given machine and work load. There just aint a one size fits all answer here, but I am grateful to have the information and debate, such that I can hold these things up to what I wish to do and make the trade off decisions for myself rather than have them imposed on me in the dark.
     
  8. Les

    Les Not associated with NotebookReview in any way

    Reputations:
    4,706
    Messages:
    5,391
    Likes Received:
    1
    Trophy Points:
    0
    For the most part, no single optimization in Win7 with an SSD is going to gain you any noticeable performance whatsoever. It can get you alot of drivespace back however. We need to look at where most of the things that we discuss in these optimizations were first created.....XP and then Vista.

    When you look at Pagefile, Indexing, SuperFetch, and Prefetch (off the top of my head) you are not gaining anything visible by turning these off. The question is that, since there is absolutely no performance gain that anyone can show with them on, why do we want all these services running in the background for no reason.

    Pagefile is simple... If you have 4Gb of RAM you will never truly need it. The mechanics of its operation say its best to turn off actually because if it is on, the system always relies on it to a certain extent even if its not needed. Myself, I prefer to force my system to use the valuable ram I paid for so its off. The RAM is much faster than the SSD. An added bonus is the new way that Win7 chooses to allocate RAM. Wanna see?? Start 50 or so programs all at once and then notice that you still have over 50% RAM available. Would you believe that this is the same even with Pagefile turned off?

    As for hibernate, I prefer it off for the space, however, see the other side of the argument where someone likes to leave all of their programs on. I choose to use sleep where I have no problem whatsoever.

    Now...getting back to performance... shutting down only one service will not gain performance but anyone who has taken the time to learn through example knows that shutting down a number of unused services decreases start time, lessens the load on the system and increases performance.

    The amusing thing in it all is the person that everyone was jumping on has '10 second start' posted in his sig if I recall. The ONLY way to ever get this is through following that in which he contested. It is physically impossible to get a 10 second started otherwise... To take that further, time to boot logon as he shows in his link does not constitue the boot time.

    Boot time is the time we press a button until the time the desktop is usable....see below but I have bettered that to 7.2 which is recorded in the appropriate thread. I know of only one other to ever post a faster time and that is Stamatisx on this site with 5 sec.



    Later gang...
     
  9. arth1

    arth1 a҉r҉t҉h

    Reputations:
    418
    Messages:
    1,910
    Likes Received:
    0
    Trophy Points:
    55
    This was correct back in the days of Windows 95/98, but isn't true for NT-based systems (including Windows 7). The paging file is used for a lot more than just an expansion of memory.

    First of all, it is used for kernel core dumps. If your machine crashes, this is where the dump gets saved, and upon reboot, it's moved to a *.dmp file which can be submitted to MS for analysis.

    Next, with Vista and 7, defragmenting of system files goes through the swap for safety. If a machine crashes at exactly the wrong time, you can risk a system file being damaged. I know, you probably never defragment either, or so seldom that you don't see this as much of a risk. But it's still one factor.

    Then, and more importantly, it is used for proactively paging out memory pages that aren't accessed. If your machine loads a an app or library that has English, Swahili, Swedish and Korean support built-in, the memory pages that point to the extra languages are probably not going to be accessed, and just take up RAM. The paging system sees that a page hasn't been accessed for a long time, and swaps it out even if you have free RAM. This increases the amount of unused RAM, which is used for disk cache. Even with SSDs, RAM is a lot faster than drive access, so increasing the amount of RAM available for disk caching is beneficial.
    If you disable swap, those memory pages that you never use will stay in RAM, and the size of your cache will be lower. And your system will run slower as a result.
    To see the result of this, boot your system, then start the Task Manager and without doing anything, look at the used memory graph. With swap enabled, it will steadily decrease, because it pages out the memory you don't use. Memory that then gets used for disk caching where it makes a difference.

    Finally, it is also used by shadow copies. If you don't have swap enabled, the shadow copies can't pre-commit a lock, and you have to wait for the disk to reserve an area. I.e. shadow copies are going to be slower. (Shadow copies are taken every time you install or uninstall something with the official Microsoft installer or InstallShield, or when you do a backup.)

    So no, don't disable swap. Feel free to reduce it, but leave at least some swap enabled, or you actually slow down your system.
     
  10. Les

    Les Not associated with NotebookReview in any way

    Reputations:
    4,706
    Messages:
    5,391
    Likes Received:
    1
    Trophy Points:
    0
    Excellent description and great knowledge.

    The difficuly still lies with me that, as nobody that I know has been able to document any type of physical performance increase with pagefile on with an SSD, we are left with the pagefile only playng the part of telling you what went wrong if you have a crash. I haven't had a crash since the early days of Vista so it doesn't help at all and would suggest that most are in the same boat. Even on that same line, very few even know how to locate the dump file much less read it.

    I would appreciate if you could find some form of proof that shows that Pagefile enabled on an SSD does in fact result in performance improvement.

    Sorry if I am a bit direct but I have been through same in dozens of forums and its always best just to get to the heart of the matter rather than all the hypotheticals that will prevail.
     
Loading...
Thread Status:
Not open for further replies.

Share This Page