Is Vivaldi eating our SSD's like Firefox does?



  • Hi everyone! I came across this article : FIREFOX IS EATING YOUR SSD – HERE IS HOW TO FIX IT [url=https://www.servethehome.com/firefox-is-eating-your-ssd-here-is-how-to-fix-it/][/url] https://www.servethehome.com/firefox-is-eating-your-ssd-here-is-how-to-fix-it/ The first comment to that article reads: [i]James SEPTEMBER 23, 2016 AT 8:00 AM Thanks, Sergei. [u]Observing similar behavior with: Vivaldi 1.5.609.8[/u] (Developer Build) (32-bit). just looking at the I/O Write Bytes column in Process Explorer. Launched this Vivaldi session ~40 ago, some of the child processes (Twitter tabs?) have written as much as 1.4 GB to date.[/i] What's the story with that? Does Vivaldi have a great appetite for SSD's? And if so, how can we extend our SSD's lifespan again? Thanks to anyone who can provide more solid information to this.



  • This is purely an academic concern. The endurance range on consumer grade SSDs is in the decades range and easily out lasts HDDs. Server class even longer.



  • Concerns about SSD wearing are just overrated.

    Anyway yes Vivaldi does a relatively high traffic, just use a Ramdisk and place your cache there



  • Hi guys.

    Thanks for your prompt replies.
    To be honest, when possible I prefer turning to the root of an issue rather than fighting symptoms.

    If you allow me to, let me give ye an example: if I continuously drive a car in a too low gear my avg. fuel consumption is higher than it needs to be. Certainly I can tweak the car's aerodynamics, change to thinner tires etc. but wouldn't shifting up be the way to go?

    So I should have asked my question a different way:

    a) (why) is there such a huge I/O load on Vivaldi?
    b) Is it a necessity (for now) and/or will this change in the future and if so, when?

    Of course could I move it to another, regular, HDD at the cost of losing the speeds SSDs provide me with.
    So 'yes' and please don't take it as an offense, I will try to get a more holistic answer from the Vivaldi team, as soon as I will have found out how to reach out to them ;-)

    Should it be the case that it won't be subject to change I guess there is no way around a RAMdisk, as I've come across another article stating that not only Firefox shows this behavior but also Chrome.
    Hopefully it's something the Vivaldi team, which, to my impression, so far is doing an outstanding (!) job providing us with a power user browser (TwoThumbsUp), can solve or improve.



  • @RogerWilco:

    Hi guys.

    Thanks for your prompt replies.
    To be honest, when possible I prefer turning to the root of an issue rather than fighting symptoms.

    Feel free to write a browser engine or to patch cromium's one… :D

    Seriously that's the engine and this is how it works, no one today has the power to tweak it to behave like the old Presto did

    If you allow me to, let me give ye an example: if I continuously drive a car in a too low gear my avg. fuel consumption is higher than it needs to be. Certainly I can tweak the car's aerodynamics, change to thinner tires etc. but wouldn't shifting up be the way to go?

    Not a good analogy, if you want to talk about cars, Vivaldi makes the chassis, GM/Ford/Whatever provides the engine.

    Vivaldi can tweak the aerodinamics, can make the chassis ligther, more confortable, with more accessories and so on.

    Vivaldi can also trim few engines settings, but the engine will be roughly, the same as the one installed under the hood of a GM car.

    a) (why) is there such a huge I/O load on Vivaldi?

    There isn't an huge traffic, there is a relatively high traffic (if you compare it with browsers from another era, like Opera 12.

    Huge traffic may happen if something goes wrong on the topsites file, and in such case deleting it and refreshing the thumbnails is more than enough to fix the problem.

    Of course could I move it to another, regular, HDD at the cost of losing the speeds SSDs provide me with.

    That would be pretty masochistic

    Should it be the case that it won't be subject to change I guess there is no way around a RAMdisk, as I've come across another article stating that not only Firefox shows this behavior but also Chrome.

    You mentioned more than 3/4 of the browser market. You could replace the sentence with "any current browser".

    That is the personal IT trend PC become faster, storage becomes faster and larger, programs becomes heavier, the average weight of a single web page is something like 1000x if compared with the situation from 2000 or 2005.

    That's not Vivaldi that's the normal evolution across the years

    Hopefully it's something the Vivaldi team, which, to my impression, so far is doing an outstanding (!) job providing us with a power user browser (TwoThumbsUp), can solve or improve.

    Vivaldi coders will surely keeps improving and optimizing their browser as they did in the last 20 months, but as I said above their control on the engine is limited. Patching it heavily would be possible, but would be a nightmare each time the upstream engine will be updated.

    And even in such case, caching a 10Kb page from 2001 was not the same thing than caching a 4MB page from 2016



  • @RogerWilco:

    Hi guys.

    Thanks for your prompt replies.
    To be honest, when possible I prefer turning to the root of an issue rather than fighting symptoms.

    If you allow me to, let me give ye an example: if I continuously drive a car in a too low gear my avg. fuel consumption is higher than it needs to be. Certainly I can tweak the car's aerodynamics, change to thinner tires etc. but wouldn't shifting up be the way to go?

    So I should have asked my question a different way:

    a) (why) is there such a huge I/O load on Vivaldi?
    b) Is it a necessity (for now) and/or will this change in the future and if so, when?

    Of course could I move it to another, regular, HDD at the cost of losing the speeds SSDs provide me with.
    So 'yes' and please don't take it as an offense, I will try to get a more holistic answer from the Vivaldi team, as soon as I will have found out how to reach out to them ;-)

    Should it be the case that it won't be subject to change I guess there is no way around a RAMdisk, as I've come across another article stating that not only Firefox shows this behavior but also Chrome.
    Hopefully it's something the Vivaldi team, which, to my impression, so far is doing an outstanding (!) job providing us with a power user browser (TwoThumbsUp), can solve or improve.

    Just curious, whats the benefit? Why spend your energy on this? Is it to save on the cost of drives? I'd also recommend avoiding HDDs as they're more error prone due to mechanical failure than SSDs. If performance is the concern, you could create a Ram disk, but Ram wears out too, eventually.



  • @The_Solutor:

    @RogerWilco:

    Hi guys.

    Thanks for your prompt replies.
    To be honest, when possible I prefer turning to the root of an issue rather than fighting symptoms.

    Feel free to write a browser engine or to patch cromium's one… :D
    >>>> OK. Where do we meet and you teach me? I'm familiar with C++, .Net and a bit of Java. What would be your contribution? ;-)

    Seriously that's the engine and this is how it works, no one today has the power to tweak it to behave like the old Presto did
    >>>> As I'm sure that you read my 1st post or maybe even the article I referred to: since when does Firefox use Blink? So it would have to be feature both share in common, right? That is if Mozilla's approach on this regard was the same.

    If you allow me to, let me give ye an example: if I continuously drive a car in a too low gear my avg. fuel consumption is higher than it needs to be. Certainly I can tweak the car's aerodynamics, change to thinner tires etc. but wouldn't shifting up be the way to go?

    Not a good analogy, if you want to talk about cars, Vivaldi makes the chassis, GM/Ford/Whatever provides the engine.
    >>>> OK, you're talking about the archtiecture itself, I was more referring to how what is there is being used. But yes, I'd like to gain an understanding what the factors are that result in this I/O load. And 'yes' - maybe some engine tweaking is possible.

    Vivaldi can tweak the aerodinamics, can make the chassis ligther, more confortable, with more accessories and so on.
    Vivaldi can also trim few engines settings, but the engine will be roughly, the same as the one installed under the hood of a GM car.
    >>>> And this is something I have never looked into: browser engines themselves. But as Blink is under LGPL / BSD license I doubt that it's a box one can't look into. Written in C++ so I'd have a good chance of understanding some of it.

    a) (why) is there such a huge I/O load on Vivaldi?

    There isn't an huge traffic, there is a relatively high traffic (if you compare it with browsers from another era, like Opera 12.
    >>>> RIght, everything is relative as far as I know. ;-) Just like everything is subjective. ;)

    Huge traffic may happen if something goes wrong on the topsites file, and in such case deleting it and refreshing the thumbnails is more than enough to fix the problem.
    >>>> OK, now we're getting to the facts. Do you have more information on this please and what is your source for this regarding the Vivaldi implementation?

    Of course could I move it to another, regular, HDD at the cost of losing the speeds SSDs provide me with.

    That would be pretty masochistic
    >>>> Correct. An option I listed but directly pointed out why I wouldn't regard it as a satisfying solution, hence my conclusion at the end of my post probably regarding setting up a RAMdisk.

    Should it be the case that it won't be subject to change I guess there is no way around a RAMdisk, as I've come across another article stating that not only Firefox shows this behavior but also Chrome.

    You mentioned more than 3/4 of the browser market. You could replace the sentence with "any current browser".
    >>>> I could, but then someone would mention that IE or Edge probably don't show that behavior. No, I don't like either of them.

    That is the personal IT trend PC become faster, storage becomes faster and larger, programs becomes heavier, the average weight of a single web page is something like 1000x if compared with the situation from 2000 or 2005.
    >>>> True if you allow a website to run all its scripts - which I don't. Same applies to plugins etc.

    That's not Vivaldi that's the normal evolution across the years

    Hopefully it's something the Vivaldi team, which, to my impression, so far is doing an outstanding (!) job providing us with a power user browser (TwoThumbsUp), can solve or improve.

    Vivaldi coders will surely keeps improving and optimizing their browser as they did in the last 20 months, but as I said above their control on the engine is limited. Patching it heavily would be possible, but would be a nightmare each time the upstream engine will be updated.
    >>>> Wouldn't that depend on the inner structure of Blink? Still, you could be right.

    And even in such case, caching a 10Kb page from 2001 was not the same thing than caching a 4MB page from 2016

    >>>> Anyhow, thanks for the hint regarding the topsites file.



  • So what is the solution? Where to change the this behavior of vivaldi in the settings?



  • @Operanix:

    So what is the solution? Where to change the this behavior of vivaldi in the settings?

    Solutions usually implies a problem, and this isn't the case

    Anyway if you want to use a ramdisk (I suggest imdisk) and move your cache there you have just to use something like this to launch vivaldi

    "C:\Program Files (x86)\Vivaldi\Application\vivaldi.exe" –disk-cache-dir="R:\Vivaldicache"

    (the example assumes that Vivaldi is installed for all users and the Ramdisk letter is R, adjust it to fit your setup)



  • @The_Solutor:

    "C:\Program Files (x86)\Vivaldi\Application\vivaldi.exe" –disk-cache-dir="R:\Vivaldicache"

    (the example assumes that Vivaldi is installed for all users and the Ramdisk letter is R, adjust it to fit your setup)

    Will this work on HDDs too? Cause that's what I have.



  • This moves the cache in a chosen location, so yes.

    The ramdisk is possibly more helpful in HDDs given a ramdisk is like 1000 times faster than an SSD, which in turn is 1000 times faster than an HDD.

    Playing a bit with the automated backup/restore functionality of many RAMdisks you can even run the whole program from the RAMdisk, and you can place there the user profile as well



  • @The_Solutor:

    Solutions usually implies a problem, and this isn't the case

    Apart from the fact that this behaviour puts load on all parts of the computer you could just imagine someone owning a basic and recently bought netbook containing a 32 GB SSD. Calculating with 1000 write cycles and 10 GB of data written every day (by Vivaldi only) gives a result of Vivaldi eating 11 % ot the SSD per annum for a more or less useless function. I do think this is a problem.



  • @ghpy:

    @The_Solutor:

    Solutions usually implies a problem, and this isn't the case

    Apart from the fact that this behaviour puts load on all parts of the computer you could just imagine someone owning a basic and recently bought netbook containing a 32 GB SSD. Calculating with 1000 write cycles and 10 GB of data written every day (by Vivaldi only) gives a result of Vivaldi eating 11 % ot the SSD per annum for a more or less useless function. I do think this is a problem.

    Not sure how you figure these writes are eating SSDs, but wear and tear is a constant shared by all mediums including HDDs; nothing lasts forever. Maybe you'll feel better about disk IO after reading this article:
    http://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead



  • @p.stephenwille@gmail.com:

    Maybe you'll feel better about disk IO after reading this article:
    http://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead

    No, I don't, as the article confirms my opinion.
    The TLC-SSD encountered the first problems after 300 TB. For a SSD of 240 GB that means 1250 writes. I calculated with the figure usually given for TLC-SSDs: 1000. That's more or less the same range.
    The more expensive MLC-variety lives longer. But even one of them stopped to allow write access after less than 3000 writes.

    Of course all this is no problem for those with a big wallet who buy huge and change there system every two or three years anyway. For those who prefer to use their systems as long as possible it is annoying to have their SSD's lifetime reduced by unnecessary data written. I know people with their nine year old netbooks still running every day. A netbook with a small SSD would never reach that point with the current behaviour of Firefox/Chrome/Vivaldi.

    https://www.servethehome.com/firefox-is-eating-your-ssd-here-is-how-to-fix-it/ speaks of 10-35 GB per day for Firefox (technically not relevant for Vivaldi) but also of 24 GB for Chrome. So we can expect Vivaldi to behave in the same range. This means about 9 TB per year.

    I never ever restore a browser session. For me this is just a useless feature with adverse effects.

    Edit: In fact the mentioned netbooks are only eight years old. My mistake.



  • @ghpy:

    Apart from the fact that this behaviour puts load on all parts of the computer you could just imagine someone owning a basic and recently bought netbook containing a 32 GB SSD. Calculating with 1000 write cycles and 10 GB of data written every day (by Vivaldi only) gives a result of Vivaldi eating 11 % ot the SSD per annum for a more or less useless function. I do think this is a problem.

    You don't need to calculate you need some real world experience, that clearly you're lacking.

    After more than 7 years since my first SSD I have yet to see a single failure, a single DOA, as single SSD marked as worn out.

    And my sample isn't huge obviously, but is large enough to value it more than "this is what I've heard"



  • @The_Solutor:

    And my sample isn't huge obviously,

    No, it is just a single experience and you surely haven't been using Vivaldi all those years. So that's statistically irrelevant real world experience. That's why there are people testing SSDs in larger numbers like in the mentioned article.

    For a normal SSD of around 250 GB the warranty will say 75 TBW (consumer product) or 150 TBW (professional line). This of course doesn't mean that the SSD dies immediately after but it is a point where problems start to occur in significant numbers for the producer to draw a line and end the warranty. If it were different the producer would give a longer warranty to induce buyer confidence.

    Tests, and the mentioned article is just one of them, show that today the majority of SSDs reach twice as much. So we've got 150 TBW for the accessible consumer SSDs and 300 TBW for the more expensive variety.

    Vivaldi, running a whole day, seems to write 24 GB of unnecessary data. Adding this up for nine years we reach 78 TBW. So if you are one ot those people who prefer longevity in products you better not use Vivaldi (most probably any Chromium-based browser) in its current state because it will eat half your SSD for nothing.



  • @ghpy:

    No, it is just a single experience

    Single experience of what?

    I'm a sysadmin and a technician, I'm talking about at least one hundred of disks, but even if they were just ten, they were ten more than what you tested personally.

    That's your problem. You're trying to demonstrate something you haven't faced not even once

    and you surely haven't been using Vivaldi all those years

    Come on! You can't be serious.



  • Ok. So according to your experience SSDs live forever or at least more than seven years, no matter how many TB are written. My opinion about SSDs' lifetime is different from yours but points of view can be different depending on the circumstances.

    A professional, be it sysadmin or businessman, will never see a disk dying. He always changes them before they get that old because it's cheaper for the company that way.
    A consumer will see a disk dying every now and then. Nevertheless he tends to avoid replacing them preventively, because it comes cheaper that way.



  • @ghpy:

    Ok. So according to your experience SSDs live forever or at least more than seven years, no matter how many TB are written.

    Sort of.

    Practically any SSD i have touched/replaced/installed is still alive and kicking no matter the usage habit and scenario.

    Which roughly translates in "SSD life is not a concern for 99.9% of the users. Period."

    My opinion about SSDs' lifetime is different from yours but points of view can be different depending on the circumstances.



  • @ghpy:

    @p.stephenwille@gmail.com:

    Maybe you'll feel better about disk IO after reading this article:
    http://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead

    No, I don't, as the article confirms my opinion.
    The TLC-SSD encountered the first problems after 300 TB. For a SSD of 240 GB that means 1250 writes. I calculated with the figure usually given for TLC-SSDs: 1000. That's more or less the same range.
    The more expensive MLC-variety lives longer. But even one of them stopped to allow write access after less than 3000 writes.

    Of course all this is no problem for those with a big wallet who buy huge and change there system every two or three years anyway. For those who prefer to use their systems as long as possible it is annoying to have their SSD's lifetime reduced by unnecessary data written. I know people with their nine year old netbooks still running every day. A netbook with a small SSD would never reach that point with the current behaviour of Firefox/Chrome/Vivaldi.

    https://www.servethehome.com/firefox-is-eating-your-ssd-here-is-how-to-fix-it/ speaks of 10-35 GB per day for Firefox (technically not relevant for Vivaldi) but also of 24 GB for Chrome. So we can expect Vivaldi to behave in the same range. This means about 9 TB per year.

    I never ever restore a browser session. For me this is just a useless feature with adverse effects.

    Edit: In fact the mentioned netbooks are only eight years old. My mistake.

    The 1250 writes is an very general average of the whole drive until one cell fails. It doesn't reflect drive longevity. As you know, SSDs employ 'wear leveling' algorithms and provide extra cells specifically to replace the failed ones, which might first start occurring at 300TB. The drive itself is fully functional for many more TB (years) of writes. HDDs fail too, and when they do, its total. The salient line in this HDD article is, "This means that 50% of hard drives will survive until their sixth birthday.". In the end, the options are slow manageable failure, or sudden and total failure. One drive for the rest of your life is not an option, but SSDs should last a lot longer than 6 years.

    http://www.extremetech.com/computing/170748-how-long-do-hard-drives-actually-live-for


Log in to reply
 

Looks like your connection to Vivaldi Forum was lost, please wait while we try to reconnect.