Increased Performance in Linux with zRam

[Icedrake]

http://www.webupd8.org/2011/10/increased-performance-in-linux-with.html

I've installed it on my comp, but I haven't really noticed a difference mainly because I have more RAM, but for those of you using <2 GB of ram, this should definitely make your computer faster (if you're using Linux that is). Based on the comments I've been reading in the article, it's gotten an extremely positive reaction, with the program making drastic speed differences in quite a few computers.

[Ishi]

I have an old computer that has only 1GB RAM in it. I have a Linux installed there and this is looking like an interesting article. Thanks for the info.

[Alan_B]

My interpretation of the article is that it gives zero increase in speed whilst RAM is available for use,

BUT a massive improvement when free RAM is exceed and it is time to use the Swap File because a normal Swap File on HDD gives a humongous penalty.

 

What you actually get is a replacement for the normal Swap File that uses space on the HDD,

and you get a VIRTUAL Swap File that uses space in the RAM you were already short off

 

Perhaps it was published in error on 2nd October - the author intended 1st April

 

The only benefit I can see is that this Virtual Swap File is compressed,

so perhaps 1 GB of Real RAM used by the Virtual Swap File might accommodate 3 GB of Notepad text in Unicode format,

but it will not accommodate more than 1 GB of incompressible files - e.g. music/video.

 

It reminds me of the amazing patents some 30 years ago for a multi-state analogue device that could hold multiple bits of data,

and this could replace a floppy disc and instantly boot up DOS with all the applications you wanted.

Merchant bankers thought they would be making a killing by investing in this new Company.

Very clever inventor fled the country with their finance before they realised their folly.

 

I am not saying that this Virtual Swap File is in any way a scam,

but do not be surprised if a free meal will cost you at the end of the day.

e.g. Windows is happy to compress old files that are rarely used

When you use a compressed file then Windows makes you take a time-out whilst it decompresses.

When your programme needs a few bytes of data in RAM that will take no time with REAL RAM,

but accessing a "block device which acts as a swap disk" will take more than a few CPU cycles, especially if it has to be decompressed.

[Icedrake]

My interpretation of the article is that it gives zero increase in speed whilst RAM is available for use,

BUT a massive improvement when free RAM is exceed and it is time to use the Swap File because a normal Swap File on HDD gives a humongous penalty.

 

What you actually get is a replacement for the normal Swap File that uses space on the HDD,

and you get a VIRTUAL Swap File that uses space in the RAM you were already short off

 

Perhaps it was published in error on 2nd October - the author intended 1st April

 

The only benefit I can see is that this Virtual Swap File is compressed,

so perhaps 1 GB of Real RAM used by the Virtual Swap File might accommodate 3 GB of Notepad text in Unicode format,

but it will not accommodate more than 1 GB of incompressible files - e.g. music/video.

 

It reminds me of the amazing patents some 30 years ago for a multi-state analogue device that could hold multiple bits of data,

and this could replace a floppy disc and instantly boot up DOS with all the applications you wanted.

Merchant bankers thought they would be making a killing by investing in this new Company.

Very clever inventor fled the country with their finance before they realised their folly.

 

I am not saying that this Virtual Swap File is in any way a scam,

but do not be surprised if a free meal will cost you at the end of the day.

e.g. Windows is happy to compress old files that are rarely used

When you use a compressed file then Windows makes you take a time-out whilst it decompresses.

When your programme needs a few bytes of data in RAM that will take no time with REAL RAM,

but accessing a "block device which acts as a swap disk" will take more than a few CPU cycles, especially if it has to be decompressed.

The way Windows and Linux manage RAM is completely different; thus comparing this program with Windows doesn't work, as this program is not intended for Windows. It's for Linux. I've installed it on my computer, and I can personally say that it has caused no problems.

Compressing pages and keeping them in RAM virtually increases its capacity. This allows more applications to fit in given amount of memory.

No offense, but it seems as if your logic is a bit incorrect.

http://code.google.com/p/compcache/

[Alan_B]

The way Windows and Linux manage RAM is completely different; thus comparing this program with Windows doesn't work, as this program is not intended for Windows. It's for Linux. I've installed it on my computer, and I can personally say that it has caused no problems.

 

No offense, but it seems as if your logic is a bit incorrect.

http://code.google.com/p/compcache/

You totally misunderstand me.

 

Your quote tells me nothing I have not already revealed, i.e.

The only benefit I can see is that this Virtual Swap File is compressed,

so perhaps 1 GB of Real RAM used by the Virtual Swap File might accommodate 3 GB of Notepad text in Unicode format,

but it will not accommodate more than 1 GB of incompressible files - e.g. music/video.

Please understand that I know there are different degrees of compression possible for different types of files,

but I thought it was self evident I was suggesting extremes.

 

My Windows System C:\ has 16 GB of used space.

7 GB of PageFile.sys and Hiberfil.sys is excluded from Partition image backups

9 GB of Executable and Data files is captured in an image backup of 5.40 GB (5,805,617,227 bytes)

 

That suggests to me that 1 GB of compressed Windows Applications would expand to about 1.66 GB when decompressed.

Another way of looking at Windows Applications is that the code information density is a little sparse and takes 66% more space than 100% efficient code.

Is Linux code fantastically inefficient compared to Windows ?

 

The initial article says indicates that a 6 year old Laptop with only 1 GB of RAM/ZRAM worked as well as a modern Desktop with 2 GB of RAM/ZRAM,

Lots of enthusiasm in the article, but not much detail. A claim that it is like adding more RAM, and no indication of any limit. Instead I see

instead of freezing after running out of RAM, the system worked like nothing happened. I didn't notice any difference at all. It looked just like adding more RAM!

Clap if you believe in fairies, and Tinkerbell, and won't die

 

Yes, I realised before I first responded to this topic, that ZRAM can hold more INFORMATION than Real RAM,

but I have seen no Linux numbers so Windows numbers are all I can go on,

therefore I would be surprised if ZRAM gave much more than a 50% boost in application storage space.

 

Have I actually misunderstood a special difference between Linux and Windows ?

Windows can decompress a compressed file on demand, but it takes extra processor cycles to do so.

This was I believe the reason why Disc Cleanup by default only compressed files that had not been recently accessed.

I guess this is not so significant if CPU speeds are improving faster than HDD speeds

but decompression time would have a major impact on the blisteringly faster speed of RAM.

 

Does Linux have special magic that it can decompress content that is held in RAM and does not need to use processor cycles to do so ?

 

I have not yet seen any numbers about Linux code information density or how much more can be held in RAM if it is used as ZRAM.

All I saw was unrestrained enthusiasm.

[Icedrake]

Good job on writing such a long post on a program that is not made for Windows... now, tell me again how you've proven zRam doesn't work?

[Alan_B]

Good job on writing such a long post on a program that is not made for Windows... now, tell me again how you've proven zRam doesn't work?

When did I say I had "proven zRam doesn't work?"

 

I will in turn ask you why you have started a topic "on a program that is not made for Windows"

 

Read me again and you just possibly might see that I do not deny its possible beneficial effects,

but the extreme enthusiasm shown by the original article for zRAM without any evidence of measurements of actual effective storage increase,

and total lack of consideration of speed performance due to compression/decompression delays,

suggests to me that it is not yet ready for discussion in a technical forum - even if it WAS applied to Windows.

[Icedrake]

When did I say I had "proven zRam doesn't work?"

My apologies if you didn't understand, but I didn't quote you directly. I'm summarizing what you've been saying in your essay; you've been coming up with reasons on as to how you believe this program doesn't work.

I will in turn ask you why you have started a topic "on a program that is not made for Windows"

If I'm reading the description of the Lounge section correctly, it allows the discussion of anything not related to the other sections of the forum here. In this case, Linux. Did I miss a part where it said that I couldn't post anything about Linux?

Discuss anything... relating to anything... software or not.

I don't know... seems like posting about Linux is allowed.

Read me again and you just possibly might see that I do not deny its possible beneficial effects,

but the extreme enthusiasm shown by the original article for zRAM without any evidence of measurements of actual effective storage increase,

and total lack of consideration of speed performance due to compression/decompression delays,

suggests to me that it is not yet ready for discussion in a technical forum - even if it WAS applied to Windows.

I'm assuming you didn't read the article correctly, because the author's enthusiasm stems from the fact that he performed multiple tests with zRam, and he found it to work effectively. Now, it's rather obvious that you've been enthusiastic about showing how this program is a scam while not performing any tests yourself to prove the validity of your claims. Simple as that. The author backed himself up with tests and results, whereas you only have claims.

[DennisD]

I took this as a simple "for info" post guys, and then it's up to anyone who wants to give it a go to make up their own minds.

 

And if they choose not to give it a go, then to just let this "for info" post slide.

 

@Ice.

 

It's really nice to see you back on here Ice, and maybe saying "could make your computer faster" instead of "will definitely" in your post would maybe have been a good idea.

 

An ending with something like ...

 

... with the program making drastic speed differences in quite a few computers, but of course you would have to make your own minds up ....

 

Just a thought by someone who's been there and done that, and had my knuckles wrapped for it.

[Alan_B]

My apologies if you didn't understand, but I didn't quote you directly. I'm summarizing what you've been saying in your essay; you've been coming up with reasons on as to how you believe this program doesn't work.

Please explain in what sort of world is it feasible to summarize me as believing "this program doesn't work" when I plainly said in post #3

"The only benefit I can see is that this Virtual Swap File is compressed,

so perhaps 1 GB of Real RAM used by the Virtual Swap File might accommodate 3 GB of Notepad text in Unicode format,"

If I'm reading the description of the Lounge section correctly, it allows the discussion of anything not related to the other sections of the forum here. In this case, Linux. Did I miss a part where it said that I couldn't post anything about Linux?

What you missed was in your own post, where you appeared to deny that my views were irrelevant due to no practical experience of Linux.

I was responding to what seemed to be a Linux take-over of a general computer discussion forum - that is all.

 

You appeared quite rude when you posted the message

Good job on writing such a long post on a program that is not made for Windows... now, tell me again how you've proven zRam doesn't work?

So far as I am aware, my Desktop has the capability of dual booting into either Windows or Linux,

therefore my hardware will support either,

and also my firmware/BIOS.will support either.

Unless there is a special trick I believe the processor would have the same instructruction set for either O.S.

therefore File compression of executables would be comparable

 

I'm assuming you didn't read the article correctly, because the author's enthusiasm stems from the fact that he performed multiple tests with zRam, and he found it to work effectively. Now, it's rather obvious that you've been enthusiastic about showing how this program is a scam while not performing any tests yourself to prove the validity of your claims. Simple as that. The author backed himself up with tests and results, whereas you only have claims.

 

The author's endorsment was 100% enthusiams and 0% facts, I quote :-

I decided to give it a try, and the result on my desktop with a quad-core CPU and 2Gb of RAM was fantastic: instead of freezing after running out of RAM, the system worked like nothing happened. I didn't notice any difference at all. It looked just like adding more RAM! Surprisingly, I got almost the same results on a 6-year-old laptop with Pentium M and 1Gb of RAM! So, I've improved the script to automatically adapt to the amount of memory in the system and automatically scale across several CPUs or CPU cores, packaged it in .deb and uploaded to PPA.

The article was followed my a very large number of comments.

The majority were negative

very few said there was any benefit and they totally failed to quantify the benefit.

 

I think my views are justified.

zRAM may give improvement but

"almost the same results on a 6-year-old laptop with Pentium M and 1Gb of RAM!" compared with

"desktop with a quad-core CPU and 2Gb of RAM"

that seems a little extreme,

and grammatically it appears he upgraded the 2 GB Desktop with zRAM and when he upgraded the 1 GB laptop it had the same sparkling performance as 2 GB with zRAM.

 

I concede what I have stated in previous posts,

there may be benefit with zRAM - but not to the degree promised, and many Linux people tried it and failed so how can you expect me to make it run ?

[login123]

Dennis, thats just the way I took it, information. Thanks, Ice, for the heads up, and Alan for the caveats.

 

I'm on "forum sabbatical" just now, time is short, but just had to pop in and comment. Addicted, I guess.

 

BTW, one has ones knuckles "wrapped" for one purpose, but "rapped" for another. Guess either might apply in this thread.

 

Just a thought.

[Icedrake]

Right now, I can't figure out if you were unable to correctly read the article and the following comments or if you just like to misinform people.

 

Now, since you insist on making claims out of your assumptions, and balantly ignoring actual tests and comments, it's rather obvious that you won't budge from your opinion. Please, if you can actually back up your assumptions by running this program on your own computer and telling us the results, I will be happy to oblige. However, discrediting something through information you've been unable to prove thus far except through more of your own claims makes it hard for me to believe your argument. I, along with many others, have used this program and many have found it to be beneficial. But hey, if you can't accept that you may be wrong, fine by me.

 

By the way, you said you had Linux on dual boot? Why not run zRam on it and see how it works out? Then maybe you can actually give some valid information.

 

@Dennis: I presumed no one would go out of their way to explain how something is false without any actual evidence. If Alan_B was able to provide any actual evidence to enforce his claims, sure, I would've easily told him that he's right. Unfortunately, he hasn't done anything of the sort except to continue conjuring up more reasons as to why this program is probably a scam. I'd say it's best to lock this thread; Alan_B has managed to ruin a thread about a legitimate program and turn it into a baseless argument.

[DennisD]

Don't worry about it Ice.

 

I'm sure folk will decide for themselves whether or not to try it, and to make their own minds up.

 

I'll leave your thread open for yourself and others, and call time on any continuing disagreements.

 

Those that want to have made their points, so I hope it's left at that.

[Alan_B]

Right now, I can't figure out if you were unable to correctly read the article and the following comments or if you just like to misinform people.

Most of the comments appended to the article were to the effect that it did not work for them.

Too many comments for me to bother counting.

Many screens without a single positive comment.

 

Now, since you insist on making claims out of your assumptions, and balantly ignoring actual tests and comments, it's rather obvious that you won't budge from your opinion.

Try me and see - show me where in that article, by which I assume the portion I quoted, there is a single measurement of a percentage improvement delivered by zRAM.

 

He said

"instead of freezing after running out of RAM, the system worked like nothing happened"

What does that mean ?

I assume his system does not run out of RAM with a limited burden of running applications.

What percentage of a larger burden was supported by the use of zRAM ?

 

He found that an old Laptop single core CPU with 1 GB RAM/zRAM gave ALMOST the same results as a newer Desktop QUAD core with 2 GB of RAM.

That IMPLIES he had a result that showed him less than a 100 % improvement.

I have already agreed to a 66% possibility - HOW IS THAT DIFFERENT from an unmeasured less than 100%

 

I admit the possibility of 66% - but the evidence does not yet prove this to me.

Pentium M was 32 bit x86

Could the modern Quad core have been 64 bit architecture.

Windows 7 installations need 1 gigabyte (GB) RAM (32-bit) or 2 GB RAM (64-bit) according to

http://windows.micro...m-requirements.

All the applications available for either x86 or x64 which I have looked at use twice the code for 64 bit systems as they do for 32 bit.

 

Can you give me a definite assurance that Linux can use the same hardware as Windows, and yet not require twice the RAM for x64 as it does for x86 ?

 

"ALMOST the same results" suggests that an old single core processor is almost as FAST as a newer quad core processor - is this another benefit of zRAM

I take this as evidence of being careless in the accuracy of his unmeasured observations.

 

Please show me measurements,

not feelings of the nature "My car goes faster now I painted go-faster stripes down the side"

 

I can accept that zRAM makes a difference - How many more times do I have to reiterate this ?

 

I have just looked through all the comments appended to that article and found this single measurement,

BUT although it suggests a 50% boost in effective RAM,

it then "damns it with faint praise" by concluding "maybe it's not having any impact on this comp."

This is the comment

To add, my swap file is now listed as 5.9GB [0 byte used] up 2 from 3.9GB [zero byte used] it always was. Could not see anything running with Top but a "sudo start zramswap" did get a "start: Job is already running: zramswap" response. Though only 3GB in working RAM, never seen the swap file being used, even with larger models of 1GB and up running, so maybe it's not having any impact on this comp.

10/03/2011 08:30 AM

 

 

Please, if you can actually back up your assumptions by running this program on your own computer and telling us the results, I will be happy to oblige.

Sorry - I do not understand, in what way will you oblige me ?

 

You discredited zRAM when YOU YOURSELF "damned it with faint praise". Let me remind you

I've installed it on my comp, but I haven't really noticed a difference ....

If you could not find any benefit what makes you think I would ?

However, discrediting something through information you've been unable to prove thus far except through more of your own claims makes it hard for me to believe your argument. I, along with many others, have used this program and many have found it to be beneficial. But hey, if you can't accept that you may be wrong, fine by me.

Please explain how it was beneficial to you when you introduced this with "I've installed it on my comp, but I haven't really noticed a difference"

By the way, you said you had Linux on dual boot? Why not run zRam on it and see how it works out? Then maybe you can actually give some valid information.

You really need to understand what I wrote.

My system has the capability,

but I do not have the inclination to learn Linux whilst I am suffering the transition from XP to Windows 7

@Dennis: I presumed no one would go out of their way to explain how something is false without any actual evidence. If Alan_B was able to provide any actual evidence to enforce his claims, sure, I would've easily told him that he's right. Unfortunately, he hasn't done anything of the sort except to continue conjuring up more reasons as to why this program is probably a scam. I'd say it's best to lock this thread; Alan_B has managed to ruin a thread about a legitimate program and turn it into a baseless argument.

I NEVER SAID THIS WAS A SCAM. I actually said

"I am not saying that this Virtual Swap File is in any way a scam,"

The subject came up because the initial claims were a little extreme and they reminded me of a major fraud some years ago where a little memory was supposed to hold a lot of information.

 

Since that comment I have NEVER referred to any deception - I just like FACTS and all that has been presented is FEELINGS.

 

The ONLY NUMERICAL RELEVANT FACT is what I have just found for myself and posted above, i.e.

"To add, my swap file is now listed as 5.9GB [0 byte used] up 2 from 3.9GB [zero byte used] it always was ... maybe it's not having any impact on this comp."

 

How sad - technical improvement 50% that was no benefit. I have no more to add

[hazelnut]

Alan enough is enough, try and keep any comments in threads like this on the much, much shorter side.

 

You seem to want to go into so much detail it makes peoples eyes glaze over.

 

Sorry Icedrake, I know you were the thread starter but I'm closing this now, hopefully the next topic you start can be discussed by certain members in a more general fashion than was the case here.