Can RAM disk speed up any portion of this workflow? [Archive]

Avisynth_challenged

11th August 2011, 19:25

Hey everyone,

I made a recent RAM purchase, 2 x 4GB DDR3-1333 to put into my desktop video processing PC.

I was wondering if configuring the system to allocate about half of the available 8GB for a RAM disk would facilitate a faster processing experience in any area of my specific workflow in converting film-based 1920x1080i @29.97fps MPEG2 to 1920x1080p @23.976fps H.264.

Here are the steps performed, and the programs I use to carry them out:

Source = U.S. HD captures from OTA via computer ATSC tuner cards saved as 1080i MPEG2 transport stream.

Step 1: Edit out commercials, with VideoRedo TV Suite 3, merge 1080i program segments into one single .mpg file.

Step 2: Demux into .m2v/.ac3 elementary streams, also with VRD TVS3.

Step 3: Run .m2v through DGIndex (I can't recall the exact version, but I think it's very recent, if not the latest).

Step 4: Load 1080i .m2v file into Avisynth (v2.57, I believe) using DGDecode plugin, and do the following processes:
- IVTC (typically using JDL Telecine, found *HERE* (http://avisynth.org/stickboy/jdl-telecine.avsi)),
- Save to intermediate AVI using Huffyuv compression in YV12 colorspace.

Step 5: Transcode IVTC'd Huffyuv AVI file into H.264 using x264. Output as raw .264 video only file.

Step 6: Mux the output of Step 5 with the .ac3 audio file using multiAVCHD (output for SDHC media).

Also, my OS and programs run on a 7200RPM SATA300 HDD, and I have 2 additional 7200RPM SATA300 HDDs as dedicated I/O devices for my video processes.

Any feedback or critique would be appreciated. Thank you very much in advance!

burfadel

11th August 2011, 22:53

Try updating Avisynth to the more recent v2.58 or even v2.58MT or v2.60 (in testing) if you like!

The MPEG2 decoder as part of DGIndex is really old, FFMpegsource is an option and is very fast, however at times there seems to be lag which may affect your encoding. Overall you probably won't gain much by using the RAMdisk, seeing that you have to copy the files over there in the first place. The best thing to do would to ensure complete updatedness of your computer, including all Windows updates, drivers, program updates, video filter updates. http://xhmikosr.1f0.de/index.php?folder=YXZpc3ludGg= is a great source for the Avisynth builds and recompiled filters. If you use any of the filters listed there, definitely use those versions instead as they have been compiled using the latest compilers and you do actually gain performance by using them (even if the code source is the same, different compilers can affect the performance quite noticeably).

Audionut

12th August 2011, 09:32

I wouldn't worry about a ram disk. Just read from one of the sata disks and save (write) to the other disk and it will be fine.

Avisynth_challenged

12th August 2011, 17:52

Thanks for the responses. Okay, so RAM disk won't speed up my process flow.

Next question, more general... is RAM disk useful for anything in a WinXP-based PC system? I've read how RAM disk can be used to create a rapidly-accessed drive volume (MUCH quicker than a HDD) where WinXP can be configured to write page/swap file data, temporary internet cache, etc. which will increase performance speed overall.

Additionally, I read that this same tweak is often used by users that have SSD as their OS drive, and this is done to minimize the amount of writes to the SSD, thus increasing the overall lifespan of the SSD (since the SSD memory elements have a limited number of times they can be written to).

Any thoughts about the validity of these concepts? Thank you once again for this discussion.

Audionut

12th August 2011, 22:30

Everyone has an opinion on the Page File. Personally, I think it should be left alone. I believe the key is to optimize the system so that the Page File isn't needed. Best and easiest way is to simply add more RAM.

RAM drives are useful. And I have used one for a SSD as you have described. It is known that SSD's have a limited life span, so it makes sense to move some of the writes to a RAM drive. However, don't fall into the trap of moving everything away from the SSD, else, you might as well not have the SSD in the first place. Some even recommend to move writes to a mechanical HDD, which just seems silly to me, I bought a SSD for a reason, not to move reads and writes back to a slow mechanical HDD.

Moving temp internet cache to a RAM drive is a good one. Generally lots of small writes there.
When it comes to a mechanical HDD, they are very slow at small writes (4KB etc), so moving all these to a RAM drive will significantly improve the speed and responsiveness of a system.

Personally though, rather than going to the effort and manually editing things, I have become rather fond of a program called Fancycache*. (http://www.romexsoftware.com/en-us/fancy-cache/)
It acts as a cache for HDD's. And can use both system memory and SSD's as the cache. It will also access extended memory above the 4GB limit of 32bit Windows.

I recently setup a PC for a friend who had a Pentium D 930, 2GB RAM and 200GB HDD. I added another 4GB RAM (6GB total) and used 2GB of that (she only plays FB games and checks emails so 4GB is way more than she needs) as a cache for the HDD. The difference was night and day.

*I recommend this even though it's only in beta stage, as the developers are working with OCZ to increase the life of SSD's, which can't be a bad thing.

hello_hello

14th August 2011, 08:49

Also, my OS and programs run on a 7200RPM SATA300 HDD, and I have 2 additional 7200RPM SATA300 HDDs as dedicated I/O devices for my video processes.

One more hard drive and you could have yourself a couple of RAID-0 volumes.
I've been running my PC using four hard drives as two RAID-0 volumes for quite a while. I could never go back to running single hard drives.
In your case I'd at least use two drives as a single RAID volume and install Windows and programs on a small partition on it, then use the third drive as a dedicated input or output drive. In my case I just put the source on one RAID volume and use the other for the output. It doesn't keep the drives very busy at all even while converting video, so I can continue to use the PC as though it's not doing anything.

ramicio

18th August 2011, 16:49

Even if you had 24 GB of RAM and barely ran any programs Windows will still use the page file. In that case it should just be disabled, not left alone. It's pointless. I have 12 GB and I have mine disabled. What can possible take up that much memory? If you have tons of things open, it's best to just shut them down because you are a human, and you can't multitask like a computer can, so you don't need so many things open.

Ghitulescu

18th August 2011, 17:14

since the storage capacities overpassed the RAM ones, the design of windows always presented the same fundamental flaw - it will use the cache no matter what settings do you have.

laserfan

18th August 2011, 18:11

... is RAM disk useful for anything in a WinXP-based PC system? I've read how RAM disk can be used to create a rapidly-accessed drive volume (MUCH quicker than a HDD) where WinXP can be configured to write page/swap file data, temporary internet cache, etc. which will increase performance speed overall.

Additionally, I read that this same tweak is often used by users that have SSD...
I have a PC with an SSD as a C: drive, and use a RAMDisk (Dataram). I have also implemented RAMDisks on my conventional PCs:

1. Not for pagefile.sys--that shouldn't get used much-if-at-all in a system with decent memory installed

2. RAMDisks are best for temporary files; I have my Environment Variables for Windows and Users' temp files set to the RAMDisk

3. We use Firefox and Thunderbird and I've set their cache locations to the RAMDisk and it speeds them up noticeably

4. Some apps don't like smallish RAMDisks e.g. a photo-editing app can make monstrous temp files. I got into trouble with a 256Mb RAMDisk initially; since then have 1Gb RAMDisks on a couple PCs and a 512Mb on another

5. On my SSD machine I set all Windows logs to output to the RAMDisk--Windows (esp 7) does A LOT of writing to logfiles!!!

6. On my SSD machine I also set the printer spooler location to the RAMDisk

7. I tried (with Windows 7) to put Users\Roaming onto the RAMDisk and got into trouble with some apps that are seemingly hard-coded to go to the C: drive or something. Also Dropbox on a PC with multiple users is very tricky to do.

If you try that route, make as big a RAMDisk as you can "afford" right up-front. Changing it later can be tricky! :)

BTW one joy of the Dataram RAMDisk is that the only writing to HDD (or SSD) occurs when you shutdown or restart (writes-out an .img file w/all files intact, then reloads from same). And there's a small time penalty on reboots as Dataram writes-then-reads from a large file on your HDD/SSD.

I don't see much (any?) use for RAMDisk in your (or my) video process--even if you made a 16Gb RAMDisk or some such you'd still hit-the-wall in processing HD video.

hello_hello

18th August 2011, 18:15

The way I understand Windows memory management working, is a program might request usage of a certain amount of RAM. It can of course change as the program requires more RAM, but the way I understand it when a program requests "X" amount of memory, Windows will generally give it "X" amount of memory even though it's probably not using all of it. What's being used is assigned to RAM, what isn't is assigned to the paging file.
So if my understanding is correct, page file usage doesn't necessarily mean the page file is actually being used for current memory requirements, and by disabling it you're possibly forcing Windows to waste RAM by assigning RAM to programs that aren't using it.

What can possible take up that much memory? If you have tons of things open, it's best to just shut them down because you are a human, and you can't multitask like a computer can, so you don't need so many things open.

Everything gets loaded into RAM. Open a couple of huge wave files with an editor and watch your RAM usage increase. Then close the wave files and if you've got plenty of free RAM, watch it's usage not decrease. Open the wave file again, and it should open much quicker.
I would have thought one of the advantages of having lots of RAM is not having to continually open and close programs.

since the storage capacities overpassed the RAM ones, the design of windows always presented the same fundamental flaw - it will use the cache no matter what settings do you have.

Or maybe it's a bad assumption based on trying to second guess Windows memory management without really knowing how it works?
The $2 question would be, if you've got buckets of RAM and you disable the paging file, do your computer tasks increase in speed, reduce in speed, or remain the same? I wonder how many people who complain about Windows using the paging file have actually tested it?

ramicio

18th August 2011, 18:29

What program loads an entire sound file into RAM? You're mistaking a system making a cache of it versus the program itself actually allocating that space in RAM and putting the audio in there. Once the file is loaded into the program completely it is gone from cache. There are still chunks left. This is why it loads quicker. If the full file was kept in RAM it would load instantaneous. The advantage of large amounts of RAM is to get rid of the page file and allow a huge head room so when you finally do use programs that take up a lot of RAM, they won't take up all of it and crash. The only think I've had crash because of running out of space is chkdsk. Microsoft insists it's good and normal behavior to load what is scanning into RAM, but I think it's a huge memory leak/hole. If your primary storage is a 64 GB SSD and you are scanning a multi-terabyte hard drive, even having a page file is going to break the scan.

Ghitulescu is not wrong. If you have loads of RAM, a mechanical hard drive, and still have a hard drive, your system is going to grind to a halt anytime you use the hard drive. An SSD makes it better, but keeping a page file on their is bad for wear. Why you would even bother to put a page file on a RAM disk is beyond me. The page file was there to cover up the fact that software requirements grew faster than what RAM could offer. In the 64-bit world it is pointless.

hello_hello

18th August 2011, 19:05

What program loads an entire sound file into RAM?

Nero Wave Editor for one.
I'm in the middle of converting a video, so I converted the DTS track to a stereo wave file.
I opened Nero Wave Editor and checked available RAM. Then I opened the 1Gb wave file. It took about six seconds and available RAM dropped by about a gig. I closed the wave file and opened it again. It was instantaneous. I then closed Nero Wave Editor completely and available memory still didn't increase. I re-opened Wave Editor and the wave file. It was still instantaneous.

You're mistaking a system making a cache of it versus the program itself actually allocating that space in RAM and putting the audio in there. Once the file is loaded into the program completely it is gone from cache. There are still chunks left. This is why it loads quicker.

So where does the file go once it's loaded into the program and out of cache? Does the program use a different RAM to perform tasks and store data?

If the full file was kept in RAM it would load instantaneous.

It does.

The advantage of large amounts of RAM is to get rid of the page file and allow a huge head room so when you finally do use programs that take up a lot of RAM, they won't take up all of it and crash.

Sounds like assumption based on not really knowing how it works.
Programs have no idea where the memory they're using comes from. Zero, zip, ziltch. As far as a program is concerned if Windows says it can have 8GB of memory, it's got 8GB of RAM to use and has no idea if 3GB of it comes from RAM and the rest is assigned to the paging file. If it happens to be using data which has been moved to the paging file, when the program requests it Windows will tell it to wait, shuffle the data around, then tell it where in RAM it can find the data it requested. The program is basically oblivious to it all.
There's no logical reason why having a huge amount of RAM should make a program less likely to crash, unless maybe if you're running Windows 98. It might though allow it to run faster.

Ghitulescu is not wrong. If you have loads of RAM, a mechanical hard drive, and still have a hard drive, your system is going to grind to a halt anytime you use the hard drive.

Well of course it'll slow down if it's got to access the hard drive, but did you completely miss my point about the paging file not actually being used for data when you have lots of RAM?

The page file was there to cover up the fact that software requirements grew faster than what RAM could offer. In the 64-bit world it is pointless.

Okay, well I guess then you know more about Windows memory management than Microsoft does. How much does enabling the paging file slow down your system? I assumed you've performed benchmarks, or tested the time it takes to perform every day tasks, or are you just assuming?
I've tried different memory intensive tasks with 4GB of RAM while changing the paging file settings and even while disabling it. I couldn't determine any of it changed the speed of my computer.

Ghitulescu

18th August 2011, 19:17

I'll give you the half-point for one correct asertion: that the program doesn't know wherefrom the requested memory comes. Yet a program can lock a certain memory block and these blocks must reside in the RAM. So a program can count on genuine RAM memory if it wants so. Any memory-intensive programs will crash, as Ramicio said, if no free RAM exists (and no load/unload mechanism is available). If they will really crash, or throw an exception or simply ignore the result, is programming dependent - and this behaviour is shared by windows (there are several memory allocation routines available under windows).

hello_hello

18th August 2011, 19:24

ramicio

18th August 2011, 19:32

hello_hello

18th August 2011, 19:44

Microsoft's products are for consumers, not power users. Even their corporate software is dumbed down. They going to keep the page file around and enabled by default, no matter what, because they don't want to hear people bitching about running out of memory, if that should happen somehow.

I asked if you've performed any speed tests yourself, not if you've got some rhetoric to offer.

Page file settings don't apply until you restart your computer. Your tests are pointless. Any computer with a hard drive and page file will all but lock up while loading a new program.

Who said I didn't restart my computer?
So if my computer will lock up when it's loading a program while using a page file, but it doesn't load the program any faster without a paging file, how do you explain it?

And Microsoft would rather hide RAM usage by putting it in a page file, that the average user won't have any idea about, than to actually show that RAM is being used by the OS...to make their bloated OSes lighter looking to the average fool.

You ever opened Task Manager?

Again, you're mistaking something for another thing. Anything you open HAS to pass through RAM. It doesn't mean Nero Wave Editor is allocating that and storing it in its own RAM space.

Once again, you're mistaking something I didn't say for something I did. I never said Nero was storing the file in it's own RAM space, but I did say that after closing and reopening Nero completely the wave file still opened instantaneously again. Mind you however it happens, I'm not sure what your point is there. Although you apparently forgot to acknowledge you were wrong about the wave file opening instantaneously.

ramicio

18th August 2011, 19:50

hello_hello

18th August 2011, 20:15

Everything locks up because it puts valuable resources in the page file. Tax the hard drive by loading a program and everything else locks up. It's not RAM, and it shouldn't try to act like it. I never said anything about the program loading faster with or without a page file, only the affect it has on the system.

Well logically if it locks up with a paging file, but doesn't load any faster without one, it's not locking up because of the paging file.

I don't need to perform any tests. I know how things work, and it's common sense.

Yep, I know lots of people who still think cleaning the registry is common sense and improves the performance of their PC too. Like you though, I've never had a single person offer test results or benchmarks to prove it. In fact, like you, most refuse to when asked so they continue to enjoy the benefits of the placebo effect.

Things are thrown out of the cache not long after the program shuts down, even if your computer sits idle. It's pointless behavior.

And things are kept in RAM after your computer shuts down? I'm not understanding your point.

The normal user looks at the bottom of the task manager at % used. They can't interpret the memory stuff on the performance page. Caching things do not change the % of the RAM used. If it does, then something is wrong.

No idea what your point is there either, nor how it relates to how a paging file slows down a PC with lots of RAM.

Ghitulescu

18th August 2011, 20:16

What program loads an entire sound file into RAM?
Nero Wave Editor for one.

So where does the file go once it's loaded into the program and out of cache? Does the program use a different RAM to perform tasks and store data?

Once again, you're mistaking something I didn't say for something I did. I never said Nero was storing the file in it's own RAM space, but I did say that after closing and reopening Nero completely the wave file still opened instantaneously again. Mind you however it happens, I'm not sure what your point is there. Although you apparently forgot to acknowledge you were wrong about the wave file opening instantaneously.

No comments, I'll let the passages speak for themselves :)

Is going any of you mods stop this troll monkeying around and hijacking the other people' threads?

ramicio

18th August 2011, 20:19

It does lock up because of the page file. Why would a program load faster with or without the page file? It still has to read from a hard drive. Hard drives suck at randomness. Memory, it doesn't matter. It doesn't need to keep things contiguous. Haha, placebo effect, what ass-ness to even mention.

hello_hello

18th August 2011, 20:29

No comments, I'll let the passages speak for themselves

Well it says more for your powers of observation than it does about the topic.
It was an ongoing conversation. I was asked which program loads an entire sound file into RAM? I explained Nero Wave Editor does, however I was simply saying that's what happens when I use Nero Wave Editor.
And no, I'm still not saying Nero loads the data in it's own RAM space. The fact that it's still in RAM after I close Nero points to it not being the case. Use your brain.

Is going any of you mods stop this troll monkeying around and hijacking the other people' threads?

Well given your last post was just a desperate attempt at point scoring and didn't contribute to the discussion, it appears if there's a troll the mods should be looking at you're it, no matter how much your belief you have a divine right to never be disagreed with keeps you in denial.

Got a link on the topic as I requested, or are they all being stockpiled with those Taiwanese made Verbatim DVDs? :D

Ghitulescu

18th August 2011, 20:30

Nero, like most other audio processing tools, keeps a peak file of several MB useful to instantly redraw the waveform without parsing again the whole file. Such a small file can be instantly loaded on todays' computers - while the real load, chunk-wise, may be postponed (on demand).

hello_hello

18th August 2011, 20:36

It does lock up because of the page file. Why would a program load faster with or without the page file? It still has to read from a hard drive.

Err... Why wouldn't it load faster without a paging file if having one causes the drive to write/read from the paging file while it's loading the program? Otherwise please explain how having a paging file cause the PC to to lock up while loading programs. You're not even making sense now.

Haha, placebo effect, what ass-ness to even mention.

That's the definition of a placebo effect. The person benefiting from it is completely unaware it's not real, while as inevitably as day becomes night, exclaiming "I don't need to perform tests".
It's a lot like those people who claim they can pick an MP3 being played in their car at 50 paces. They never want to perform listening tests either.

hello_hello

18th August 2011, 20:41

Nero, like most other audio processing tools, keeps a peak file of several MB useful to instantly redraw the waveform without parsing again the whole file.

If you read my posts properly, you'll see the file I loaded was far larger than several MBs and it reloaded instantly even after closing and re-opening Nero.

Such a small file can be instantly loaded on todays' computers - while the real load, chunk-wise, may be postponed (on demand).

You'll also see I never claimed it'd remain in RAM for an eternity. Naturally if it's no longer being used Windows will use the RAM for something else if need be.
Where's this postponed, imaginary real load coming from when I reopen a wave file? And for that matter, why would the "real load" happen any quicker if I tried to open a wave file at the same time which wasn't already in RAM?

ramicio

18th August 2011, 20:54

If you have memory on a mechanical drive, accessing the drive with someone else is going to force the computer to wait until the head seeks the page file part of the drive. If you don't get it, I can't help you. A page file is 100% unnecessary for the normal user having a minimum of 6 GB of RAM. I don't care if it's stuff that's rarely accessed, there is no reason to put anything from RAM onto a storage device. Windows obviously chooses to put things in the page file that it needs, which is bad behavior. There is no need for any sound editor to have a file in RAM. It just needs to analyze the peaks and draw a rough waveform.

I don't need to do tests for a brand new computer with a hard drive, small amount of RAM, and a page file to tell me why it all but locks up when launching programs. I don't care how super everything in a computer is. If it's a mechanical hard drive, and there is a page file, this behavior is inevitable. It's common sense, and it's been a huge and annoying bottleneck of computers for years. Now we have 64 bit OSes to be able to have tons of RAM. The page file should be dead by now. If you don't want a lot of RAM, you should stay with a 32-bit system.

hello_hello

18th August 2011, 21:10

If you have memory on a mechanical drive, accessing the drive with someone else is going to force the computer to wait until the head seeks the page file part of the drive. If you don't get it, I can't help you.

I get it! Which part of not accessing the drive with someone else failing to speed up drive operations are you not understanding? If it doesn't speed up when there's no paging file, then there's no lock-up and the paging file isn't slowing things down. It's not that hard, surely.

I don't care if it's stuff that's rarely accessed, there is no reason to put anything from RAM onto a storage device. Windows obviously chooses to put things in the page file that it needs, which is bad behavior.

No, you're assuming that's what it does. I argued that with plenty of RAM it'll only assign memory to the paging file which isn't being used, therefore having no detrimental effect on data access.

There is no need for any sound editor to have a file in RAM. It just needs to analyze the peaks and draw a rough waveform.

Well I'll confess you made me laugh out loud with that one.
So what happens when I play it? Is the data loaded directly from the hard drive to the CPU? What if I compress it. Does the program perform calculations on the data while reading directly from the hard drive? Where does it store the new data? What's the point of having 12GB of RAM if you're not going to load anything into it?
Anyway, whether there's a "need" to or not, I've shown that's what happens.

I don't need to do tests for a brand new computer with a hard drive, small amount of RAM, and a page file to tell me why it all but locks up when launching programs. I don't care how super everything in a computer is. If it's a mechanical hard drive, and there is a page file, this behavior is inevitable. It's common sense, and it's been a huge and annoying bottleneck of computers for years. Now we have 64 bit OSes to be able to have tons of RAM. The page file should be dead by now. If you don't want a lot of RAM, you should stay with a 32-bit system.

My PC doesn't have huge amounts of RAM, it does have a paging file, and it doesn't all but lock up when loading programs.
The placebo effect is an powerful thing. Common sense should tell you that too.
I'm not arguing disabling the page file will have an adverse effect on PC performance if you have buckets of RAM, but I am unconvinced having one will slow it down by comparison, and I do believe not having one would potentially waste RAM. You've not offered any common sense reasons to explain why it would, aside from some belief that because Windows shows the paging file is being used, it must contain data required by programs.

ramicio

18th August 2011, 21:21

If it's putting data in the page file that's not being used, it will be slow to load that data. I don't care if it's data that's going to be accessed a second later, or an hour later. You're referring to Superfetch, anyway, not Windows' memory management.

PUTTING THINGS MEANT FOR MEMORY ON A MECHANICAL DRIVE SLOWS EVERYTHING DOWN. How hard is this to understand? It's not placebo that getting more RAM and disabling the page file speeds things up. You are eliminating a slow, mechanical device completely from the equation.

Ghitulescu

18th August 2011, 21:26

Since the first day of PC-DOS, then MS-DOS and finally Windows, there were always 3rd party software aiming at improving the environment. Names like Norton were well-known to the earlier users of the PC. Slowly but steadily, all the functionality that those 3rd party tools offered has been incorporated into windows. Good or bad, in the end the enduser could avoid unnecessary costs for a reasonable performance. Of course, s/he could have 3rd party tools with a slightly better performance (MS always hindered these tools, but it was hard to be proved, those who tried, like DR-DOS were driven out from business) for a nice price. And the antitrust trials, always opposed by M$, couldn't stop this trend.
Conclusion: I would let the RAM management entirely to windows. Only the super experts may tweak the last drop of power form the system. Like everywhere in this world.

hello_hello

18th August 2011, 21:44

If it's putting data in the page file that's not being used, it will be slow to load that data. I don't care if it's data that's going to be accessed a second later, or an hour later. You're referring to Superfetch, anyway, not Windows' memory management.

For the hundredth time, if you've got lots of RAM it's only assigning unused memory requests to the paging file. Memory which has been requested but which the program isn't using. It contains no data.

PUTTING THINGS MEANT FOR MEMORY ON A MECHANICAL DRIVE SLOWS EVERYTHING DOWN. How hard is this to understand?

Which part of "it's not putting data on the mechanical drive" can't you grasp? Download and install a utility that monitors page file usage, enable the paging file, then post back with some facts.

It's not placebo that getting more RAM and disabling the page file speeds things up. You are eliminating a slow, mechanical device completely from the equation.

Who said adding more RAM doesn't speed things up? Bundling it with a disabling of the paging file proves nothing about the effect of disabling the paging file on it's own (after you've added the RAM) and whether you can measure a further non-placebo speed increase.
FFS. I added more RAM to my PC, I didn't disable the paging file, therefore not disabling the paging file makes my PC faster. Doesn't make sense either, does it?

ramicio

18th August 2011, 22:12

Adding more RAM only makes things faster when there stuff running that can max out what you don't have. Having 192 GB of RAM versus 6 GB is going to absolutely nothing for running barely anything. How does this unused data get back to the RAM? It has to transfer from the hard disk. Which is slow. Is your computer going to intuitively know maybe 10 seconds before you do something when you are going to want to load data that has been thrown to the page file? NO. Computer are calculators, not brains. Computers are nowhere near being as capable of any brain on earth. It can do a calculation faster. It's what it's built for. Calculating. Not thinking. In your world the page file is some magic wand that can read your mind.

In the 64 bit world a page file is useless, because you should know what you're doing and why 64 bits is beneficial over 32 bits, and you should be building a system with tons of RAM. If you don't plan to have > 3 GB of RAM then you shouldn't even run a 64 bit OS. There is no performance benefit, whatsoever.

Groucho2004

18th August 2011, 23:12

If you don't plan to have > 3 GB of RAM then you shouldn't even run a 64 bit OS. There is no performance benefit, whatsoever.

If you are talking about 32 Bit apps running on a 64 Bit OS - yes. Applications written to take advantage of a 64 Bit OS and processors will run faster. x264 is one example.

hello_hello

18th August 2011, 23:18

In the 64 bit world a page file is useless, because you should know what you're doing and why 64 bits is beneficial over 32 bits, and you should be building a system with tons of RAM. If you don't plan to have > 3 GB of RAM then you shouldn't even run a 64 bit OS. There is no performance benefit, whatsoever.

And around and around and around we go....
You can repeat the same thing from now to eternity, but it does nothing to show any data is being written to the page file simply because you have one. Windows might be using it, but not necessarily for data.
Why would Windows write data to the paging file if there's still plenty of free RAM? You're making an assumption which doesn't make sense.

When it comes to a 64 bit OS, or whatever theory you've invented for yourself, having a paging file may very well be useless in terms of it being required to swap out data... I never said otherwise. If however, it can be used to swap out memory which programs have requested but aren't actually using, then it's not useless in terms of saving RAM from being wasted instead of being put to use.

It's a fairly simple concept you continue to ignore in preference to repeating yourself.

Groucho2004

18th August 2011, 23:38

Why would Windows write data to the paging file if there's still plenty of free RAM?

As soon as you minimize an application, Windows will swap most of the memory allocated by that app to the page file. There are also some system services and processes which use the page file by default. It doesn't matter how much physical memory is still available.

ramicio

19th August 2011, 00:16

x264 is a program of lots of calculations and fancy math. It requires much more math than most other CPU uses. And there are new and improved instruction sets for the 64 bit processors versus the old 32 bit ones, so that helps more so than being able to calculate to 64 bits of accuracy.

What do you mean it doesn't use it for data? Does it use it to store food, water? I don't understand what you mean by not data. A computer is data and calculations. Windows randomly throws things into a page file to keep RAM free and not look like a bloated OS. Now this stuff you have cached in the hard drive has to load from the hard drive. It's no faster to send the cached data from the hard drive to the RAM than to just load it.

hello_hello

19th August 2011, 02:21

Windows randomly throws things into a page file to keep RAM free and not look like a bloated OS.

Why would it have to? Even if the RAM is 100% full of data, why does Windows have to randomly dump some of it? It could just as easily replace the unused data in RAM with the data that's actually needed, or swap it out when new data needs to be loaded, without having to randomly clean some of it first. Even if Windows tries to keep some RAM free, if it's going to swap unused data out to the paging file and there's no paging file, wouldn't it just delete it instead and you'd still have to load it from the hard drive again.
It sounds like you're making big assumptions.

http://support.microsoft.com/kb/2267427

In modern operating systems, including Windows, application programs and many system processes always reference memory using virtual memory addresses which are automatically translated to real (RAM) addresses by the hardware.

Processes and Address Spaces
All processes (for example, application executables) that run under 32-bit versions of Windows get virtual memory addresses (a Virtual Address Space) that range from 0 to 4,294,967,295 (2*32-1 = 4 GB), no matter how much RAM is actually installed on the computer.
In the default Windows configuration, 2 GB of this virtual address space is designated for private use of every process, and the other 2 GB is shared between all processes and the operating system. Typically, applications such as Notepad, Microsoft Office Word, and Adobe Acrobat Reader use only a fraction of the 2 GB of private address space. The operating system assigns RAM page frames only to virtual memory pages that are in use.

The way I understand it, only the portion of the requested virtual address spaces which is in use would be assigned RAM. The rest is assigned to the paging file and the program is happy about the amount of memory it can use. Hopefully that's correct....

When it comes to 64 bit operating systems, I'm pretty sure the sizes of the address spaces are larger.

Even from just reading the page I linked to though, there's obviously so much going on "under the hood" when it comes to memory management, there's no way I'd simply disable the paging file and assume it's making my PC run faster, no matter how much RAM I have. I'd be testing it extensively before coming to any conclusions, because I'd be astounded that with plenty of RAM the existence of a paging file will slow the PC down.

ramicio

19th August 2011, 02:28

Again, anything having to read from a physical spinning disk (or even fetching data over a SATA link) is going to slow something down, somewhere. I don't care how useless or unimportant YOU think the data is that it doesn't deserve to be loaded as fast as from RAM, the page file is a bottleneck. It just extends the virtual memory space onto a disk, and it even throws some inactive stuff on there. But again, a computer can't assume when that inactive stuff is going to be accessed. It can't read your mind. Why don't you try putting your page file on a optical RW drive with a UDF volume, or some slow CF drive, and tell me nothing is slow.

hello_hello

19th August 2011, 07:55

Well, I won't bother banging my head against a wall any longer. The concept of unused memory (memory which is reserved for a program but it's not using because it contains no data) being assigned to the paging file instead of wasting RAM is obviously not in line with the placebo effect you currently enjoy.
If you're happy disabling the paging file has increased the speed of your PC, while claiming you don't need to test it, yet now telling me to perform tests, I'm more than happy for you.

Ghitulescu

19th August 2011, 08:11

The speed of a well done RAID array of modern HDDs is aproaching fast to that of the RAM. What people knew some 10-15 years ago seems to fall into oblivion today - the real bottleneck of any PC architecture is the internal bus - everything will run close to specs on both north and south bridges independently, however, mixing them will force data to pass through the "CPU strait" with serious reduction of traffic. More RAM cannot help, faster ones again not (if the memory controller is slow).

hello_hello

19th August 2011, 10:32

The speed of a well done RAID array of modern HDDs is aproaching fast to that of the RAM.

Comedy gold!!

The old RAM in this PC (DDR2 800) has a real transfer speed of 3920MB/s according to NovaBench. My (also old) pair of hard drives running as a RAID-0 array top out at about 89MB/s according to the NovaBench benchmark on the fastest part of the drive (HDTune's standard benchmark puts them at about 155MB/s).

A well done RAID array would need to consist of over 100 hard drives to match my RAM speed.
A well done SSD array would probably need 12 hard drives or more.

I wonder what defines a badly done RAID array?

Ghitulescu

19th August 2011, 11:27

Groucho2004

19th August 2011, 12:11

Loading a 4MB MP3 file from memory would be as instataneous as from a defragmented RAID.

Again you're providing evidence that you have no interest in what the OP posted and just argue for the heck of it.
This is about moving large amounts of data around, 20 or even 50 GB. I can assure you that the end user would notice the difference between 200 MB/s and 20 GB/sec.

hello_hello

19th August 2011, 12:17

I was talking about the speed the enduser sees. Not about specifications, nor "theoretical" benchmarks, that are good only for benchmarking but have no practical use.
Loading a 4MB MP3 file from memory would be as instataneous as from a defragmented RAID. In the end whether it takes 400ms or only 2ms is the same for the average Joe.

And it's still a load of twaddle.
Loading an MP3 doesn't just involve opening a file, it involves opening an audio player and potentially lots and lots of additional files. And if you don't have a lot of RAM, possible swap file involvement as well. And as Groucho2004 said, the thread is really about moving large amounts of data, not opening a single MP3. If I save a single video file to a single drive I can easily see the speed difference in many seconds, even minutes, compared to saving it to a RAID volume which is still only twice as fast as a single drive and nowhere near the speed of RAM.
I'm so used to running a RAID-0 volume, using a PC with a single hard drive is painfully slow.... even an average Joe like me doesn't have much trouble telling the difference.

My first computer, a 200MMX was faster than anything else until the PIII appeared (in audio processing), yet it scored in half or one third of a PII in (that time) Sisoft Sandra. Benchmarks are [mostly] for marketing purposes, real life computers are way different.

Terrific. I'll let my hard drives know while I'm using them in the real world. I guess it'll make them load and save files faster....

ramicio

19th August 2011, 15:43

A RAID array of SSDs can't even touch RAM speeds yet. Saying benchmarks are useless is the losing battle cry of the AMD fanboy. Intel and AMD have much different price points. AMD goes after value versus performance. Intel goes after all-out performance at cost. Those who actually want the best will pay for it. Even so, Intel has been killing AMD in the low-power/low-cost arena lately, anyway.

Why can't you get this through your head that the page file is useless if you have enough RAM? Why is it useful if the system isn't putting anything in it, which you claim? Unused is different than free space. The way you're describing it, the system puts into the page file cached stuff that is older. So I ask you, what is point? Disabling the page file is going to keep that stuff in RAM. There's your speed. Why is it so hard to understand that anything put into a mechanical drive is going to be slower? Again, how? I don't care what you think needs to be read fast, or not. How you can say a hard drive acting as RAM is going be just as fast as RAM is beyond me. Again, it's common knowledge that Windows does use the page file for stuff even if you have plenty of RAM free. So again, how would this not slow down a system? Your lack of common sense baffles me.

Ghitulescu

19th August 2011, 16:54

Again you're providing evidence that you have no interest in what the OP posted and just argue for the heck of it.
This is about moving large amounts of data around, 20 or even 50 GB. I can assure you that the end user would notice the difference between 200 MB/s and 20 GB/sec.
I answered already the question. See here the conclusion (you may want to read the whole statement):
Conclusion: I would let the RAM management entirely to windows. Only the super experts may tweak the last drop of power form the system. Like everywhere in this world.
Concerning the off-topic statements, please notice that it was someone else that slightly detourned the argumets' flow, and he did this with the experience an "off-topicer" has.

Not for small amounts of data, like those used in avisynth scripts, I agree. It may add, maybe a millisecond here, another there, so the speed adds so in the end a faster system will be fast.
The issue I raised was that avisynth doesn't work with huge data blocks (definitively not with a whole Blu-ray at once), and in the end, all these data chunks are kept in the live RAM by windows, thus a RAM disk being useless. The speed of avisynth scripts does not depend on the HDD access times - unless there's only a NULL filter (direct copy). One should not enter the world of video processing with minimum specifications.

About one having 50GB of RAM (to keep a whole file of yours in a RAM-drive), I'll be damn :) never heard of in the PC world.

A RAID array of SSDs can't even touch RAM speeds yet. Saying benchmarks are useless is the losing battle cry of the AMD fanboy. Intel and AMD have much different price points. AMD goes after value versus performance. Intel goes after all-out performance at cost. Those who actually want the best will pay for it. Even so, Intel has been killing AMD in the low-power/low-cost arena lately, anyway.

Why can't you get this through your head that the page file is useless if you have enough RAM? Why is it useful if the system isn't putting anything in it, which you claim? Unused is different than free space. The way you're describing it, the system puts into the page file cached stuff that is older. So I ask you, what is point? Disabling the page file is going to keep that stuff in RAM. There's your speed. Why is it so hard to understand that anything put into a mechanical drive is going to be slower? Again, how? I don't care what you think needs to be read fast, or not. How you can say a hard drive acting as RAM is going be just as fast as RAM is beyond me. Again, it's common knowledge that Windows does use the page file for stuff even if you have plenty of RAM free. So again, how would this not slow down a system? Your lack of common sense baffles me.

If one reads my statements, he'll probably notice that I haven't say RAM = HDD in terms of speed, but that the speed difference became smaller. The access speed of the modern HDDs is more than 200+x that of the former models, the bottleneck being here the interface and the file system. The speed of HDDs is still vastly inferior to that of RAM, whose access speed did not improved 200x :), but a mere ~20x (compared to the RAM modules of the equivalent PC-era). The time passes by and the difference will become less and less important, especially when one compares the prices :).

SSDs were initially advertised as almost reaching the speed of RAM, yet the mapping algorithm killed this speed, so there are mechanical HDDs faster than SSDs.

I had experienced RAM-drives (long time ago, when they had sense). Has any of you used them or just speak out of theoretical perspective (RAM is faster than HDD therefore RAM-drives should be faster than the real ones :) )?

ramicio

19th August 2011, 16:58

RAM drives are fast, but you will never see, as a file system, the speeds RAM can provide. You're better off getting an SSD than using up RAM for something silly like that. Patience is a virtue. Why people are getting more and more impatient and need everything NOW, instead of waiting a little bit, is beyond me. I'd like to smack some people back into when computers sucked in the 90s to show them how good they have it now. What HDD is as fast as an SSD? Yes there are slow as balls SSDs out there still. Those still represent what first came out. There is not a single HDD that can outperform a common-era SSD. It would take at least 4 HDDs in a RAID 0 array to even come close, and then you are still losing out pathetically in the IO department.

Ghitulescu

19th August 2011, 18:07

Nope, never use a SSD as a temporary directory. It will be distroyed in few weeks, when its cells reached the maximum write cycles.
The memory management of windows is still the best the untrained user can get. Other than a handful of fake accelerators on google (that take the money and run) nothing is available for windows in this respect. If one has enough RAM that he thinks of a RAM-disk, then he has enough memory to let windows works as default.

ramicio

19th August 2011, 18:11

You're terribly misinformed about SSD reliability. With wear-leveling (TRIM) it would not be unheard of for an SSD last decades as something always being written to and erased (a temporary drive for video production, photo stuff, anything really.) The bigger the drive the longer it will last.

hello_hello

19th August 2011, 19:44

Concerning the off-topic statements, please notice that it was someone else that slightly detourned the argumets' flow, and he did this with the experience an "off-topicer" has.

I don't get it. Do you deliberately lie all the time or are you in complete denial?
Look at the thread. ramicio first mentioned the paging file in post #7. You provided your typical misinformation on the topic in post #8. I was third in line, post #9. We've all been discussing it ever since. The only reason you kid yourself it was me who detoured the topic's flow is because you can't stand anyone disagreeing with you.
What's even funnier, is the fact that once you'd performed a liitle back-flip, we've been in agreement on this topic. At least when it comes to disabling the paging file if you have plenty of RAM. I don't think you should disable it either. :)

The issue I raised was that avisynth doesn't work with huge data blocks...

No you never mentioned AVISynth, you referred to playing MP3s.

If one reads my statements, he'll probably notice that I haven't say RAM = HDD in terms of speed, but that the speed difference became smaller.

No, you said "The speed of a well done RAID array of modern HDDs is approaching fast to that of the RAM." Then you went on to offer some nonsense about the real bottleneck being the internal PC BUS.

The access speed of the modern HDDs is more than 200+x that of the former models, the bottleneck being here the interface and the file system.

Nonsense. Not when it comes to mechanical drives. Hard drives have been spinning at 7200rpm for years. How has the access time improved 200+x if the platters are still spinning at the same speed?
I just looked up the specs of an old 20MB Seagate drive.
http://www.seagate.com/ww/v/index.jsp?vgnextoid=b9df99f4fa74c010VgnVCM100000dd04090aRCRD
Average latency 5.6ms.
Newer Seagate SATA3 drives, average latency 4.16ms.
http://www.seagate.com/ww/v/index.jsp?vgnextoid=b9df99f4fa74c010VgnVCM100000dd04090aRCRD
How far back are you going for your comparisons? The dark ages?

I had experienced RAM-drives (long time ago, when they had sense). Has any of you used them or just speak out of theoretical perspective (RAM is faster than HDD therefore RAM-drives should be faster than the real ones :) )?

Yep, I've used RAM drives plenty. Please tell me they're not much faster than hard drives, because I could use a good laugh.

Nope, never use a SSD as a temporary directory. It will be destroyed in few weeks, when its cells reached the maximum write cycles.

More misinformation. Unless you're assuming we're all still living in the early 90's too?

http://www.storagesearch.com/ssdmyths-endurance.html
"In most applications, high endurance flash SSDs can have a reliable operating life which is typically 3 times as high as that of a hard drive."

ramicio

19th August 2011, 19:53

Hard drives have gotten faster transfer speeds over time, because that's just physics. You have an equal size disc spinning the same speed, but you pack data closer together, it's going to read faster. This is just sequentially. It's exactly like optical media. They don't spin any faster over time, but they have gotten faster just because of the data being packed closer together. Access times have improved, but only because of advancements in materials, not anything to do with data density. They have also barely increased.

ramicio

19th August 2011, 19:55

The page file should be disabled if you have a lot of RAM. No one will change my mind. I don't like when people disagree. I will go many steps forward and say I think people who don't agree with me deserve to die for that.

hello_hello

19th August 2011, 20:23

ramicio

19th August 2011, 20:26

hello_hello

19th August 2011, 20:58

As soon as you minimize an application, Windows will swap most of the memory allocated by that app to the page file. There are also some system services and processes which use the page file by default. It doesn't matter how much physical memory is still available.

Not according to Task manager. I've currently got four programs open. Firefox using the most memeory at 203,080K.
One by one I maximised them, then opened Task Manager. Page file usage, 646MB. I then minimised each program one by one. By the time the only non-minimised application remaining was Task Manager, page file usage was 646MB.
I did a little more surfing before submitting this post. Opened and closed quite a few tabs. Firefox is now using 235,752K. Page file usage 694MB. Minimising Firefox again didn't change that.

hello_hello

19th August 2011, 21:06

Why should anyone fold on their beliefs? To do that is weakness, and it means whatever new comes along will be lost just as easily. I can't stand these open-minded liberal hippie douche bags. I only even look at evidence when it doesn't defy logic.

This isn't a religious discussion, it's a discussion regarding something which can easily be tested and proven. It's nothing to do with weakness, which has no direct relationship to being open minded.
You can believe all my arguments defy logic if you wish, it does nothing though to prove disabling your paging file has increased the speed of your PC.
Me, after I upgrade this PC shortly, I'll probably put it to the test myself, because leaving the blinkers on would be my definition of weakness. If I'm wrong and disabling the paging file speeds up normal tasks, I'll admit it, because I'm not too weak to do so.

hello_hello

19th August 2011, 21:33

Why can't you get this through your head that the page file is useless if you have enough RAM? Why is it useful if the system isn't putting anything in it, which you claim?

Sorry, I missed your question....
The way I'm describing it, which is the way I understand it, a program might request "X" amount of RAM but will use less than "X". If what isn't being actually used by the program is assigned to the paging file, RAM is not being wasted.

Maybe my understanding is wrong.... anything's possible... I'll have to do some more research on it later. You could be correct and you've got so much RAM your system runs fine without a paging file, but I'm just not convinced having one will slow it down.

ramicio

19th August 2011, 21:40

Beliefs have only a small part to do with religion, so cut that crap out.

What you state has nothing to do with speed, it has to do with allocation. You're talking about the commit charge. How would it be not be slow to stuff away unused memory space on a hard drive? When the program needs that space it would write it to the hard drive. Even if it brought that space back to RAM, it's empty beforehand. This would also count as pointless behavior, and it's not how it works. You would be better off not committing more than required. A hard disk is not made for memory. Even the interface is too slow for RAM purposes. Yes, Windows does still unnecessarily use the page file for stupid crap even if you have plenty of RAM. It's pointless to keep a page file with plenty of RAM. It was originally called a swap file. It would swap memory pages that are currently unused to a hard drive to make way for currently running stuff. But I ask you again, when the computer again needs that data, how is it not slow to have to pull that from the hard drive? It's best to just keep memory for memory, and storage for storage.

hello_hello

19th August 2011, 21:52

Beliefs have only a small part to do with religion, so cut that crap out.

You're the one who made it sound religious.
Thanks for the advice, I will out the crap. I didn't bother reading the rest of your post.

ramicio

19th August 2011, 22:06

I didn't bother reading the rest of your post.

:thanks:

hello_hello

19th August 2011, 22:33

The pleasure... and it made quite a noticeable difference... was all mine.

Groucho2004

19th August 2011, 23:09

Not according to Task manager. I've currently got four programs open. Firefox using the most memeory at 203,080K.
One by one I maximised them, then opened Task Manager. Page file usage, 646MB. I then minimised each program one by one. By the time the only non-minimised application remaining was Task Manager, page file usage was 646MB.
I did a little more surfing before submitting this post. Opened and closed quite a few tabs. Firefox is now using 235,752K. Page file usage 694MB. Minimising Firefox again didn't change that.

Try another program, IE for example. I don't know why Windows doesn't page out the memory allocated by FF. I suppose FF is designed this way to be more responsive.

laserfan

19th August 2011, 23:31

...RAM-drives... any of you used them or just speak out of theoretical perspective
Had you paid attention, you'd know at least one member of this thread has stayed on-topic! ;)

It does seem you guys have managed to lose the OP! :devil:

Ghitulescu

20th August 2011, 10:50

hello_hello

20th August 2011, 12:06

I know about TRIM. But the writing cycles still remain the same.

And the misinformation continues.....
No they haven't. They've improved substantially over the years. It's 2011. ;)

http://www.storagesearch.com/ssdmyths-endurance.html
"In the early days of flash SSDs managing this was a real headache for oems and users. The maximum number of write cycles to an address block - the endurance - was initially small (about 10,000 write cycles in 1994, rising to 100,000 in 1997)."

Write endurance rating:
"2 million cycles. (The typical range today for flash SSDs is from 1 to 5 million. The technology trend has been for this to get better. "

"AMD marketed "million cycle flash" as early as 1998."

Ghitulescu

20th August 2011, 12:58

Again an off-topic issue, from an experienced off-topic-poster, but to please Groucho2004:

A million write cycles memory won't need wear levelling algorithms, not otday, not at the current technologies changes (has anyone, besides me :), older than 5 years HDDs or computers?). The rest is statistics, and figures juggling and so on.
Also one needs to come back to real life. Start first with some relatives, the SD cards (or CF or xD or or or or even memory sticks). They are of two major types SLC and MLC. MLC being the worst, by far. A SLC memory stick (industrial) costs in Germany about 10x the price of a MLC one (consumer), from the same label (SanDisk) and similar size and ~ the same speed. I prefer to buy 10 MLCs than 1 SLC, unless I have to provide reliable data to my customers (should I have customers).
Now, what are the odds that a 1 million write cycles SSD would cost the same as a 10000 one? What are the chances that the unsuspecting customers pays the premium price for an ancient technique (10k) thinking it's really 1M? I'm talking about nonames, for names one needs to pay the name too, and still not 100% sure he'll get what he paid for. He's just a stupid consumer, not a pro one, he has no means to check the validity of the marketing dept's claims. I mean here, I've never seen in my life that SLC or MLC being actually written on the product. When the product sold in the '80ies came usually with full specs and a schematics, those sold in the '90ies only with specs, now one should be lucky to see today the specs on the package. Everything else is marketing balooneys, with little support in the practice. Can you really prove that your HiFi has a lower than advertised noise floor? Hardly. And even if you do, you'll be immediately confronted with the lack of testing conditions.

hello_hello

20th August 2011, 13:54

Again an off-topic issue, from an experienced off-topic-poster, but to please Groucho2004. A million write cycles memory won't need wear levelling algorithms, not otday, not at the current technologies changes (has anyone, besides me , older than 5 years HDDs or computers?). The rest is statistics, and figures juggling and so on.....

Again more nonsense, from a an experienced inverter of fairy tales.
Can you really prove any of what you said isn't anything more than another one of your fantasies?
Is it a similar fantasy to the one where I posted about off-topic issues while replying on the same subject you were discussing with someone else.... or is it even more magical?
How does discussing on off-topic issue please Groucho2004?

Ghitulescu

20th August 2011, 15:18

Why would a SSD those each cell withstands 1M write cycles need TRIM? No other similar, related or even distant technology needs TRIM, yet 1M or even better 5M cells need TRIM?
Have you actualy written 1M times something?

I worked with RAM-disks long time ago, when my computer had more RAM than HDD (yes), for a particular reason - I needed to compile something and the time lost with opening each small file the compiler needed took a very long time on HDD, yet much faster on RAM-disks. Small files, here's the killer advantage of a RAM-disk, opening speed, not [only] the transfer rate. Larger files, like the music ones, were still faster transferred, however the paging subsystem faulted the speed advantage.

hello_hello

20th August 2011, 16:05

Why would a SSD those each cell withstands 1M write cycles need TRIM? No other similar, related or even distant technology needs TRIM, yet 1M or even better 5M cells need TRIM?
Have you actualy written 1M times something?

If I'd copied all the rhetoric you've posted and saved it each time, I would have gotten fairly close.
That's your evidence? The fact you haven't saved something a million times? I knew you were making it up as you went again.

Nope, never use a SSD as a temporary directory. It will be distroyed in few weeks, when its cells reached the maximum write cycles.

When you've made up your mind about the required number of write cycles, please let me know. As much as you don't seem to realise it, you can't have it both ways. Either SSDs don't require a lot of write cycles and will be destroyed in a matter of weeks, or they do so they won't be.

A quick search of my C drive (which contains only the OS and programs, very little user data, no temp internet files) indicted over 250 files have been modified today. My D partition, which contains nothing but user data, including my Firefox profile, another 44 files. That's around 300 today. How many times they've been modified today, I don't know.

I worked with RAM-disks long time ago, when my computer had more RAM than HDD (yes), for a particular reason - I needed to compile something and the time lost with opening each small file the compiler needed took a very long time on HDD, yet much faster on RAM-disks. Small files, here's the killer advantage of a RAM-disk, opening speed, not [only] the transfer rate. Larger files, like the music ones, were still faster transferred, however the paging subsystem faulted the speed advantage.

Point?

ramicio

22nd August 2011, 13:58

Where are you all getting your 1-5 million write cycles number? MLC ranges from 1,000 - 10,000, and SLC can be up to 100,000. Most people go for MLC, and SLC is fading out, even in enterprise. So when you have only thousands of write cycles, your need for TRIM becomes apparent. If SSDs did, in fact, offer in the millions of write cycles, then TRIM makes less sense. But it makes sense anyway. Why just trash certain areas of the SSD and leave some never touched? That's moronic, even if you could do a trillion write cycles.

Ghitulescu

22nd August 2011, 19:36

Our guru said it, by linking one of the external pages. :)
My next quest would have been, if not off-topic, why he still uses mechanical HDDs instead of SSds ;)

hello_hello

22nd August 2011, 21:40

Our guru said it, by linking one of the external pages.

Are you envious of the fact I at least look for information and try to provide links to back up what I'm saying, rather than make it up as I go and cry wolf every time someone disagrees with me?

Well if it's wrong, it's wrong. I haven't done any more research on it yet but I'm not precious about it like some, if it's incorrect then so be it. I'll admit I posted something which was incorrect just to demonstrate to you how easy it is for an adult to do it.

What's quite funny, is our resident pseudo expert on everything, argued about the necessity of 1 million write cycles, made up a bunch of stuff as to why, yet obviously had no idea whether the information he was arguing about was correct or not.

My next quest would have been, if not off-topic, why he still uses mechanical HDDs instead of SSds

As you said, it's off topic, typically irrelevant, and if you want to know the answer you should start a new thread. :p

Ghitulescu

23rd August 2011, 07:56

Are you envious of the fact I at least look for information and try to provide links to back up what I'm saying, rather than make it up as I go and cry wolf every time someone disagrees with me?
You must live in a fantasy world. Not only that everyone here knows you by now, but reading back all topics in which I posted, s/he will see that you challenge my statements. Not vice-versa.
And this is not a challenge, it's all about how can one help the original poster on his/her original question.
Not only that you hijack every single topic with off-topic questions related to your topics - but also you manage to reverse the ratio between the original topic and your issues. Like here, where more space was allocated to your SSD off-topic issue than with the RAM-drive.
I have to admit, you're extremely experienced in doing this (well, it took you 2-3 striked accounts, but victory at last), without having a mod stopping you.
Well if it's wrong, it's wrong. I haven't done any more research on it yet but I'm not precious about it like some, if it's incorrect then so be it. I'll admit I posted something which was incorrect just to demonstrate to you how easy it is for an adult to do it.
How can you give advices to a person when you don't know about it?
What's quite funny, is our resident pseudo expert on everything, argued about the necessity of 1 million write cycles, made up a bunch of stuff as to why, yet obviously had no idea whether the information he was arguing about was correct or not.
That was hypothetical construction, to prove you that you have no idea what you're talking about. It's called reductio ad absurdum. Unlike other "experts" around here I do not write about things I am clueless at.

hello_hello

23rd August 2011, 08:53

Comedy Gold! Speaking of living in a fantasy world.....

Your last post was just more yada, yada, yada..... look back over the thread, even Groucho2004 mentioned your arguing just for the sake of it.

Where your lie falls apart, is it takes two to argue. Feel free not to participate in everything you're accusing me of. I can't argue with myself, although at least if I did, it'd be rational. Try answering questions on the topic for a change instead of incessantly whining about them being asked. You're probably boring everyone else with it as much as you're boring me. If you're the expert you think you are, we might all learn something.

That was hypothetical construction, to prove you that you have no idea what you're talking about.

LOL! well that backfired on you big time, didn't it? :)

Unlike other "experts" around here I do not write about things I am clueless at.

That's obviously where we differ. I'm able to make the distinction between fact and fiction.
Found that huge stockpile of Taiwanese made Verbatim DVDs yet? :p

There's one in every forum. Even expects mods to intervene to protect their divine right to never be disagreed with. Here.... apparently you're it.

ramicio

23rd August 2011, 15:24

You're both pretty irrational, and I, for one, am sick of you two jumping into threads just to start stuff with people, and then turn the threads into your own world in which to argue. Go serve your own forum and be the only members and just argue day and night. Or rather, you two should meet in real life and get married.

hello_hello

23rd August 2011, 18:13

Just like you jumped into this thread and argued the same topics with us? What a joke...
The only thing which makes our "arguing" any different is the amount of space Ghitulescu takes up in every thread with insults and whining about not being agreed with.

ramicio

23rd August 2011, 18:24

Yes, I brought up the page file :rolleyes: Oh, and yes, there are tons of threads with me and the same person just arguing to argue :rolleyes:

YOU are a joke. When Ghitulescu jumps in on a thread to HELP people, you jump in and try to start an argument with him. You jumped into one thread I was in and tried to start an argument on why I think eac3to uses better processing than something else. You should go away. You help no one with anything. You would at least be respected if you knew what you were talking about one bit. At least Ghitulescu mostly knows what he's talking about. You're both idiots just for that fact that you keep falling for each other's mental torment and just keep flooding the forum with arguments.

It's easy to prove you're an idiot. Your whole argument with me, in this very thread has to do with something in which you have no first-hand experience. You stated you have a small amount of RAM, and have a page file. Yet you tell me I need to take time out of my day to do tests to prove something to YOU? Then you go on to argue about SSDs, another area where you have, AGAIN, no first-hand experience.

hello_hello

23rd August 2011, 21:32

Yes, I brought up the page file :rolleyes: Oh, and yes, there are tons of threads with me and the same person just arguing to argue

Ahhh... I'm enjoying watching the hypocrisy unfold. I wonder how many posts you'll waffle on about this topic for. Sorry.... I didn't mean to upset you.

I often write essays here to help people, so look at some of my posts and let me know when you know what you're talking about. Many of which have been interrupted by Ghitulescu posting irrelevancies just to argue with me.
I didn't try to start a fight with you about eacto. I actually posted in that thread with an intention to at least try to help, I just added a little joke to my second post given we were currently discussing the paging file in this thread while your blinkered attitude had obviously carried over to that one. Get over yourself.

Yes, you bought up the subject of the paging file.... in a thread regarding RAM drives.

The fact you've used a PC without a paging file and your blinkered attitude has you enjoying the placebo effect doesn't prove me wrong, it only proves that you refuse to be. You don't have to prove anything to me. You're welcome to believe what you want. If you want to post about it in a forum though, don't expect everyone to accept it as fact.
Nobody's going to take something like that seriously when it comes from the type of person who thinks they can always tell if a music source is MP3 the second they hear it. Something else you claimed but refused to test, even when it was made easy for you.

ramicio

23rd August 2011, 21:42

You are the type of idiot who thinks anything in math is absolute and real, and actual observations need to be proven. I see nothing from you lately, except arguments with Ghitulescu. Speaking of hypocrisy, it's so funny you say I have a narrow view of everything. Look who's talking, calling everything a placebo. Your logic is just so laughable. You continue to tell me I'm enjoying the placebo effect, while you still don't have any first-hand experience whatsoever with what we're talking about. That's funny. You are likely the kind of person who, if you're into cars, benchmark races. When did I claim to be able to INSTANTLY pick out ANY lossy file from something lossless? The concept of lossy is just completely moronic to me, and it's impossible to compress without audible artifacts. They may not be all over the place and constant, but why bother to ruin good sound by even 1 artifact, when disk space is so cheap? Cheapness, and elitist, mathematical attitudes. YOU have a blinkering attitude.

hello_hello

23rd August 2011, 22:00

And the hypocrisy continues.....
Maybe you should have just copied and posted your previous post, it would have saved you retyping much of the nonsense.

"When did I claim to be able to INSTANTLY pick out ANY lossy file from something lossless?"
Selective memory is generally part of the placebo effect.
http://forum.doom9.org/showthread.php?t=160068&highlight=mp3 (#9)
"Anyone without damaged hearing under the age of 80 can easily hear the difference. It's easier to lie about not being able to hear a difference than lying about hearing a difference."

Enjoy the placebo effect.

And no, I don't have a blinked attitude. I also rip everything to a lossless format. I just happen to own an MP3 player which only has 8GB of memory.