![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Paging file
- Thread starter Glyphic
- Start date
Are you being intentionally obtuse? I said that oral counting and a stopwatch are not going to be objectively observable or recordable, thus won't work. I used "count mississippis" in the same figurative manner I was hoping it was originally used. If I was mistaken, then allow me to clarify: find a way to rule out human error when counting, or else it won't work. While I'll give the benefit of the doubt as far as ethically relaying the results, I wouldn't suggest trusting even my own counting or recording with a stopwatchit's not going to be accurate or objective enough.
BobSutan
[H]F Junkie
- Joined
- Apr 5, 2000
- Messages
- 12,121
GreNME said:
This makes sense in that it introduces the human factor and could currupt the precision timing needed to make the test as error-free as it possibly could be. A built-in timer in the benchmarking program (like some timedemo and stress testing apps have) would be preferable.
So you two don't think, even as an inaccurate human being, I couldn't tell the difference between 10 or so seconds and close to 30 seconds? Again I will state my point. That is the difference I count on MY system without a page file versus a pagefile. This thread is losing any semblence of legitimacy due to ridiculous arguments. Stated ONCE more....my machine is a gaming system. If disabling my pagefile speeds up my level loads times dramatically....someone PLEASE explain to me how this is hurting my performance!!!!!!!!
You are not using reproducable parameters outside of whatever your claims are. In other words, we just have to take your word that you did not mistake one thing for another. If results are not objectively recordable and observable, then they are prone to be biased. If you are used to running a testing environment, you should know this fully well.
GreNME said:
Again...as I stated from the very beginning. I am testing my OWN system, and I am quite capable of counting seconds. This last time I even used your preferred "mississippi method". With no page file, I counted 11 mississippis for the load button in Stalingrad to turn to Play. I exited the game, created a 300 MB static pagefile, rebooted, and did the same. I counted 27 mississippis before the button turned to Play.
I may not be Stephen Hawking, but everytime I reproduce this, I get similar results. We can argue this until tomorrow morning when I leave on my much needed vacation....but those are the facts in the realm of my system. May I remind you, you have yet to show me any proof that my performance would be hurt by disabling my pagefile.
Gah! How many times do I have to reiterate: I am not saying it hurts performance, I'm saying it does not improve performance. You are essentially asking me to prove a negative, which is not a reasonable request. And when I lay out a scenario in which to provide proof for the positive, you keep giving un-provable and unobjective methods.
This is like debating the existence of a god with someone.
This is like debating the existence of a god with someone.
SJConsultant
2[H]4U
- Joined
- Jan 14, 2004
- Messages
- 3,599
While disabling the pagefile may yield *some* performance in *isolated* cases and system specifications, it's not going to work for everyone in all circumstances as some people claim or lead others to believe.
The whole problem with "system tweaks" is that they need to be evaluated on a case by case basis and not made into a generalized tweak that can be applied to everyone's system and work with equal results as in this case.
Just my .02
The whole problem with "system tweaks" is that they need to be evaluated on a case by case basis and not made into a generalized tweak that can be applied to everyone's system and work with equal results as in this case.
Just my .02
SJConsultant
2[H]4U
- Joined
- Jan 14, 2004
- Messages
- 3,599
djnes said:
I'm still skeptical about it without some way to measure the difference repeatedly and accurately and not basing it off of someone's "claims" as most people put here.
GreNME said:
I haven't made a single baseless claim. I'm told you the results of my testing on my own machine. Your not going to convince me to suddenly re-create my page file. I am quite happy with the system as it is. Frankly, I could care less at this point. I've actually done something on my end, which is more than I can say for you. I've already moved on to my vacation, which mentally, I am ready for thanks to dealing with your earlier lack of rationale and logistics. I am quite capable of counting seconds, and while it wouldn't stand up at a scientific institute, it clearly shows a change in performance. if I was comparing 10 to 12 seconds, then I wouldn't have this argument. But, it's over twice the time to load the same exact level, WITH a page file. I've tried several levels, with and without, and have gotten a large difference in times for both. If this is too complicated, then I'm out of ideas on how to explain it to you. Your the one who flew off the handle earlier because you couldn't POSSIBLY deal with a system getting better performance without a page file. I don't need to prove anything to myself or anyone else. Can't say the same about you. Consider this my last post in this thread, on this matter.
You've exhausted my tolerance for lack of reasoning for the week, and it's only Monday night.
GreNME said:
I haven't backed out of shit. I HAVE done some level of testing. I won't be near a computer until next Saturday, but I'm waiting for you to give me what will make you happy, your highness. This seems to be bothering you to the point maybe you should consider getting a life. Why does it bother you so much that my level load times have increased without a page file???
I'm not the one who seems angry here. Also, as I've already said numerous times, I am not dictating terms here. I am asking for an agreed-upon consensus of criteria and methods, so that once the results are posted, no party can complain.
As I've also pointed out, which you keep claiming I've said otherwise, I have already done this testing on systems in my own lab, as well as in a thread here on this very forum where the results showed no improvement. So your continued accusations of no research and observation on my end are pretty much hyperbole on your part.
I've done this before, including on this forum. I've backed up my claims publically before. Now you are invited to do the same. All you are responding with are venomous remarks and baseless accusations.
As I've also pointed out, which you keep claiming I've said otherwise, I have already done this testing on systems in my own lab, as well as in a thread here on this very forum where the results showed no improvement. So your continued accusations of no research and observation on my end are pretty much hyperbole on your part.
I've done this before, including on this forum. I've backed up my claims publically before. Now you are invited to do the same. All you are responding with are venomous remarks and baseless accusations.
I'm a little pissed off because you can't accept and admit one basic truth of ALL computing. Different systems yield different results. Just because it doesn't happen in your little "lab" doesn't mean it isn't happening on my system. With over 200,000 systems on our corporate network, I think I know a little something about differing performance and results, even on exact same hardware and OS versions.
Your the one who continues to push the issue, continues to tell me I am incapable of counting seconds, that I must just be making this up, or pulling it out of my ass. You have that level of thinking, and you expect me to sit and rationally agree upon something with you? Are you out of your freakin mind? That shows an unbelievable amount of ignorance to sit and argue with someone about a concept you have no proof on. You weren't sitting next to me when I did my testing...but yet you continue to argue with me and say it's not possible. Other knowledgeable people have posted that one may be true on one system, isn't a given on another, but yet you continue to argue with me about the performance of MY computer, that you have never seen! How foolish is that? I am wearing black Nike sandals right now....are you going to argue with me on that as well?
Furthermore, who gives a shit if I run a page file on my system or not? Will you sleep better tonight if I was to be converted to your thinking and create a page file? I disabled mine and have found it to speed the loading and swapping of level data on my system. his is on my system...MINE. I alone use this system, and I will tweak it to my tastes. I will ask again...why does it bother you so much that my system does have a noticeable measurable difference without a page file? Tell me what bullshit you want to hear to make your moons re-align?
Your the one who continues to push the issue, continues to tell me I am incapable of counting seconds, that I must just be making this up, or pulling it out of my ass. You have that level of thinking, and you expect me to sit and rationally agree upon something with you? Are you out of your freakin mind? That shows an unbelievable amount of ignorance to sit and argue with someone about a concept you have no proof on. You weren't sitting next to me when I did my testing...but yet you continue to argue with me and say it's not possible. Other knowledgeable people have posted that one may be true on one system, isn't a given on another, but yet you continue to argue with me about the performance of MY computer, that you have never seen! How foolish is that? I am wearing black Nike sandals right now....are you going to argue with me on that as well?
Furthermore, who gives a shit if I run a page file on my system or not? Will you sleep better tonight if I was to be converted to your thinking and create a page file? I disabled mine and have found it to speed the loading and swapping of level data on my system. his is on my system...MINE. I alone use this system, and I will tweak it to my tastes. I will ask again...why does it bother you so much that my system does have a noticeable measurable difference without a page file? Tell me what bullshit you want to hear to make your moons re-align?
And you seem to not understand one very important facet of computing: it is not magic or voodoo. Everything happens for a reason, and every real reason is reproducable.djnes said:
Once again with the insults. How does the fact that this has been observed publically, on this very forum, mean anything to you?djnes said:
Ahh, once again with the pissing contests. How about I mention my many clients, in numerous situations, ranging from the medical field to accountants to financial institutions to corporations (yes, corporations), with a few home or home-business users in there? How about I mention that I've seen those "many different configurations" and the results therein, as well as loads of like machines with identical specs and apps?djnes said:
In other words, you're not impressing me with your attempt to claim validity because you work somewhere. I do too, and I'm not impressed with your story.
No, I am refusing to just "take your word for it" when I have seen and displayed these results myself, both in my own personal testing lab and on this very forum. I am giving you a chance to back up your "trust me" story, by giving you a scenario where an agreement is accomplished so that every side of the issue can feel satisfied in contributing to the process and methods.djnes said:
You are refusing. That is your right. Your constant attempts to portray me in a manner I have not presented myself is ridiculous.
I have displayed proof here on the [H] before. Ask Phoenix, I am not lying here. If the search didn't suck, I would find the thread for you to look at yourself.djnes said:
More straw-man arguments. I did not say that. I said that you were experiencing a placebo effect and are convinced there is an increase when you have no proof, no repeatable objective methods, and have admitted to calling it your opinion before.I am not arguing your opinion. You are entitled to it. I am arguing fact.
Then don't take part in the comparison of results with the rest of us. Continue to do whatever you want to do, regardless of anything I've pointed out, and go on your merry way. However, whenever you claim on this forum that removing the page file creates a performance increase, expect someoneeven if it isn't meto point out that you are wrong. You are entitled to believe whatever the hell you want, but when you try to pawn opinion off as fact, don't get pissed off when the facts are pointed out... especially when they contradict your statement.djnes said:
If you don't want to take part, just say so. I repeat:
If you don't wish to take part, just say so.
Whichever you decide, could you please stop the posturing now? You are impressing no one, and bringing this thread closer to being locked.
You know what GreNME? I don't wish to participate. My thoughts on the page file are shared by many on here, as I've gotten many comments recently saying so. It's also supported by many on here as well. I'm not trying to impress anyone, as I'm not as insecure as you, turning to an online forum for acceptance. I have much more important things to do with my time than try to rationalize with you. I run many projects at work, and your attitude and lack of open mindedness and reasong would have you off my teams faster than you can figure out what happened. Your really naive if you think that all computers act the same when given the same tweaks. Seriously, get a life dude...there's more to life than trying to impose your will on some forum, that everyone has to think your way or their wrong. Myself and a few others on here are getting quite a laugh at you.
Your quite foolish to argue with me over my own system. I do not wish to deal with such foolishness and ignorance anymore. My only hope is that one day you will mature enough to see that people can have a different opinion than you, and not be wrong! In my 4 plus years here, I have never EVER seen anyone as ridiculous as your reasoning.
Don't take this as my conceding the argument. It makes a difference on my system, and that's all I care about right now. I just don't really give a shit about proving it to you right now. Tomorrow, at this time, my fiancee and I will be sitting on a white sandy beach....and that's all I care about right now. If you don't want to believe me, or how my system acts....I DON'T GIVE A HALF, ONE, OR TWO SHITS IF YOU DO! I really don't...and I say all that with a big smile on my face!
Your quite foolish to argue with me over my own system. I do not wish to deal with such foolishness and ignorance anymore. My only hope is that one day you will mature enough to see that people can have a different opinion than you, and not be wrong! In my 4 plus years here, I have never EVER seen anyone as ridiculous as your reasoning.
Don't take this as my conceding the argument. It makes a difference on my system, and that's all I care about right now. I just don't really give a shit about proving it to you right now. Tomorrow, at this time, my fiancee and I will be sitting on a white sandy beach....and that's all I care about right now. If you don't want to believe me, or how my system acts....I DON'T GIVE A HALF, ONE, OR TWO SHITS IF YOU DO! I really don't...and I say all that with a big smile on my face!
See, all you had to do was say you didn't want to take part. Instead, you take time to insult me andonce againdo the pissing contest thing.
Thanks for risking getting this thread closed. Now I can continue to hash out a system with Phoenix and any others who want to take part (what do you say, SJConsultant?).
Mods, please don't close the thread yet.
Thanks for risking getting this thread closed. Now I can continue to hash out a system with Phoenix and any others who want to take part (what do you say, SJConsultant?).
Mods, please don't close the thread yet.
GreNME said:
Good...I'm glad...really! I hope you find the answers that make you sleep better at night. I really do. This was an incredible waste of time, trying to instill some logic in you. Maybe someone else can try, because I give the hell up. Anyone who thinks all computers will act the same is beyond logical and understanding. Enjoy!
Glyphic said:
I owe you an apology on behalf of myself and GreNME. I let his foolish rantings get to me, and your thread went no where. I don't deal with close-minded, know-it-all people like him. You want my official suggestion? Unless your using an app that specifically requires a page file, and you have enough system memory, you can try it out either way. For a pure gaming system, I recommend shutting it off. If it's a multi-purpose system, pick a decent size, and set it to be static at that size.
Dude, don't apologize for me. You seriously can't seem to cut it out with the personal attacks and pissing contests, can you? Go ahead and get your last word in and be done with it, because I'd like to continue to set up this co-op test with other, willing candidates.
Honestly, unless your goal is to get the thread locked, why do you persist? I offered a chance to back up your story, and you declined. Get over it.
Honestly, unless your goal is to get the thread locked, why do you persist? I offered a chance to back up your story, and you declined. Get over it.
YellowPeril
Gawd
- Joined
- Nov 20, 2003
- Messages
- 666
memory can experience both fragmentation and r/w errors too i believe.Glyphic said:
returning to the origional point of the thread, i recently disabled my page fileon my desktop with a gig, and it did seem a little quicker. i tried it on my laptop with 512, and i ran out of memory very quickly. right there is a testament to the differing hardware reactions to a page file.
also, someone mentioned earlier that disabling the page file doesnt stop paging. i admit that i was confused when i saw page file in the task manager. so what is really going on with that. what are page faults, and how does disabling the page file change the system. if i have read through this correctly, you are just moving the page file to memory? i think that i am getting hung up on the definitions of PF and VMM.
thanks guys, and take deep breaths...
Fark_Maniac
2[H]4U
- Joined
- Feb 21, 2002
- Messages
- 2,438
Congratulations djines on the engagement!djnes said:
For everyone else...since we are mostly looking for proof...I'm getting a copy of WinBench 99 that does not contain video testing tools. we'll see how this works. The company that GrenME had given me pages back has been bought out and the full/updated version should only be had on cd's. I think WB'99 will be out of date...but we'll at least have a look.
The hard drive will still be doing work without a page file, even if you have two gigs of RAM.Glyphic said:
Listen, guys, you seem to be under the impression that if a page file is removed, that the hard drive isn't used for memory management. You are very much incorrect. There is still paging going on, it is simply not happening on the page file. And believe me, if you really completely did get rid of all the things that your hard drive does with memory management, none of your programs would work.
You know, despite what some people may have said, No one said different hardware configs will not have different results. Some people just need to learn to read. Yes, with 512 MB and under, you will likely experience averse effects when removing the page file. However, even if you have 1-2 gigs, I have maintained that you will not gain any speed from removing the page file. The only argument made against me has had to do with "it feels this way" and "that's my opinion," both of which are subjective. I am talking about actual fact, not opinions, not how you feel when you run it.YellowPeril said:
Now you are beginning to learn. Yes, even when you disable the page file, it does not stop paging. What it does stop is storing it in the page file you had originally. Instead, the paging is done in a more hit-or-miss manner, and yes, more of what you have loaded stays in the RAM. However, you did not read correctly, and you are not moving the page file to RAM when you turn it off. You are just turning it off. The thing is, even when you turn off the page file, actual paging will continue to take place.YelloPeril said:
This is not a bad thing, mind youif the system did not do that, then no matter how much memory you have, you would very quickly run out of space in the RAM and lock up.
As for the difference between the PF and the VMM, the answer is fairly simple: the PF is part, or only one component, of the virtual memory subsystem. Without the page file, VMM still exists. If the VM subsystem were gone, the PF would be gone as well.
You mean the full version can only be bought through CD, right? Also, I want to make sure that whatever we use as a testing tool can be had by everyone without breaking licensing agreements. So, let's find things that are going to be accessible.Fark_Maniac said:
SJConsultant
2[H]4U
- Joined
- Jan 14, 2004
- Messages
- 3,599
GreNME said:
I'd be willing to participate in testing, just need to come up with some way of accurately benchmarking the results.
Fark_Maniac
2[H]4U
- Joined
- Feb 21, 2002
- Messages
- 2,438
I think this is what we are looking for...but a cd needs to be ordered for 5 bucks a pop.
http://www.veritest.com/benchmarks/bwinstone/default.asp?visitor=
http://www.veritest.com/benchmarks/reqfrm.asp
http://www.veritest.com/benchmarks/bwinstone/default.asp?visitor=
http://www.veritest.com/benchmarks/reqfrm.asp
No, no, and no, it's both. We already know good methods for testing overall performance, I suggest we use the [H]'s current method. Fraps+timedemos.
However, as tested before, and it's not quite a tweak for overall performance. The tweak is aimed at improving PF performance, when the system is paging. This is absolutely key. If our results prove that overall performance isn't generally affected(system may not page during tests), but the system is faster, WHEN PAGING, it's a bonus to the end user, and a good tweak. There are two things to Provo/disprove. Is the system faster overall, does it increase ave/min/max fps, or calculation times? I think no, it shouldn't affect these operations much at all, unless the system is paging...
I don't think anyone is recommending that, I always get the use to find their peak memory usage and compare it to the amt. of physical RAM they have before making this recommendation.
Yes I remember the post, but it didn't test the claims made by most people here (faster load times, and not noticing lower performance when paging). It did show little differences within 'margin of error' between the different runs. It didn't test while paging. This is akin to saying 'more RAM doesn't help a machine' when the test is a pi calculation... Of course it doesn't, you need a faster CPU for that...
I think the forum is eating/ate posts (during the outage?). I did a search on posts created by me and my RAM vs Swapfile thread is also AWOL.
Have fun!!!
I'm willing to shell out a few clams for a good benchmark, but we need to make sure we are looking at the right tool. I think it will require a tool to monitor performance, and a process to force paging. This is where I draw a blank. How do we induce paging?
We need to induce paging, and test under that environment, otherwise, we may not be 'noticing' the results in a test. However, a user, working on a machine for hours, will have the system page large amounts of data, as determined by the OS. This will be noticable. If the tweak helps load times too, we should try to verify that.
No, it's not as simple as prove/disprove. There are claims that it increases performance, and as I said, there are three possible outcomes: improvement, decreased performance, or nothing.
Also, that the system still needs to page is important when discussing whether performance is improved or not. Without a page file, there is not already a set place for unused memory data to go. This obviously is not a problem for the OS to handle, but if the machine runs into this situation, it can be problematichence not suggesting running without the PF below 512MB in the first place, yes?
Like I said earlier, even with 2GB of memory to hold information, you would still run out of space soon enough when using programs. So the data has to go somewhere, n'est ce pas? When not in RAM, the data is on the drive. While you can get less incidents of paging with larger amounts of RAM and no swap file, this is still no guarantee of increaased performance, which is what I've been arguing from the beginning.
Now, all other things equal, a machine with 2GB of memory is going to perform better than one with 512MB of memory, but this also does not lead logically to the 2GB without a page file performing better than a 2GB with page file. Therein lies the part where I am claiming that people are assuming things without a basis in fact.
Whether a machine pages or not is not going to significantly (meaning recordably) affect performance. Those who have claimed it has are doing so without any way to prove it. That's my point.
But it did use some benchies that had to load separate files into memory to run. You are correct in that none of them forced paging and then gauged performance, and I don't know of a single bench (or method, honestly) out there that can.
Yeah, that sucks. That just means that if we go through with this one, that we each save a copy of the results page, or one of us host it off the board as well as here.
We'll show those forum gremlins.
This is going to be the tricky part. I have some ideas, though...
Get a regular document viewer or something (I'm thinking Adobe Acro Reader) and load a rather sizeable document. For testing different stress loads, we could possibly choose multiple documents under the same program.
Take a static web page (that has been downloaded) and load it in an agreed-upon browser. This will help take up memory space.
Then we have to come up with either a script or small program that can simultaneously put itself in focus and out of focus while running a counter/timer. What would be even better is if we could find something that gauges page faults while doing so.
However, even with all that, it's still not guaranteeing paging, and there is still margin for error. Maybe we could put our heads together and write something ourselves?
Fark_Maniac
2[H]4U
- Joined
- Feb 21, 2002
- Messages
- 2,438
would a good test of paging be something like firing up a game of FarCry...loading a saved part (this generally takes 30-40 seconds)...play for about 5 min...get killed...then restart what you had just played. Reloading only takes 5 sec or so with my gaming rig w/o pagefile. I'm just thinking of heavy loading and reloading of large files. After the level is loaded into the pagefile...when the level is reloaded...shouldn't it reload out of the pagefile?
Granted this test would be vastly different on each person's machine, there would be repeatable times on repeated tests on each person's machine. Granted here we won't have an hard proof...
Rig Specs
Granted this test would be vastly different on each person's machine, there would be repeatable times on repeated tests on each person's machine. Granted here we won't have an hard proof...
Rig Specs
Now you guys are thinking on the right track... This is also why so many factors affect this (to induce paging), things like startup programs, services, and physical RAM could affect when a system pages. I really think this will have to be done in a very strict environment, esp. the amt. of RAM... Using out normal machines to run a test won't fly, we would have to make a 'test' OS load for this. I'll post more after lunch.
Fark_Maniac
2[H]4U
- Joined
- Feb 21, 2002
- Messages
- 2,438
GreNME said:
By hand...I could output to a vcr and record that way...I have no other way..