Squabbling cores

[Bill_Bright]

The opinion expressed by Bill is not shared by a great many very experienced people.

Two is not a great many. There are a great many, however, who say to just leave it and let Windows manage it for most users.

And more importantly, even those who suggest it is okay to delete it or set a fix size, typically say for most people, letting Windows manage is still best for them.

 

[ThrashZone]

Hi,
I still have a few crucial mx100 256gb in use mostly on z490 oc os's they are actually first ssd's I ever bought still kicking win-10 & 11

Only 3 year warranty on them

 

[lexluthermiester]

Two is all I'm going to make time for. I have better things to do with my time than spend hours on a dissertation of an explanation.

Mr Shrek is a smart person, smart enough to know that to see the results, he has to TEST them. However, it does NOT take a smart person to know that if a user moves a high usage file, like a pagefile, off the OS drive, said drive will see much less wear & tear(if the drive is an SSD) and will improve performance by dividing drive accesses between two drives. What this means is that while the system is accessing the pagefile it can also, simultaneously, access other files and data from the OS drive without waiting for reads/writes to/from the pagefile to complete.

 

[ThrashZone]

Hi,
I'll look at the milage on the 10 year old mx100 later tonight
They've had system image after system image written to them over and over so again they should be dead long time ago.

The 850 pro was a main os ssd I did oc on it a bit but the mx100's always did better so I abused them

 

[lexluthermiester]

They've had system image after system image written to them over and over so again they should be dead long time ago.

You'd be surprised how long those MLC based drives last.

 

[ThrashZone]

You'd be surprised how long those MLC based drives last.

Hi,
Indeed I had to check how long the warranty was and surprised it was only 3 years

It's past it's warranty three times, on heavy oc use to not just normal day to day horseshit although it likely doesn't have as many usage hours

 

[Bill_Bright]

if a user moves a high usage file,

LOL Love the "if" statements.

It is funny how some people pick examples that are other than the typical norm, pretend those exceptions make the rule, then use those less common examples in a futile attempt to justify their case.

It is like today's politicians who vote against a new law because 5 people will be adversely affected by it - even though it will benefit 100 million.

No doubt, some will be able to poke a hole in anything said with some extreme exception. But exceptions do not make the rule. And there are several rules here. These include the fact that the vast majority of users are better off just letting Window manage the Page File. And SSDs no longer suffer from wear issues for the vast majority of users. And that SSDs are ideal locations for page files.

It is also a rule that if you decide to dink with the page file, and pick some arbitrary number, or worse pick some number tossed out by someone who has NOT analyzed your specific system, and how you use it, then who is the fool? Then I have to wonder, who is the fool? The one that leads, or the one that follows?

It is also a rule, with exceptions of course, that the vast majority of computer users do NOT use their computers for one task, and one task only. Therefore, resource utilization will NOT be constant. Consequently, neither will the ideal PF settings - meaning, it is not a "set and forget" setting.

Two is all I'm going to make time for.

Too bad you didn't even pay attention to those two. Linus noted starting around 2:10, do NOT disable the PF. Then around 2:40 he says to "leave the size to the system itself". The only thing Linus said is that you can move the PF to a secondary drive if you have one - but also said to leave a PF on the boot drive Windows is smart enough to use them both optimally.

Then look at Leo's video. What does he say around 4:20? That the default (let Windows manage it) is "a fine default setting" unless for some reason you know want to do something different. Now at 7:40 in discussing disabling the PF, Leo, with emphasis clearly states, "It's NOT something I recommend you do."

Even as far as dinking with the size, he clearly states you must know your virtual memory requirements now, and in the future.

So if you are only going to pick two, I recommend you pick two that support your claims instead of recommending against them.

 

[ThrashZone]

Hi,
Yep mx100 has a few more host writes on it

 

[Aquinus]

I'd expect that excessive paging would more than half the speed of a process; my Son recently went wild on Minecraft mods and I could see the hard drive go nuts and things could take up to 10 mins before Minecraft would even respond (he had 12GB of RAM, since upgraded to 16GB).

That makes a lot more sense. The reality is that if you run out of memory, the only real solution is to buy more memory, or stop running other programs that use the memory you need. Paging is okay if it swapped out memory for unused tasks, but if you start swapping Minecraft while it's being used, performance is going to tank hard, probably more than by half.

 

[Shrek]

That is what got me thinking about the idea of withdrawing certain tasks for later and then I wondered if that is already done in modern operating systems or if it might be a thing of the future.

So I asked here, and it paid off.

 

[Aquinus]

That is what got me thinking about the idea of withdrawing certain tasks for later and then I wondered if that is already done in modern operating systems or if it might be a thing of the future.

So I asked here, and it paid off.

The OS will attempt to swap out whatever it can swap out if memory pressure is high. The two best ways to avoid that are by limiting what you keep open at any given time (because background processes can very easily wake up and swap an application back in when it doesn't have to,) or by buying more memory. Modern operating systems are usually pretty good about this, but depending on how often the system is swapping pages, no OS will make up for swapping to slower storage. There is no getting around running out of memory. The OS will do the best it can, but when push comes to shove, swapping is expensive and the more it happens, the slower the machine will get and no amount of OS optimization will fix that.

On the opposite side of the spectrum, if you have considerably more memory than you actually use, the OS will use that extra memory for stuff like disk caching, which can actually improve performance. So my recommendation is to plan for more memory than you need. At worst all it's used for is caching. At best, you'll need it and it will prevent you from swapping. Running out of memory hurts performance more than almost anything else though (aside from thermal throttling perhaps.) You really want to avoid it.

 

[Bill_Bright]

Pretty sure the OS may swap low priority, less often used pages to the PF too - even if there is physical RAM left. That is not a bad thing, especially if the alternative is to "close" the file and save it back to disk. In other words, even with tons of system RAM, the OS may still use the PF. And for 99% of us 99% of the time, no problem because the high priority, often accessed pages are still in the much faster RAM - or maybe even in one of the super fast on-die "L" caches.

 

[Aquinus]

Pretty sure the OS may swap low priority, less often used pages to the PF too - even if there is physical RAM left. That is not a bad thing, especially if the alternative is to "close" the file and save it back to disk. In other words, even with tons of system RAM, the OS may still use the PF. And for 99% of us 99% of the time, no problem because the high priority, often accessed pages are still in the much faster RAM - or maybe even in one of the super fast on-die "L" caches.

I think that might be specific to Windows. MacOS and Linux won't typically hit swap unless you run out of physical memory. My machine has been on for 10 days.

 

[Bill_Bright]

I think that might be specific to Windows.

I have essentially zero experience with Macs. I used to maintain UNIX and Solaris systems, but as a hardware tech, so not much exposure to the inner workings of the OS compared to my years with Windows based hardware. But assuming Linux works the same, and taking your word about Macs, I will assume you are spot on.

Interesting how, with a massive 64GB installed, you still have 3.8MB swap usage - even if just a drop in the ocean.

 

[Wirko]

if a user moves a high usage file, like a pagefile, off the OS drive, said drive will see much less wear & tear(if the drive is an SSD) and will improve performance by dividing drive accesses between two drives

You're assuming that the pagefile is a high usage file. It should not be. Heavy use of PF means the system's memory usage is exceeding the amount of physical memory, or is close to that. At that point, you don't care about performance or SSD wear, you just want your system to remain stable and finish its tasks without out-of-memory errors.

 

[lexluthermiester]

You're assuming that the pagefile is a high usage file. It should not be.

Ideally, sure. However, the pagefile exists to be used and often is. Moving it off the main drive is always a good thing.

Heavy use of PF means the system's memory usage is exceeding the amount of physical memory, or is close to that.

Also, true.

At that point, you don't care about performance or SSD wear, you just want your system to remain stable and finish its tasks without out-of-memory errors.

Again, true.

None of those points negate what I was saying earlier. The pagefile as it still exists is a relic of an age when system RAM was extremely expensive and usually had a minimal presence in an average PC. That started changing in the early to mid 2000's and today is a non-issue. The continued presence of the pagefile in modern OSes is only as a "safety net" type function. A great many people know this and it's why they manually manage the pagefile, or disable it altogether.

As was pointed out above(Bill always has to get his epeen on), the ONLY reason Linus and Leo suggested leaving it be is simple fear of liability and ignorance of "unknown" potential issues that "might" take place. The simple fact that they took the effort to make video's on the subject and show people in detail how to manage or disable the pagefile says very clearly that they acknowledge the benefits of utilizing such configurations.

 

[Wirko]

Pretty sure the OS may swap low priority, less often used pages to the PF too - even if there is physical RAM left.

Why would the OS do that? It doesn't seem beneficial. Poorly written applications, however, can trigger that. If a process reserves and commits* a huge amount of memory, even briefly, without actually needing that much, it can result in swapping to the PF.

* I'm not pretending to understand much of that. Apparently, before a process can actually use pages of memory, it first needs to reserve them (this does not consume any memory, it's just some kind of declaration). Then it commits the pages, which does consume the memory. All this probably means that a well-behaved application does not commit more memory than it needs. But it can reserve a lot more.

 

[Bill_Bright]

Windows does the divide access but it is not equally. Windows will automatically choose the drive that responds first - which typcially is best performing drive, the SSD. So if the secondary drive is an SSD and the primary a HD, assuming sufficient free space on both Windows managed drives, Windows will pick the SSD every time. (Source)

But just the thought of wear and tear on SSDs these days is misplaced. Wear is not a problem with today's SSDs. Even data centers use them for their most often accessed data.

As was pointed out above(Bill always has to get his epeen on), the ONLY reason Linus and Leo suggested leaving it be is simple fear of liability and ignorance of "unknown" potential issues that "might" take place. The simple fact that they took the effort to make video's on the subject and show people in detail how to manage or disable the pagefile says very clearly that they acknowledge the benefits of utilizing such configurations.

Oh bullfeathers! Total nonsense! You cite two sources pretending they back your claims. Then when it is pointed out they actually contradict your claims, you claim they only said that out of fear of being sued - with you now pretending you know why they made those videos in the first place.

It is funny how some, incapable of supporting their claim with true facts, resort to puerile personal insults, as if that makes them appear more knowledgeable. Pretty sad.

By default, the PF is system managed and the size is adjusted dynamically. If the nonsense spewed above was true, Microsoft could have easily coded that system managed PF to set itself to 0Mb. But it didn't. Why? Well I suspect Lex would claim because the folks at Microsoft are stupid morons who don't want to get sued.

And then Lex totally ignores the other primary purpose for page files - dumps! Disable the PF and no dumps, unless the admin sets up a dedicated dump file - not a "normal user" task.

Why would the OS do that? It doesn't seem beneficial.

Why not? I mean why keep low priority pages in fast RAM if not immediately needed? What's the alternative? Keep it in fast RAM? What if a higher priority call comes in? The OS would have to move that low stuff out first. That could cause delays. The other alternative is to close the file. That could cause even more delays if needed again.

 

[lexluthermiester]

Why would the OS do that?

It doesn't. Unix/Linux/BSD based OSes still use the swapfile as a "scratchpad" to write temp data for referencing various functions. Windows does the same thing which is why a small(256MB IIRC) swapfile still exists on the "C:" drive even after the pagefile is fully disabled.

 

[Kapone33]

You'd be surprised how long those MLC based drives last.

The only bad SSDs were the OCZ and Kingston first Gen drives. With Kingston especially their USB drives. It is why Sandisk was the best because they had had the best NAND chips until Samsung did their rise.

 

[Wirko]

The continued presence of the page file in modern OSes is only as a "safety net" type function.

Well, that's not so bad.

What's the worst thing that can happen if you have a PF, and the worst thing if you don't?

If there's a PF: the file expands due to applications that eat RAM like it's free (as in free beer). In an extreme case, you run out of disk space. Windows probably isn't stupid enough to allow that on the system drive but I don't know. The system also becomes painfully slow, but less so with an SSD. It's on you to decide how often you can tolerate that.

If there's no PF: applications crash. OS can crash too. It's on you to decide how often you can tolerate that.

And what's the best outcome in each case? With a PF, the file is hardly ever written to, doesn't take up much disk space (2.5 GB for 16 GB of RAM) and doesn't contribute to SSD wearout. Without a PF, you save those gigabytes.

I checked my Win 7 system with 16 GB of RAM, and discovered a 1 GB fixed size page file. Oh well. Looks like I like to experiment, then forget. Can't recommend that to others.

 

[Kapone33]

Well, that's not so bad.

What's the worst thing that can happen if you have a PF, and the worst thing if you don't?

If there's a PF: the file expands due to applications that eat RAM like it's free (as in free beer). In an extreme case, you run out of disk space. Windows probably isn't stupid enough to allow that on the system drive but I don't know. The system also becomes painfully slow, but less so with an SSD. It's on you to decide how often you can tolerate that.

If there's no PF: applications crash. OS can crash too. It's on you to decide how often you can tolerate that.

And what's the best outcome in each case? With a PF, the file is hardly ever written to, doesn't take up much disk space (2.5 GB for 16 GB of RAM) and doesn't contribute to SSD wearout. Without a PF, you save those gigabytes.

I checked my Win 7 system with 16 GB of RAM, and discovered a 1 GB fixed size page file. Oh well. Looks like I like to experiment, then forget. Can't recommend that to others.

Here's one for you. What if you put the Page file on a RAID 0 PCie 4 array?

 

[Wirko]

Here's one for you. What if you put the Page file on a RAID 0 PCie 4 array?

Can't say. DRAM random access time is about 50 ns, which calculates to 20M IOPS. Two SSDs in RAID 0 achieve ~1M random IOPS but only if both can be read/written to simultaneously. So if the PF is accessed in small chunks, mostly randomly, you get "RAM expansion" that's 20 times slower. If there's a lot of sequential access then it's not that bad. Too many ifs to predict anything, and that RAID 0 array may be used for other stuff too, not just the page file.

 

[phill]

Just so I'm clear and I'm not being a total donut here, but we are speaking about the page file under this right??

Generally I set to 2GB and move on with life. As long as the system has enough RAM (which if I recall is 4GB from the replies in the post?? OS is Windows 7??) there's generally not a problem??

And 4TB or even 13TB writes, isn't anything in my books to be worried about on an SSD.. Mines at 6.5TB ish and showing no issues that I can tell of...

If you can upgrade the PCs with some more memory and if required an SSD (if they don't have one already) then that'll be all you need I think I go on the assumption for at least 1GB of RAM, per thread. If you can make it 2GB per thread, gives you masses to breath

 

[lexluthermiester]

The only bad SSDs were the OCZ and Kingston first Gen drives.

I never used any of the first gen Kingston SSD's, but the OCZ models that I used were rock solid. That was a LONG time ago..

Here's one for you. What if you put the Page file on a RAID 0 PCie 4 array?

Really fast pagefile performance. While means if your system does have to drop into the pagefile, the performance will be very good compared to most other systems. If that option is available to you, I say go for it, you have nothing to loose.

Just so I'm clear and I'm not being a total donut here, but we are speaking about the page file under this right??

Yes.

Generally I set to 2GB and move on with life.

See, you go with even less than I generally do. I usually alternate between 3GB or 4GB depending on the amount of system RAM installed. >8GB? 3GB. <8GB? 4GB.

I think I'm going to experiment with 2GB just to see the effect, if any..