How to disable compressed memory on windows 10? Am I reading task manager correctly?

[FreshJR]

This is a two part question.

I'd like to disable compressed memory. In my opionion ram is supposed to be blazing fast. I do not want to wait for the CPU to decompress whatever it compressed or waste CPU power on that task. From what I read on google, I would have to disabled the super fetch service.

From what I remember, the superfetch service does more than compress memory. From windows xp and onwards it would preload "cache" stuff I may open in the future into memory instead of fetching it from storage at runtime. That feature was nice and I wouldn't want to lose that if possible.

Part2: Am I reading memory usage correctly.

-The total addressable space is 25.9GB (24gb real memory and 1.9gb page file).
-Out of 24 gb's of ram, I have 7.4gb actively used and 11.9 gb cached (by superfetch?). The 4.7gb remainder is truly unallocated.

Additionally there are two parts of 568mb + 340mb that have been moved into the page file. Do i subtract that 568+340mb from real in use phsyical memory?

The open applications requested to use up to 20gb memory but are using 11.4gb (7.4gb used + 4.2 compressed). The remainder of their commit ram is 8.6gb. Ideally if that 8.6gb is all static data to be loaded from the harddrive it has already been read and stored inside the cached portion of memory? Yet it could also be ungenerated runtime data? Is their also a possibility that the remainder of the committed ram is remains on harddrive to still be loaded instead of the cache?

 

[TheELF]

For memory compression start powershell as admin and use these commands
Disable-MMAgent -mc
Enable-MMAgent -mc
to turn it off or back on again.

The rest I have no idea what you are talking about or why anybody would even care.

 

[FreshJR]

For memory compression start powershell as admin and use these commands
Disable-MMAgent -mc
Enable-MMAgent -mc
to turn it off or back on again.

The rest I have no idea what you are talking about or why anybody would even care.

Thanks !!

I know that learning how to read the memory tab wont change anything, but I'd rather not be ignorant and try to learn.

Just like additional gauges on a car. Sure gages like oil pressure, boost, knock, A/F ratio, trans temp are unnecessary to the general population, yet they are important indicators to your cars performance that some people want to look at. Using the same logic, you don't even need RPM, or engine temp.

 

[TheELF]

I'm sure you can find the info on ms's technet site somewhere if you search enough.
https://social.technet.microsoft.com/Search/en-US?query=task manager memory&ac=4

 

[Micrornd]

OP - You should try to understand this better before jumping to conclusions.
Watch this video and then turn memory compression back on - https://channel9.msdn.com/Blogs/Seth-Juarez/Memory-Compression-in-Windows-10-RTM

 

[FreshJR]

OP - You should try to understand this better before jumping to conclusions.
Watch this video and then turn memory compression back on - https://channel9.msdn.com/Blogs/Seth-Juarez/Memory-Compression-in-Windows-10-RTM

I will be observing my commit overtime. If it rarely exceeds 24gb, then I do not want compression for sure. They can say what they want, but compressing 4gb takes a chunk of power, yet if my use rarely exceeds phsyical available memory, then it can be completely avoided. If I find myself consistently over, then I'd rather do an upgrade to 32gb

 

[Elixer]

Yeah, really not worth disabling it.
They did a good job here.
You can blink and miss the decompression.

 

[TheELF]

I found out about it because some of my games tanked in performance,disabling the compression made a world of difference.

 

[mikeymikec]

I found out about it because some of my games tanked in performance,disabling the compression made a world of difference.

Could you quantify this in some useful way please? Specs, benchmarks, etc. Because I'm fairly sure that if Win10 sucked significantly for gaming, reviewers would have spotted significant differences at least a year ago already and posted their findings.

 

[Elixer]

I found out about it because some of my games tanked in performance,disabling the compression made a world of difference.

Yeah, really wondering what games that compression decreased performance that much.
OP has 16GB, and I can't think of any game that it would have trouble running on 8GB, let alone 16GB.

The whole point of the compression is to save page faults, and it does that quite well.

 

[TheELF]

It was some frostbyte game I think it was mirrors edge catalyst.
Could just have been a fluke though a one time deal, anyway that's when I noticed compression using huge amounts of CPU in task manager.

 

[FreshJR]

Yeah, really wondering what games that compression decreased performance that much.
OP has 16GB, and I can't think of any game that it would have trouble running on 8GB, let alone 16GB.

The whole point of the compression is to save page faults, and it does that quite well.

With my results, I feel that input lag has gone down significantly. I do not have the tools to measure it, but I feel like I shaved off 20ms.

Sure your eyes cannot see the difference, but you feel the difference just moving the mouse.

Input lag can be significant, even with fast response displays. My pc which is probably twice the speed of my laptop in cpu power and infinitely stronger than intel IGPU graphcis, is 16ms+ behind in response compared to the laptop. My pc screen has a 4ms display. Wish I knew the source (most likely AMD drivers or time to transfer data from CPU to GPU over PCIE?).

This is easily seen in games that require frame specific actions. I know that it is off at least 1 frame (16ms) but most likely 2-4 frames. Eliminating memory compression has seem to speed up the whatever happens between mouse input and result.

 

[Elixer]

With my results, I feel that input lag has gone down significantly. I do not have the tools to measure it, but I feel like I shaved off 20ms.

Sure your eyes cannot see the difference, but you feel the difference just moving the mouse.

Input lag can be significant, even with fast response displays. My pc which is probably twice the speed of my laptop in cpu power and infinitely stronger than intel IGPU graphcis, is 16ms+ behind in response compared to the laptop. My pc screen has a 4ms display. Wish I knew the source (most likely AMD drivers or time to transfer data from CPU to GPU over PCIE?).

This is easily seen in games that require frame specific actions. I know that it is off at least 1 frame (16ms) but most likely 2-4 frames. Eliminating memory compression has seem to speed up the whatever happens between mouse input and result.

This doesn't make sense, Window's compression / decompression routine is highly unlikely that it was causing that, it would be practically impossible to have that many hard faults per sec to have issues with 16GB. The routine can decompress 1GB/sec per core, and if you think it was causing 20ms lag for mouse (only for mouse?), that would mean some unbelievable thrashing on your system, all your I/O would have spiked, while Windows was trying to read the data again.

The compression feature also will not touch anything that is accessed frequently, as a game would do. As I said, it is only meant to prevent hard page faults by compressing things that aren't currently being used, so the OS don't have to dump that to the page file, and later, when it is needed again, go to the page file and retrieve that info.

As for what really is actually causing your input lag, no idea, there are a ton of things that could cause this, drivers, other things being run at the same time, the game recording crap, and so on...

Running performance monitors on my two systems with AMD & nvidia GPUs, there was no perceivable hit between enabling it, and disabling it, on multiple AAA games, from FPS to RTS games.

I have seen posts about disabling compression floating around, but, nobody has been able to verify the results, (as in do XYZ, and you can reproduce it almost every time), and it doesn't happen for the majority of the people, which is why, this seems to be a red herring type of thing.

 

[razel]

Watch this video and then turn memory compression back on - https://channel9.msdn.com/Blogs/Seth-Juarez/Memory-Compression-in-Windows-10-RTM

You know they're doing real work when the white board is no longer white.

 

[Ketchup]

Reminds me of when I put two IDE IBM spinners together in RAID-0 and thought it made the computer faster. After a while things slowed down and I realized it wasn't the RAID at all, it was Windows getting slower.

 

[FreshJR]

This doesn't make sense, Window's compression / decompression routine is highly unlikely that it was causing that, it would be practically impossible to have that many hard faults per sec to have issues with 16GB. The routine can decompress 1GB/sec per core, and if you think it was causing 20ms lag for mouse (only for mouse?), that would mean some unbelievable thrashing on your system, all your I/O would have spiked, while Windows was trying to read the data again.

The compression feature also will not touch anything that is accessed frequently, as a game would do. As I said, it is only meant to prevent hard page faults by compressing things that aren't currently being used, so the OS don't have to dump that to the page file, and later, when it is needed again, go to the page file and retrieve that info.

As for what really is actually causing your input lag, no idea, there are a ton of things that could cause this, drivers, other things being run at the same time, the game recording crap, and so on...

It had to be the game data that was compressed. Where else would it find 4GB of data out of the total 12GB used?

Game loads textures, doesn't use them for a while, windows starts compressing it, then you get delay once those textures are needed. I discounted the fact that there was a delay to compress the data, since it could just use free cpu cycles.

Sure it can do 1GB/sec, but that still adds up to ms's being added upon a accessing a few MB of compressed data. I still haven't exceeded 12GB used out of my 24GB. So placebo effect or not, I do not need the compression, makes more sense when memory starved.

This wouldn't show up in performance monitors. FPS was fine, I am talking about input lag. You would need an LED in series with a keyboard key then use the slow motion camera on your phone to measure the frame times

I have no data, so yes, I am not arguing that. You can list it as a placebo for the time being. I might have an old keyboard, if I can figure out how to wire space in series with one of the numlock LED's, I can provide hard data

 

[Elixer]

I am just trying to say that Windows wouldn't even touch game data with 24GB in your system.
Yes, obviously, something was eating up RAM, and I suppose the game itself (which game was this?) might have explicitly requested paged memory, but even then, it still would be faster than the old way.

Contrary to how it seems, the compression/decompression is in fact FASTER than what normally would happen, not slower.

That is why I am puzzled by this, it shouldn't have any result on your system, until those hard page faults happen, then you would see a improvement over the older way.

 

[TheELF]

We aren't talking about standard conditions though,wen I had the problem the compression would completely consume one whole core,which is a problem if you don't have a multi core beast.