Path: csiph.com!fu-berlin.de!uni-berlin.de!individual.net!not-for-mail From: Andy Burns Newsgroups: alt.comp.os.windows-11 Subject: Re: memory usage vs fan usage Date: Thu, 17 Apr 2025 19:57:08 +0100 Lines: 46 Message-ID: References: Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-Trace: individual.net zRUoCoIzo8PEetUFcOk09AttIIIbgx7og74nCph4zHEzwscKkL Cancel-Lock: sha1:Z0rXLYeNp4hj6iUT7EFOZovOGh8= sha256:bimzVBdrbhjqYc3uKXKAL0+dt7fPfLsEoniXKBEyI1U= User-Agent: Mozilla Thunderbird Content-Language: en-GB In-Reply-To: Xref: csiph.com alt.comp.os.windows-11:18465 Paul wrote: > Andy Burns wrote: >>> I've always been a bit suspicious of the MemoryCompression >> service, so I'm experimenting with it turned off for a bit. >> (powershell disable-mmagent -mc) > > Makes sense, but the first question you have to ask, is why > the Memory Compressor is sucking cycles, on machines which > are not down to their last byte of memory. > > On W10, between 250MB in use all the way up to 1GB in use, > the compressor goes from "railed" at the how end, to barely > doing anything at all on a machine equipped with 1GB of RAM. > I did some tests in a VM, to see how it behaves. This was > back when the Memory Compressor was visible in Task Manager. > > You have 16GB or 32GB, and the Memory Compressor should be > taking a snooze in such a case. Is that its defined behaviour, or just what a normal person would hope it does? > Only when the memory use is > at 90-95%, should the compressor be picking up again. > > On Linux, something called "swappiness" defines some of > the memory management behavior. Linux distros, some of > them start to swap at half full, others allow the OS > to get within 50MB of exhaustion, before doing > something about it. Does Windows perhaps have a similar > policy ? Yes, I'm aware of that "dial" > I used Process Explorer to watch the Memory Compressor, then > ran my malloc64.exe test program (home-brew code), and the > Memory Compressor only started to consume cycles near the > very end of the run. There was no sign of any changed > behavior that would account for your memory compressor > to be running. When you're down to 250MB of memory left, > you should start to see a response. I know it doesn't fully hide from taskmgr or procexp, but why does it have to hide what it's doing from them? I mean it must have a .exe or .dll somewhere, how much CPU usage does it do behind a curtain?