Forums before death by AOL, social media and spammers... "We can't have nice things"
| alt.comp.os.windows-11 | Steaming pile of horseshit Windows 11 | 4,852 messages |
[ << oldest | < older | list | newer > | newest >> ]
| Message 4,072 of 4,852 |
| Paul to sticks |
| Re: Windows fans: tell me where the narr |
| 26 Jan 26 16:02:11 |
From: nospam@needed.invalid On Mon, 1/26/2026 10:22 AM, sticks wrote: > On 1/25/2026 8:42 PM, Paul wrote: > > ---snip--- > >> # In Summary >> >> Windows Update’s “slow phase” is dominated by: >> >> | Stage | What Happens | Why It is S ow | >> |-------------------------|---------------------------------- ------|----------------------------------| >> | Manifest parsing | Load thousands of XML mani ests | Many tiny files, heavy parsing | >> | Supersedence evaluation | Compare each component to WinSxS | Graph traversal, pointer chasing | >> | Applicability rules | Evaluate prerequisites and dependencies | Constraint solving | >> | Transaction planning | Build install plan | Large metadata structures | >> >> This is fundamentally a **metadata computation problem**, not an I/O problem. >> >> Your observation about L3 cache is spot-on: the workload is m mory-latency-bound. >> >> ********************* End: CoPilot Answer ************************ >> >> Paul > > Thanks for the very interesting work to show us exactly what the process all involves. Very informative and helpful! I just thought it was weird, that the machine with more L3, seems to be doing updates faster. My original conclusion was that there was cheating of some sort going on :-) Paul --- SoupGate-Win32 v1.05 * Origin: you cannot sedate... all the things you hate (1:229/2) |
(c) 1994, bbs@darkrealms.ca