The thing is, the user may notice. 60ms vs 180ms is the time allocated per thread. If it's just one user, he/she will not notice a difference. But as more users log into the box, so do the number of threads (provided the users actually do something, like run an applications. Otherwise, what was the point of them logging on). Think of it like this: You're off to grocery store to buy a package of toilet paper. Two checkout lanes are open (as in two CPUs), both lanes have a queue of 3 carts (threads) each. All 6 carts are filled to the brim. It would take an average of 5 minutes each shopper to complete their order. Your item would only take 30 secs to process. You're looking at a 15 minute wait to purchase single item. What if the store had a quanta limit. Say 2.5 minutes per cart, at the end of 2.5 minutes what's been processed is saved and the remaining items in the cart go to the end of the line. So with 3 shoppers in front, that's a 7.5 - 8 minute wait (don't for get to add time for cashier to save current shopper's information and switch to the next shopper (Context Switch)). Now let's say the quanta limit is 1 minute per cart. Now it's just a 3-3 1/2 minutes wait. Granted, the shoppers with a full cart will have to wait longer, but those with smaller orders will not. Hopefully, if your grocery store is smart, it will give you a priority boost and open an express lane. If the queue gets too long, it will open new registers and/or higher faster cashier. Maybe add some Hyperthreading and add a bagger at each lane. If business is good and the store is small, it maybe time to remodel and upgrade the store. Or perhaps it's time to build add another store. Tuning and Optimization of a terminal server is a must, there are many well known and well documents tweaks. Personally, I would stick to those. You may understand the internal workings of windows better than anyone else but that's because you're special. Someday, those servers maybe someone else's responsibility and you may not be around to help. Joe -----Original Message----- From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On Behalf Of Jim Kenzig Sent: Tuesday, September 16, 2003 7:49 AM To: thin@xxxxxxxxxxxxx Subject: [THIN] Re: thread quanta and chkroot.cmd thingy And do you really think a enduser will notice the time difference between 60 ms and 180 ms? And one other thing I want to know...how do you figure in the human error of the testing of these things... I mean who is stopping and starting the stopwatch to time this? Are there computer programs you are using to do this? If you blink isn't that a few hundred milliseconds lost? How can you be sure it is accurate? -- Regards, Jim Kenzig Quoting Brian Madden <brian@ > My completely theoretical and totally unsubstantiated theory: This setting > does not matter on a Terminal Server. > > Brian > > Brian Madden > 202.302.3657 ******************************************************** This Week's Sponsor: ThinPrint http://www.thinprint.com ********************************************************** Useful Thin Client Computing Links are available at: http://thethin.net/links.cfm For Archives, to Unsubscribe, Subscribe or set Digest or Vacation mode use the below link: http://thethin.net/citrixlist.cfm ******************************************************** This Week's Sponsor: ThinPrint http://www.thinprint.com ********************************************************** Useful Thin Client Computing Links are available at: http://thethin.net/links.cfm For Archives, to Unsubscribe, Subscribe or set Digest or Vacation mode use the below link: http://thethin.net/citrixlist.cfm