If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#21
|
|||
|
|||
What's the difference between these two memories ?
On Thu, 27 Dec 2018 14:51:48 -0500, Flasherly
wrote: On Thu, 27 Dec 2018 08:42:23 -0200, Shadow wrote: Sorry Flasherly, I got lost after the first phrase. I sync to an atomic clock on startup, so it should be within half a second of "real" time at the end of the day. But I can live with 30 seconds off. I was worried it might be hardware failure, but since Linux does not have the glitch, I'm pretty sure it's just XP having fits with the amount of CPU and memory at its disposal. The first written phrase, I provided, corresponds to after your startup, and that is specific to Dimension 4, in the user settings, for defining how often Dimension 4 performs an Atomic Clock synchronization. The Dimension 4 polling interval, I checked, just after the prior post, and I've determined that my computer is about 8 seconds faster than a standalone La Crosse radio receiver atomic clock, on the wall, behind this monitor. But the whole point is that if you can go beyond an OS start-up synchronization event, a more frequent interval timed synchronization may improve your computer's chronographic accuracy. OK...checking with the second hand to the Casio. This is my watch: https://en.wikipedia.org/wiki/File:C...ave_Ceptor.jpg I'm about 4 seconds fast on the computer, thus my computer is not as accurate as the Casio. (I've checked the Casio to other standards, such as a short-wave band radio receiver and England's "Big Ben" timed broadcast. So my computer is not within 4 seconds near to a chronograph, nor is Windows or, apparently, Dimension 4.) Checking the BIOS clock may also be an indicated course to account;- as might running Linux shed further light on narrowing in on a cause of the discrepancy for accurate time keeping. Between a computer crystal-derived signal reference and a WEB software interface to poll a reference Atomic Clock server, there's no excuse for this behavior. I believe it would be safe to assume, that you not build such as a rocket-ship to blast off to the moon, not if you're designing that trajectory based on a computer's ability to keep timed accuracy. Oh, it's a freeware time-sync utility. I thought you were talking about science fiction. So is Neutron. http://keir.net/neutron.html 7Kb small, and has an INI file so you can change the servers. (the included ones are not working). If the first server doesn't respond it tries the second, etc, for a total of 14. I can give you my list if you want. Keir writes good software. []'s -- Don't be evil - Google 2004 We have a new policy - Google 2012 |
#22
|
|||
|
|||
What's the difference between these two memories ?
On Thu, 27 Dec 2018 18:54:56 -0200, Shadow wrote:
Oh, it's a freeware time-sync utility. I thought you were talking about science fiction. So is Neutron. http://keir.net/neutron.html 7Kb small, and has an INI file so you can change the servers. (the included ones are not working). If the first server doesn't respond it tries the second, etc, for a total of 14. I can give you my list if you want. Keir writes good software. []'s I'd have asked if you tried decreasing a N Interval to poll for a lower integer (more often) than a default periodicity, or manually, but there may not be that provision if there's nothing more to configure it than the illustration. http://keir.net/resources/scrn_neutron.png I'd personally want that feature automated - within a definable N Integer - as a part of a clock-aide program options. Or, you may simply write it yourself, perhaps easily enough, with AUTOIT scripting language for Windows. Fiction occurs as an assumption of limitations, that a computer is a instrument of capabilities to measure time at some discrepancy from a scientific principle atomic clocks provide. Yet computers may perhaps play that role in specialized instances. In data network transfers as a part of secure protection protocols augmented to account a high-precision timing event. Perhaps augmented with specialty time-references or equipment outside of a common desktop build. I wouldn't rule it out offhand. .. . . Temperature is still only part of the story. Thermal noise is the ultimate limitation on crystal oscillator performance (crystals are by far the most common type of "digital clock"). The crystal itself has Brownian noise due to dissipative effects of air resistance, anchor loss, and thermoelastic damping. Brownian noise creates a random force that acts to disturb the crystal vibration. This force creates random fluctuations in the exact oscillation frequency of the crystal. The electronic oscillator circuit responsible for compensating for the energy dissipation due to mechanical damping also adds noise that has essentially the same effect. This still isn't quite the whole picture. Random changes in frequency lead to a random walk in the period between two zero crossings. You can think of this process as what happens when you flip a quarter and keep track of the total heads and tails count. The odds of any flip giving heads or tails is 50%. If you add 1 to your count for every head and subtract one for every tail, the standard deviation of the count is unbounded as time increases toward infinity. Similarly, random fluctuations in the oscillation frequency "accumulate" over time to lead to timing drift of the reference. https://docs.microsoft.com/en-us/win.../accurate-time Impact of increased polling and clock update frequency In order to provide more accurate time, the defaults for polling frequencies and clock updates are increased which allow us to make small adjustments more frequently. This will cause more UDP/NTP traffic, however, these packets are small so there should be very little or no impact over broadband links. The benefit, however, is that time should be better on a wider variety of hardware and environments. For battery backed devices, increasing the polling frequency can cause issues. Battery devices don’t store the time while turned off. When they resume, it may require frequent corrections to the clock. Increasing the polling frequency will cause the clock to become unstable and could also use more power. Microsoft recommends you do not change the client default settings. Domain Controllers should be minimally impacted even with the multiplied effect of the increased updates from NTP Clients in an AD Domain. NTP has a much smaller resource consumption as compared to other protocols and a marginal impact. You are more likely to reach limits for other domain functionality before being impacted by the increased settings for Windows Server 2016. Active Directory does use secure NTP, which tends to sync time less accurately than simple NTP, but we’ve verified it will scale up to clients two stratum away from the PDC. As a conservative plan, you should reserve 100 NTP requests per second per core. For instance, a domain made up of 4 DCs with 4 cores each, you should be able to serve 1600 NTP requests per second. If you have 10k clients configured to sync time once every 64 seconds, and the requests are received uniformly over time, you would see 10,000/64 or around 160 requests/second, spread across all DCs. This falls easily within our 1600 NTP requests/sec based on this example. These are conservative planning recommendations and of course have a large dependency on your network, processor speeds and loads, so as always baseline and test in your environments. It is also important to note that if your DCs are running with a considerable CPU load, greater than 40%, this will almost certainly add noise to NTP responses and affect your time accuracy in your domain. Again, you need to test in your environment to understand the actual results. |
#23
|
|||
|
|||
What's the difference between these two memories ?
On Thu, 27 Dec 2018 20:07:37 -0500, Flasherly
wrote: On Thu, 27 Dec 2018 18:54:56 -0200, Shadow wrote: Oh, it's a freeware time-sync utility. I thought you were talking about science fiction. So is Neutron. http://keir.net/neutron.html 7Kb small, and has an INI file so you can change the servers. (the included ones are not working). If the first server doesn't respond it tries the second, etc, for a total of 14. I can give you my list if you want. Keir writes good software. []'s I'd have asked if you tried decreasing a N Interval to poll for a lower integer (more often) than a default periodicity, or manually, but there may not be that provision if there's nothing more to configure it than the illustration. http://keir.net/resources/scrn_neutron.png I'd personally want that feature automated You could just use Task Scheduler, though I don't like it running in the background, so I disabled the service. http://www.blackviper.com/windows-se...ask-scheduler/ []'s -- Don't be evil - Google 2004 We have a new policy - Google 2012 |
#24
|
|||
|
|||
What's the difference between these two memories ?
On Fri, 28 Dec 2018 08:29:01 -0200, Shadow wrote:
You could just use Task Scheduler, though I don't like it running in the background, so I disabled the service. http://www.blackviper.com/windows-se...ask-scheduler/ Learning an OS well enough to get around can seem at times less daunting than defending or hardening it. And getting around, in the case of Microsoft, has been as much better left to 3rd-party developers, generously approximated by free or near program utilities. Hardly an ingenuous claim or premise when, subsequently abandoned or rendered obsolete from manipulation by tech industry, intent on wholesaling individuals into an advertising market of cornered cattle yards stocked with handheld-sets of social media. Auto-It is somewhat elegant for running on top of Microsoft in a script form. I've even run into programs, subsequently distributed for freeware, written initially within Auto-It language conventions, as is adequately explained within the distribution source for an included documentary file. A recursive call to an timed event and a configuration allowance for a clean exit should be relatively simple;- worst that could happen is it could get "gummy" during an OS shutdown. |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
8x8, 16x16, 1M x 36 P -ram memories | Majki Majk | General | 2 | November 28th 08 05:53 AM |
RLP memories | Guillaume Dargaud | General | 0 | September 23rd 08 07:47 PM |
Using different types of DDR memories together | Hamza | General | 1 | August 25th 07 06:56 AM |
2 memories 256 MB and 512 MB | thewhitebsd | General | 2 | January 28th 05 01:40 PM |