View Single Post
  #16  
Old December 24th 18, 06:53 PM posted to alt.comp.hardware.pc-homebuilt
Shadow[_2_]
external usenet poster
 
Posts: 195
Default What's the difference between these two memories ?

On Mon, 24 Dec 2018 15:49:49 -0200, Shadow wrote:

On Mon, 24 Dec 2018 09:25:13 -0500, Paul
wrote:

Shadow wrote:


Last night I synched my time, turned off my network, closed
windows. When I started up today I opened Wireshark, turned the
network back on and time was spot-on (nothing showed on Wireshark
other than the usual ARP stuff and the Neutron query).
After it was on for 3 hours I checked the time and it was 4
seconds off. I disconnected the network again, etc et al, and when I
had rebooted the time was exact, no delay. I did NOT correct the time
before rebooting.
IOW the BIOS time seems to be working fine.
I always assumed Windows used the BIOS hardware clock and not
some "internal software clock".
I read somewhere that BCLK is used for overclocking. My BIOS
settings are set to default. What could be altering the BCLK time by
so much ? And could adding memory have affected it somehow ?
TIA
[]'s


Well, it's either the absolute frequency of BCLK which
is off, or, something is preventing clock tick interrupts
from being serviced.

The best references I've seen on the various clocks in a
computer, is on the VM hosting software company web sites.
They usually explain what clocks are inside a real PC,
and how the virtualized environment provides those same
clocks as "fakes". But it also teaches you about how
clocks work on the host itself.

If you have a modern multi-core Intel processor, you
can try locking the cores together temporarily as a test.
The machine can save power if the cores operate
independently on frequency, but it also causes some
complications when handing clock information from
one core to another. My other machine, the core clocks
are locked, and turbo is disabled (to prevent overheat).


Well, you seem to on to something. My computer was on for
nearly 3 hours running simple "one thread" programs and didn't lose a
second.
So I put it to convert a video with Wondefox Video Converter,
which uses all cores on my AMD-FX8300 (it says it's using 7 cores).
The conversion took under 5 minutes and my clock is now 12 seconds
behind.
So it has something to do with using more than one core, AND
it never happened before I added memory to the second memory bank (I
can't remember clock ever being so late but I never actually tested
it).

I'm using XP, if that matters.
I'll now boot into Linux and see if the same thing happens
there.


No. In Linux even using all 8 cores at near max there is no
slowdown of the clock. I converted 3 videos using ffmpeg. The CPU
ventilator sounded like a helicopter taking off. So it must be an OS
thing.
Funny thing, I tried ffmpeg in XP and the clock went FORWARD 8
seconds.
I suppose I'll just have to carry on using Neutron at Startup,
since the time is never off by more than a minute, it won't affect me
as a user.
[]'s

I just checked with system monitor, and although it does use
core 1 most of the time, there are occasional small spikes in all the
other cores. I thought it would only use the other cores when under
stress ....
--
Don't be evil - Google 2004
We have a new policy - Google 2012