Driver to drive?

Skybuck Flying wrote:
I am also quite annoyed with AMD and their X2 processor.

The multiplier is locked as some of you know... so properly
underclocking is not possible.

The motherboard had two nice features:

CPU Clock Multiplier or something like that.

and

HT Multiplier.... (hyper transport)

First I tried HT... later I figured it was the wrong setting.

I didnt notice any performance degradation with HT running at 1x instead
of auto/5x ?

Maybe HT is for core to core communication... but I think it's main roll
was memory speed/bandwidth.

Didnt notice a thing... quite odd.

The HT multiplier was unlocked according to CPU-Z

Unfortunately the CPU multiplier not :( setting it to 5x or 4.5x or
something instead of 10x had no effect :(

I am not going to mess with voltages cause I dont know how to do that
properly... and certainly don't want to risk damage ;)

Just gonna wait for the night too cool down before I continue my gaming :)

Or maybe I do a low power game ;) :)

Bye,
Skybuck.
With Cool N' Quiet enabled, you can change the CPU multiplier.

If you entered the Windows control panel "Power", if you select the wrong schema,
it can leave the CPU running at full speed. Selecting a different schema,
gives Cool N' Quiet voltage and multiplier changes automatically. This
does not change the peak heat output, but improves thermal behavior
at idle. If you run Prime95, it will get just as hot.

If you use RMClock, you should be able to see the multiplier options.
If no options are visible (as in this picture), check your BIOS settings.

http://1.bp.blogspot.com/_pEfSMUbljUA/SpL8iTa_d7I/AAAAAAAAACY/Kj3D5Qj4L-U/s1600/t2400_rmclock.jpg

You can get RMClock here.

http://cpu.rightmark.org/download.shtml

With a little work, I expect you can create an RMClock schema, that
prevents the top multiplier from being used. And then, even running
Prime95, the temperature could be reduced. But you're not going to
like the reduced CPU performance.

AMD used to have a spec, where for each processor it showed the P-states.
That would explain the minimum and maximum multipliers used, when the
processor was tested at the factory. Most other listings you can find,
only list the maximum multiplier (for top CPU speed).

As an example, I can use a cpu-world.com entry. A 3800+ has a 2GHz CPU
internal clock, and a 200MHz base clock, for a peak multiplier of 10X.

http://www.cpu-world.com/CPUs/K8/AMD-Athlon%2064%20X2%203800%2B%20-%20ADA3800DAA5CD%20(ADA3800CDBOX).html

"In the minimum performance state (P-state) the processor runs at
1000 MHz and 1.1 Volt core voltage. Thermal Design Power in the
minimum P-state is 40.1 Watt"

That suggests the minimum CPU multiplier is 5X. And Cool N' Quiet
can vary the multiplier from 10X to 5X, while Windows is running.
The RMClock dialog box would show six rows of info, for the
allowed multiplier values.

*******

The HT multiplier, sets the clock for the Hypertransport bus. That
is the bus between the CPU and Northbridge. Setting the multiplier
to 1X instead of 5X, reduces bandwidth to your video card by a
factor of five or so. There is a second Hypertransport bus segment,
between the Northbridge and Southbridge. And you can fool around
with that if you want, as well.

The HT multiplier doesn't affect memory, since the memory controllers
are right on the CPU itself. Only the chipset bus bandwidth is affected
by HT.

A legitimate reason for reducing HT multiplier, is if you are
overclocking by raising the base clock on the CPU. If you raise the
base clock from 200MHz to 250MHz, you drop the HT multiplier from 5X
to 4X. Keeping a constant 1000MHz clock on the HT bus. That's the
basic idea. Hypertransport transfer rates, differ from one socket
to another, so my values listed here, are just to illustrate the
concept of staying below the limit for the bus. Some later AMD processors,
use a faster HT bus.

Paul
 
Yeah good thinking Paul.

The AMD X2 3800+ processor does have some kind of ability to scale
down/throttle down, just have to find how to do it. It's not via the bios
that's for sure... but this could be a benefit.

Apperently it's automatically supported, by the hardware/bios/windows 7.

Start->Control Panel->System & Security->Power Options->Power Saver->Change
Plan Settings->Change Advanced Power Settings->Processor Power
Management->Maximum Processor State->Setting

^ change this from 100% to 50% or 5% or so... 50 or 5 for this processor
doesn't matter it will go from 2 ghz down to 1 ghz which is nice... this
should make the processor run a lot cooler.

This is absolutely necessary for me for the coming days cause europe is
about to get the biggest heat wave since years. Probably 32 to 33 degrees
celcius tomorrow.

It will be interesting to see if my PC will boot at all or immediatly
shutdown... so I have to do this now... before I can't reach the setting
anymore... which is kinda interesting.

The bios method would have been faster... if I can't reach the windows
settings in time... then the PC would be stranded in heat.

Also the svhost.exe process consuming 50% energy during boot has not been
solved yet... I think you had some command for it to run each services in
it's own svhost.exe to see which is the culprit. Doesn't mean yet that the
problem can be solved but at least it would give me some idea where to start
looking.

It's either RPC, Power, Plug & Play... I know that much.

So that nasty bug could cause excessive heat leading to early shutdown.

I couldn't use live mail in 1 GHZ mode while that process was consuming all
50% so I put it back up to 2 GHZ so I can type this mail.

Usually the svhost.exe problem goes away after a few minutes. Then I will
turn down the setting to 5% or so... I will also try sleep mode instead of
shutdown... to try and prevent that service from rebooting and causing a
problem tomorrow.

Bye,
Skybuck.
 
And to reward you for your help, here is a little interesting screenshot:

http://www.skybuck.org/Winfast/TemperatureReadingsWith1GHZMode.png

The build-in cheap temperature meter in my Desktop Clock might be a few
degrees of when comparing with this screenshot which claims 30 degrees
temperatures as inputs.

Anyway... the svhost.exe has died down... so now it's very interesting to
try and game a bit and see what that does to game performance and
temperature ! ;)

Bye,
Skybuck.
 
On 22/07/2013 6:32 AM, Skybuck Flying wrote:
I didn't bother re-appling the therman paste interface material as far
as I remember (just used the stuff that was on it... not believing the
little bubbles bs) or maybe I did...

I think I pulled of the whole thing for cleaning or so... and then put
it back on... or maybe not...

I don't think it's a thermal paste interface material probably because
on idle the temperature drops back to 32 degrees celcius or so... which
to me seems to indicate it's running as it would if it was applied fresh.

Bye,
Skybuck.
The temperature elevation above ambient is the dissipated power divided
by the thermal resistance to ambient, so when it's idle, the temperature
elevation won't be high anyway. For example, if the thermal resistance
is twice what it should be, and (just pulling numbers out of the air for
illustrative purposes) the normal elevation on idle would be 2.5
degrees, then it will actually be 5 degrees - not much difference. But
if the elevation under computing load should be 20 degrees, it will
actually be 40.

I'm not clear what you mean by "little bubbles bs", but if you're
talking about literal bubbles in the thermal compound, it's clear that
they will indeed raise the thermal resistance, simply because they
reduce the area avaiable to conduct the heat, air not being a good
conductor.

Sylvia.
 
On Sun, 21 Jul 2013 22:43:19 +0200, "Skybuck Flying"
<Windows7IsOK@DreamPC2006.com> wrote:

Right now it's 27.5 degrees celcius 0.5 degrees down from 28 degrees
(...)
This future is not on the top of my list for a next/new motherboard ;)
Instead wasting everyone's time writing all that irrelevant crap,
perhaps some temperature measurements and some calculations might be
useful to determine which part of the thermal chain is causing your
alleged overheating? It's almost like you would prefer to complain,
rather than fix your machine.

I can only hope.

--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
 
Micro/nano bubbles.

^ I don't really believe in it until proven ;)

I dont think there is difference between cleaning + smearing it vs or
pulling it off + smearing it a bit again + and pushing it onto it again.

The force is probably going to squash lot's of tiny little bubbles if there
are any at all.

Bye,
Skybuck.
 
"Jeff Liebermann" wrote in message
news:2t4pu8pranllnd9q8sh5qj021f28ovrt3u@4ax.com...

On Sun, 21 Jul 2013 22:43:19 +0200, "Skybuck Flying"
<Windows7IsOK@DreamPC2006.com> wrote:

Right now it's 27.5 degrees celcius 0.5 degrees down from 28 degrees
(...)
This future is not on the top of my list for a next/new motherboard ;)
Oh I see I made a little typo there... a common mistake.

I ment to write:

This Feature is now on the top of my list for a next/new motherboard ;)

The shutdown is quite nice... it alert me that there is a big heat related
problem ! ;)

Now it's solved for the mean time..

Played some frozen throne and company of heroes, so far no shutdowns...

And stutter/slow down is barely noticeable at least in a 2 v 2 or 1 v 1
hihihi.

What I would like to know is how WATT is transferred into HEAT.

Mostly for CPU and for GPU.

Anything else like memory chips and harddisks is extra.

For example: I wonder how much heat a top model of nvidia puts out ? It
claims it needs 300 watts of power... does this mean it outputs 300/84 times
that of a 84 watt haswell processors ?

Bye,
Skybuck.
 
En el artículo <2t4pu8pranllnd9q8sh5qj021f28ovrt3u@4ax.com>, Jeff
Liebermann <jeffl@cruzio.com> escribió:

Instead wasting everyone's time writing all that irrelevant crap,
Skybuck is a troll, best ignored. You and Sylvia are wasting your time
responding to him.

--
(\_/)
(='.'=)
(")_(")
 
On Mon, 22 Jul 2013 07:00:55 +0200, "Skybuck Flying"
<Windows7IsOK@DreamPC2006.com> wrote:

For example: I wonder how much heat a top model of nvidia puts out ? It
claims it needs 300 watts of power... does this mean it outputs 300/84 times
that of a 84 watt haswell processors ?
No, it means that the card converts 300 watts of power to heat when
running at full blast. It only burns 130 watts at "idle". I hope you
can tolerate the associated fan noise.
<http://community.futuremark.com/hardware/gpu/NVIDIA+GeForce+GTX+690/review>
<http://www.pcper.com/reviews/Graphics-Cards/NVIDIA-GeForce-GTX-690-Review-Dual-GK104-Kepler-Greatness/Power-Consumption-T>
<http://www.geforce.com/hardware/desktop-gpus/geforce-gtx-690/specifications>
<http://www.anandtech.com/show/5805/nvidia-geforce-gtx-690-review-ultra-expensive-ultra-rare-ultra-fast/16>
Note that the GPU can run at up to 98C max.

Only $1000 for the GTX 690 card. Truly a bargain for a machine that
you probably spend more time troubleshooting than using.


--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
 
Note that the GPU can run at up to 98C max.

What about the graphics card PCB and other components ?

What about the motherboard and components around it ?

Seems like this graphics card is designed to fry one's computer ?

Bye,
Skybuck.
 
On 22/07/2013 2:58 PM, Skybuck Flying wrote:
Micro/nano bubbles.

^ I don't really believe in it until proven ;)

I dont think there is difference between cleaning + smearing it vs or
pulling it off + smearing it a bit again + and pushing it onto it again.

The force is probably going to squash lot's of tiny little bubbles if
there are any at all.
Making the air go where?

But the real issue is whether there's enough compound to ensure thermal
contact across the pad.

Sylvia.
 
Skybuck Flying wrote:
Micro/nano bubbles.

^ I don't really believe in it until proven ;)

I dont think there is difference between cleaning + smearing it vs or
pulling it off + smearing it a bit again + and pushing it onto it again.

The force is probably going to squash lot's of tiny little bubbles if
there are any at all.

Bye,
Skybuck.
But you don't smear it a bit. You use the "middle dot" method.
Placing a dot of paste in the center of the CPU, and compressing
the cooler on top, makes the paste spread radially, and helps
force out air.

http://www.arcticsilver.com/amd_application_method.html

http://www.arcticsilver.com/PDF/appmeth/amd/md/amd_app_method_middle_dot_v1.3.pdf

You can do a calibration run first, wipe off excess, apply a dot of
paste, see how much it spreads. Then, take it apart, wipe off the
paste, and apply an amount scaled so that it'll spread and cover
the entire surface.

Paul
 
On Mon, 22 Jul 2013 05:25:43 -0400, Paul <nospam@needed.com> wrote:

Skybuck Flying wrote:
Micro/nano bubbles.

^ I don't really believe in it until proven ;)

I dont think there is difference between cleaning + smearing it vs or
pulling it off + smearing it a bit again + and pushing it onto it again.

The force is probably going to squash lot's of tiny little bubbles if
there are any at all.

Bye,
Skybuck.

But you don't smear it a bit. You use the "middle dot" method.
Placing a dot of paste in the center of the CPU, and compressing
the cooler on top, makes the paste spread radially, and helps
force out air.

http://www.arcticsilver.com/amd_application_method.html

http://www.arcticsilver.com/PDF/appmeth/amd/md/amd_app_method_middle_dot_v1.3.pdf

You can do a calibration run first, wipe off excess, apply a dot of
paste, see how much it spreads. Then, take it apart, wipe off the
paste, and apply an amount scaled so that it'll spread and cover
the entire surface.

Paul

Air bubbles are bad. They reduce the amount of the medium the thermal
numbers are based on. They reduce thermal efficiency, regardless of
size.

There are other effects in other industrial segments to further
illustrate this.

We found early on that potting HV circuits without vacuum reduction of
any and all air introduced during the mixing and pouring process of the
potting medium would always eventually result in failure of the circuit.
With those air voids being the culprit.
 
Dot in the middle seems flawed to me.

It will not reach the corners and might leave big air gaps.

Which can then lead to excessive heat build up around corners and damage
capacitators around the cpu.

Bye,
Skybuck.
 
"Skybuck Flying" <Windows7IsOK@DreamPC2006.com> wrote in message
news:ab299$51ed9535$5419b3e4$6131@cache90.multikabel.net...
Dot in the middle seems flawed to me.

It will not reach the corners and might leave big air gaps.

Which can then lead to excessive heat build up around corners and damage
capacitators around the cpu.

Bye,
Skybuck.
Wrong.

"capacitators"? *sigh*
 
Skybuck Flying wrote:
Dot in the middle seems flawed to me.

It will not reach the corners and might leave big air gaps.

Which can then lead to excessive heat build up around corners and damage
capacitators around the cpu.

Bye,
Skybuck.
This is why you use an inspection mirror, and
after the cooler is clamped into place, look at
the joint between the cooler and CPU.

You should see a white line on all four sides, which
tells you the paste made it out to the edge. That's
how you check that you have used enough paste.

By doing the calibration step, and applying too little
the first time, I haven't managed to apply so much in
the second attempt, that it went all over the place. Doing
a trial installation, knowing you are using too little
paste, helps you estimate how much is really needed.

If you don't see the "white line" of the paste oozing
out, then apply a little bit more (a bigger dot), and
try again.

Paul
 
On Mon, 22 Jul 2013, Chris S. wrote:

"Skybuck Flying" <Windows7IsOK@DreamPC2006.com> wrote in message
news:ab299$51ed9535$5419b3e4$6131@cache90.multikabel.net...
Dot in the middle seems flawed to me.

It will not reach the corners and might leave big air gaps.

Which can then lead to excessive heat build up around corners and damage
capacitators around the cpu.

Bye,
Skybuck.

Wrong.

"capacitators"? *sigh*


To be fair, it's a common error for beginners. I'm sure I thought they
were "capacitators" at one point early on, and then I heard kids at school
use "capacitator", so there must be something that causes people to
misread the word. Perhaps an expectation that it should sound different.
Chances are good that until/if someone gets to electronic school, they
won't hear the word. I misinterpreted "astronomy" when I was a kid,
precsiely because I was only reading it, not hearing it, and once I got
the right pronunciation, it was still hard to shift. It likely is
compounded by others mispronouncing "capacitor" so the beginner may hear
what they already thought it was.

Michael
 
What do you do if you need to try again ?

Remove everything and a new dot ?

Or just add a little bit more ?

In last case that kinda defeat the whole purpose eh ? ;)

Bye,
Skybuck.
 
Air is a good insulator that's common knowledge.

The question however is how many micro bubbles there would be when pulling
of the heatsink ?!

Also how many would remain if smearing/whirling it a bit and then pushing
the heatsink back onto it ?!

I remain skeptical ;) :)

Bye,
Skybuck.
 
Skybuck Flying wrote:
What do you do if you need to try again ?

Remove everything and a new dot ?

Or just add a little bit more ?

In last case that kinda defeat the whole purpose eh ? ;)

Bye,
Skybuck.
1) One objective, is to remove any air bubbles.
That requires cleaning off the stuff on there already.
Wipe off enough, until a white haze is all that's left.
You don't have to clean it until it is shiny again. Just
wipe off the excess.

2) Apply your calibrated dot of paste, then squash
the cooler onto the dot, and fasten the clip.
You're done. Inspect from the side (with a mirror),
and check that the joint between CPU and cooler,
shows a bit of paste. That proves the installation worked.
If you can't see paste on the joint between CPU and cooler,
you need more paste (bigger dot of paste).

If the CPU is cool enough now that the machine is not
shutting down, you don't have to do anything.

But if you feel the computer is running hotter than other
people using the same processor, then fixing the paste
might help.

Your processor is probably good to at least 65C.
Just a guess. That's 65C Tcase.

Also, have you checked thar BIOS setting yet ?
When the cursor is placed on the shutdown item,
do the "+" or "-" keys do anything ? If you can
adjust the BIOS-controlled shutdown, again, maybe
you don't need to do anything except disable the
shutdown.

You see, that BIOS shutdown, is not the only shutdown
feature. There is still THERMTRIP (hardware protection).
So you are always protected against a damaging level of heat.
The BIOS feature, would be in addition to THERMTRIP.

Paul
 

Welcome to EDABoard.com

Sponsor

Back
Top