Overclocking & cpuz info

[anotherpaul]

The "old PCs" thread got me into trying overclocking on my 2007
Gigabyte GA-P35-DSR3 rev 2 with the Intel E6550 & 8 gigs of
Corsair ram.

Looked up "howtos" & did it, sortof(?). The cpu freq is at
400 mhz & the multiplier at 7x; had the cpu vcore set to 1.400.

Now the bios shows cpu speed of 2.8 ghz. BUT cpuz & the
aida64 shows 2.8 ghz off/on as mostly it is at the 2.4 ghz!
The cpu voltage seem to stay at around the 1.36v & 1.376v
& never get to my setting of 1.400v. Oh, the C.I.A.2 is in
the default "disabled" & the performance enhance set to
standard as dictated.

Cpuz has the title at the bottom of "Clocks (Core #0), which doesn't
change. So is it displaying alternate core info; if so, any idea on
having both cores running at the same speed?

 

[Flasherly]

The "old PCs" thread got me into trying overclocking on my 2007
Gigabyte GA-P35-DSR3 rev 2 with the Intel E6550 & 8 gigs of
Corsair ram.

Looked up "howtos" & did it, sortof(?). The cpu freq is at
400 mhz & the multiplier at 7x; had the cpu vcore set to 1.400.

Now the bios shows cpu speed of 2.8 ghz. BUT cpuz & the
aida64 shows 2.8 ghz off/on as mostly it is at the 2.4 ghz!
The cpu voltage seem to stay at around the 1.36v & 1.376v
& never get to my setting of 1.400v. Oh, the C.I.A.2 is in
the default "disabled" & the performance enhance set to
standard as dictated.

Cpuz has the title at the bottom of "Clocks (Core #0), which doesn't
change. So is it displaying alternate core info; if so, any idea on
having both cores running at the same speed?

Got a neat-o, nifty little tool that brings my 3.2 P4 to its knees.
StereoTool (it's a plug-in for AIMP2, Russian sound player).
According to the many compression, bass and loudness sound-processing
settings, and how they're set, severe artifacts due to processor
limitations can be introduced. Way I initially "tested it out" was to
find all the settings needed, turning off some, and exactly how much
needed if an adjustable setting for processing increased load. I then
fed a 4-channel output into two amps true to 100watt and cranked them
to their max (have a power wattage output meter connected in parallel
with the speaker wires). I tend use percussion at stage volumes as a
final determinate.

Had I two cores, I'd disable one while running the program on each
consecutively;- then both simultaneously for further attempting to
raise StereoTool's processing levels past what either prior single
core was, on average, capable;-- Discounting overclocking as a concept
now includes unlocking possible hidden cores for some level of gained
processing to some level of lower standards than an amount of revealed
processors.

Another thing. Looking at some newer CPU reviews I recall someone
saying with multicores benchmarks utilities, final results and
indications were overall less indicative of an empirical sense once
provided the single core, that going from results obtained by a "seat
of the pants" feeling given the individual by pushing familiar
applications to their limits may be more pertinent to spreading, over
cores, applications and threads for pertinence to a desirably
optimized outcome.

The last would be Joe Blowhard's opinion. Personally, I'm just an
observer on this world largely contained in programs limited to single-
core programming. No multicores in any of my boxes. These days, I
don't encode video and haven't any games, either, for giving frame
rates. Just a snap to drums going balls to the walls.

 

[anotherpaul]

did you try underclocking the cpu to see if the speeds match ?

Thanks. I had thought I had the Intel stepdown disabled, but it turned
out to be elsewhere.

The variance of the speed of the cpu was due to the Intel stepdown
control; win7 does report both the rated speed & the overclocked speed.
The speeds do show the original & overclock alternately with cpuz.

Now overclocked at 3.00 mhz on the 2.33 mhz rated speed; left the
system voltages as "auto". Running very nicely in both win7 & linux.

 

[Paul]

Thanks. I had thought I had the Intel stepdown disabled, but it turned
out to be elsewhere.

The variance of the speed of the cpu was due to the Intel stepdown
control; win7 does report both the rated speed & the overclocked speed.
The speeds do show the original & overclock alternately with cpuz.

Now overclocked at 3.00 mhz on the 2.33 mhz rated speed; left the
system voltages as "auto". Running very nicely in both win7 & linux.

Have you load tested yet ? Check with Prime95, as an example.
If an overclock is well adjusted, Prime95 won't throw errors.

http://mersenne.org/freesoft/

When you run that, it'll ask "Join GIMPS ?". You answer "No, just testing"
to carry out tests. The modern versions of Prime95 run a thread per
core, to give a good load test.

The Prime95 test does math with a known answer. If the test discovers
a "rounding error", that might be interpreted to mean the CPU isn't
working soundly, or the memory doesn't like the speed you're running it at.

IF the computer crashes completely, a second after the test starts,
then you weren't even near stability. Lots of work to do in that case.

As a final acceptance test, I like to run Prime95 at the same time as
I'm running a game with 3D content. That provides a nice mix of stressors,
to uncover issues with the setup. Once the load testing is complete, you
can use the computer with confidence, that less stressful operations will
work correctly.

Paul

 

[anotherpaul]

Have you load tested yet ? Check with Prime95, as an example.
If an overclock is well adjusted, Prime95 won't throw errors.

http://mersenne.org/freesoft/

When you run that, it'll ask "Join GIMPS ?". You answer "No, just testing"
to carry out tests. The modern versions of Prime95 run a thread per
core, to give a good load test.

The Prime95 test does math with a known answer. If the test discovers
a "rounding error", that might be interpreted to mean the CPU isn't
working soundly, or the memory doesn't like the speed you're running it at.

IF the computer crashes completely, a second after the test starts,
then you weren't even near stability. Lots of work to do in that case.

As a final acceptance test, I like to run Prime95 at the same time as
I'm running a game with 3D content. That provides a nice mix of stressors,
to uncover issues with the setup. Once the load testing is complete, you
can use the computer with confidence, that less stressful operations will
work correctly.

Paul

Got the Prime95 & ran it for about 1 hr 18 min; no warnings & no errors
while doing some web surfing. Probably could do better but checking
the temp with hwmonitor the core temps got up around 68 C with the
Zalman fan running around 2000 rpm. I do notice some people run
the tests for hours.

The core temps quickly lowered to around 44 C & is now bouncing between
55 C & 43 C !!??!!

I believe that it's stable enough for I use the computer for these days
but will check again when I ever get to ripping some videos to the
hard drive after I finish doing the CDs. The only games I play are
the pinball games; too much work for the other types.

 

[anotherpaul]

Got the Prime95 & ran it for about 1 hr 18 min; no warnings & no errors
while doing some web surfing. Probably could do better but checking
the temp with hwmonitor the core temps got up around 68 C with the
Zalman fan running around 2000 rpm. I do notice some people run
the tests for hours.

The core temps quickly lowered to around 44 C & is now bouncing between
55 C & 43 C !!??!!

I believe that it's stable enough for I use the computer for these days
but will check again when I ever get to ripping some videos to the
hard drive after I finish doing the CDs. The only games I play are
the pinball games; too much work for the other types.

Would like info on people's cpu temp reporting. HWmonitor shows 2
types of temp; 1) cpu, 2) core temp #0 & #1. The cpu temp is usually
lower by 5 deg than the core #1 temp & could be 10 deg lower than
core temp #0.

So, is it usually the "cpu" temp that is stated or one of the "core"
temps. Puzzled as the Intel spec sheet says max is 72 deg C but
which temp is the max referring to?

 

[SC Tom]

Would like info on people's cpu temp reporting. HWmonitor shows 2
types of temp; 1) cpu, 2) core temp #0 & #1. The cpu temp is usually
lower by 5 deg than the core #1 temp & could be 10 deg lower than
core temp #0.

So, is it usually the "cpu" temp that is stated or one of the "core"
temps. Puzzled as the Intel spec sheet says max is 72 deg C but
which temp is the max referring to?

According to this link to your CPU, the Tcase is 72C, which is the casing of the CPU and not the core(s):

http://ark.intel.com/products/30783/Intel-Core2-Duo-Processor-E6550-(4M-Cache-2_33-GHz-1333-MHz-FSB)


On my T8300, the max temp is listed as Tjunction, which I believe is the core temp(s):

http://ark.intel.com/products/33099/Intel-Core2-Duo-Processor-T8300-(3M-Cache-2_40-GHz-800-MHz-FSB)

I use a program called Core Temp since it shows me how far I have to go before the CPU reaches that a TJ Max of 105C.
HWMonitor shows my CPU as 35-40F (yep, Fahrenheit) at idle. There is no way it's running that cool in a laptop with
factory cooling

There's also a program called Real Temp http://www.techpowerup.com/realtemp/docs.php that I've been playing with
since I'm not too sure of the Core Temp output either. After going through the document in the link, near as I can tell
by using a digital thermometer in the laptop's vicinity, my readings are 5-10C off. I don't worry about it too much
though since I've never had my laptop shut down from overheating (and hopefully I never will).

 

[anotherpaul]

According to this link to your CPU, the Tcase is 72C, which is the casing of the CPU and not the core(s):

http://ark.intel.com/products/30783/Intel-Core2-Duo-Processor-E6550-(4M-Cache-2_33-GHz-1333-MHz-FSB)


On my T8300, the max temp is listed as Tjunction, which I believe is the core temp(s):

http://ark.intel.com/products/33099/Intel-Core2-Duo-Processor-T8300-(3M-Cache-2_40-GHz-800-MHz-FSB)

I use a program called Core Temp since it shows me how far I have to go before the CPU reaches that a TJ Max of 105C.
HWMonitor shows my CPU as 35-40F (yep, Fahrenheit) at idle. There is no way it's running that cool in a laptop with
factory cooling

There's also a program called Real Temp http://www.techpowerup.com/realtemp/docs.php that I've been playing with
since I'm not too sure of the Core Temp output either. After going through the document in the link, near as I can tell
by using a digital thermometer in the laptop's vicinity, my readings are 5-10C off. I don't worry about it too much
though since I've never had my laptop shut down from overheating (and hopefully I never will).

Thanks. Will also checkout Real Temp. There is a difference in the
core temps as reported by hwmonitor & core temp 1.0. Core Temp
does show the e6550 has a Tj max of 95C; hope that is true.

There are advices that one should not go over 60C for the e6550;
no idea where that is coming from or how that number relate to the
Tcase of 72C.