I'm just OCing and stressing a new build for a mate. Asus P5K-Deluxe WiFi-AP
with a G0 Q6600 topped with a Thermaltake Big Typhoon. The case isn't the
best in the world for ventilation (Gigabyte 3D Aurora) but it's also not the
worst. 1 x 120mm in the front, 2 x 120mm in the back as well as the PSU and
8800GTX exhausting out the back.
I have it clocked at 3.2GHz at the moment, (400 x 8) stressing with Prime95
ver. 25.5. It's been running for over an hour now with CPU-Z giving a vcore
under load of 1.280V. (These P5K BIOS settings are all over the place, you
have to set it, then check in Windows to see how far out it is. It's et to
1.375 in BIOS.)
The hottest it's got, in a 20°C room, according to CoreTemp is 68°C. What do
you guys (and gals?) think of that? Should I give it to him like that? He's
a gamer but I don't think he has any games that would put the load on the
cores that Prime does. I had the vcore in BIOS one setting lower and Prime
threw up an error. Seems good as it is though.
He wants it overclocked, mainly for bragging rights I suspect. He has gamer
friends who know jack about OCing. He has a tame OCer in me, and gives me
his cast-offs when he upgrades for building his new machines. I got a
7800GT, a 120GB IDE HDD, and IDE LG DVD writer, a Samsung CD writer and his
old case (average) for doing this build. It's been a PITA too, he got a
bolt-thru kit but not until after I'd already built it. I had to remove the
mobo to fit it, then do the screws up with pliers (15° turns at a time,
heatpipes everywhere)as the cooler is too big for me to get a screwdriver
near the screws. The 8800GTX is a *very* tight fit, extremely tricky to put
in and take out.
Sorry, I digress. Temps? Ok or not? Also, why does CoreTemp give a Tjunction
of 100°C for the Q6600 and "only" 85°C for my E4500? Does that mean it can
run hotter?
~misfit~ wrote:
>
> The hottest it's got, in a 20°C room, according to CoreTemp is 68°C.
> What do you guys (and gals?) think of that? Should I give it to him
> like that? He's a gamer but I don't think he has any games that would
> put the load on the cores that Prime does. I had the vcore in BIOS
> one setting lower and Prime threw up an error. Seems good as it is
> though.
>
> Sorry, I digress. Temps? Ok or not? Also, why does CoreTemp give a
> Tjunction of 100°C for the Q6600 and "only" 85°C for my E4500? Does
> that mean it can run hotter?
>
What you are seeing is the maximum allowed value of Tjunction, not the
actual temperature value. When Tjunction hits approx. 20°C higher than
Tcasemax the processor will throttle.
"In the event of a catastrophic cooling failure, the processor will
automatically shut down when the silicon has reached a
temperature approximately 20 °C above the maximum TC. Assertion
of THERMTRIP# (Thermal Trip) indicates the processor junction
temperature has reached a level beyond where permanent silicon
damage may occur. Upon assertion of THERMTRIP#, the processor
will shut off its internal clocks (thus, halting program execution) in
an attempt to reduce the processor junction temperature."
~misfit~ wrote:
> I have it clocked at 3.2GHz at the moment, (400 x 8) stressing with Prime95
> ver. 25.5. It's been running for over an hour now with CPU-Z giving a vcore under load of 1.280V. (These P5K BIOS
> settings are all over the place, you have to set it, then check in Windows to see how far out it is. It's et to 1.375
> in BIOS.)
>
> The hottest it's got, in a 20°C room, according to CoreTemp is 68°C. What do you guys (and gals?) think of that?
> Should I give it to him like that? He's a gamer but I don't think he has any games that would put the load on the
> cores that Prime does. I had the vcore in BIOS one setting lower and Prime threw up an error. Seems good as it is
> though.
FWIW, those are the temperatures I saw at 3.2 GHz with identical vCore.
'~misfit~' wrote, in part:
| The hottest it's got, in a 20°C room, according to CoreTemp is 68°C. What
do
| you guys (and gals?) think of that? Should I give it to him like that?
He's
| a gamer but I don't think he has any games that would put the load on the
| cores that Prime does. I had the vcore in BIOS one setting lower and Prime
| threw up an error. Seems good as it is though.
_____
For what it's worth, my experience with my E4300 is that a TAT temperature
of about 72 C is the limit for overclocking this individual E4300. When the
room ambient temperature is low enough, I can increase the CPU speed and
have it remain stable as long as the TAT temperature is below about 72 C;
at that point nothing more is possible. With a room ambient ~ 22 C the TAT
temperature and 100% load X 2 reaches ~ 72 C at about 3.2 GHz. As winter
comes on I'll be able to try with room ambient temperatures down to 10 C or
lower.
From my experience (and from reviewing posted anecdotal information) ANY
temperature is 'safe' for an Intel CPU. The worst that can happen is that
the system locks up, the CPU cools down, and you can just reboot with no
harm done.
Since you are building this system for a friend, you have to consider the
same problem that Dell or any system builder has; you don't have any
control on the conditions of use. A lower CPU core temperature means the
end user can let more dust bunnies collect on the heatsink fins, place the
system case in the sun, etc. It'd be easy enough to just tell your friend
that and have him decide if more performance is worth taking a little more
care in the conditions of use.
Please continue to post information about your quad core experience. I
figure that the 'Penryn' prices will be reasonable about the same time as my
last kid is out of university and I will be able to afford a new CPU B^)
As you can see from my 'Centech' post, a not-contact infrared digital
thermometer can be real useful for checking temperatures in various parts of
the system case. Your 3 X 120 mm fans seem more than adequate for cooling,
even in NZ. I've got two 120 mm fans blowing in (one over the two hard
drives and one over the CPU/memory/GPU slot 1 area) and two rear panel 80 mm
fans to blowing out (plus the PS fan blowing out.) With the ThermalTake 7i
heatsink fin orientation, the front panel 120 mm fan gives a straight
through air flow to cool the CPU while the side 120 mm fan adds air flow for
the GPU slots and chipset heatsink. I removed the 'faceplate' around the
rear panel connectors to increase the air flow across the heatsink for the
DC to DC voltage downconvertors-regulator. The case runs at a slight
positive pressure, so a filter in front of each 120 mm fan takes care of
dust. I'm still using the two Enlight 8950 server cases I bought about
seven or so years ago for my personal system builds - nothing like a big
roomy case for overclocking.
Phil Weldon
Phil Weldon
"~misfit~" <misfit61nz@yahoot.com.au> wrote in message
news:4757f77e$1@news2.actrix.gen.nz...
| Hey,
|
| I'm just OCing and stressing a new build for a mate. Asus P5K-Deluxe
WiFi-AP
| with a G0 Q6600 topped with a Thermaltake Big Typhoon. The case isn't the
| best in the world for ventilation (Gigabyte 3D Aurora) but it's also not
the
| worst. 1 x 120mm in the front, 2 x 120mm in the back as well as the PSU
and
| 8800GTX exhausting out the back.
|
| I have it clocked at 3.2GHz at the moment, (400 x 8) stressing with
Prime95
| ver. 25.5. It's been running for over an hour now with CPU-Z giving a
vcore
| under load of 1.280V. (These P5K BIOS settings are all over the place, you
| have to set it, then check in Windows to see how far out it is. It's et to
| 1.375 in BIOS.)
|
| The hottest it's got, in a 20°C room, according to CoreTemp is 68°C. What
do
| you guys (and gals?) think of that? Should I give it to him like that?
He's
| a gamer but I don't think he has any games that would put the load on the
| cores that Prime does. I had the vcore in BIOS one setting lower and Prime
| threw up an error. Seems good as it is though.
|
| He wants it overclocked, mainly for bragging rights I suspect. He has
gamer
| friends who know jack about OCing. He has a tame OCer in me, and gives me
| his cast-offs when he upgrades for building his new machines. I got a
| 7800GT, a 120GB IDE HDD, and IDE LG DVD writer, a Samsung CD writer and
his
| old case (average) for doing this build. It's been a PITA too, he got a
| bolt-thru kit but not until after I'd already built it. I had to remove
the
| mobo to fit it, then do the screws up with pliers (15° turns at a time,
| heatpipes everywhere)as the cooler is too big for me to get a screwdriver
| near the screws. The 8800GTX is a *very* tight fit, extremely tricky to
put
| in and take out.
|
| Sorry, I digress. Temps? Ok or not? Also, why does CoreTemp give a
Tjunction
| of 100°C for the Q6600 and "only" 85°C for my E4500? Does that mean it can
| run hotter?
|
| TIA,
| --
| Shaun.
|
|
On Thu, 06 Dec 2007 17:18:22 GMT, Fishface <invalid@ddress.ok?> wrote:
: ~misfit~ wrote:
: > I have it clocked at 3.2GHz at the moment, (400 x 8) stressing with Prime95
: > ver. 25.5. It's been running for over an hour now with CPU-Z giving a vcore under load of 1.280V. (These P5K BIOS
: > settings are all over the place, you have to set it, then check in Windows to see how far out it is. It's et to 1.375
: > in BIOS.)
: >
: > The hottest it's got, in a 20°C room, according to CoreTemp is 68°C. What do you guys (and gals?) think of that?
: > Should I give it to him like that? He's a gamer but I don't think he has any games that would put the load on the
: > cores that Prime does. I had the vcore in BIOS one setting lower and Prime threw up an error. Seems good as it is
: > though.
:
: FWIW, those are the temperatures I saw at 3.2 GHz with identical vCore.
:
:
I'd be thrilled with those temps. I was up around 75 at that clock
but I needed more vcore for stability (39C inside the case)
~misfit~ wrote:
> Hey,
>
> I'm just OCing and stressing a new build for a mate. Asus P5K-Deluxe WiFi-AP
> with a G0 Q6600 topped with a Thermaltake Big Typhoon. The case isn't the
> best in the world for ventilation (Gigabyte 3D Aurora) but it's also not the
> worst. 1 x 120mm in the front, 2 x 120mm in the back as well as the PSU and
> 8800GTX exhausting out the back.
>
> I have it clocked at 3.2GHz at the moment, (400 x 8) stressing with Prime95
> ver. 25.5. It's been running for over an hour now with CPU-Z giving a vcore
> under load of 1.280V. (These P5K BIOS settings are all over the place, you
> have to set it, then check in Windows to see how far out it is. It's et to
> 1.375 in BIOS.)
>
> The hottest it's got, in a 20°C room, according to CoreTemp is 68°C. What do
> you guys (and gals?) think of that? Should I give it to him like that? He's
> a gamer but I don't think he has any games that would put the load on the
> cores that Prime does. I had the vcore in BIOS one setting lower and Prime
> threw up an error. Seems good as it is though.
>
> He wants it overclocked, mainly for bragging rights I suspect. He has gamer
> friends who know jack about OCing. He has a tame OCer in me, and gives me
> his cast-offs when he upgrades for building his new machines. I got a
> 7800GT, a 120GB IDE HDD, and IDE LG DVD writer, a Samsung CD writer and his
> old case (average) for doing this build. It's been a PITA too, he got a
> bolt-thru kit but not until after I'd already built it. I had to remove the
> mobo to fit it, then do the screws up with pliers (15° turns at a time,
> heatpipes everywhere)as the cooler is too big for me to get a screwdriver
> near the screws. The 8800GTX is a *very* tight fit, extremely tricky to put
> in and take out.
>
> Sorry, I digress. Temps? Ok or not? Also, why does CoreTemp give a Tjunction
> of 100°C for the Q6600 and "only" 85°C for my E4500? Does that mean it can
> run hotter?
>
> TIA,
A tool like "RMClock" can display whether throttling is occurring. I don't know
if RMClock is quad compatible or not.
The digital temperature measurement inside the processor, is measured with
respect to Tcase_max. That means, as the temperature gets up to Tcase_max,
the register containing the temperature is just about reading zero. To get
the absolute temperature, versus the relative temperature register, any measurement
tool must know what the Tcase_max is for the product.
If you had access to that "register value", at the instant the reading hits
zero, you should see throttling start to occur. From an end user point of
view, that is the thing you want to avoid, from a cooling design standpoint.
So with your quad core Prime95 test running, you want enough cooling in place,
so that throttling is not seen in RMClock, or the raw measurement from
a tool that can access the HECI temperature measurement, is not hitting the
zero value.
The Big Typhoon is optimized for silent cooling, as opposed to the absolute
best cooling. You can change the fan on it, to do better, at the expense of
extra noise. Or switch to something like a Tuniq Tower.
The Zalman CNPS9700 is listed here as having thermal resistance of 0.12 to
0.16 °C/W, which is better. To compute temperature rise, theta_R * power_in_watts
gives temp rise in degrees C. The Q6600 is 95 watts, but you have to scale that
by the degree of overclock, and also scale by the ratio of voltages squared.
(If stock was 1.28V and you actually applied 1.375, then 1.375/1.28 squared is
a factor of 1.154x more.) 95W * (3.2/2.4) * 1.154 = 146W. And 146W times the
0.12 of a CNPS9700 would give a temp rise of 17.5C above case air temp.
There are plenty of choices out there, and always something new.
To aid a good CPU cooler, a high volume exhaust fan near the exit also
helps. A performance CPU cooler can't do the job, if there is a cloud
of hot air around the cooler. That is one of the reasons, I use one of
these :-) I power this from a Molex connector, not a motherboard header.
I also have a home made voltage reducer, as this is too noisy at a
full 12V. It also helps to use a few extra holes on the front of the
case (like remove some drive bay covers), to get enough air to feed
that thing. I only bought this fan locally, as an impulse purchase,
when I saw the fan had a metal frame rather than the usual plastic frame.
Somewhere on teh interweb Fishface typed:
> ~misfit~ wrote:
>> I have it clocked at 3.2GHz at the moment, (400 x 8) stressing with
>> Prime95 ver. 25.5. It's been running for over an hour now with CPU-Z
>> giving
>> a vcore under load of 1.280V. (These P5K BIOS settings are all over
>> the place, you have to set it, then check in Windows to see how far
>> out it is. It's et to 1.375 in BIOS.) The hottest it's got, in a 20°C
>> room, according to CoreTemp is 68°C.
>> What do you guys (and gals?) think of that? Should I give it to him
>> like that? He's a gamer but I don't think he has any games that
>> would put the load on the cores that Prime does. I had the vcore in
>> BIOS one setting lower and Prime threw up an error. Seems good as it
>> is though.
>
> FWIW, those are the temperatures I saw at 3.2 GHz with identical
> vCore.
Somewhere on teh interweb Howard Goldstein typed:
> On Thu, 06 Dec 2007 17:18:22 GMT, Fishface <invalid@ddress.ok?> wrote:
>> ~misfit~ wrote:
>>> I have it clocked at 3.2GHz at the moment, (400 x 8) stressing with
>>> Prime95
>>> ver. 25.5. It's been running for over an hour now with CPU-Z giving
>>> a vcore under load of 1.280V. (These P5K BIOS settings are all over
>>> the place, you have to set it, then check in Windows to see how far
>>> out it is. It's et to 1.375 in BIOS.)
>>>
>>> The hottest it's got, in a 20°C room, according to CoreTemp is
>>> 68°C. What do you guys (and gals?) think of that? Should I give it
>>> to him like that? He's a gamer but I don't think he has any games
>>> that would put the load on the cores that Prime does. I had the
>>> vcore in BIOS one setting lower and Prime threw up an error. Seems
>>> good as it is though.
>>
>> FWIW, those are the temperatures I saw at 3.2 GHz with identical
>> vCore.
>
> I'd be thrilled with those temps. I was up around 75 at that clock
> but I needed more vcore for stability (39C inside the case)
His case temp is around the same, (although I don't know where the
thermistor is that reads it). I've advised him to get a 120mm cut-out above
the Big Typhoon 120mm fan and fit a case fan to feed the cooler with ambient
temp air. That shoud decrease core temps maybe 10°C. Whether he does it or
not remains to be seen.
--
Cheers,
Somewhere on teh interweb Brett Kline typed:
> ~misfit~ wrote:
>>
>> The hottest it's got, in a 20°C room, according to CoreTemp is 68°C.
>> What do you guys (and gals?) think of that? Should I give it to him
>> like that? He's a gamer but I don't think he has any games that would
>> put the load on the cores that Prime does. I had the vcore in BIOS
>> one setting lower and Prime threw up an error. Seems good as it is
>> though.
>>
>> Sorry, I digress. Temps? Ok or not? Also, why does CoreTemp give a
>> Tjunction of 100°C for the Q6600 and "only" 85°C for my E4500? Does
>> that mean it can run hotter?
>>
>
>
> What you are seeing is the maximum allowed value of Tjunction, not the
> actual temperature value. When Tjunction hits approx. 20°C higher than
> Tcasemax the processor will throttle.
>
>
> ftp://download.intel.com/design/proc...s/31559205.pdf
>
> "In the event of a catastrophic cooling failure, the processor will
> automatically shut down when the silicon has reached a
> temperature approximately 20 °C above the maximum TC. Assertion
> of THERMTRIP# (Thermal Trip) indicates the processor junction
> temperature has reached a level beyond where permanent silicon
> damage may occur. Upon assertion of THERMTRIP#, the processor
> will shut off its internal clocks (thus, halting program execution) in
> an attempt to reduce the processor junction temperature."
>
> CPUID = 06F7h Tcasemax= 71°C
> CPUID = 06FBh Tcasemax= 62.2°C
>
> http://softwarecommunity.intel.com/I...ad/980465.aspx
> http://softwarecommunity.intel.com/i.../30228130.aspx
>
>
>
> I do not know of any current software which reads a variable Tjunction
> value.
<two hours later>
Thanks for all the links, my mind just boggled.
I fully realised that Tjunction does not change, that it's constant for any
given processor. My question, which I was still unable to answer after
reading those links (and links) was related to the difference between the
Tjunction of the Q6600 (100°C) and the E4500 (85°C). Does this mean that the
Q6600 can run hotter than the E4500?
I see CoreTemp has the ability to change from displaying current temp to
"Delta to Tjunction". The displayed output then says (in my current case)
"39C° (sic) to Tjunction remaining". This leads me to believe that Tjunction
= maximum temp for that CPU. Hence the question: Why is the Tjunction on the
Q6600 15° higher than the E4500? Does that mean that the Q6600 is designed
to handle higher temperature? I couldn't find the answer to that in the
links you provided. I know, I can Google it, probably in less time that it
took reading those links (and links). However I hoped someone here knew the
(simple) answer to that question off the top of their head.
--
Thanks,
~misfit~ wrote:
>
> I fully realised that Tjunction does not change, that it's constant
> for any given processor. My question, which I was still unable to
> answer after reading those links (and links) was related to the
> difference between the Tjunction of the Q6600 (100°C) and the E4500
> (85°C). Does this mean that the Q6600 can run hotter than the E4500?
The Q6600 G0 and the E4500 M0 can be ran to 73°C.
The Q6600 B3 stepping can be ran to 63°C.
Versions of Core Temp prior to .95.4 read Tjunction of the Q6600 as 85°C.
Version 95.4 reads it as 100°C.
The simple answer is don't concern yourself with Tjunction at all. Keep
Tcase as low as you can; stress test, and ship it out. Any C2D will throttle
or simply halt before damage is done.
Q6600 "safe" temps? 
Linear Mode
