Prev: Replacing DVD drive
Next: Question Bios Chip
From: Paul on 26 Jan 2010 19:09
Ian D wrote:
>> Doesn't the 920 throttle at 68C ? I looked in the datasheet (320834) and
>> the throttle temperature is listed as 68C on page 73 (temp at 130W).
>>
>
> The Tj Max for the i7 920 is 100 C. The throttling starts
> at around 92 C, so in theory, it should never reach 100 C.
> Those are junction temps, so the package temps are usually
> cooler. CoreTemp and HWMonitor only read the junction
> temps, but Everest also reads the die temps, which are
> usually about 10 deg cooler. At idle the core 0 temp is
> usually about 3 or 4 deg above the others.
>
> With the stock cooler set at the most relaxed cooling,
> I've tested the 920 in the mid 80s C without any issues.
> I normally run the fan at the most aggressive setting.
> That way I usually stay below 70 C on heavy loads.
> The maximum CPU power usage I've seen on my system
> is 152 watts.
>
> There are those who run their 920s in the 70s to low 80s
> without issue. In the long term the CPU may be able to
> handle that, but the surrounding components, especially
> capacitors, could be affected by those temps.
>

OK, I missed that. On previous generations, I'd just glance at
the chart and pick off the max temp at the max power. But on
the Core i7, it is using Tjmax rather than a Tcasemax. The
Tcasemax graph I was looking at, is artificially created
for the purposes of verification by placing a thermocouple
on the top of the processor. The Tjmax is what they're relying
on for control purposes.

I was suggesting using RMClock, or any other utility that
can monitor for PROCHOT, as a means of determining whether
any processor performance is being lost by throttling. As the
cooling solution used, should be able to prevent throttling
when a good stress application is being used.

One of the reasons I haven't been that interested in this
topic in the past, is finding threads on some of the
enthusiast sites that got into endless arguments about
temperature measurements and how meaningful they were.
And I'd just drift off to doing something else.

(Example of endless arguments, near the bottom)
http://www.xtremesystems.org/forums/showthread.php?p=4117662

One nice thing about using a well designed aftermarket
cooler, is it can use better quality fasteners than the
Intel push-pins. Some people have trouble getting the
push-pins working properly, even when the bundled heatsink/fan
is brand new. The Noctua uses screws, and as long as you
don't strip them (which is a problem with aftermarket coolers),
may work out better in the long run. I have a cooler with
screws on my LGA775 system, and prefer that to having to deal
with plastic.

Paul

From: Jim Cochrane on 29 Jan 2010 17:40
On 2010-01-27, Paul <nospam(a)needed.com> wrote:
> Ian D wrote:
>>> Doesn't the 920 throttle at 68C ? I looked in the datasheet (320834) and
>>> the throttle temperature is listed as 68C on page 73 (temp at 130W).
>>>
>>
>> The Tj Max for the i7 920 is 100 C. The throttling starts
>> at around 92 C, so in theory, it should never reach 100 C.
>> ...

Thanks to everyone who posted here for your good info and pointers.

I decided to buy a cooler to replace the stock cooler, even though the
stock cooler would probably have sufficed for non-overclocking use -
because I suspect I will want to overclock at some point, and (since it
appears from reading some reviews that coolers can be hard to install) I
think it's better that I have the shop building the computer for me
install the cooler rather than hardware-clumsy me do it later.

I decided to go with the Noctua NH-U12P SE2 120mm SSO CPU Cooler, which
I believe Paul recommended. According to the reviews I read, it not
only does a good job of cooling, but is also very quiet.


Jim

--- news://freenews.netfront.net/ - complaints: news(a)netfront.net ---
First  |  Prev  | 
Pages: 1 2
Prev: Replacing DVD drive
Next: Question Bios Chip