 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>>>> Apparently it's really hard to go any faster than that for some reason.
>>>
>>> One might suspect the speed of light.
>>
>> I'd suspect either heat or signal interference, personally... but I'm
>> not an Engineer. ;-)
>
> Could be that too, but 4GHz is only 7.5cm in a vacuum, so on silicon I'd
> expect significantly less.
How the hell do you even know that?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Those temperatures aren't that hot, I had a P3 system that ran around
similar temperatures, I had the BIOS alarm set at 85 and it went off
sometimes during the summer. IIRC CPUs can work up to 90 or 100 degrees.
> It does seem to run somewhat faster, now, though. Still waaay too hot for
> my comfort, though. Maybe I'll bolt a peltier cooler on the case fan ;)
Really it shouldn't need all that if you are not overclocking. Assuming you
bought the correct heatsink and fan for the CPU you shouldn't be anywhere
near the maximum temperature. Maybe the temperature reading is wrong for
some reason?
> Fun fact: The screws to the heatsink mount can back out and loosen over
> time. I only discovered this after I installed the new heatsink and fired
> up the computer, only to have it shut itself down in the middle of POST!
> After a few cycles of this and the occasional BIOS complaint that
> overclocking failed (huh? I've never overclocked my system...) I finally
> ripped of the new heatsink and noticed the screws were rather loose.
Did you check in the BIOS that the bus, CPU frequencies and voltages are all
set correctly?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> Could be that too, but 4GHz is only 7.5cm in a vacuum, so on silicon I'd
>> expect significantly less.
>
> How the hell do you even know that?
Google "c/4GHz"
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
scott wrote:
>
> Those temperatures aren't that hot, I had a P3 system that ran around
> similar temperatures, I had the BIOS alarm set at 85 and it went off
> sometimes during the summer. IIRC CPUs can work up to 90 or 100 degrees.
>
AFAIK that depends on the CPU and where the heat sensor is actually
located. The surface tends to be pretty heavily cooled, so the nearer
the surface the sensor is, the less it shows on the same core
temperature. These days there's also lot of stuff that can care of the
heat - at least multiple cores (not in the case of P4, but on newer
ones) and caches.
-Aero
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> AFAIK that depends on the CPU and where the heat sensor is actually
> located.
I thought all modern CPUs had the sensor actually on the CPU die, presumably
in an intelligent place to best detect when the CPU was getting too hot.
Maybe I'm wrong on that though, I know in the old days the sensor was on the
motherboard, so you'd get huge variations from PC to PC, even with the same
case and CPU.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
scott wrote:
>
> I thought all modern CPUs had the sensor actually on the CPU die,
> presumably in an intelligent place to best detect when the CPU was
> getting too hot. Maybe I'm wrong on that though, I know in the old days
> the sensor was on the motherboard, so you'd get huge variations from PC
> to PC, even with the same case and CPU.
>
Yes, at least it's on the CPU itself nowadays, I dunno the exact
location of the sensor of each chip. But especially with multiplying
cores, increasing caches etc the heat generation might be unbalanced
(eg. 1 core runs at 100%, other one idles), which makes the sensor
unbalanced (no matter where it actually is), resulting in a point that
has more heat than at the sensor, so the tolerances need to be higher. 1
core of dual-core processor ain't enough to fire up the proc when the
fan is running, but if the fan is controlled by a sensor not right
beside that particular core, there needs to be more tolerance for the
heat (eg. the core is 90°C, the sensor says 70°C and the fan checks
"ooh, 70, I'd better get running" instead of waiting for the 90).
This is all just my thinking, not scientifically proven, but I think
Intel has given a moment to think the throttling temperatures :).
-Aero
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Provided the cores are on the same die, you won't have much temperature
difference. Maybe a few degrees. Silicon is a pretty good thermal conductor.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Provided the cores are on the same die, you won't have much temperature
> difference. Maybe a few degrees. Silicon is a pretty good thermal
> conductor.
A CPU die is *really* thin though, that severely reduces the thermal
conduction between cores...
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
scott wrote:
>> Provided the cores are on the same die, you won't have much temperature
>> difference. Maybe a few degrees. Silicon is a pretty good thermal
>> conductor.
>
> A CPU die is *really* thin though, that severely reduces the thermal
> conduction between cores...
AFAIK, Core 2 Quad is a dual-die solution.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> A CPU die is *really* thin though, that severely reduces the thermal
>> conduction between cores...
>
> AFAIK, Core 2 Quad is a dual-die solution.
Even more thermal resistance between cores then.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
