p4 throttling

[hydrata]

Ok, I just read through malakai's post on the gripes and moans section, so I hope you will please consider me as a noob that's trying to learn and not just going for the quick overclock...after all, i've been reading this forum everyday for a month before i finally buy :) anyway, on to the point....i understand that the term "throttling" means that the processor clocks itself down....does this happen on desktop p4's, what factors cause it? how can you tell if it happens? just general information on "throttling" would be appreciated, thanks. See malakai, i'm trying to learn and become as knowledgeable as you and others :)

[Yodums]

Throttling doesn't mean going down completely it means the speed is fluctuating, constantly moving. Yes this happens on the desktop CPU. We don't talk much about mobiles/labtops here since they're so limited. Heat causes it. There's an on-die sensor which controls the throttling. When the on-die sensor hit beyond 55 degrees it will throttle back and fourth to get a certain speed to match 55 degrees or so. When your benchies are extremely low at a high speed there's a sign. It's most likely when your chip is at 50 degrees it's throttling due to the fact the insocket thermistors (The diode which is placed under the chip so the end-user can monitor) isn't corect.

[imgod2u]

Are you sure it's 55C? I would imagine it'd be higher. As the technical max operational temperature of copper/silicon is about 60C. Either way, in most cases, throttling won't be a problem. As P4 Northwoods rarely get up to that temperature. I've done some overclocks on Northwoods (alas, not for my own rig) and the full load temp has never gone beyond 52C.

[Yodums]

Originally posted by imgod2u
Are you sure it's 55C? I would imagine it'd be higher. As the technical max operational temperature of copper/silicon is about 60C. Either way, in most cases, throttling won't be a problem. As P4 Northwoods rarely get up to that temperature. I've done some overclocks on Northwoods (alas, not for my own rig) and the full load temp has never gone beyond 52C.

Pretty sure it's 55. I'd have to disagree about the part that it doesn't happy often, it happens very often. The Northwood put out alot of heat. Considering they put out as much watts as an Athlon XP at the same speed, however the P4 is a 0.13 micron processor and it putting out the same amount of heat when the XP is a 0.18 micron processor means it has more heat in one area then a XP would thus being harder to cool. That's why you see them so hot.


Most Northy's are just over 45. As I mentioned, do not forget the in-socket thermistors aren't correct. The sensor controlling the throttling is an on-die sensor located in the CPU Core which takes the temperatures at the hottest place in the core.

Yodums

[Colin]

You may find this (http://www.intel.com/technology/itj/q12001/articles/art_4.htm) interesting. The diode that controls clock throttling is in a much hotter portion of the chip than the diode your motherboard reads the temps from. One has to wonder why? Marketing perhaps?

[imgod2u]

As I recall from an article over at aces, power consumption of a 2.2 GHz Northwood was somewhere around 60W in contrast to an Athlon 1.67 at near 70W. However, power consumption and heat generation really isn't directly affecting the way the throttling controls work.
As for whether the thing throttles back or not. As there is no real difference between a 2.4 GHz stock P4 and a 2.4 GHz overclocked P4 asside from microvariances in the trace pathes which may cause a critical path. This really doesn't have much effect on heat generation and, provided the same cooling, dissipation. So if a stock 2.4 GHz P4 can work fine on stock cooling than an overclocked 1.8a at 2.4 will work fine. If it is throttling back, then all the benchmarks you see of the stock 2.4 would be the result you get with the overclocked 2.4. And from what I understand, the benchmarks are pretty impressive.