"1.7GHz. It's there. Unless you need it."

ledzepp98

Distinguished
Dec 31, 2007
223
0
18,680
cool, great link and a recommended read for all...but, for those who are lazy, here is the section of the article that BHC's great post is referring to:

Pentium 4 Power Management – A Performance Limiter

Last year, as details of the Pentium 4 were emerging, Intel's elaborate thermal and power regulation requirements for P4 systems raised many eyebrows. Platforms required new power supplies capable of pumping out more current, and enormous copper heat sinks for CPU cooling. Humorous jabs ensued… "Turns your PC into a toaster oven," some said.

But just prior to the P4 launch, Intel sought to confront the issue by publicizing power specifications for the soon to be released CPU. To many people's amazement, the 1.5GHz Pentium 4 was said to consume only 54.7 Watts - a maximum power dissipation rate well below the fastest Athlon.

Numerous independent product reviews have quoted this value (and still do) particularly in contrast to AMD's latest 73 Watt, 1.33 GHz Athlon. The only problem is that the quoted Pentium 4 power dissipation figure is wrong - or at least the number is misleading.

These two specifications are defined under entirely different conditions. AMD reports the true absolute maximum power dissipation rates without constraint. Intel on the other hand, publishes a compromised figure that is open to lots of interpretation. Page 70 of Intel’s P4 datasheet shows the dissimilarity.

Since most applications are ‘unlikely’ to cause the processor to consume its absolute maximum power, Intel quotes a figure that looks more like a ‘not-to-be-exceeded’ figure. In theory this power level could easily be exceeded under stress tests, benchmarks and particularly challenging application loads. If this were to happen, a thermal diode (required on all P4 platforms) would trigger a power management mechanism that instantly cuts CPU performance, and allows it to cool down.

The P4 spec reads, “The Thermal Monitor feature … is intended to protect the processor from overheating when running high power code that exceeds the recommendations in this table,” (54.7 Watts for the 1.5GHz model).

Looking forward, if you buy a 1.7GHz P4, it will run at that speed when it is idle, or under light loads, when CPU utilization is nominal, or in applications that don’t really need a 1.7GHz CPU. But when you drive it to extremes, or wish to extract all available performance from the processor, you may find yourself spontaneously and unavoidably power managed to a lower effective clock speed. Intel’s motto… “1.7GHz. Its there. Unless you need it.”

Update: Intel’s Thermal Design Guide has revealed that the absolute maximum power dissipation of the 1.5GHz P4 is actually 72.9 watts. This is 33% higher than the published system design specification, and essentially identical to the 1.33 GHz Athlon. In order to prevent the CPU from exceeding 54.7 watt, thermal throttling is used. If performance critical applications drive CPU power above its artificially low 54.7 watt limit, the CPU is halted with a 50% duty cycle (alternating 2 microseconds on; 2 microseconds off) until it cools down. This effectively turns your 1.5GHz processor into a 750MHz processor – just at the moment you demand peak performance. On the other hand, you will probably still be able to check your email at 1.5GHz. This scheme is described on page 23 of Intel’s P4 Thermal Design Guide. Commentary is already floating around the web that perhaps Intel feels guilty about selling 750MHz CPUs in 1.5GHz clothing, and thus has decided to cut the price by 50% as well.
 

Raystonn

Distinguished
Apr 12, 2001
2,273
0
19,780
That article is full of misnomers. The Thermal Monitor feature can be used in a number of different ways. Its use is controlled by the motherboard manufacturer and can be added as a BIOS setting.

For most cases, the thermal control circuit merely provides an added level of safety against unsafe temperature levels. This means it will only activate if the CPU reaches an unsafe temperature, not a certain wattage level. If you provide adequate cooling, you can run at maximum wattage forever at full speed ahead and never trip the protection mechanism. It's based on temperature, not wattage. The writer of that article needs to improve his reading comprehension skills.

Personally, I think it's a great idea. It's much better than having your CPU fry the way Athlons do. It's better than the best of the old methods, which simply shut down your CPU in event of temperature overload. At least this lets you continue to operate and save your work while informing you of the overheating situation.

To see the technical details yourself, read here: ftp://download.intel.com/design/Pentium4/guides/24920301.pdf

-Raystonn

-- The center of your digital world --
 

bhc

Distinguished
Apr 14, 2001
142
0
18,680
Raystonn,

Thanks for the providing more info. I read the section on thermal monitoring. However, I found it is curious that the trip-off temperature is never specified other than it is pre-calibrated in the factory. Since Intel actually specs wattage (e.g. 54.7W for 1.5 GHz P4) but no temperature, could it be that is wattage a P4 would reach the trip-off temperature under normal (Intel's reference) cooling conditions?? If that is the case, this wattage is way below the absolute max power (72.9W for 1.5 GHz), which means under normal cooling, a P4 may not be allowed to take in a full load. In that case, Bert McComas is still correct in his assessment.

Of course, with "adequate" cooling, one can run at the max wattage at full speed. However, if this "adequate" means "extraordinary", then it is pointless since with "extraordinary" cooling, one can even run 1.3 GHz Athlon at 1.6 GHz.
 

bhc

Distinguished
Apr 14, 2001
142
0
18,680
By the way, I agree that having a safety switch is better than not having one. AMD and others really need to do something in this area. However, a poorly-designed safety switch may drastically reduce the max performance of a CPU can produce, and to think people pay a premium in order to get that performance. That would be very annoying as well.
 

dhlucke

Polypheme
This is just another reason not to buy a P4. What are they doing about this?

<font color=red>Computers are useless. They can only give you answers.</font color=red>
Pablo Picasso
 

Raystonn

Distinguished
Apr 12, 2001
2,273
0
19,780
As mentioned in the document to which I previously linked, the point where the thermal control circuit goes active is specified in the processor datasheet. You will find the P4 datasheet here: ftp://download.intel.com/design/Pentium4/datashts/24919802.pdf

The maximum allowed temperature for a 1.5GHz P4 is 72C. If this temperature is exceeded, the thermal control circuit will become activated to save your CPU from becoming damaged. You will also be notified via ACPI if you are running an operating system which supports this. (Windows does support ACPI.)

I feel much better about an investment in a P4 knowing I cannot fry it. This will keep countless overclockers from accidentally frying their chips. 72C is extremely hot.

-Raystonn

-- The center of your digital world --
 

AmdMELTDOWN

Distinguished
Dec 31, 2007
2,000
0
19,780
hmmmm, makes me wonder if the ppl who run the benchmarks on the P4 know this and intentionally trick the P4 into throttling down, hmm...just makes me wonder.





"Amd cpu...Gone in 2 secs flat, it truly is a fast chip!"<P ID="edit"><FONT SIZE=-1><EM>Edited by AmdMeltdown on 04/14/01 04:36 AM.</EM></FONT></P>
 

bhc

Distinguished
Apr 14, 2001
142
0
18,680
Raystonn,

Thanks again. 72C is indeed high. Now can you clarify what 54.7W (for 1.5 GHz) means? Is it indeed "not-to-exceed" (under normal cooling) or just a marketing lie? Furthermore, if a P4 runs at the absolute max power (72.9W for 1.5 GHz) with normal cooling (Intel's reference heatsink and fan), what will be the temperature? Can you or someone shed some light for us? Thanks.
 

cutepunk

Distinguished
Feb 14, 2001
26
0
18,530
A quote from Page 28 of Intel's Pentium 4 Thermal Design Guidelines :
(ftp://download.intel.com/design/Pentium4/guides/24920301.pdf)

<<A chassis cooling solution designed to the TDP as specified in the Datasheet will adequately cool the processor to a level where activation of the Thermal Monitor feature is very rare or non-existent. Various levels of performance versus cooling capacity are available and must be understood before designing a chassis. Automatic thermal management must be used as part of the total system themal solution.>>

In other words, this feature will prevent you from frying your CPU when you make a mistake (of installing heatsink, when the fan doesn't work, etc)
 

Raystonn

Distinguished
Apr 12, 2001
2,273
0
19,780
This is the wattage that is not exceeded by the CPU during the execution of 99.9% of applications, including intensive benchmarking. A similar number is given by AMD for their CPUs. Neither company's CPU actually has this number as a hard limit on power consumption. Both can go much higher. These occasional higher numbers are called spikes and generally do not contribute greatly to the average wattage used by the CPU.

It has been shown that you cannot supply adequate cooling for Athlons that run at their maximum power usage. Luckily, this maximum power consumption only lasts a few microseconds during the most intense calculations imaginable. Otherwise there would be many upset Athlon owners with a great deal more keychains.

Intel CPUs aren't much different in this regard. (You wouldn't expect them to be. They're built using the same materials.) The only difference is that in the off chance it's already very hot and one of these intense spikes from an extremely merciless calculation comes around and actually lasts for a while, you can be guaranteed that your P4 will not be damaged. The same circumstances will damage an Athlon and even a P3.

This feature in the proper handling of power consumption/heat spikes is actually a great advancement in CPU design. A side affect of mentioning the advancement is actually mentioning the problem that necessitated the fix, about which the general public was completely unaware. This could have the negative PR impact of people relating the power/heat spike issue with Intel rather than properly crediting them with fixing the problem that has plagued all modern processors to date from all companies.

-Raystonn

-- The center of your digital world --
 
G

Guest

Guest
It's possible that Northwood will solve the problem. The 0.13 Micron process should produce much less heat, so this 'ceiling' should be hit less often. Hopefully this is just a temporary feature (like socket 462) that will dissapear or become irrelivant when the 'real' p4 is introduced.

Why has Intel introduced the p4 at this stage, when it's clearly very immature? I don't know. Maybe they felt that after the 1.13GHz disaster, they needed to keep up with AMD before the Northwood is released (otherwise their reputation could be damaged substantially). If the p4 hadn't been released yet, Intel would be in serious doo-doo, comparing the 1GHz p3 with the 1.33GHz Athlon!

Did anyone also read the stuff about the poor bus design of the p4 meaning less actual data throughput than the p3?


~ I know the human being and fish can coexist peacefully ~
 

Raystonn

Distinguished
Apr 12, 2001
2,273
0
19,780
"It's possible that Northwood will solve the problem."

What problem?

"Hopefully this is just a temporary feature..."

Heck no. Hopefully it's a permanent feature. This is not a problem, but a feature. Read into the posts above more carefully.

-Raystonn

-- The center of your digital world --
 

bhc

Distinguished
Apr 14, 2001
142
0
18,680
Raystonn,
From what you said, it appears the number like 54.7W is the nominal max power (3 sigma??) and there are power excursions will exceed it. Given that the time response of the current comparison circuit should be very fast, I can see if there is a local (on-chip) temperature rise from one of those excursions, the trip-off may be instantly activated, which is not the intended purpose since it really means to be a safety switch. Even if it only happens in very small percentage of cases, it's not good, especially those may be the cases users would want MORE performance, not 50% less. For many years, when we buy a DESKTOP CPU with a certain clock speed, you can be sure the CPU will deliver that speed ALL the time. If a 1.5 GHz P4 does not deliver 100% of the time, what about a (hotter) 1.7 GHz??

I totally agree having a safety switch is nice, but Intel, AMD and others need to work to reduce the power consumption in addition to the safety switch. I hope once the 0.13 micron technology ramps up, with much smaller die size and lower supply voltage, the power consumption will go way down and make this discussion moot. Of course, power consumption also increases linearly with the clock speed. So, in another year, the problem may show up again.

That is why I rather see a more efficient CPU architecture than something good for pumping up the clock speed (like P4). If the CPU core is not efficient, you need HIGHER clock speed just to keep up with the competition. But, higher clock speed produces higher power consumption. Eventually, you will have to pay the price. So far, Intel has the advantage of a more advanced IC processing technology, so AMD chips even at lower clock speed do not consume less power. However, AMD is currently into the advancement of SOI (silicon-on-insulator) technology, which greatly reduces the junction capacitance (C) and hence power consumption, and Intel is not. Therefore, Intel may not have the advantage for much longer. Let's wait and see in a couple of years if the P4 design is wise or not.
 

Raystonn

Distinguished
Apr 12, 2001
2,273
0
19,780
There will be the occasional moment that this power spec is exceeded, perhaps once in a year. Since the thermal circuit is based on temperature and not wattage, it would have to last for a while and create an exorbitant amount of heat in order for this cirtuit to become active. We're talking 70 degrees Celsius maximum here. That's highly excessive. A spike for a small moment will not cause the CPU to raise in overall temperature to 70C.

The average user will never, ever, see his CPU temperature reach this level with the standard provided heatsink/fan, even at full throttle running intense benchmarks for a month straight. This thermal technology is meant only as a protection against improperly installed, or broken, heatsinks and fans. Intel took a look at the fry rate of the Athlon and took steps to ensure it would not happen to the P4, ever. This is important from a product quality standpoint.

-Raystonn

-- The center of your digital world --
 

AmdMELTDOWN

Distinguished
Dec 31, 2007
2,000
0
19,780
and AMD took steps to insure that the "fry rate" on their chips go up so that they can sell twice as many in half the time.

now it's comming all together...hahahahahahahathlon!



"Amd cpu...Gone in 2 secs flat, it truly is a fast chip!"
 
G

Guest

Guest
Meltdown, you are so irrational. The only way you fry an AMD cpu is if you take it and overclock it by double its manufactured clock speed, install the heat sink wrong, or you don't check and make sure the fan works. I leave the cover off my case all the time so I can see if it works. Oh wow, Intel's CPU's have thermal protection feature on it, who cares! If you overclock and compute responsibly, you won't fry anything.

Aklein
Life is hard...Live with it
 

mjdunn

Distinguished
Feb 23, 2001
228
0
18,680
What is the normal running temp on the P4? I didn't see an answer in this thread. 70C is very high but if the CPU idle's at 55C is 70C very high?
 

peteb

Distinguished
Feb 14, 2001
2,584
0
20,780
<<< It has been shown that you cannot supply adequate cooling for Athlons that run at their maximum power usage. Luckily, this maximum power consumption only lasts a few microseconds during the most intense calculations imaginable. Otherwise there would be many upset Athlon owners with a great deal more keychains. >>>

WTF? So those of us that crunch seti fulltime (from pc powerup to shutdown) and crunch DivX etc. at the same time are not running our systems flat out? I can assure you that my system copes with the heat output of my 1.33@1.6Ghz Athlon quite admirably thank you very much.

-* This Space For Rent *-
email for application details
 

Grizely1

Splendid
Dec 31, 2007
7,810
0
30,780
My 900@1.13 is nice a frosty with my watercooler when i run seti all day. (has been running about 19 hours straight now i think).

-----------------

All your RAMBUS are belong t............ ahh screw it
 

AeroSnoop

Distinguished
Dec 31, 2007
121
0
18,680
AmdMeltDown you are missing something here. You can't have it both ways. Either the P4 is throttling down to a dinky 750mhz when you really need the power, or the P4 really puts out 70watts just like the Athlon. If the later is true that means Intel's marketing is just pure bullshit that only narrow-minded people like you would believe. So which one is it?
 

bhc

Distinguished
Apr 14, 2001
142
0
18,680
Raystonn,
Sorry to be skeptical here. However, I know the heat transfer coefficients within the chip and from the chip to outside are finite. During a power excursion, there could have been a substantial temperature gradient for a short time. Since the thermal monitor circuit is ON CHIP, it will react instantly (well, the speed of electron), but the average temperature of the chip and heatsink etc will hardly change much if the duty cycle of those excursions is quite short as you suggest.

Let me give you an example. One of the ways to make ultra-shallow transistor junctions for deep submicron transistors is laser anneal. The temperature on silicon surface will reach greater than 1200C, but the laser pulse is short, and total amount of energy is low, the silicon substrate temperature is therefore much, much lower. Because of the finite heat transfer coefficient of silicon, a huge temperature gradient up to 1000C is established to activate impurity dopants and make ultra shallow junctions. The power excursion is not an extremely localized laser pulse. However, the basic idea is the same, and we don't need a 1000C temperature gradient to trip the switch (30C??).

I don't have any hard evidences to show the throttle down of P4 occurs frequently, but physics is there. All I say is a simple on-chip thermal switch may do more than it is intended. I am open to you to provide some hard evidences to show it is not the case. Thanks.
 

kal326

Distinguished
Dec 31, 2007
1,229
108
20,120
Good point, being that it is a "thermal diode" one would think that the diode would cut off at a temp, not a wattage. I agree with the errors of the article. In addition it is usually possible to addjust how much and when the cpu is throttled back in BIOS.

"You're just a heartless computer" <font color=red>"That is correct. What is the problem?" Z.O.E.</font color=red>
 

Raystonn

Distinguished
Apr 12, 2001
2,273
0
19,780
"... but the average temperature of the chip and heatsink etc will hardly change much if the duty cycle of those excursions is quite short as you suggest."

That's the key. The thermal circuit is not triggered until the average temperature on the chip exceeds 70C. These short excursions that do not adversely affect the CPU will not cause the thermal circuit to activate. Only heat that falls just barely short of that required to damage the CPU will activate the thermal circuit. The trigger temperature is set at the maximum safe operating temperature of the CPU, not at soem artifically lowered number..

You'll never see this triggered unless you have improper cooling such as a broken or inefficient heatsink/fan, a vent is blocked with dust, etc. Would you prefer a fried CPU? You will be notified via ACPI in the event the circuit is triggered, so it's not a silent operation. You will definately know if it's happening. You can then take corrective action for your cooling solution instead of unknowingly shortening your CPUs life.

-Raystonn

-- The center of your digital world --