[gumby5647]

Hey, i need a real data here, not a guess.

How much power does the P4 draw? i thought it was around 50 watts....can someone please tell me exactly?

same with the G4, i thought it topped out at 17 watts.


thanks for your help.

------------------
AIM: gumby5647
ICQ:41746288
[email protected]
http://homepage.mac.com/gumby5647
iMac DV +, Indigo,
GO PHANTOM REGIMENT! All the way in '01!

[Nimisys]

The official specs for the P4 are 54.7watts... reality is that under peak usaeg it can hit as high as 100w. thats why it has clock throttling in it for those reasons. goto http://www.inqst.com/articles/athlon4/0516main.htm and read their latest info on the P4 for its heat matters. Intel really screwed the pooch with this one.

the T-bird is around 78w... but it still runs stable at 140f
Palimino is around 58.4W without speedstep enabled

i don't know what the G4 is but because of its much lower transitor count, i am guessing it is much less, but its heat sink is huge! it's the size of a PC atx power!

A 733 MHz MPC 7450 has a maximum power dissipation of about 20.7 Watts. 33 million transistors, 106 sq mm die, 1.8/1.9V, 483 CBGA, 0.18u design rules (Moto HiP 6). Add 1 or 2 Watts for L3 cache, but it's no big deal.

As I recall, a 500 MHz MPC 7410 has a maximum power dissipation of about 12 Watts. 10.5 million transistors, 57 sq mm die, 2.1V, 0.18u design rules (Moto HiP 6).

As I recall, a 500 MHz MPC 7400 has a maximum power dissipation of about 17 Watts. 10.5 million transistors, 83 sq mm die, 2.15V, 0.22u design rules (Moto HiP 5).

I have no idea why Moto doesn't design their processors to run at lower voltage. Something like 1.6V would translate to about 30% reduction in power usage compared to 1.9V.

I would hazard a guess the reason why Apple uses such large heat sinks is that the El Capitan case (the Power Mac case) simply isn't suited for large power dissipation. It looks to be about 40 Watts for all processors is the best it can handle.

[gumby5647]

what about the G3 series....old and new ones....?

------------------
AIM: gumby5647
ICQ:41746288
[email protected]
http://homepage.mac.com/gumby5647
iMac DV +, Indigo,
GO PHANTOM REGIMENT! All the way in '01!

[Jsnuff1]

Originally posted by THT:
I would hazard a guess the reason why Apple uses such large heat sinks is that the El Capitan case (the Power Mac case) simply isn't suited for large power dissipation. It looks to be about 40 Watts for all processors is the best it can handle.

actually the g4 case is has one of the best cooling systems there is, i would say the reason for the large heatsink is that it cools the proccessor better, therefore increases performance

[Randycat2001]

The G4 heatsink has no fan, am I correct? Honestly, I don't know.

[Nimisys]

Originally posted by Jsnuff1:
actually the g4 case is has one of the best cooling systems there is, i would say the reason for the large heatsink is that it cools the proccessor better, therefore increases performance

i doubt you got any real concept on cooling then.that case doesn't have much vents to allow fresh air inside of it, nor does the plastic breathe much. a much smaller hs with a slower fan (3500RPM) would probably cool better. a HS only transfers heat when there is enough difference between it the cpu core and ambeient temperature. in this case with a lack of ventalation the HS performs worse than if there was a constant airflow. Secondly the cpu being cooler doesn't add to performance in any way. if the copu was overheating then it might help stability by lowering the temperatures, but it would not help performance. Look at the PC people running their CPU's at 1 or 2 degrees C or less... they don't go any faster at the clock then anyone else. having a cooler CPU just allows you to run it at higher speeds if you want (OverClocking) but will not provide any extra performance aat the stock speed.

Go and learn at HardOCP and the HardForum

[This message has been edited by Nimisys (edited 05-18-2001).]

[Jsnuff1]

Originally posted by Nimisys:
i doubt you got any real concept on cooling then.that case doesn't have much vents to allow fresh air inside of it, nor does the plastic breathe much. a much smaller hs with a slower fan (3500RPM) would probably cool better. a HS only transfers heat when there is enough difference between it the cpu core and ambeient temperature. in this case with a lack of ventalation the HS performs worse than if there was a constant airflow. Secondly the cpu being cooler doesn't add to performance in any way. if the copu was overheating then it might help stability by lowering the temperatures, but it would not help performance. Look at the PC people running their CPU's at 1 or 2 degrees C or less... they don't go any faster at the clock then anyone else. having a cooler CPU just allows you to run it at higher speeds if you want (OverClocking) but will not provide any extra performance aat the stock speed.

Go and learn at HardOCP and the HardForum

[This message has been edited by Nimisys (edited 05-18-2001).]

yea and u obvously have no concept of the g4 case, fine i agree with you that there is litte performace increase wiht cooler cpu, but the g4 case is far supirior than any other wintel tower ive seen. No ventalation? half the back of the g4 has air inlets, more than enough airflow. and a larger heat sink equal more surface area therfore heat disipatation is much faster, so i doubt a smaller heatsink with a less powerfull fan would do better

[Nimisys]

having holes doesn't matter if you got no method for circulating the air into and then out of the case. Having to use only an HS ontop of that doesn't help any more as a HS needs plenty of air circulation to function properly, surface area or not.

As for PC HSF's they can keep a 70watt chip that burns out in 8secs under 140F at 5000RPM or under 100F with a 7000RPM fan. now these suckers are like 1/6 the size of your brick and seem to have no problem cooling hot hings. i got a feeling they would perfrom equally weel ontop of a G4. If mac were to reduse the brick in half and puit a slower 2500RPM fan on it, they wouldn't have the thermal issues holding back their DP setup.

as for PC cases the ALuminum ones look nice (i like the look of brushed metal) but more importantly are the best for keeping thie internals as coola sthe outside. http://www.pcmods.com/aluminumcases.htm

[davidflas]

Originally posted by Randycat2001:
The G4 heatsink has no fan, am I correct? Honestly, I don't know.

I don't know about the 733Mhz G4's, but the heatsink in my 466Mhz G4 has no fan on it.

------------------
Jesus chills with a 40

[Randycat2001]

Originally posted by Nimisys:
having holes doesn't matter if you got no method for circulating the air into and then out of the case.

The entrance vents you are worried about are at the bottom of the G4 case, ASFAIK. They are that way on a BW G3 case. Anyways, big heat sink/natural cooling vs. small heat sink/forced cooling can be argued either way. If it means one less fan, it means that much less fan noise, so choosing a big heat sink/natural cooling is not a bad decision (this isn't a race for smallest heat sink size- Apple knows this and heat sink size isn't a specific feature to draw consumers- but noise reduction can be). A high-clocked x86 chip, on the other hand, may not have much choice as far as heat sink size. A natural cooling heat sink for that kind of dissipation probably would not fit in the case in a practical manner at all. So fan-cooling (with the commensurate small heat sink) is the only feasible way to go for a PC.

Everybody feel the underside of your G4 tower toward the edge along the long dimension- feel those slots? If you are really sensitive you can feel a gentle breeze being drawn into the case from there.

[cube-dude]

http://forums.macnn.com/cgi-bin/Foru...ML/000430.html

[Blakhawkg3]

Originally posted by Nimisys:
Look at the PC people running their CPU's at 1 or 2 degrees C or less... they don't go any faster at the clock then anyone else. having a cooler CPU just allows you to run it at higher speeds if you want (OverClocking) but will not provide any extra performance aat the stock speed.

yes that is true.....the other thing a lower cpu/case temperature does is increase the life of your entire system

[AirSluf]

Heat dissipation off a heatsink depends on exactly 2 things.

1. Surface area squared of the HS

2. Heat transfer off HS to Air.

A larger HS has the square of the area to dissipate heat so the temp difference ratio from the HS to the air follows a curve dependent on the difference in squared sizes. In short, the larger a heat sink gets, the faster the surface temp goes down. As surface temp goes down the efficiency raises because there is a larger difference in ambient temp and surface temp. The actual thermal transfer follows some sort of exponential curve so more temp difference means much more (or quicker) heat transfer.

This all means a huge heatsink can pump out lots of heat without having to rely on as big or fast (read noisy) fan to get the needed difference in ambient vs heatsink temp that a small heatsink would need.

[Randycat2001]

Actually, there is a third factor- surface temperature of the HS.

Ideally, you want the HS to be really hot- that's when heat transfer is most efficient. If it is just barely warm, there isn't going to be much potential to enable appreciable heat transfer, surface area or not. A larger heat sink will tend to be cooler due to its greater thermal mass. Unit heat transfer will likely be low, but this will be made up somewhat by the increased surface area that comes with the greater size. A small heatsink will tend to run hotter so the unit heat transfer will be better, but the lesser surface area will drastically reduce the overall heat transfer (hence the addition of the fan to artificially increase the unit heat transfer to make up for the small surface area).

So the choice of appropriate heatsink is a balance between surface area, heat transfer to air, and HS operating temperature. The heat transfer to air and HS operating temperature are closely coupled. The surface area, size, and HS operating temperature are also closely coupled.

[OreoCookie]

The G4 733 is the only one with a fan (maybe the 667 shipped with a fan, too). For all other models, a passive cooler suffices.

[Chimpmaster]

Just a point of logic rather than technical fact.

G4 notebooks are great. They have been around for a few months now.

Havent seen a p4 notebook yet. Wonder if we will (smirk).

And dont give me any bull about a high end p3 being better than a g4 - thats absolutely rubiish and we all know it.

There isnt a p3 on the market that would give a G4 a run for its money.

[AirSluf]

Actually, there is a third factor- surface temperature of the HS.

Ideally, you want the HS to be really hot- that's when heat transfer is most efficient. If it is just barely warm, there isn't going to be much potential to enable appreciable heat transfer, surface area or not. A larger heat sink will tend to be cooler due to its greater thermal mass. Unit heat transfer will likely be low, but this will be made up somewhat by the increased surface area that comes with the greater size. A small heatsink will tend to run hotter so the unit heat transfer will be better, but the lesser surface area will drastically reduce the overall heat transfer (hence the addition of the fan to artificially increase the unit heat transfer to make up for the small surface area).

So the choice of appropriate heatsink is a balance between surface area, heat transfer to air, and HS operating temperature. The heat transfer to air and HS operating temperature are closely coupled. The surface area, size, and HS operating temperature are also closely coupled.

Well, this goes to show how dangerous a little knowledge can be. Randy, you have the first part exactly backwards, and don't seem to realize you support my first point!!

Assuming a good thermal bond with the CPU, you want to design a the heatsink to be as cool as possible for the specific heat transfer rate required. Hotter metal normally indicates poor internal thermal transfer and storage of heat vice radiation away from the heatsink, or a dismally under-designed surface area which severely hampers the ability to dissipate heat. A large heatsink with a good thermal bond and using the correct materials is cooler because of the math.

Surface area squared is radiating heat into the atmosphere, while internal volume is a cubed factor. So surface temp is reduced by a difference-squared factor before you even begin while the equivalent heat storage capacity for the same temp is difference-cubed larger. The lower HS temp actually makes it more efficient at extracting heat away from the CPU where it can then be distributed to a much greater internal volume and dissipated over a much larger area.

The larger HS of the same material will always have a lower air to surface temp difference and be more efficient at extracting heat from the CPU. While the higher difference in air to surface temp on the smaller HS will have a greater heat transfer per unit area, the additional squared-difference in area of the larger HS will always pump out more total heat than the smaller one. (A HS that is 2x longer on all sides will have 4x the surface area, roughly including the fins, and nearly 8x the internal volume.) The heat dissipation curves never cross, it is not possible with the laws of physics as they are currently understood. The differences diverge rapidly in favor of more surface area.

Smaller heat sinks require the extra overhead of a fan to push more air past the smaller surface area, gaining "virtual" surface area. A design tradeoff that works fine until the fan dies, then it is only minutes until curtains for the poor CPU with an undersized HS.




[This message has been edited by AirSluf (edited 05-26-2001).]

[Nimisys]

Originally posted by Chimpmaster:
Just a point of logic rather than technical fact.
.

Havent seen a p4 notebook yet. Wonder if we will (smirk).

And dont give me any bull about a high end p3 being better than a g4 - thats absolutely rubiish and we all know it.

There isnt a p3 on the market that would give a G4 a run for its money.

p4 mobile is one quarter off when intel gopes to .13 micron and the northwood core

maybe not a P3, but an t-bird will give your G4 a run for it's money and then some... even at same speed the t-bird is only about 5-10% slower, so it doesn't take much of a higer clock to be faster even with altevec factered in.

the Athlon4 is available in notebooks only right now and it runs 5-15% faster than the t-bird with 30% less power (54.7w). its dynamic clock adjustment hardware (PowerNow!) allows it to have a battery life over 4 hours under normal conditions.

Somehow, AMD meant in when it said it had the fastest mobile CPU out, even including the G4 and latest P3 mobile at 1.13ghz (again)

[olePigeon]

Somehow, AMD meant in when it said it had the fastest mobile CPU out, even including the G4 and latest P3 mobile at 1.13ghz

I'd like to see that logic. It's 5-10% slower than a G4 except when in a latptop? Huh? Oh yea, can you show me some benchmarks from AMD and 1 3rd party source regarding the "t'bird" processor? I've been curious as to why every PC zealot has been boasting about it. Thanks.

[Randycat2001]

Originally posted by AirSluf:
Well, this goes to show how dangerous a little knowledge can be. Randy, you have the first part exactly backwards, and don't seem to realize you support my first point!!

Take it easy there! I wasn't trying to counter your argument- just add my 2 cts. A hotter heat sink does transfer heat to ambient air more efficiently. The h-value of a heat-sink is not constant. It is nonlinear across a range of temperature differentials (between sink to ambient) and degree of natural/forced convection. That is how you can have a hotter heat sink shedding more heat energy than a same-size sink with lower temperature (not to mention that higher temperature differential to air automatically means higher heat transfer- it's in the heat transfer equation after all). While it's true the chip on the hotter heatsink is probably running hotter, as well, there's no harm/no foul as long as it is still safely within its operating range.

Your disposition lies with the temperature disparity between the source (the CPU) and the heatsink, not the heatsink and ambient air which I was addressing. I absolutely agree that a cool heatsink will suck more heat from a hot chip than a cool heatsink. It makes sense to couple the chip and the heatsink intimately for maximum heat conductivity, so sooner or later the heatsink core is going to approach the temperature of the chip. So technicially, there is no such thing as a hot chip/cool sink scenario. The heatsink must stabilize in line with the temperature of the chip. So in the end, you have a hot chip and a relatively hot sink (or at least warm) and then heat dissipation will occur to relatively cool ambient air.

Typically you want the heatsink to run at some elevated temperature, but only to the point where the chip is still running comfortably. That way you end up with a heatsink option that dissipates heat efficiently and stays relatively compact in size. The fan-blown/small heatsink combo is a perfect match in that the heatsink size is very compact, yet the heatsink runs at an elevated temperature for truly efficient heat dissipation (also assisted by the forced cooling of the fan). Assuming you can deal with the fan and air noise, you cannot argue that this type of heatsink does an excellent job of shedding 50-70W in such a compact profile. The efficiency speaks for itself.

On the other hand, this G4 heatsink appears to approach heat dissipation physics on the opposite extreme, and there is absolutely nothing wrong about that. It uses a large, naturally-cooled heatsink. So core heatsink temperature never really get that hot and the neither does the chip. Heat dissipation efficiency isn't that great since the heatsink doesn't get appreciably hot (low h-value and low temperature differential to drive the transfer) compared to the ambient air temperature and no forced-convection as with a fan-blown type. However, it makes up for this by brute size and thus ultimate surface area, so the cumulative heat dissipation stays appreciable. The size is big, relative to the total amount of heat it dissipates (less than 20 W?), but it doesn't need a fan and is thus quieter. It's just a different set of priorities which leads to a different set of heatsink parameters- that's all. I'm not trying to say one way is better than the other. I'm all for quieter computers- trust me! If a big heatsink is what it takes, I have no problem with that. ...but make no mistake that the same size heatsink with the addition of a fan(s) has the potential to dissipate as much as 10x more heat energy, assuming you are allowed to run at an elevated temperature w/o ruining the chip. That is the thermal efficiency I speak of.



[This message has been edited by Randycat2001 (edited 05-26-2001).]

[Nimisys]

Originally posted by olePigeon:

I'd like to see that logic. It's 5-10% slower than a G4 except when in a latptop? Huh? Oh yea, can you show me some benchmarks from AMD and 1 3rd party source regarding the "t'bird" processor? I've been curious as to why every PC zealot has been boasting about it. Thanks.

for g4/t-bird comparison you got to look for the linux benches and i believe i have only seen the k7/g4 ones, and the t-bird out does the k7, just as the g4+ is faster than the g4. as for the notebook athlon re-read my post... i explained it clearly, it was a new core, the palomino, and it runs fater than t-birds by 5-15%, but has only been released in notebook format. the server ones will be faster with more cache and they come out with the 760MP chipset next month. then a faster still desktop version will release in quarter3, once they get the yeilds up on the workstation model.

for t-bird benches look at any of the major PC sites... ([H]ardOCP, TomsHardware, SharkyExtreme, AnandTech, ect) it is the quickest chip out there and dam cheap, a 1.33ghz will run about $250, and overclocks like a mofo, with the AXIA cores hitting 300+mhz with standard cooling/voltage. it is remarkably similar to your G4 also, and if the core were to be fed its native RISC, it would be able to beat the G4 clock for clock. Arestechnica has the article in their cpu theory and praxis section.

[iPad]

1.2Ghz Thunderbird-C's are also going for around $150, your best bet if you want an AXIA or AVIA batch (as i believe, most 1.2+Ghz Tbirds are within these two batches). If you have the cooling, you can get 200+, but i'm probably expecting around 1.4Ghz for mine since i'm just gonna have a Thermosonic Thermoengine with a YS Tech 6800RPM fan. I'm building my system next week:

Antec SX1030B ATX Case
1.2Ghz AMD Thunderbird-C OEM
Iwill KA266 Mobo
256MB PC2100 Crucial CL2.5 DDR RAM
45GB IBM 75GXP 7200 RPM ATA/100 (just gonna pull this out of my mac and put in a cheapo maxtor 20gb in my mac)
MSI Starforce 8831 Geforce 2 Pro 5ns 200/400 64MB DDR
CDRW (Plexwriter 12x/10x/32x) /DVD (Aopen 12x DVD) /Zip (NEC Zip internal OEM) / Floppy (generic)
SoundBlaster Live! 5.1
Altec Lansing 5 piece gamer
Various Cooling Materials (2 80mm Panflos, 1 Blorb for vid card, Arctic Silver Goo - adhesive and paste)
Generic 10/100 BaseT ethernet card
Windows 2000 Professional

Total Cost: $1080 (heh, what a coincedence, the price just happens to be the first batch of the 10+ nVidia NED drivers)

But I already own quite a few parts etc, so the total cost that i'll have to pay will probably be around 830, then i'll pick up a copy of Giants, Tribes2, and some other stuff, and game all summer

Oh, but lets see, we're talking about power consumption, sorry got carried away , let's put it this way for those people saying how G4's will ultimately be l33t, with an analogy.

The G4 is a more effiecient processor, however, just because it is more efficient, does not mean it's faster. It's kinda like comparing wind power to..hydro, etc, although wind power isn't that efficient (i dunno, someone back me up), it does not generate more electricity and typically costs more than if you were to go dam power (Thunderbird)..BUT, a dam is more likely to break down, because it requires more maintainence and produces..heat? (heh), those thunderbirds can give you scaldings. However, it seems more of the world is on dam power, simply because its cheap...and it works fine. And in certain situations, when it's damn windy and the waves are calm (which almost never happens ), wind power generates more power..(altivec), and vise versa (SSE with the Palomino).

Anyways, that was pretty screwed..my apologies

I'm just sick of Apple right now, and i'd rather have the system that i've posted instead of a mid end iMac or a refurbished G4/450.

[This message has been edited by iPad (edited 05-27-2001).]

ooops. were are now entering the land of really bad analogies. Dam power==many $$$$ more than wind power. You forgot to factor in the initial cost [paid for by taxpayers, while the water and power get given away at less than market rates to big users] and the other hidden costs that are usually not counted by the Army Corp and other dam lovers like habitat destruction and eventual siltation that could render the dam useless [depends on location]. Sorry to go OT, but west is littered with dams and they have cost many $$$$.

[iPad]

Yeah, that's what i stated, "Anyways, that was pretty screwed..my apologies", therefore, i was trying to relate it to something in a situation like that, although not true..the fact is:

The G4 is more efficient
The Thunderbird is faster

Altivec makes a difference...but even so, the Thunderbird will still have a small edge in apps such as that, if you wanna talk about dual processors, the Thunderbird will still take the lead as the Tyan 760MP board has already started shipping for about $550 US.

The only reason you get a mac nowadays is because of the OS. End of statement.