Why do the FX's put out more heat?

[madpete]

Hey guys,

In the quest to get 3ghz prime stable, I have been through 5 CPU's in the last 6 months ( FX-53, 3200+ winchester, 3800+ venice, 4000+ San Diego, and now FX-55 San Diego).

I've noticed on both the FX's that they put out more heat than the other chips, why would this be? Esp as 4000+ SD is almost the same chip as FX SD isn't it?

Temps seem to be around 5-10 degrees higher on compareable voltage, and pwm temps also go up with FX's (in my case from 34 degrees to 40 degrees. I've only noticed this because I've gone through a lot of CPU's in a short time.

Oh and still can't hit 3Ghz prime stable (even tho she will boot @3ghz on default volts), this FX does 2.98ghz....so close and yet so far

[Crankster]

There is no reason that a 4000+ claw should draw less than a FX-53.
They are the same chip but with unlocked multis.

Compared to the other chips the 0.13 nm and 1 mb cache makes the difference.

[madpete]

Yeah thats what I thought, yet it does happen on my system, i can't really explain it.

[ozzimark]

i have a suspicion that AMD uses more faster/hotter transistors in the FX line.

[IYP]

yeah i second ozzimarks comment...they expect people to overclock with the fx so they mite give it a little more headroom

[frostedflakes]

FX-55 Clawhammer supposedly had some implementation of strained silicon, which could explain the higher heat. And correct me if I'm wrong, but doesn't FX-55 San Diego have a default voltage of 1.5v? If so, the reason for higher heat should be obvious.

[madpete]

FX-55 Clawhammer supposedly had some implementation of strained silicon, which could explain the higher heat. And correct me if I'm wrong, but doesn't FX-55 San Diego have a default voltage of 1.5v? If so, the reason for higher heat should be obvious.

Does it? I didn't think it did.

If i set mine to auto it reads out as 1.35 ish in smart guardian, which usually means its actually 1.4v.

But what I was talking about was at the same volts, the FX's are hotter, and they also seem to heat up the PWM more too...as if theres more strain on it.

[madpete]

Is the OPN on your processor ADAFX55DAA5BN? This correlates with a 1.5v part.

Frosted, off the top of my head I don't know, but I will check (I always forget to write it down). I was not even aware that there where FX's with 1.5v as default volts, I thought the AA meant variable voltage???

Edit: Your post appears to have vanished

[[XC] leviathan18]

4000sd venice winnies all are 90nm fx are 130 so more heat...

[madpete]

My FX-55 SD is 90nm

But anyway, the FX-55 SD, runs hotter than the 4000+ SD at same volts with same cooling, and the PWM temps are also higher (they are both 90nm). There must be something differant about the FX's, just don't know what it is.

[frostedflakes]

Frosted, off the top of my head I don't know, but I will check (I always forget to write it down). I was not even aware that there where FX's with 1.5v as default volts, I thought the AA meant variable voltage???

Edit: Your post appears to have vanished

Yeah you're right, after I posted I did some searching and realized I messed up, so I went ahead and deleted the post.

[[XC] leviathan18]

maybe is dsl sSOI ????

[Frodin]

Maybe the temp- sensor on the FX chips has been calibrated differently to further remove the chance of the "cold bug" appearing?
Just a thought...

[SIOUX]

The defult volt for a 90nm Sd is 1.35-1.4v ...............

[ben805]

possibly better memory controller and lower latency cache..... have you try to bench both 4000 SD and FX55 SD side by side at the same clock speed on same voltage and see if there's any difference?

[madpete]

possibly better memory controller and lower latency cache..... have you try to bench both 4000 SD and FX55 SD side by side at the same clock speed on same voltage and see if there's any difference?

Ben, no not conducted any side by side benchies, just an observation on my system. It could be as you say, it could also be the calibration of the temp sensor as has also been mentioned but that does explain higher PWM temps too.

[ben805]

if someone can do a side by side comparison and confirm the FX to be superior to regular SD...which I think it should, then you got your answers.

[ozzimark]

maybe is dsl sSOI ????

DSL = dual stress liners = fancy name for strained silicon

frodin: i don't think the temp sensor has anything to do with the cold bug. after reading around, i think it's the memory controller failing when it gets really cold. this is just my theory though

[Dumo]

@madpete, what week is your fx55 sd?

[madpete]

@madpete, what week is your fx55 sd?

Dumo,

0513 mate...

ADAFX55DAA5BN - CABCE 0513MPAW

[Dumo]

Dumo,

0513 mate...

ADAFX55DAA5BN - CABCE 0513MPAW

Interesting, mine is 0511 CABGE-MPMW. Lets compare notes