Use TCaseMax to predict overclockability and heat

B

buzzly

Limp Gawd
Joined
Feb 17, 2006
Messages
225
Get you AMD64TcaseMax program here.
http://www.thecoolest.zerobrains.com/forums/viewtopic.php?t=83

Description from the authur:

This is a new type of program, which checks AMD64 CPUs for their maximum rated case temp or TCaseMax and reports the TDP which corresponds to the processor in question.
On most AMD processors TCaseMax and TDP values are constant, and depend on CPU type and model, but on the E Revision chips, both this values and the default voltage are variable.
These are the chips that have variable values:
Athlon64 (Venice, San Diego)
Athlon64 FX (San Diego)
Athlon64 X2 (Manchester, Toledo)
All Rev E Opterons and Dual Core Opterons
What this all means is this: On 90nm manufacturing process, the power leakage of transistors on individual processors differs greatly, this is why AMD implemented TCaseMax. Each CPU has a different TDP rating, here's how it works:
A program reads the TCaseMax value off the CPU, then depending on the processor (CPU type like A64, Opteron;CPU Rating like 3500+, 146) it finds the processor's Thermal Profile (If you download the "AMD Opteron™ Processor Power and Thermal Data Sheet" and look at pages 10-11 you'll see the different thermal profiles). When the thermal profile is determined, the program finds the TDP rating that corresponds to the CPU's TCaseMax value.
So in a few words, the higher the TCaseMax, the higher the TDP of a processor will be (the transistor leakage is higher), so the processor will run hotter. Several people have reported that the Opterons rated at 71C TCaseMax were pretty hot. On the other side these processors will be able to reach higher clock speeds.
That is why we see that on average processors with higher TCaseMax can usually reach higher CPU speeds when overclocked.
Every CPU type (A64, FX, X2, Opteron, DC Opteron) has a different thermal profile, that is why you see that on an FX a TCaseMax rating can be "only" 57C and it'll overclock way better than an A64 with a 57C TCaseMax rating. That is because at 57C TCaseMax, an FX has a TDP of 77W while the A64 has a TDP of 44W.

Right now we're working on getting a php page up where the user will be able to upload his CPU stats to the server and be able to see what other people could manage with the same chips, or look at other CPUs with different ratings and see what they're capable of.

We believe this will be a great way to be able to roughly predict the overclockability of a CPU even before overclocking.
We hope you find this software useful.



----------------------------------------------------------
I stole the rest from Silentpcreview.

Cut-and-paste from 2 polls Silentpcreview is running.

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

By now, everyone who has an AMD A64 CPU should be aware that individual TDP info is embedded in recent "E" stepping processors. This is a poll to get an idea of the range of TDPs in A64 single-core processors. (pre-"E" stepping processors are not included in this poll.)

Use AMD64 TCaseMax v1.18 to obtain the info about your A64 processor. Please use the format below to report your info. Periodically, all the results from this list will be compiled and summarized in edits of this post.

There are several A64 "E" stepping processors at SPCR right now:

3000+, 939, Venice, Rev JH8-E3, 57°C, 44.1W
3500+, 939, Venice, Rev JH8-E3, 65°C, 67.0W
4000+, 939, San Diego, Rev JH8-E4, 59°C, 50.0W

----------------------
SUMMARY @ late night, Apr 12/06 -
Ordered by power, low to high

3000+, 939, JH8-E6, 53°C, 32.4W
3500+, 939, DH8-E3, 55°C, 36.4W
3200+, 939, DH8-E6, 55°C, 38.2W
3000+, 939, JH8-E3, 57°C, 44.1W
3000+, 939, JH8-E4, 57°C, 44.1W
Op144, 939, JH8-E4, 57°C, 44.1W
3700+, 939, JH8-E4, 57°C, 44.1W
4000+, 939, JH8-E4, 59°C, 50.0W
3500+, 939, JH8-E3, 65°C, 67.0W

--------- Second Poll ------------------------------------------------------

By now, everyone who has an AMD A64 CPU should be aware that individual TDP info is embedded in recent "E" stepping processors. This is a poll to get an idea of the range of TDPs in X2 processors. (pre-"E" stepping processors are not included in this poll.)

Please use AMD64 TCaseMax v1.18 to obtain the info about your A64-X2 processor. Please use the format below to report your info. Periodically, all the results from this list will be compiled and summarized in edits of this post.

There are 2 X2 processors at SPCR right now:

X2-3800+, Toledo, Rev, JH9-E6, 63°C, 1.35V, 65.6W
X2-4800+, Toledo, Rev, JH9-E6, 59°C, 1.35V, 85.0W

----------------------------------------------------
Summary, late night April 12/06

Optrn 180, JH9-E6, 49°C, 1.35V, 35.0W
Optrn 165, JH9-E6, 51°C, 1.35V, 45.0W
X2-3800+, JH9-E6, 63°C, 1.35V, 65.6W
X2 3800+, BH9-E4, 63°C, 1.35V, 65.6W
X2 3800+, BH9-E4, 63°C, 1.35V, 65.6W
X2-3800+, JH9-E6, 61°C, 1.35V, 59.4W
X2-3800+, JH9-E6, 61°C, 1.35V, 59.4W
X2 3800+, BH9-E4, 61°C, 1.35V, 59.4W
X2-3800+, JH9-E6, 67°C, 1.35V, 78.1W
X2-3800+, JH9-E6, 65°C, 1.35V, 71.9W
X2-4800+, JH9-E6, 59°C, 1.35V, 85.0W
X2-3800+, BH9-E4, 71°C, 1.35V, 89.0W
Optrn 165, JH9-E6, 63°C, 1.35V, 105.0W (<- My HUNKA HUNKA BURNING LOVE)
 
Unfortunately it doesnt work in all cases, ive known several people with very low maxtcase temps with insane oc's, and vise versa.
 
Yeah in my experience it doesn't really tell you anything.. all 0550vpmw Optys I've seen have had 49C, and we all know how they clock.
 
Confused... what the hell does this mean? since the tcasemax is higher it should overclock slightly higher?
My x2 3800 says TCASEMAX 63c TDP 65.6Watts
 
fenderltd said:
Confused... what the hell does this mean? since the tcasemax is higher it should overclock slightly higher?
My x2 3800 says TCASEMAX 63c TDP 65.6Watts
Click to expand...

TCaseMax is the maximum temperature an imaginary dot centered on the top the heat spreader can get and still have the chip run inside all the specifications.
At higher temperatures, integrated circuits dissipate more power (mostly from increased leakage current), and have higher electrical resistance in the interconnects (wires between trasistors).
Generally a chip with a high Tcasemax have low leakage current, and ample time (at stock speeds) to propogate signal across the interconnects. These are both generally very good things for overclocking.
Now as others have said, there is only a corespondance between good overclockers and high TCaseMax, one does not imply the other. It is entirely possible that with good cooling and airflow (or just other random factors) a low TCaseMax chip can overclock much better than a high Tcasemax chip. But I also tend to believe that high TCaseMax chips will - statistically, on average - be better overclockers than low TCasemax chips.
 
Thank you for helping me out. A upload database will be cool once its complete
 
FreiDOg said:
TCaseMax is the maximum temperature an imaginary dot centered on the top the heat spreader can get and still have the chip run inside all the specifications.
Click to expand...
actually, i'm not entirely sure that's true... that would be the maximum junction temperature (TControl Max), which is 70c last i checked.

what i believe happens, is there is a very rudimentary heatsink, if one at all. the cpu is powered up for a short period of time. the temperature after X time is the TCaseMax.. and obviously, if that value is hotter, the amount of power the chip is drawing is higher, thus why there is a linear relationship between TCaseMax and the "real" TDP ;)
 
From the AMD power and thermal datasheets:
1. Tcase max is the maximum case temperature specification which is a physical value in degrees Celsius. This value
is programmed into Rev D and later processors. Refer to the appropriate functional data sheet, and the
THERMTRIP Status Register in the BIOS and Kernel Developer’s Guide for AMD Athlon™ 64 and
AMD Opteron™ Processors, order# 26094.
2. Tcontrol max (maximum control temperature) is a non physical temperature on an arbitrary scale that can be used
for system thermal management policies. Tcontrol max represents the value at which the processor has reached
Tcase max when measuring the thermal diode with a dual sourcing current temperature sensor. Refer to the
appropriate functional data sheet, and the THERMTRIP Status Register in the BIOS and Kernel Developer’s
Guide for AMD Athlon™ 64 and AMD Opteron™ Processors, order# 26094.
Click to expand...

It doesn't specify what part of the case there, but from the mechanical / cooling design guides the center of the heatspreader is where they say to play the thermal sensor. It could mean the bottom of the CPU package, but that doesn't seem pratical (or useful) in an FCPGA design.
 
well, either way, it'd be odd if it was the TCaseMax that is the highest temp it's rated to do stock, as the rating goes down to 49c for some cores... :p
 
IMHO, tcasemax and tdp is a reasonably arbitrary look into ther performance potential of a die. Both of my best OC'ers have exhibited very high tcasemax and tdp values in comparison to the other 3-4 dual cores i have played with. Now, there are other things which tcasemax cannot account for like the durability of the integrated memory controller, so it is prohibitvely minute in its prediction at times. I would give the analog that tcasemax is like rating a turbo for a car; it gives a hypothetical indication of its performance ability. But that turbo is limited by its capacity once connected to the rest of the vehicle system (exhaust, intake, etc.).
 
fenderltd said:
Confused... what the hell does this mean? since the tcasemax is higher it should overclock slightly higher?
My x2 3800 says TCASEMAX 63c TDP 65.6Watts
Click to expand...


Mine reports 67c max.......
 
Crap.... does this mean my proc sucks for OCing?

TcaseMax = 61C
TDP = 59.4W

It actually is true... running at 2.4GHz at 1.35, but I needed to bump it to 1.45v to hit 2.5GHz stable. Haven't even been able to hit 2.6 with this POS.
 
No, as stated there are plenty of cpu's with very low maxtcase numbers and OC beautifully.
 
I still don't understand this. But just look at the big difference in TDP among these three Opteron's. The hot one is mine and it can do 2.83ghz @ stock 1.35v. And even when it is under-volted to just 1.1v (2.2ghz), it's idle temp is still @ 38c.

Optrn 180, JH9-E6, 49°C, 1.35V, 35.0W
Optrn 165, JH9-E6, 51°C, 1.35V, 45.0W - CCBWE 0551UPMW
Optrn 165, JH9-E6, 63°C, 1.35V, 105.0W - CCB1E 0608MPMW

Edit: Add stepping info. The 0551 is not a good OC'er and my 0608 is. May be this TcaseMax/TDP thing does make sense. I just don't understand why the BIG difference in TDP within the same product line.
 
raxen said:
Crap.... does this mean my proc sucks for OCing?

TcaseMax = 61C
TDP = 59.4W

It actually is true... running at 2.4GHz at 1.35, but I needed to bump it to 1.45v to hit 2.5GHz stable. Haven't even been able to hit 2.6 with this POS.
Click to expand...



That is such a bad attitude. ANY overclock beyond the manufacturer's RATED speed is a bonus, your cpu is NOT a POS. But please, DON'T rma it because it won't overclock as high as you THINK it should. I can run stock vcore to 2.4ghz or a bit more, it takes 1.45v to get it stable at 2.6, but it takes a whopping 1.6v to get it about 80% stable at 2.7 no matter what else I do....so I run it 2.6ghz 24/7, Folding, gaming, video crunches, and whatever else, it is like a BRICK HOUSE baby! :p
 
TheRapture said:
That is such a bad attitude
Click to expand...
exactly why i've come to hate things like this

"oh no, my cpu must suck because of it's TCaseMax"
"oh, i have a shitty stepping"
"blah! my ram is holding me back"
"grr, my heatsink sucks and i could get so much further with cooler temps"


truth is.. none of those really matter. cooling is the most benefitial, but i'd be surprised if someone got more than 100-150mhz from stock cooling to an awesome waterblock

</rant>
 
TheRapture said:
That is such a bad attitude. ANY overclock beyond the manufacturer's RATED speed is a bonus, your cpu is NOT a POS. But please, DON'T rma it because it won't overclock as high as you THINK it should. I can run stock vcore to 2.4ghz or a bit more, it takes 1.45v to get it stable at 2.6, but it takes a whopping 1.6v to get it about 80% stable at 2.7 no matter what else I do....so I run it 2.6ghz 24/7, Folding, gaming, video crunches, and whatever else, it is like a BRICK HOUSE baby! :p
Click to expand...

Don't worry. I'm not going to RMA it.

My CPU IMO is a POS, and that is my opinion. The new LDBHE have been able to hit 2.6 pretty easily, and compared to those, my chip is significantly worse.

I'm not ranting that I can't OC my chip. When I make a thread going "WHY CAN"T I OVERCLOCK PAST 2.5GHZ!@!!!@$@!%@#%#@" then you can yell at me. I'm merely just supporting the fact that my results have been coinciding with the theory that lower TCaseMax cannot overclock well.

In case you're wondering, here's what I've been able to achieve with the chip.

2400 - 1.350v
2450 - 1.400v
2500 - 1.450v
2600 - ??? It wasn't 1.5. Wasn't gutsy enough to try 1.55.
 
x2 3800+ BH9-E4 TCaseMax 61*C TDP 59.4
It gets 2.74 at 1.5 (maybe a hair or two more but dont wanna use beta bios to pass 275htt).
 
TheRapture said:
raxen.....no harm intended.....it's all just good spirited discussion....
Click to expand...

Sorry... I was pretty confrontational with the reply.
 
You must log in or register to reply here.
Back
Top