QX9100 outruns 920XM

I've tried to explain it twice now and I really don't know how to explain it better, TDP is a specification/rating, not the power that the CPU is running at.

Changing the TDP limits, as done by Throttlestop, changes when the processor throttles turbo bins. The normal way for them to work is by turning off turbo when the TDP is exceeded. By increasing the limits turbo throttling happens at a higher power limit and the CPU could be run higher than it's rated TDP. You seem to be suggesting that I don't believe TDP could be exceeded yet in both previous posts I already stated that it can.

Sure, there are times when TDP might be exceeded for ~20us (~20 millionths of a second) but in this case it's effects are AFAIK considered inconsequential. The older C2D use to run IDA on both cores for these short times to lessen the effects of switching VID/FID and is probably why the fulltime IDA bug can be made to work, by which all cores can use the IDA frequency 100% of the time. However this does not IMO make TDP an average power. FYI here is a quote from the i7 920xm and sundry Intel datasheet.


And of course it could be easily exceeded by overclocking.

As for the older AMD processors burning up AFAIK that was because of thermal runaway due to inadequate protection.

Thanks moral. I don't know of anyone personally overvolting a laptop i7 but since voltage mods can be done on the VRM it should not matter what the CPU is. Perhaps cutting tracks isn't received well

I don't have an i7 mobile but do have an i7 desktop. On my desktop CPU current and Vcore can be read by software to likely produce a truer CPU power reading than having to use a kill-o-watt. Even without any power reading a good idea can also be had by the temperature relatively and a good guess made if familiar with the cooling solution ie °C/W. Temperature measuring by PECI shows the hottest core of 4 cores. Note that the TDP is 95W and that the desktop mainboard hardware is wired to tell the CPU it's using much less power than it really is so no problems with turbo throttling when exceeding TDP, in this case by as much as ~50%. Always amazed at how much current goes through these CPUs.



If people want to believe TDP goes up and down with CPU usage instead of being a specification/rating then so be it, I tried to explain but seems I failed miserably. I guess I'm just not so good at explaining things. Better if I STFU and let you get on with it.

 

I would quote myself...but that should not be necessary.....i stated that the TDP is the intel spec.....look at the top of page 10

 

Oh, no, we both understand that. This whole thing started from scook9 making the mistake of using TDP when he meant thermal load. What we've been arguing afterwards, I think, is more what you might call the "usefulness" of TDP as a number.

 

I have several times voiced my opinion on the whole TDP conversation. I don’t care for it much. With the core line of CPU’s it was a little easier to understand the operating point of the processor, you know how many cores it has, and what frequency it’s running at. With the i series you no longer even have that. Intel does not specify the number of cores or at what frequency they are running at. The spec is contradictory in it specifies TDP as a maximum, then in the next sentence, that TDP is not the worst case. (I am accustomed to seeing on spec sheets the absolute worst case on IC’s as a min/max kind of thing that covers voltage, power dissipation, temperature, etc..) Also it is totally ambiguous what constitutes realistic worst case and synthetic worst case applications. Intel also does not give any types of thermal resistance figures for die to case (in this case the sheathing they put over the core) case to ambient. So really in the end, like the spec says design to the TDP for heatsink sizing. There has to be a better way to specify the power a processor uses and under what exact conditions is operating in.

 

where did you find that nifty tdp/tdc monitor??