Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Are all CPU's created equal?

C

circuitt

macrumors member
Original poster
Nov 3, 2016
88
11
My 2015 15" MBP 2.5 repeatedly (15 times in a row) gets over 13800-13900+ Geekbench score Multi core.
Looking at the score chart, it's running as if it's a 2.8. Are all CPU's created equal ? regardless of what MHZ stamped on the chip? Back when I was building PC's (overclocking) chips varied chip by chip on how high they could be oc'ed while being stable.
 
Last edited:
Well, different CPUs (same chip spec) can have different thermal characteristics, based on impurities and chance. E.g. of two allegedly identical CPUs running at the same frequency, one can be substantially cooler. But two instances of the same chip at the same frequency should actually show very similar performance.

What are you RAM scores?

Edit: or even better, can you post the entire benchmark (including the detailed results)?
 
IMG_1795.PNG

I'm at work so can't run it now, but I'll do it again tonight when I get home.
Here's a screen shot.





leman said:
Well, different CPUs (same chip spec) can have different thermal characteristics, based on impurities and chance. E.g. of two allegedly identical CPUs running at the same frequency, one can be substantially cooler. But two instances of the same chip at the same frequency should actually show very similar performance.

What are you RAM scores?

Edit: or even better, can you post the entire benchmark (including the detailed results)?
Click to expand...
 
cmaier said:
Huh? What impurities? What chance?
Click to expand...

Impurities in the silicon crystal the chip is made from. Although 99.9999999% pure silicon, that 0.0000001% is an impurity that can affect the electrical properties at baseline. Defects in the crystal structure itself (might also have an effect at the CD of modern chips. CD = critical dimension :https://www.quora.com/What-is-the-d...nm-and-critical-dimension-in-photolithography
And although the fabrication itself is ridiculously precise, there can still be unintentional material placed in the wrong place if everything is not working perfectly in the manufacturing process -> these are also impurities that affect electrical properties.
 
Is there a CPUZ equivalent for OSX, perhaps I can monitor the CPU's actual MHZ in real time?
 
The Mercurian said:
Impurities in the silicon crystal the chip is made from. Although 99.9999999% pure silicon, that 0.0000001% is an impurity that can affect the electrical properties at baseline. Defects in the crystal structure itself (might also have an effect at the CD of modern chips. CD = critical dimension :https://www.quora.com/What-is-the-d...nm-and-critical-dimension-in-photolithography
And although the fabrication itself is ridiculously precise, there can still be unintentional material placed in the wrong place if everything is not working perfectly in the manufacturing process -> these are also impurities that affect electrical properties.
Click to expand...

I have a Ph.D. in electrical engineering/semiconductor physics, designed microprocessors for 15 years, so I understand how doping works. But these variations result in the Schmoo distribution and are taken care of in the binning process. There will not be any substantial difference in heat generated between two identical processers, and certainly not due to any variation in "impurities" or crystal defects.
 
  • Like
Reactions: T'hain Esh Kelch
cmaier said:
I have a Ph.D. in electrical engineering/semiconductor physics, designed microprocessors for 15 years, so I understand how doping works. But these variations result in the Schmoo distribution and are taken care of in the binning process. There will not be any substantial difference in heat generated between two identical processers, and certainly not due to any variation in "impurities" or crystal defects.
Click to expand...

I was obviously referring to the manufactured chips before the binning process. And it depends on how thoroughly they bin the chips afterwards. Its fairly common to have two CPUs that are sold as the same model, but show slightly different thermal characteristics and overclocking potential.
 
leman said:
I was obviously referring to the manufactured chips before the binning process. And it depends on how thoroughly they bin the chips afterwards. Its fairly common to have two CPUs that are sold as the same model, but show slightly different thermal characteristics and overclocking potential.
Click to expand...

Except you were referring to two sold chips, one being "substantially cooler" which is very different than "slightly different thermal characteristics."
 
so in layman terms, two chips spec'ed for the same MHZ, one could turbo boost higher than another?
As in the single core clock speed seems to be on par.
 
cmaier said:
I have a Ph.D. in electrical engineering/semiconductor physics, designed microprocessors for 15 years, so I understand how doping works. But these variations result in the Schmoo distribution and are taken care of in the binning process. There will not be any substantial difference in heat generated between two identical processers, and certainly not due to any variation in "impurities" or crystal defects.
Click to expand...

I don't think we are disagreeing here - I never mentioned heat, I was only answering the question "what are impurities?".
 
I am quite sure the amount of thermal paste has a much bigger difference on thermal levels than CPU manufacturing differences, when comparing different machines.
 
  • Like
Reactions: cmaier
T'hain Esh Kelch said:
I am quite sure the amount of thermal paste has a much bigger difference on thermal levels than CPU manufacturing differences, when comparing different machines.
Click to expand...

That is possible but more likely that it is every chip gets its own VID or default voltage it runs at, since the power drawn (thus heat produced) is a square of the voltage applied the difference in heat can add up quickly, or combination of both. Then you could add in case that covers the cpu and it not being nice and flat between the two processors, the thermal paste below that not applied consistently ... the list can be long. Although sometimes now I am thinking about that last part they go with bare cpu no cover in laptops. Or in Apples case with the dual 2009 MP in there to mess people about on easy upgrade possibility.
 
MacUser2525 said:
That is possible but more likely that it is every chip gets its own VID or default voltage it runs at, since the power drawn (thus heat produced) is a square of the voltage applied the difference in heat can add up quickly, or combination of both. Then you could add in case that covers the cpu and it not being nice and flat between the two processors, the thermal paste below that not applied consistently ... the list can be long. Although sometimes now I am thinking about that last part they go with bare cpu no cover in laptops. Or in Apples case with the dual 2009 MP in there to mess people about on easy upgrade possibility.
Click to expand...

No, they run at the same voltage. They have voltage regulators to ensure that. They have plls lock the frequency. So the only dynamic power factor that varies is the switching capacitance, but that varies much more based on what the processor is doing than on physical differences between the processors.
 
cmaier said:
No, they run at the same voltage. They have voltage regulators to ensure that. They have plls lock the frequency. So the only dynamic power factor that varies is the switching capacitance, but that varies much more based on what the processor is doing than on physical differences between the processors.
Click to expand...

All computer boards have voltage regulators still does not change the fact that the chips that go into them have a default VID set at the factory. If you ever have the chance around a PC or three with the same motherboard/cpu model in them hit the delete key go into BIOS and you will see the VID is going to be different. Then again in the highly unlikely event that Apple is paying some monster cash to Intel for the chips they get off the same assembly line every one else gets them from for the exact binning of identical VID chips then it could be possible.
 
MacUser2525 said:
All computer boards have voltage regulators still does not change the fact that the chips that go into them have a default VID set at the factory. If you ever have the chance around a PC or three with the same motherboard/cpu model in them hit the delete key go into BIOS and you will see the VID is going to be different. Then again in the highly unlikely event that Apple is paying some monster cash to Intel for the chips they get off the same assembly line every one else gets them from for the exact binning of identical VID chips then it could be possible.
Click to expand...

No, the CHIP has multiple voltage regulators that locks the on-chip VDD-VSS to whatever the cpu needs. Any VID variation is a consequence of the binning process - a chip on the left of the schmoo may be binned above its speed class and demand a higher voltage to compensate. It's rare when this is a significant factor, but more importantly devices that have this issue will also have lower static power dissipation and the higher voltage will reduce leakage, which more than compensates for any power difference caused by the voltage component of dynamic power dissipation.

As a previous poster stated, if one is seeing a substantial heat difference under identical workloads, it has nothing to do with any supposed 'impurities' difference, but is almost certainly caused by packaging differences (different amount or distribution of thermal paste, etc.)
 
cmaier said:
As a previous poster stated, if one is seeing a substantial heat difference under identical workloads, it has nothing to do with any supposed 'impurities' difference, but is almost certainly caused by packaging differences (different amount or distribution of thermal paste, etc.)
Click to expand...
Yeah, but sometimes some chips really do run cooler. At least on the desktop side back when I was into overclocking and all that, I would see 10ºC differences with two supposedly identical chips, even after reseating the heatsink multiple times. The cooler running ones typically had lower default VIDs.

Now in the case of the Geekbench scores, I suspect that more comes down to some people not closing all background processes and some people running on battery pulling the averages down.
 
terraphantm said:
Yeah, but sometimes some chips really do run cooler. At least on the desktop side back when I was into overclocking and all that, I would see 10ºC differences with two supposedly identical chips, even after reseating the heatsink multiple times. The cooler running ones typically had lower default VIDs.

Now in the case of the Geekbench scores, I suspect that more comes down to some people not closing all background processes and some people running on battery pulling the averages down.
Click to expand...

In the "old" days the schmoo's were extreme, and often times they would actually DOWN bin chips frequently for the desktop. Those days are long since past.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back