Performance/Voltage Index

[robberbaron]

Either way, I doubt most people here know so much about the inner workings of the microprocessor's components or the manufacture process to make any claim as to this having anything to do with the "quality" of the transistors.

If a transistor is able to switch at a high speed but not require much voltage to do so, I would say that such a transistor is of "good" quality. It's why FX-55's are able to do such high speeds around stock voltage. They are made with the best dies and thus best transistors.

 

[zeebs]

Touche! Can't say I am fully sold but sounds logical to some extent.

 

[(cf)Eclipse]

sounds logical eh? this might be a bit in depth, but...

basically, a cpu can run run stable up until the speed of the slower part. just like how a chain is only as strong as it's weakest link. thus one cpu can be astounding, aside from a small flaw in the cache area (quite typical given the size of the cache). any defects/imperfections will lower the overall speed of the cpu. read this for more info.
typically, a part of the core that is of lessor quality needs more voltage to switch at a higher rate without failing, given a constant temperature. as the temperature rises, the transistor performance decreases due to higher resistance and gate leakage (which adds noise, and further increases heat)
i believe that it is due to this temperature and gate leakage increase from higher voltages that it appears that the VID is higher at slightly lower speeds and voltages.

moral of the story? more mhz you can do at a given speed, the better the slowest link in your cpu is, thus having 'higher quality' transistors in the sense that the defects are minor/non-existant

 

[Zinn]

Hmm my Fauxhammer 3500 is getting 2210 stable at 1.264... that's a decent 1748. My PSU wigs out when I overclock (can OC cpu or memory but not both) so I figure for now I'll undervolt and enjoy 32 idle/39 load temps.

 

[zeebs]

^^^ Nice explanation above. I am familiar with all that stuff, guess I just needed to see it again. Nice article too. Microprocessors really are amazing!

 

[el rolio]

well let me throw mine in there:

3200+ 939
2400/1.34 (this thing fluctuates between 1.33 and 1.35)
= 1791

thats about it folks. havent tried to go higher cuz i just dont have ze parts i suppose i could run a divider and check it out tho...

 

[Frallan]

sounds logical eh? this might be a bit in depth, but...

basically, a cpu can run run stable up until the speed of the slower part. just like how a chain is only as strong as it's weakest link. thus one cpu can be astounding, aside from a small flaw in the cache area (quite typical given the size of the cache). any defects/imperfections will lower the overall speed of the cpu. read this for more info.
typically, a part of the core that is of lessor quality needs more voltage to switch at a higher rate without failing, given a constant temperature. as the temperature rises, the transistor performance decreases due to higher resistance and gate leakage (which adds noise, and further increases heat)
i believe that it is due to this temperature and gate leakage increase from higher voltages that it appears that the VID is higher at slightly lower speeds and voltages.

moral of the story? more mhz you can do at a given speed, the better the slowest link in your cpu is, thus having 'higher quality' transistors in the sense that the defects are minor/non-existant

OR IN MY CASE A SMALL FLAW IN THE SECOND CHANNEL PART OF THE MEMCONTROLLER ok sry for screaming just venting a bit here....

 

[(cf)Eclipse]

yup, i have that problem. can only get to 234mhz reliably with two sticks in. any higher =

 

[centvalny]

more voltage + more cooling = more mhz..