Tweet
"The link between the frequency and the powerconsumption of a CPU"
Hi everyone,
New member, here! I'm trying to help out an advisee of mine, Daniel, an international high school student doing an "Extended Essay" on the topic listed above. If you'd be willing to help out, that would be great!! Please email me at [email protected]
Daniel is 16 years old, and very interested (and wicked smart ) in computer engineering. Last year he built a pretty impressive DIY gaming computer by himself. Could you please offer tips / suggestions / direction?
I've emailed a friend at Google, who has replied thus:
My bigger concern on this would be that it's not as cut and dry as Daniel might expect. Faster CPUs with less to do may not always use more power than slower CPUs where that same load is closer to the capacity limit. Also, every generation of CPU includes a variety of power-saving technology that means that it's using less power per cycle. Also modern computers constantly reconfigure CPUs at varying different speed based on workload. So it might be a far more complicated matter than he expects. But the data should be there - there's entire industries that study this daily. Some of them are tight-lipped about it, others aren't. Not all the data is going to be as easy to find, but it should be there. And with some time and knowledge of Linux, and benchmarking, it may not be that hard to come up with original data
He just started his research:
I have found a couple of papers on cpu power efficiency, and I am currently reading through them for useful information. I have also been thinking about the experiment, and I have decided to do 2 tests for the performance: one multi-threaded and a single threaded. I will use the performance of the cpu in the default setup as performance "1.00", then I'll calculate the performance of the other setups in comparison to this. I continuing to look for research, mostly focused on how a transistor works.
If you could offer any tips or would be willing to help, we'd both be incredibly grateful.
New member, here! I'm trying to help out an advisee of mine, Daniel, an international high school student doing an "Extended Essay" on the topic listed above. If you'd be willing to help out, that would be great!! Please email me at [email protected]
Daniel is 16 years old, and very interested (and wicked smart ) in computer engineering. Last year he built a pretty impressive DIY gaming computer by himself. Could you please offer tips / suggestions / direction?
I've emailed a friend at Google, who has replied thus:
My bigger concern on this would be that it's not as cut and dry as Daniel might expect. Faster CPUs with less to do may not always use more power than slower CPUs where that same load is closer to the capacity limit. Also, every generation of CPU includes a variety of power-saving technology that means that it's using less power per cycle. Also modern computers constantly reconfigure CPUs at varying different speed based on workload. So it might be a far more complicated matter than he expects. But the data should be there - there's entire industries that study this daily. Some of them are tight-lipped about it, others aren't. Not all the data is going to be as easy to find, but it should be there. And with some time and knowledge of Linux, and benchmarking, it may not be that hard to come up with original data
He just started his research:
I have found a couple of papers on cpu power efficiency, and I am currently reading through them for useful information. I have also been thinking about the experiment, and I have decided to do 2 tests for the performance: one multi-threaded and a single threaded. I will use the performance of the cpu in the default setup as performance "1.00", then I'll calculate the performance of the other setups in comparison to this. I continuing to look for research, mostly focused on how a transistor works.
If you could offer any tips or would be willing to help, we'd both be incredibly grateful.