News for hackers. Stuff that matters.
Re:Ryzen 8000G review: An integrated GPU that can beat a gra... by zamadatix on Tuesday January 30, 2024 @10:08AM (#39190932)
Along with comparing what the others have said about comparing TDPs between vendors (or even different product lines sometimes) TDP is about multi-core draw so you'd need to also normalize the multithread performance differences between the two, not look at the single thread performance being near equal (that'll be significantly less than 65 Watts for each).

I think the 8700G is probably more efficient but it won't be anything like 65 vs 181 watt would leave one to believe.

Re:Ryzen 8000G review: An integrated GPU that can beat a gra... by marmaduke on Tuesday January 30, 2024 @10:29AM (#39191219)
CPU mark scores are 32k and 39k respectively. It's at least twice as efficient
Re:Ryzen 8000G review: An integrated GPU that can beat a gra... by zamadatix on Tuesday January 30, 2024 @02:44PM (#39194710)
Assumming:

- PL1/PL2 from Intel can be compared to TDP from AMD

- CPU Mark scores alone are the best indicator of what full utilization looks like vs another test

You'd be accurate. You can't really assume either of these things though. The better comparison to TDP from AMD is probably closer to "base power" of 125W than PL1/PL2 of 181 on the Intel side. Other benchmarks, like encoding using all cores efficiently with matrix extensions, are in the 25%-30% difference. Combine these two facts and it's probably closer to 1.5x efficient (instead of the 2.5x the raw numbers above suggest, 3x the relative improvement), but you'd have to actually measure active power doing the same amount of total work to know for sure. Even ignoring one of the facts the encoding benchmark differences are enough to guarantee it'd actually be less than twice as efficient, not at least twice as efficient.