I
I've redone the power numbers on the MSI MPG Z390 Gaming Edge AC. There was an issue with the ASRock Z370 motherboard supplying 1.47 volts at load. Benchmarking seems unaffected, but power numbers look a bit better, around 166W for the 9900K and 123W for the 9700K.
Looking at the benched power consumption for various CPU:s only causes confusion.
I assume a correct approach to measure the TDP would to bench all CPU:s when they have their turbo boosts disabled, so that they run on base clock all the time, with adequate cooling so that no throttling occurs.
Yet I only see a mix of strange results, where a i9 9900K CPU draws 168W, where a i7 8086K draws 94W, and where i7-8700K draws 150W (despite being basically the same CPU as the i7 8086K). All of them should basically have the same TDP. And so on.
I think the correct way to bench it is to put a big enough cooler on it and just run it at stock with no setting changes.
You shouldn't handicap it because then users will get unexpected results when they use the cpu.
Yeah, but I am pretty sure that turbo boost is not part of the TDP formula. A CPU will boost higher than the base clocks for some time if the cooling is good enough, until further thresholds is reached in temperatures and power consumption. But the advertised TDP is, to my understanding, only valid for base clocks. Or who knows? I guess TDP isn't even measured the same between AMD and Intel.
Why should you modify your cpu so it matches the TDP number? What purpose would that achieve?
You should be able to know what power consumption you have out of the box without any setting changes so you can choose a corresponding cooler and psu.
TDP number is just a number they put out. It's not really useful now that they call the 9900k a "130" watt TDP. This is why you should only rely on third party testing.
I am curious under what conditions the TDP can verified. Because when looking at the power consumption benchmarks from various reviewers, it just gets confusing, as I mentioned in my first post. The CPU manufactures can't just say "it's a number". It must be verifiable, and have some meaning. Otherwise it's utterly pointless to everyone.
I am curious under what conditions the TDP can verified. Because when looking at the power consumption benchmarks from various reviewers, it just gets confusing, as I mentioned in my first post. The CPU manufactures can't just say "it's a number". It must be verifiable, and have some meaning. Otherwise it's utterly pointless to everyone.
Ah I understand now. It used to verifiable. Previously, before 8700K, most of intel's CPU would run at their TDP at a demanding load like cinebench. Nowadays, it's just a number.
Here is tom's hardware review of the 7700K. Note that it runs pretty much at 95W on the demanding loop. On a higher demand one, in the recently reviewed 9900K, the 7700K uses 115W on the AVX torture test, pretty much worse case scenario. But the 9900K uses about 200W. The 7700K was tested at stock with no settings modifications.
So yeah. It's literally just a number from intel nowadays. It used to be pretty accurate. A 95W CPU will use about 95W on demanding loads.
I theorize that intel doesn't want to look less efficient than AMD, but at the same time wants to completely overtake AMD in performance. So they clocked their 9900K up the wall but still rate it as "95W" TDP to look good. After all, it's just a "design" number. Not an actual specification.
So yeah, it's pointless lol. NVidia on the other hand, is pretty accurate with their TDP if you buy close to stock cards.
Read this page which might be educational about Intel's TDP.
https://www.anandtech.com/show/13400/intel-9th-gen-core-i9-9900k-i7-9700k-i5-9600k-review/21
Oh it does dictate how big the cooler needs to be, to run it at base clocks. That's like cars that only achieve their specced amount of performance with race quality fuel and slick tires.In any case intel's TDP is no longer relevant for these newer gens. Thermal Design Power that doesn't dictate how big of a cooler you need? Now that is useless.