Testing for chip nanufacturing deviances?!

[DonCornholio]

Hi there,

did some of you see the latest Video from german overclocker (der)8auer on youtube where he tests 13 Ryzen 7600 to see how much they deviate in performance and power consumption and so on? 
It's actually pretty astonishing, espcially when you compare cpus, and i thought that this is exactly the kind of thing i wished that LTT would also do on a larger scale.
I really would like to know if those deviations are found in all price ranges, how they change from generation to generation and obviously also wether amd or intel do a better job at this.
Do you guys/gals think that this would be a video you'd like to watch?
Does LTT have something like this already planned ? 
Are there already other reviews where someone did this at a larger scale? 

 

[Budget DIY]

I watched this particular one, and I felt it slightly lacking. I can't see LTT do much better then this. Other then just ordering a tray of CPU's. However, I would like to see Gamers Nexus do these tests. Because with THIS particular vid, I'll ask myself: how well is the CPU/cooler mount. Because a bad mount can impact performance. But I also wonder about how reliable the data is, that the mobo reports. Gamers Nexus would probably measure power draw from the wall, or to the mobo and whatnot. Not something I see Linus doing.

[RONOTHAN##]

3 minutes ago, DonCornholio said:

how they change from generation to generation and obviously also wether amd or intel do a better job at this.

Power differences have always been different between chips, FX chips were pretty notorious for having some that would take back a ton of power for the same voltage compared to a relatively smaller amount on another chip. The difference in chip leakage (the amount of current losses for a given frequency) from chip to chip has seemed to have gotten a lot more consistent over the years, but at the same time both sides have implemented individual CPU V/F curves, so while current draws at a given frequency have gotten more consistent, the amount of voltage a particular CPU requests is less consistent so CPUs do still vary power figures a fair bit. 

 

10 minutes ago, DonCornholio said:

I really would like to know if those deviations are found in all price ranges

They are in both low and high end ranges. There's less data available for AMD chips, but IgorsLab has done binning statistics for most of the recent CPU releases (13th gen articles linked at the end of the post). They use ASUS's SP statistic which is based on the aggressiveness of the V/F curve on a particular chip, so higher SP value means less voltage is necessary, and thus a higher SP value means it would draw less power, and could perform better depending on how the motherboard sets power/thermal limits. Higher end chips are usually binned better and therefore have a higher floor for how bad they can be, but that doesn't necessarily mean that there's not still a large range in which they can fall. 

 

Spoiler

https://www.igorslab.de/en/the-last-and-fastest-of-their-kind-intel-13th-gen-raptor-lake-binning-part-3-i9-13900ks/

https://www.igorslab.de/en/intel-13th-gen-raptor-lake-binning-auch-klein-kann-fein-teil-2-i7-13700k-kf-und-i5-13600k-kf-2/

https://www.igorslab.de/en/intel-13th-gen-raptor-lake-binning-over-500-cpus-tested-part-1-i9-13900k-and-13900kf/

 

[svmlegacy]

Silicon Lottery used to give out some numbers based on their findings... Too bad they shut down some time ago.

 

Some of the info they used to give: https://www.tomshardware.com/news/amd-ryzen-3000-cpus-silicon-lottery-binning,40157.html

[freeagent]

A lot of the time you can see this on overclockers forums. I used to get jealous sometimes