@lilhasselhoffer
i dont care to read between the lines, type something or dont.
otherwise its assumptions i have to make, likely to be incorrect.
@
DirtyDingusMcgee
but then you usually dont care much about speed (above sweetspot), and more about things like EEC.
You...are...not...reading...so...why....should...I...care?
Let me be more abrupt, you are incapable of understanding words, and read your own alternative meaning into them. Cool, you just get put on my "bad faith actor" list internally, and I let you have the win you want to have. It's stupid, but the much stupider thing would be for me to try and make you understand that words have a meaning, so fine. I'm not a masochist, so have that wonderful phyric victory. Waldorf, the puppet/muppet, would me proud.
I'll have to just laugh at what I think is an appropriate exchange between Statler and Woldorf.
S: Is your hearing aid fixed?
W: No.
S: Then how do you know what I'm saying?
W: I don't!
S: Oh. (does double take)
Also...let's end with a math check. What is 2500? Well, over the course of Q4 2019 to Q4 2024, the average PC sales is about 70 million (
source). I'm going to give you the benefit of doubt, and suggest 2500 per quarter is what you pulled from thin air as "statistically significant." That's about 0.0036% of the entirety of all PC sales...and making any conclusion as 100% factual from that is silly.
Why then suggest about 36 as a discussion point? That seems like a complete load...until you consider most people building that many are not building a bespoke product. Too few to be volume, too many to be niche, so the most likely situation here is outfitting a bunch of similar PCs for a deployment. That means the same hardware, purchased in bulk, from likely stable sources. That's still not a great statistical slice, but it could give you a view on some aspects worth talking about.
It's almost like having to deal with this crap on a regular basis gives you an appreciation for how little some things are controlled, and how much goes into what sound like it might not be complex. I mean, a QVL tests memory sticks on a platform that introduces variation, with a CPU that introduces variation...thus meaning a QVL is not a guarantee but proof they checked some units with some hardware. Given this isn't exactly full blown six-sigma level analysis, you only get what you pay for. Motherboards don't cost a grand...mostly...so the expectation is equally low in that the QVL is a check but not a warranty or guarantee. That lack of explicit or implied performance should be enough to tell a reasonable person that it is not anything special, which is something which might escape you if the difference between anecdote being personal experience and fact being something borne out of statistically significant samples is something you want to fight over.
Alternatively, 2500 over 5 years is about 0.00018% of the sample...or 1 in 560,000 units. That's...wow. It's like saying grabbing a cheap Bestbuy floor model will give you insight into a Falcon Northwest PC...which should let you know how outlandish that is. ShrimpBrime got it right in saying that a few million examples would definitely give us a clue...but your umbrage is kinda fun to see. Not useful, but fun in that sort of way a car wreck is interesting.
