Are Review Sample CPUs Cherry-Picked? $H!T Viewers Say 1

Linus Tech Tips ·Linus Tech Tips ·2018-05-06 · 1,254 words · ~6 min read
Floatplane YouTube

Transcript

JSON SRT VTT 107
0:00 engineering samples or review samples
0:03 they're cherry picked so that the media outlets
0:07 like Linus tech tips will see inflated performance and overclocking numbers so
0:12 that you yes you jerry will get suckered
0:15 into buying a new CPU every year right
0:19 well that's what people keep telling us and
0:23 this common knowledge has been around for years now so today in our inaugural
0:29 installment of [ __ ] viewers say we are going to either prove that there
0:35 is something to this or debunk this myth once and for all
0:42 thanks to zotac by the way for bringing us today's video the mech one pc is an
0:47 ultra slim desktop built for gaming featuring a sleek robotic themed design
0:52 check it out at the link below to learn more
1:02 so here's how we decided to run our test we grabbed our sample core i5 8400 and
1:09 not to play favorites our sample ryzen 5
1:12 1600x then we went to the local nci
1:24 sorry memory express to buy retail chips of the same to
1:28 compare them against for your peace of mind we will upload the entire uncut
1:35 clip of us buying the chips and running
1:38 our tests on them we're gonna have that linked below anyway let's talk about why
1:43 it might make sense or might not for a company to seed reviewers a superior
1:49 sample now the obvious answer we hear all the
1:52 time is that they would get articles boasting of better performance numbers
1:57 or a better overall user experience in
2:00 that initial oh so important wave of
2:03 launch videos and articles enticing
2:06 consumers to open up their wallets all
2:10 while not actually having to provide that level of performance or polish at
2:15 retail but conspiracy theories aside would this
2:19 actually make business sense imagine this you've got a lemonade stand where
2:24 you sell the freshest tastiest lemonade on the block
2:28 on day one and then you swap it out with
2:31 lemon pledge now a lot of people are gonna show up to
2:35 buy on the second day but as soon as your reputation for selling what is
2:39 literally furniture polish spreads the kids across the street who were selling
2:44 cheap powdered lemonade the whole time are gonna take the bulk of the market
2:48 share now we're not saying this has never
2:51 happened think back to watch dogs and no man's sky at e3 both ubisoft and sony
2:57 respectively showed off beautifully complex worlds with tons of graphical
3:01 detail and then both of them ended up being
3:05 significantly simplified upon release
3:08 now while in some industries it is possible to get away with this kind of
3:11 bs in general the negative pr has the
3:15 potential to completely stall future sales so it is not a risk to be taken
3:20 lightly and in electronics the up to
3:24 multi-billion dollar investment that goes into designing and fabricating a
3:29 processor design makes this kind of switcheroo an economic impossibility
3:35 like it's not a different paint job so then let's look at one of the other
3:39 theories maybe they're just taking golden samples off of what is otherwise
3:45 the regular production line now that one is much more likely
3:50 but it doesn't really hold up to closer scrutiny either from my experience and
3:56 others earlier cpus are produced at the tail
4:00 end of the manufacturing learning process when viable yields are lower and
4:06 compared to the more mature units that come later
4:09 can actually require more voltage to
4:12 maintain a specific clock speed not to mention that they might have
4:15 poorer memory compatibility or might output more heat at the same performance
4:20 levels even the couple of weeks between early
4:24 sample shipping out and retail availability can result in improvements
4:30 now there have been and there always will be exceptions to this but it is far
4:37 more likely that we are dealing with the same silicon lottery as any other CPU
4:42 because you got to remember even on the same wafer no two processors will have
4:47 identical characteristics so if it doesn't make any sense why do people
4:52 still think it's happening most of the time as far as we can tell
4:56 it just seems to be a fanboy argument justifying why someone disagrees with
5:01 the results being published by the dishonest
5:04 media and frankly we're just so tired of hearing
5:08 it that we not once but twice
5:12 the first time we bunged up the footage so we had to start over including buying
5:16 two more cpus Anthony
5:20 not once but twice went out and bought
5:23 these cpus so we could show you this
5:28 tldr for gaming we found basically no
5:31 variance in frame rates even in directx 12 mode while synthetics actually
5:36 favored our retail chips as for productivity tests on the Intel
5:40 side our chips traded blows back and forth but it's not a clear-cut victory
5:45 for either side and we put this down to normal chip or even run to run variants
5:49 due to the dynamic and reactive nature of turbo boost and then as for team red
5:54 it was pretty much even across the board so for better or for worse AMD appears
6:00 to be shipping exactly the same product today that they were in april of last
6:05 year and then to back that up we even tried overclocking both of our AMD chips
6:10 our Intel ones were locked but both of our AMD chips hit identical clocks at
6:15 identical voltages and identical temperatures they even crashed at
6:18 identical settings so in summary then
6:23 this conspiracy theory is busted for the
6:26 two major CPU players and while nothing that we've said today
6:30 is going to prevent people from making these arguments in the future what i
6:34 hope is that it has at least armed the rest of you with the ammunition that you
6:38 need to shut them down once and for all
6:42 it doesn't make business sense it doesn't make engineering sense it just
6:46 doesn't make sense
6:50 but you know what does make sense switching to ting ting is the mobile
6:54 carrier that's focused on customer service and customer satisfaction when
6:58 you call ting if you have an issue for whatever reason you do not speak to a
7:02 robot you get put through directly to a person and you don't pay extra for the
7:07 privilege in fact the average ting bill is only 23 bucks a month per device
7:11 because of the way they bill you pay only for what you use now if you're
7:15 stuck in a contract and you switch to tank they'll cover 25 of your
7:19 cancellation fee up to 75 bucks and if
7:22 you use our link Linus.ting.com to try out their savings
7:26 calculator and see if you'll save money you can actually get another 25 bucks in
7:30 service credit or towards a new device so go try it out
7:35 so thanks for watching guys if this video sucked you know what to do right
7:38 that one yeah but it was awesome get subscribed hit that like button or check
7:42 out the link to where to buy the stuff we featured in the video description
7:45 maybe literally these ones maybe we'll put them on ebay i don't know also link
7:48 down there is our merch store which has cool shirts like this one and our community forum which you should totally
7:53 join we have so many processors now