How did AMD go from being the poor man's choice to dominating the CPU market?
>>107769958idk man it happened 35 years ago
>>107769958The competition killed itself, that's how, AMD is still shit, just better than the non existant competition.
>>107769958>zoomer wasn't alive in the Athlon XP era
>>107769958TSMC
>>107770120I member, but as soon as the c2duo started rolling out people switched pretty quickly.As for op's question, no one gave a shit before the x3d chips came out recently.
>>107769958X3D.so they just used the newly developed chip stacking offered by >>107770402
>>107769958Phememe I & II and Bulldozer flopped so hard that it was a do or die situation for them.>>107770402Zen and Zen+ didn't even use TSMC.
>>107769958AMD used to dominate early 2000s with the Athlon XP's and Athlon 64's vs Pentium 4. And Athlon X2 vs Pentium D. However Intel beat AMD for over a decade once Core 2 Duo and Quads came out.Reason why AMD won was because Intel got lazy. For 7 straight generations the best consumer CPU's were 4C8T. If you wanted 8 cores it was $999 for an i7 6900K or $330 for a Ryzen 7 1700. Or you wanted 6 cores well then the Ryzen 5 1600 was $219 while the i7 6800K was $434.There's reason why the next generation Intel released after Ryzen 1000 had more cores. For 7 straight generations Intel did not increase core counts but coincidence the generation after Ryzen 1000 came out Intel released CPU's with more cores. If Intel had affordable 6 and 8 core CPU's when Ryzen 1000 came out it would have flopped.
>>107770767Zen and Zen+ also werent that good, that started from Zen2
I recall the only real CPU pre ryzen that was worth a damn was the 8350 and uh... 6300 for lower end builds? Ryzen 1 was ok but by the time Ryzen 2 series came out shit was tight.
>>107769958socio-economic factorswait, what was the question?
last amd chip was an atlhon x2.i remember considering them later, but you were required to install some cpu 'patch' to run adobe programs at the time. 'into the trash it goes'intel has been fucking up the last few gens.still, the longevity of amd chips have always been questionable, and i prefer to keep my systems over a decade.
>>107770865Zen and Zen+ had the advantage of being the cheapest way to get a CPU with 6 or 8 cores. If Ryzen 1000 never released Intel i7's still would be 4C8T CPU's and i5's 4C4T. Intel was so stupid. If they responded to Ryzen 1000 by making 8th gen Intel basically the 10th gen by having those CPU's have those core counts they wouldn't have lost so bad. Then they would have caught up with the core counts and they still had better single core.By the time Intel caught up with Ryzen in core counts Ryzen had caught up with Intel in single core performance with Ryzen 3000.
>>107770767Phenom II was great and they even had an affordable 6 core that came out a year later, and supported both DDR2/3 for extra flexibility.
>>107769958By being less jewish than IntelAviv.
>>107771283Yep. Back in like 2009 a Phenom II was a great buy. You could either get a Phenom II X3 or a Core 2 Duo. And the Phenom II X3 aged far better than the Core 2 Duo due to the extra cores. Non hyperthreadded dual cores did not age well at all.And if you won silicone lottery you could even unlock the Phenom II X3 into a Phenom II X4.
>>107771316*back in 2009 Phenom II was a great buy as a budget gamer
>>107769958intel forgot to keep trying once they reached the top
>>107769958mfw this thread
>>107769958>>107770120/thread
>>107769958better last level cachehigher bandwidth, lower latency, (optionally) larger
>>107769958Intel died.
>>107770402Arrow Lake is TSMC built and yet it sucks dick. worse than its rust prone predecessor in some work and most games
>Intel is on top>Rather than add more cores to CPU's decide that nobody needs more than 4>After 6 years AMD releases Ryzen. While worse single core performance it offers more than 4 cores for 1/3rd of the price of Intel's HEDT offerings>Intel drip feeds more cores rather than instantly matching Ryzen>Ryzen catches up on Intel in single core speeds with Ryzen 3000>Only after Ryzen 3000 does Intel's CPU's match AMD's core countsAnd that's how Intel lost
>>107769958By copying, as the Chinese always do.Intel (American company) innovatedAMD (Chinese CEO) imitated
>>107773622What did Intel innovate from 2010 to 2017? How 4 cores is enough and 2 cores is enough for laptops?
>>107770877PhenomII X2 let you unlock two more cores for free. While it ran hot it was good enough for gayman.At work we built a server farm of Athlon X4s as the only limit was the amount of RAM when shitting out webpages.The one drawback of these chips were media applications tuned to Intel, SSR 4.2 in particular.
>>107773622I would argue that Intel has a worse track record of innovation than AMD overall, if you look at their entire histories
>>107769958intel fumbled every major opportunity for the last 15 years, basically havent done anything significantly better than their competition since zen2 launched, fumbled gpus (larrabee)fumbled memory (optane)fumbled AI (ponte vecchio)fumbled cryptoetc etc.
>>107769958The person in the lead decided it was far enough ahead that it can take a break. It is the classic tortoise and the hare story.
>>107774392bingo
>>107773983Intel bet the entire farm on 10nm being seamless a decade ago and Cannon Lake -> Ice Lake roadway. None of the materialized until much later and AMD (CPU)/TSMC (foundry) simply outclassed them. Intel never had recovered since.
It's going to take every ounce of my being not to fomo into the 9950x3d2, 192mb cache on a 9950x3d sounds so coolI guess I'll need a 420mm arctic freezer iii though
>>107773622Wrong, Intel scoffed at EUV and pretend DUV still had life in it. They never imagine being outclassed by another foundry. TSMC gambled on EUV taking off and got massive capital from Apple. Intel still pretend that chiplets/modules were still unnecessary and monolithic chips were the way to go. They decided to mess around with side gigs that led to nowhere and neglect their core business. Intel is just paying the price for this.
>>107772523smartest nigga alive
>>107769958intel is better for laptopshybrid architecture is finally getting goodMeteor Lake is amazing, comparable to arm in battery life
>>107769958Intel got lazy after making Sandy Bridge.
>>107769958techtuber marketing
>>107769958Intel rested on their laurelsEveryone got sick of their shit and moved on to greener pastures (i.e. apple).AMD did its own thing as well and ended up winning by default.AMD is having its own Intel moment with Radeon thougbeit. Reduced to an irrelevant segment only used by a small niche of loyal gaymers and only has its console agreements keeping the division afloat.
>>107769958by innovating
>>107769958>>107772523it took 20 fucking replies for someone to mention jim keller the jew killerthis site is dead
>>107769958when intel decided to install spy chips for the surveillance community and take orders from shitreal.
>>107775968This, it used to be at Intel every other generation the node got shrank. After haswell that ended we just got the 14nm+++++ nonsense and ever increasing TDP for several generations.AMD was smart to not be in the business of both designing silicon and fabbing their silicon. This freed AMD to use what ever node was the best at the time.
>>107769958>BCMBig Cache Memory.
>>107772523>>107772571>>107776329>it took 20 fucking replies for someone to mention jim keller the jew killerAnd the funny part is that he's responsible for both times AMD kicked Intel directly in the dick; with both AMD64 and Zen.
>>107773622>Intel (American company)
>>107777943>Intel hires him because AMD is out innovating them>develops the Royal Core project to get rid of hyperthreading which were single thread, single core but could split into dual threads when needed>Pat Gelsinger cancelled the Royal Core project
>>107769958intel got too cocky
>>107769958a tradition of engineering excellence.semi unironically sotheyve been created with the assistance of intel, as a result of an anti trust lawsuitand theres been a couple times theyve been at the top cpu-wiseduring the pentium era with their k6during the pentium 4 era with their barton 2500+and when they introduced the amd-64 specification which we call today x86-64>poorfag solutionpoorfags have to buy good shit because they dont have the monies to make bad purchases
>>107769958By betting the company on a ground-up redesign that happened to provide better performance at a lower price. It worked out for them.I love AMD more than my children, and have since the 386DX/40 - but I'm under absolutely no illusions that, for every timeline like our one, there are 50 where what remains of AMD is three men in a basement of Samsung or Qualcomm's New Dehli regional HQ.
>>107775921I don't think that would have saved them, 10nm was shit for sure, but they backported icelake to 14nm as rocket lake which was underwhelming even to the still skylake derived comet lake. . if amd was competing against icelake in 2017 , core counts still would have been lower than what amd was offering, I suppose being on 10nm would have allowed them to do 12 cores instead of 8, but even so, because amd used chiplets, amd could have simply just added another die since there is enough space on the am4 package to do it, and you're in the same pickle again.
>>107778199>Pat Gelsinger cancelled the Royal Core projectFuck me, intel really are a black hole for innovation, how they lucked into their position is beyond me
>>107769958I think the switch happened because intel was retarded enough to include stupid E-cores in their cpus, had a batch of high end cpus to "normally" reach 100 C and have a half life as a result, and some intel cpus needed a cpu frame due to the imperfect temperature allocation they were getting from certain motherboards cpu cases. Not entirely sure about every detail on the last one though
