What went so right with AMD and Ryzen? Especially considering Phenom and FX were flops.

>>105857386 (OP)
Nothing, it's shit

Jim Keller

itsj ust cheaper. if it was sold for the same price ppl would buy intel

>>105857386 (OP)
Not what went right with AMD but what went wrong with Intel.

>>105857392
>Nothing, it's shit
cope harder faggot

>>105857386 (OP)
X3D cache

>AM4
I hate that dainty faggot socket so much its unreal

Intel's LGA is a real rock solid lock and load sumbitch. AMD is only ok now because they started copying it.

TSMC

>>105857829
post nose.

>>105857400
spbp

>>105857829
shalom

>>105857386 (OP)
Cheap while being pretty good. It's just the price really.

>>105857386 (OP)
god punishing intel for all the evil shit they did in 80s 90s and 00s

>>105857386 (OP)
they just stopped making dogshit products
now if only their gpu division could figure this out

>>105857400
Meme. Jim Keller was primarily working on K12, which was canned and resulted in him leaving. Mike Clark was the chief architect on Zen.

>>105857386 (OP)
I don't know but I want it repeated forever, the AM4 socket has been a miracle in a market that really needed one.

>>105857386 (OP)
They invested in a technology for if moore’s law/ transistor shrinking started slowed down. (Chiplets, allowing for easy scaling of more cores.)
While also partnering with the one company that has been the best at pushing transistor sizes down. (TSMC)
TLDR the prepared for the worst and got the best.

Intel for comparison invested in ??? technology (fuck all)
And they got stuck on 14nm for ten years.
They didn’t prepare at all and got the worst.

Phenom I was ok but the cache bug killed the multithread performance and the node was bad.
Phenom II was good but it got delayed to fix those problems.
Phenom X6 was a waste of time for anything that isn't rendering or video encoding, and athlons could to just that for less cost
FX was AMD being retarded, they kinda fixed it but never releasing a high end cpu with those improvements.

>Zen
In large part TSMC instead of GF, early Zen 1 and 1+ wasn't that good compared to intel, they're optimized for low power consumption (ínfluenced by the big cats AMD it's based on) but not for high performance per core. With TSMC and good decisions (chiplets, caches, what to improve) after Zen 2 they could compete.

>>105859642
>Intel for comparison invested in ??? technology
good morning sir

THICC CUNNY MUMMY

They were right about bump23.

>>105857444
>Not what went right with AMD but what went wrong with Intel.
>>105857657
>X3D cache
Intel totally dropped the ball on this and knew extra cache made gaming benches go brrr a decade ago.
It took AMD releasing Zen 3 (2020/2021) to be able to comfortably beat in gaming an Intel CPU (i7-5775C) released in (2014/2015).
>https://www.anandtech.com/show/16195/a-broadwell-retrospective-review-in-2020-is-edram-still-worth-it

>>105863683
Ryzen 3000 is Zen 2 and it was released in 2019. And in those times AMD still had problems with some software because ICC was still popular...

>>105863683
Nah, it was Zen2 that really started the gayming crab. Zen3 was the final nail in the coffin. Zen1/Zen+ killed Intel's "cheap HEDT" segement.

>>105857386 (OP)
AMD gambled on chiplets being the future for CPUs due to economic realities on the foundry side. They happen to get windfall of TSMC's foundries (fueled by capital from Apple).

Intel refused to adapt and bet their foundry barn on 10nm node. It flopped really hard and they were stuck with 14nm and clocking the shit of it to keep gayming types enticed. Meltdown destroyed Intel's prestige ("Nobody get fired for getting Intel") in the enterprise segment.

upgrading cpus feels like such a scam anyway. 5600x -> 9600x is only a 20% speedup

>>105865090
In games, for other tasks is faster.

>>105857400
JIM "INTEL DEAD IN THE CELLAR" KELLER

>>105865171
>>105857400
Jim worked for intel too...

>>105857386 (OP)
>Phenom II
>Flop
Phenom II was based as shit, literally competed with high end Intel 45nm for way less money, and FX was killer for budget builds, but what gave it a bad rep is that it could never take the high end, it was killer for multithreading but the gaymers didn't like it because it competed with i3s in gaming.

>>105857829
They both suck; give me CPU pins I can bend back with a mechanical pencil or solder back on

>>105857829
>AMD is only ok now because they started copying it
AMD didn't copy shit they've been using the LGA socket on there server boards for more then a decade you fucking idiot cope harder

>>105866421
Intel's first LGA socket was more than two decades ago

>>105859642
transistor hasn't been shrinking since 20 years ago.
They just keep figure out new ways to stack transistors differently.

>>105866435
>my beloved LGA 775 is old enough to drink
Holy shit nigga

>>105866435
So was AMD's keep coping

>>105866584
Intel's was 2004, AMD's was 2006. I'm not coping at all, I hate Intel with a passion. Facts don't lie though

>>105865457
Phenom II was good for poorfags and FX was so bad that it couldn't even compete on price.

>>105866948
FX was amazing price for price for low end budget builds, why build a core i3 rig when the FX-6300 is quite good? It's when you get to the midrange things really fell apart, yeah an FX-8350 is a beast and comes in right around the cost of a non overclockable i5 3470, but you need a decently high end motherboard to run an FX-8350 without it throttling and crashing on you because it's a 125W chip, they just couldn't take it to Intel, but AMD FX-6300 was such a killer budget build back in the day man, you could put it on the cheapest fucking 780G motherboard possible and it would run full speed all day long under prime 95 and get better performance than Intel's budget stuff for less money.

>>105857657
Not everyone is Gaymer

>>105863683
I remember i7-5775C practically being a paper launch, I think they're unicorns in my country. Our biggest retailer continued to sell Devil Canyon until Skylake launched.

>>105867102
>Our biggest retailer continued to sell Devil Canyon until Skylake launched.
22nm chips lived such a long ass life at retailers, same with AMD FX, I'm thinking because we were kinda in between DDR3 and DDR4, and there not being such a huge increase at the time, (still stuck on 4C8T until Coffee Lake), that it made sense to built a 4790K rig and a 970 3.5GB card, that was a sick computer for the money at the time.

does anyone else until a platform matures before jumping in and buying? went with 5950x on AM4 now just waiting for the last equivalent for AM5

>>105867618
zen 4 basically is the mature platform, it a beefed up zen 3, especially now that were on second gen motherboards.

>>105867618
>does anyone else until a platform matures before jumping in
I've done that unintentionally and it does pay off, because later revisions of motherboards on the same socket tend to be better than the revision 1.0 stuff.

>>105867618
I just buy the best that's around when I need an upgrade, though if I know something new that's supposed to be significantly better is going to come out soon I usually wait for that.

>>105867686
yes i know theres always gonna be better things for individual components but the platform that holds it all together needs to be stable. saw a friend RMA so many AM5 shit it must be a headache.

>>105857386 (OP)
i loved my phenom 2 x6

lasted me almost a decade

but i loved my ryzen 8/16 even more (3rd gen i think)

and my ryzen ai9 hx370 in my asus proart px16 is the fastest cpu ive ever had; it even stays pretty cool so long as the system has good space around it; i keep it on a cute little stand so i can use the touch if i need to

>>105867768
>lasted me almost a decade
Based, AM3/AM3+ chips are eternal as shit, I used my FX from 2014 to 2023 and it's still an occasional use computer hooked up to the TV.

>>105867712
Eh, I never had to RMA anything.

>>105857386 (OP)
Intel stopped developing anything new and just released the same fucking thing year after year with an extra 0.2 niggahertz. Then for some reason they thought people wanted laptop CPUs in their desktops, not sure what crackpipe they got that idea from.

>>105857444
Jeets

How long do you guys wager a 5950x is going to last? Thinking of making one final upgrade to my x570 motherboard.

>>105869793
I have a 5900X and it's chugging along just fine, I don't plan on upgrading until last gens of CPUs that use DDR5 since with 12c/24t I don't really have anything that it can't handle, transcoding and compiling is the only things that actually use the CPU 100% for me.

>>105869668
>Intel stopped developing anything new and just released the same fucking thing year after yea
intel developed new nodes and transitor designs, but as of now failed to bring them into risk production. I mean if youre at the spearhead of technology, not every attempt is sucessful, Like Ayyymd nearly went bankrupt due to their foundry buiseness not hitting goals and ultimately they split up. The yearly half baked rehash is the consequence and not the cause.

>>105871146
Yeah but if you're gonna make a bet then make a good one. Like the laptop CPUs for desktop is best case fucking retarded and now they're close to running the company into the ground because of it. While I think they've made some retarded choices and should suffer for it I really don't want to see them die off cause we both know that if AMD is the only option they'll become as lazy or even worse than Intel when AMD did the FX line of CPUs and I really don't want another long stagnation period.

>>105871146
AMD (CPU division) relied on their own fab until 2020-2021, even if after creating GF they were forced to only buy the most advanced waffers from GF. Pre Ryzen 7000 the IO chip was still made by GF.

>>105871290
yea and thats a good thing. Most of the ICs today are made on 20-40 year old nodes. It allows for cheaper but still up to the task shit.
Cutting edge is fuck expensive, limited in capacity and a gamble, it is reserved for the products that actually benefit from it while still making a profit

Intel got lazy and shat the bed for multiple years because they owned the market. Nvidia will put itself in that position if they don't wake the fuck up

>>105871352
njeetdia doesn't need to do anything because the AI bubble will keep them on top even when it bursts
desktops are basically just background noise for them now

>>105857386 (OP)
Life is composed of swinging pendulums.

>>105871314
Cutting edge isn't optional for high performance computing chips. AMD found a loophole for the GF affair without harming the Ryzen or finances.

>>105857386 (OP)
Intel just shat itself in the same time frame.
Ryzen is good, but not a lightyear technological leap good.