How did AMD go from a laughing stock to completely dominating the PC CPU market?

>>105933364 (OP)
how did your ass go from twp to bbc on the daily

just as every empire one day falls, another takes its place

>>105933364 (OP)
>Be AMD
>Make one failed product
>Your opponent sees you are weak and starts creating monopoly deals with OEMs
>Your product don't get sold and you can't get contracts with big boys
>Sue Intel
>Win
>Damage already done
>Rewamp product
>If this fails we are bankrupt
>Shit sells like hotcakes because intel was more stagnant than you because they thought they already killed you

Simply the competition was stagnant and overpriced so they managed to make a lineup that's pretty close in performance, with more cores and at a very cheap price. They effectively undercut Intel for years and stole their marketshare

>>105933364 (OP)
>dominating the PC CPU market?
Intel still >70% on steam. I swear you guys are ridiculous.

>>105933364 (OP)
How many times a week do we need this thread?

>>105933416
>ignoring that a huge amount of steam reports are craptops from 10-15 years ago

>>105933416
Which you realize 80% of those are toaster laptops when you see the clockspeeds for the vast majority of Intel don't go above 3ghz.

>>105933364 (OP)
>Competition
>Intel sold out to NSA
>Intel introduced cuck cores
>AMD keeps their Chad performance cores
>Intel cut corners on production, 14gh gen has high failure rate
>AMD is cheaper for better or equal performance
>AMD introduces cool 3d vcache
>AMD has many pcie5.0 lanes, Intel does not(This means you can't have more than one 5.0 nvme without bottlenecking the 2nd)
>Intel throws out their well known and understood naming convention for something that is more complicated, for the obvious reason of tricking customers
>Intel calls customers consumers
Gee idk it's a mystery anon

>>105933453
>>Intel introduced cuck cores
i noticed this shit with arrowlake. it'll say something like "24-Core (8P+16E)" is that what you mean? what does that shit even mean.

>>105933484
Performance and efficiency cores. Performance cores are how you imagine cou cores to be and have been for many years. Efficiency cores are low power cucked corea

>>105933364 (OP)
AMD cut the fat from the Fabs, and now use TSMC. The fabs didn't needed to do cutting edge and now are very successfully. But intel supressed AMD for all the 10's until bulldozer nearly killed it. After buldozer AMD went for intel jugular and took the top at Datacenter and HPC.

>>105933364 (OP)
Intel stagnated. Allowing AMD to catch up.

>>105933484
Intel adopted big.LITTLE:
https://en.wikipedia.org/wiki/ARM_big.LITTLE

>>105933549
https://arstechnica.com/gadgets/2021/11/intels-alder-lake-big-little-cpu-design-tested-its-a-barn-burner/

>>105933364 (OP)
Through a state-of-the-art global marketing and shilling campaign.

>>105933542
I'm old enough to remember when Intel sheep were coping with
>i-intel have sekrit weapons(tm) and will crush amd next generation
The suicides were epic when you responded with "oh, so Intel have been holding out on you and charging you top dollar for >4 cores all these years?"

>>105933557
Go home LoserBenchmark, you're drunk.

>>105933549
>>105933555
The primary push for Intel to adopt big.LITTLE is increase multithreading performance. For the same die space of a single big core, you can fit 4 LITTLE cores. the space for 8 big cores you can have 32 LITTLE cores.

>>105933364 (OP)
AMD dominated Pentium with Athlon 64 and Athlon XP you fucking child

>>105933578
Now I'll make you feel like a young'n: AMD dominated Intel with the 386DX-40. Even low-speed 486s (at twice the price) got their asses kicked by it - their 25 and 33MHz buses paled next to 40MHz - the 486s better IPC was worthless if you couldn't keep it fed with data from memory and I/O.
This was eventually solved by throwing caches all over the motherboard to supplement the 486s 8KB internal cache, but not before there was an AMD 386DX-40 on nearly every desk in the world.

>>105933574
Did it work?

Don't know but they should share their secret with the GPU division

>>105933444
Could Steam even differentiate base clock speed and boost clock speed?
Modern processors set the base clock speed below even 2 Ghz

>>105933364 (OP)
jim keller.
risc-v is next

>>105933670
E-cores excel in heavily parallelized tasks.
Stock 14900:
>In Cinebench R23 multi-core:
Stock i9-14900K: ~37,000-40,000 points.
>With E-cores disabled:
~22,000-24,000 points
>Hypothetical 10 P cores:
~27,500-30,000 points
E-cores are ~50-70% as efficient as P cores per core in multithreaded loads but add massive parallelism. A stock 14900k effectively performs like a ~14 P core equivalent in heavy workloads. Difference is, far less power for the stock 14900k with its 8 p and 16 e cores vs a hypothetical 14 p core only variant. ~350 stock vs ~400 for the 14 p core. roughly speaking, a single p core can use up to around 30 watts of power. 30 * 14 = 420 watts. while an entire e core cluster (4 e cores) uses 8 watts of power.

>>105933761
Don't we use GPUs for high parallel tasks?

>>105933771
no. you do not compile your gentoo with your gpu. no all (most) paralllel tasks are good for the gpu. only specific ones are.

>>105933771
GPU's are only good for ~some parallel tasks. Not everything scales well on GPU's. If GPU's were the end all, be all of parallelism, then AMD and Intel wouldn't be pumping out 100+ core count CPU's and being able to charge upwards of $20,000 for them.

Realistically why can't Intel catch up?
They've had almost a decade to counter Ryzen, but the most they've done is come up with Estrogen cores.

>>105933376
MEDS, AT ONCE!

the writing was always on the wall

>>105933887

>>105933610
Most people here aren't old enough to have lived through that time and the mid 90s-early 2000s. There was no point in buying Intel shit compared to AMD's offerings at that time either. I still fondly remember my Gen1 AMD Thunderbird and the ability to OC from 1Ghz up to 1.4Ghz on the stock fan they provided. Much better than what I had before with the 600Mhz P3 slot processor that could barely eek out 800Mhz. Sure the AMD chip was a generation later but even the K6s were better than the P2/P3 back then and at a much lower price point. I had a K6 machine back during the P2-era and I loved it so much. I only jumped over to slot P3 later because I fell for the shilling campaign and how it'd be better in certain situations like video encoding. It really wasn't.

What I miss is the variety we had with GPU manufactures back then. I loved my Geforce 256 and later Geforce 2. But I'd give anything to go back to having options like Voodoo, TNT and NEC's Power stuff. I really miss when GPUs were black boxes and there was real competition in that area.

I used that AMD Thunderbird+Geforce 2 system I built as my main workstation well into the Dual-core era of CPUs. It hung with the best of them for over a decade. I even continued editing and encoding video on it. Although, eventually I had to move on to keep up.

I also really really miss proper dual-socket motherboards. These modern ones just don't offer the kind of features they did back then. I'd give anything for a good EATX motherboard that could work with "consumer" AMD CPUs that had tons of pcie lanes and the ability for both sockets to share the same massive pool of RAM without delay.

I keep wanting to build dual-socket system but now that they've nerfed all but the highest end server CPUs in that department I can't justify it. Plus the features on the motherboards are ass (hardly any PCIe slots for starters).

I'd kill for a POWER10+x86-64 dual-socket motherboard.

>>105933921
lol my boomer family members bought a shitton of intel stock during this time since it was depressed compared to AMD. They became rabid intel shills for the next few decades. By the time PC building became mainstream with the RGB light and shit, they would throw an autistic hissyfit worthy of /g/ if you so much as implied you wanted to put any AMD parts in your build since they thought it was making their moneyline go down. Hell, more recently, they even coped by claiming that the 14th gen processors self destructing was akshully a good thing since it means people would have to buy more intel CPU's to replace the broken ones.

>>105933840
Intel has to essentially replace their performance core that's based on the Pentium 3 (P6), re-creating a new one from the ground up or completely reengineer their current P6 performance core, drastically. To such a degree its the equivalent of tearing a house down to its studs.
The foundation of the performance core was built with scaling only going up to 4 cores with the release of the core series in 2006. Resulting in they never went out of their way to keep the power consumption down because they figured desktop would stay on 4 cores for mainstream until the sun explodes.

125 watts isn't bad. 125 watts for 4 cores isn't good. But when AMD was a drooling retard from 2006 till 2017, Intel got complacent.

>>105933364 (OP)
TSMC

>>105933921
matrox was the gpu everyone wanted nobody had

>>105933707
It reads the base clock that the is reported to the OS. Only laptops have a base of under 3ghz these days, unless you're going for toaster-tier PC.

>>105933941
Frankly, even if AMD started to suck again (and I'm not a huge fan of it anymore or any big tech company) I'd still not buy Intel. Just because who owns it. I do have an Intel workstation still with an i7 in it. It's the last version of the i7 that doesn't allow you to do stuff like GPU passthru. Even though the i5 version did. Since they took that away if you wanted something you could OC. I needed the extra CPU speed so I grabbed the i7 without thinking much about it. But then a few years later I wanted to add another GPU to that workstation and run Windows VMs with passthru. Which is obviously impossible despite the fact that could be easily enabled through microcode/software update. They're greedy bastards.

Same goes for both companies when it comes to their consumer vs. server line ups. But I digress. Modern hardware is all shit anyway so why even bother to argue who's wallowing in what shit.

I'd really love to go over to POWER9/10. But I'm always going to want some kind of AMD64 chip in my workstations. Since a lot of old shit I'll never get source code for needs it for performance reasons and I just happen to do most of those things.

So dual socket system with the secondary AMD64 chip along with its pool of RAM is something I highly desire. That or maybe AMD64 chip+lots of RAM on an PCIe card. But no one really offers either. I've found some PCIe cards in the past but they're very expensive and don't offer great specs.

Imagine running whatever OS you like built from source code on POWER10 CPU then having a second AMD64 CPU+GPU you could run legacy Window VMs on or AMD64 binaries either natively or through shit like wine. It'd be the ultimate answer to the modern workstation problem. You have all the legacy software at native speeds with the peace of mind that it's isolated from the main OS/system/everything else running natively on the POWER10 CPU+its GPU.

>>105933948
For Intel to seriously compete with AMD, they need to get their performance cores down to at least 15 watts of power. A 50% reduction in their current draw. And that only makes them marginally competitive because at this point, AMD has gotten their power consumption down to around 10 watts per core.

Intel does have the performance. What they lack is power efficiency with their performance cores. E cores keep them competitive in multithreading but do nothing in single threaded performance. Arrowlake has helped, but that mostly came from axing hyper threading off the performance cores. Most of the improvements with Arrow was with their E cores. Arrows E cores rock for multi. Their efficiency and performance is amazing in multi, but their performance cores are practically no different than raptor lake in performance and only offer upward of 15% less power. Thanks to killing off hyper threading...

>>105933999
Intel doesn't have to compete with AMD really. All they have to do is continue churning out overpriced Xenons and consumer level chips to OEMs and large server farms. Most of these places are under strict contracts and will never offer AMD at all or will refuse to offer it as their so-called flagship offering.

The industry I work in refuses to buy AMD CPUs at all even though the modern consumer level ones are better for our needs and work loads. They simply can't shop around and try out other vendors because they're under strict contract via all the other companies we do business with, our partners and Intel itself to buy Intel and buy them on a regular schedule.

In return we get reduced prices on the hardware, direct support from Intel and privy to any CVEs and bugs well before the general public.

>>105933364 (OP)
Apple kept making the same product and then hired all indians

>>105933999
If I care that much about multi-threaded performance, I'm going to get a CPU designed with it in mind, not some jack-of-all-trades with compromises.

>>105933707
Steam should be parsing data A LOT MORE. Fucking add country filters for fucks sake and add if the chip is mobile, add if the monitor is mobile. The steam survey is absolute ass for knowing anything because 50% of it is third worlders

>>105934070
Steam shouldn't be collecting that data at all.

>>105934077
It's voluntary anon.

>>105933364 (OP)
Marketers started running Intel. AMD has Engineers running AMD.

>>105933887
I love how cores don't grow linearly with cores.

>>105934077
It literally asks your permission, and then shows you what it's gathering before you hit the OK to submit it.

>>105934070
Would be fucking nice. Remove the CPU core list and just make it part of the Intel vs. AMD breakdown in the other part of the chart would be nice. Same for resolution vs. monitor size, since the vast majority of those 1080p users are either laptops, or 24" display users (1440p/4K is a waste at that size).

>>105933416
There's a long tail to the upgrade process. Intel was the preferred choice for a very, very long time and it's not like people are going to throw out all their perfectly workable computers running Intel CPUs just because the latest models suck. As far as modern upgrades & newly built systems go, AMD most certainly seems to be dominating.

>>105934128
I'm sure they have all the data that cpuz gets from your system but they refuse to release anything but these cucked charts that tell us nothing. There should be way more data than is published and it should be able to be filtered by reigon, we should have a breakdown of age of cpu vs age of gpu, way more generational data, clockspeed data on the speed the cpus and gpus are all running at, and most importantly country of origin by percentage for each part. There is no fucking way people in the US are running 1080p on their desktops at a rate over 40%.

>>105934131
If intel didn't switch sockets every 1-2 years I would consider them a viable option. But they don't. So I consider them a laptop only cpu maker.

>>105934165
Integrated graphics laptops is the one thing intel can do.

>>105933453
>Intel throws out their well known and understood naming convention for something that is more complicated, for the obvious reason of tricking customers
>Ryzen AI Max 71626
AMD has been muddying their product stack naming for a pretty long time, especially in the laptop segment.

>>105934269
>names their gpus 9070 and 9060
What in the fuck were they thinking? Is it because the cpus had the same names? Now I'm constantly saying 9070ti or 5070xt on accident

>>105934165
Do you upgrade your CPU that often? I have 10600k and I see no reason to upgrade.

>>105934308
I imagine the people who bought 1st gen Ryzen would be very happy to be able to just drop in a 5950X or 5800X3D on the same mobo and same socket, huge performance increase right there.

Also I have a 7950X3D on an X670E board, so 1st gen AM5. The mobo supports PCIe 5.0 and has everything I need. I could see myself just dropping in a faster AM5 CPU once another gen or 2 are out on the same socket and the perf difference is bigger compared to what I already have. It's very nice to be able to upgrade.

>>105934288
Before you speak or type just pause and allow yourself to catch up. After a while people will stop thinking you're dimwitted.

Being so shit that they were forced fabless helped a lot.

>Not investing at AMD in 2018 when stock only $2

>>105934347
Personally I bought a 3600 for $200 amd then a 5700x3d for $200. I also bought 16gb more ram for $40. Buy at entry/mid but still an upgrade from current then upgrade again when it's cheap and memory is cheap.

>>105934367
I can't, I post all day at work on my phone and can be interrupted at any moment

isn't china still making money off of chips from like 1985?

why can't intel just corner markets like that and do it better and create its own branch of tech with simpler retro chips? this is our chance to go back to 2004! do it intel!

>>105933364 (OP)
lisa su unironically.

>>105933453
1)no it was intel being intel
2)yeah lets stop pretending that amd doesnt hav backdoors they are required by law
3)cuck cores were ok the problem started when intel also introduced a new ISA specifically for them while amd made the c cores which was basicly just a packed redesigned more efficient regular cores

>be 5900X
>bsod weekly
>scream random WHEA shit
>bsod daily
>bsod hourly
>die
after 2 cpus, 2 mobos and 3 kits of memory i finally have stable non-OC system. Thanks AMD.

>>105934505
Those aggressive high clocked parts by AMD seem laughably unstable

>>105934505
Damn, I have 4 different Ryzen computers and every single one of them is rock solid. It must suck to be (You).

>>105934505
>oh look the same retard that tries to pass the motherboard makes pushing voltages beyond of what amd suggested as an amd problem

what are the odds

>>105934625
what part of non-OC you don't get?

>>105934505
>>scream random WHEA shit
Huh, is that what was happening to my PC? Was sure it was a W11 problem since I never had an issue with W10. Though haven't had an issue lately either. Not sure if the undervolting helped or not.

>>105933422
Every morning saar

>>105934347
>5950X
Yes, package just arrived, will be swapping out my 1800X.

>>105933364 (OP)
They had to compete. Complacency brings death. It will happen to Nvidia soon.

>>105933416
>he thinks opt-in hardware survey that only faggots do is a realistic summary of hardware
Oh no no no.

>>105935232
yeah bro totally
any day now Da wAy iTs MeAnT To bE PlAyEd nVidia will suffer backlash for stagnation and people will buy the radeon product instead (even though radeon has had better price/performance in every segment for around 35 years now
>amd has no drivers xD

>>105933364 (OP)
Because Intel doesn't need the DIY part market to makes exponentially more money than AMD.

>>105934628
everyone knows what went down just stop it already we know the truth bitch

>>105935420
>backlash for stagnation
They're not stagnating, though. They're just increasing prices. The cards are still getting faster.

>>105933364 (OP)
Real answer Intel lost 20%(or more in some tasks) performance to spectre and meltdown patches and are still dealing with it

>>105935756
>niggerbrain thinks that everyone that has cpu shit on themselves is because they do some gaymurmeme niggerhshit.
nigger, i like stability and i expect computer to work and stay on 24/7 when you don't fuck extra with it, that means mobo with default bios settings and ram with jedec spec.

>>105936275
>motherboard vendors finally admits that they pushed voltage beyond what agesa allowed
>that single idiot that always thinks he is above and beyond says otherwise

By allowing glowie-PSP to be implemented.

i got this and im happy

>>105935989
mitigations=off

>>105934802
unless you've overclocked to the max and become heat limited, undervolting always makes silicon more unstable

>>105933364 (OP)
well you see it wasn't that Zen was great
it's merely adequate
Intel failed in designing a new core. also simultaneously failed to make improvements in their fabrication methods.
even on a more dense process and without smt Lion Cove is larger than a Zen 5 core, doesn't clock higher and doesn't have higher SPECint 2017 Rate 1T results. So it's using more pricier transistors to do effectively the same. then they don't have any options to add more cache for gamers

>>105936761
>then they don't have any options to add more cache for gamers
incorrect. they can add l4 cache like they did with a line of broadwell chips.
example:
>Broadwell i7 5775C with 128MB of L4 cache (eDRAM)
It was actually faster than a Skylake 6700K, even though, on paper, Skylake had a faster architecture and higher clock speeds, but thanks to the L4 eDRAM cache, it was faster in games.

Intel can do, they just don't bother, because ultimately, the massive power consumption drives "gamers" away when AMD has competing or faster products that use 50-70% less power.

>>105937063
that's not what I mean by option. I mean there is no SKU doing that.
theoretically they could just do the same thing AMD is doing too. or just make a bigger die. they have many technical ways to do that but they offer the consumers no options

>>105936438
you literal fucking ignorant monkey, those voltage fuckeries are problem if you are running docp/expo and even there its majorly problem with am5 platform due am4/ddr4 handles higher voltages than am5/ddr5 by spec. You are free to point to any verified resources that show that mobo manufacturer has fucked with voltages on am4.

>>105937565
>stop believing reported truth
>believe my fanboy, head cannon instead

>>105937644
literally all verifiable shit is from am5 and even there only with expo enabled, so its your side with fanboy head cannon

>>105934148
>There is no fucking way people in the US are running 1080p on their desktops at a rate over 40%.
I believe it, unlike the switch from 720p to 1080p, we haven't seen a full shift to 2160p or even 1440p.

>>105937681
>my head cannon is real
>i swear!
is lisa su in the room with you right now?

>>105937690
Nobody I know uses 1080p outside of laptops. If you built your computer you aren't still running 1080p when all the tvs in the house are already 4k. It's unbelievable. In the old days computer screens were far higher resolutions than tvs. I had a 16:10 2560x1600 in like 2009.

>>105934379
>invested in Nvidia before their price went up ten times
>only had 1 share
If only I hadn't been poor.

>>105937716
as i said, you are free to give any materials that confirm your dreamworlds existence

Intel stopped giving a shit, sort of like AMD with GPUs.

>>105937717
>Nobody I know uses 1080p outside of laptops
Well everyone I know still uses 1080p, even coworkers. It's being slowly replaced but nothing compared to how quickly 1080p caught on.

>>105937752
>claim his head cannon is real
>>demands over refute his head cannon
fascinating. what is lisa su telling you?

>>105933364 (OP)
Intel sat on their asses with nearly a decade of 1-2% ipc improvements. They funded hundreds of millions to billions for DEI programs. Meanwhile AMD worked their ass off nearing bankruptcy and continued to deliver double digit IPC per generation. AMD's Innovation comes from necessity. Intel's "innovation" came by asking OEM to not sell AMD products, but they got fined for that. AMD needs to keep innovating. Bring us true UNIFIED architecture, combine CPU-GPU combine RAM-VRAM. No more extra steps, just bring us a single unified structure.

>>105933416
>but muh dominated by ching-chong internet cafe PCs hardware survey!

>>105937757
This disgusts me. 1024×768 came out in 1990 when tvs were 720x480. Pc should always be better than tv.

>>105937717
vast majority of desktops, even for gaymers, are 1440p and less. 1080p is still the dominate. 4k is still to much of a struggle to run, especially for high refresh rate.

>>105933364 (OP)
They threatened Intel with an anti-trust lawsuit, so now they have to pretend to suck so AMD can catch up in about 10 years.

>>105933718
>tfw can't invest in tenstorrent

>>105937767
Like an APU? Those have existed for decades

>>105937837
Monolithic was their downfall. Still they're claiming nova lake will be a big return to form.

>>105937834
I doubt this in first world countries. I've used 4k since 2016.

>>105937859
>Like an APU
No. APU still reserve separate pool of ram from the system memory. Its not unified in the sense that CPU can read/write from that reserved pool. Apple's approach is all memory is shared across. No offlimits to CPU/GPU. So they dont need to transfer from one shared memory to another as they do in AMD APU. APUs have the same problem of discreet GPU v system memory with regards to multiple steps necessary for CPU-GPU to read/write data. The extra step causes tons of latency/slows down the computation.

Further, Apple's current approach is great, but I think they're potentially going to have a single unified cpu-GPU core scaling. Right now they have separate CPU-GPU cores. AMD/Intel are doing bigSmall arc with their CPU die, but this could be further extended to blur the lines imo. Further Apple uses ARM chips so they use RISC instruction set, so it scales very well with cores with minimal overheads. So they potentially have benefits to going a unified CPU-GPU core system. If not standard ARM with standard RISC instruction, they could do custom instruction set to optimize their hardware to their need.

>>105937878
>my head cannon is real

>>105938022
Thanks for the insight anon

>>105937878
Maybe for TVs. The majority of Euros and Muricans will not spend that much money on a PC monitor.

>>105938715
Why are the pc monitors so overpriced?

>>105938855
They are not that much in demand.
The common denominator doesn't care about high fps, adaptive sync or color accuracy,

In comparison any basic cheap 4k tv will sell well despite being shitty

>>105937763
Post source any time now ;)

>>105933484
>24-Core (8P+16E)
8 good cores and 16 dogshit ones

>>105933484
It means one monolythic die with 8 cores and 16 shitty chiplets straped onto it to say they have cores

>>105933364 (OP)
>How did AMD go from a laughing stock to completely dominating the PC CPU market?
Intel didn't think they needed to innovate while amd was down, so they basically were caught up with their pants down and basically the best they had were 4 core furnaces and since they were stuck in 14nm++++++++++++ they couldn't react fast enough to cram more cores into a single die.
And let's be real, if the current intel CEO doesn't follow Pat's path into becoming a huge foundry able to compete with samsung and TSMC, they will be doomed forever.

>>105938022
>APU still reserve separate pool of ram from the system memory
For legacy software support, its not a true partition. Its entirely virtual and arbitrary, AMD has an entire document detailing this. APUs can snoop memory pointers between CPU and GPU, they have a virtual page file, and flat address space between CPU and GPU.
Try reading programming reference manuals before pretending to be an expert, kid. Everything you told that anon was wrong.

>>105937717
maybe in pisslow you and your hardstuck buddies are running 4k displays with pitiful hz. actual fraggers care about having a solid 240hz with low response times to benefit from all the frames they get running games off configs with minimal visual bloat.

>>105939508
I don't need 240fps for childnight and pubg because I do not play them. I haven't played counterstrikeb since source. I want a cinematic experience.

>>105934379
it was like 12-13 dollars in 2018

>>105933484
You have 8 cores that work and 16 that will bottleneck your processes when they try to synchronize.

>>105933441
>>105933416
>>105933444
>>105934131
>>105935237
>>105937787
>>105933364 (OP)
All you have to do is look at quarterly revenue to realize that AMD is still not even in the same league as Intel. Intel has had an absolutely dogshit past 6 months and yet it still nearly doubles AMD in revenue (and that includes AMD's GPU too lmao). In 2024 Intel doubled what AMD makes.

>>105939882
Huh I thought it split or something but you're right, the other guy just made up a number.

>>105939953
All that matters is growth unless you want to buy the entire company for some reason.

Intel has decaying, not growing, sales.

>>105939953
Brainlet noninvestor.
Intel's straight gross revenue figure is inflated because they're paying themselves as their own foundry customer.

>>105937848
thank god investors are cancer

>>105939953
Intels primary revenue is AWS and Azure

>>105939021
They're really the last of the old vertically integrated CPU vendors. It's like how computer hardware manufacturers used to write their own OS and word processor to sell with their microcomputer.

>>105939993
He has private investors doe

>>105939994
AMD doesn't have very good DC products. They're really primarily a consumer and maybe workstation brand right now.

>>105940017
Dumbest post in this thread right now

>>105933718
Risc-V is pleasant to program for but it seems like it's not a very performant ISA. idk enough about processor design to know why that is though.

>>105940024
Oh yeah? What's AMD's best datacenter CPU?

>>105940017
Aws buys threadrippers as well. They also have their own arm style processors for certain workloads now.

>>105940026
Probably because it takes 6 micro ops to do what X86 or ARM both can do with 3

>>105940033
Trolling is an art, kid

>>105940034
>AWS
>Threadripper
You have to be 18 to post here

>>105940034
Sorry, I meant epyc not threadripper. And the aws arm chip is called graviton

>>105940045
Yeah I just forgot what they called their version of xeons.

>>105940045
In anon's defense, Epyc is a really retarded name for a product intended for corporate use.

>>105940081
Branding almost doesn't matter at all. Virtually no one in a company of hundreds or even thousands of people knows anything about hardware infrastructure. Its invisible to the average employee using it.

>>105940004
Let's be real, the fact that neither amd nor nvidia have their own foundries has made so TSMC can ask for literally any price they want per waffer and they'll pay it since they have no competition, I'd wager that half the reason gpus are unobtanium nowadays is purely tsmc's doing, since every generation they use bigger chips.

>>105940104
Epyc is better for virtualization and xeon for databases. Aws making their own arm chips doe ec2 had to be a huge blow for amd.

>>105940104
Branding is almost more important in a corporate setting because you'll have people arguing with eachother using slide decks. Bad branding is kind of like sending troops into battle with bad equipment.

>>105940108
>TSMC can ask for literally any price they want per waffer
TSMC is not remotely unique. Contract manufacturing is highly competitive and if they raise their prices much everyone will just use a slightly worse node elsewhere until elsewhere catches up.

>>105940199
Which one is better at AI?

>>105940011
obviously not the problem pandering to the general public investors is they are the reason for enshittification but hey enjoy your totally legit manifest v3 thank god google saved us from ourselves

>>105940471
Imagine having what is essentially Reddit but for boomers and they can send you to jail as your boss.

>>105940698
steam/valve says hi

>>105940706
God I wish I was one of the lucky 336 that work there

>>105936528
Works on my machine, but I also used the safe settings.

>>105933364 (OP)
>dominating
I don't use that crap and I never will you massive poofter

>>105933364 (OP)
good hardware guys
it's a shame the software-side is still utter garbage
amd would also be worth 4 trillions if they had good firmware, drivers and user-space software

>>105933414
>pretty close
Zen has outperformed every Intel generation since it came out
the current Epyc chips have absolutely dominating performance on every benchmark in both single and multithreaded worloads

>>105933364 (OP)
iirc they used to get too hot but now they don't

>>105940108
Samsung foundries are around to make TSMC not doing that

>>105933364 (OP)
>laughing stock
Laughingstock is one word ranjeesh
>completely dominating the PC CPU market?
Intel decided it would be a good idea to make the same CPU for ten years

>>105938022
The only thing Macbook does is using the fastest available soldered RAM on the market. Something that HP Zbook also does.
They didn't invent the concept of using expensive component.

>>105933364 (OP)
amd didnt do anything, intel just went woke and collapsed

>>105933364 (OP)
a combination of chiplet ingenuity and intel going full retard and sitting on its lead

>>105933416
there's been a surge of cheap refurbished LGA 1366/2011 kits that deliver great performance per buck with intel server CPUs, they are here to stay for a while. they're a great buy for global south gamers along with tray AM4 CPUs

>>105933416
guess what cpus and gpus asian internet cafes will only buy, for whatever reason

>>105939508
4k with 120hz is enough for me.

>>105933364 (OP)
NEVER OBSOLETE
I will upgrade on am6 in 5 years, see you later virgins

>>105933364 (OP)
they used tsmc. /thread
>>105933376
fpbp

>>105944576
Many such cases!

>>105934505
5900x going strong. Definitely a You problem

>>105934379
So, buy intel stocks now.

>>105933364 (OP)
Intelavive has more indians.

>>105933761
Cuck cores.

>>105933364 (OP)
Because the more cores the better.

>>105933364 (OP)
by realizing PC gamewrs are retards and will buy playstation tier CPUs

>>105933840
They already started that with 10th gen, when they finally moved past 2c4t/4c4t/4c8t

>>105937717
>In the old days computer screens were far higher resolutions than tvs
Yeah, because TVs were generally 13 or 27 inches, maybe 32 if you were rich, and every one of them was 480i, or 576i if you live in memeland.
Now you just have poorfags refusing to update from 1080p for their computers, with slightly less poor poorfags coming up with fake excuses about high refresh rate even though you can trivially get 165hz @ 1440p below the $300 price point as long as you have a mere DP1.4/hdmi2.1 connection.

If you're not buying up Intel stock and way out amd puts you're fucking retarded.
The white man (Intel + Nvidia) will always win in the end, we're just genetically better than you.

Intel-Aviv sucks. Estrogen cores, 14nm+++++, etc

>>105939508
actual fraggers quit playing modern videogames because they're all fucking trash

dumbass zoomer troons would get melted on quake and unreal tournament

>>105944576
as someone that came from an 8350 fx to a 5700x i must agree i dont see the point of upgrading for at least 7 years

>>105941787
Zen is good but there's no reason to lie about it, the first two gens didn't outperform Intel.

>>105933364 (OP)
Intel hedged everything on their 10nm node and thought EUV and chiplets were still unnecessary.
AMD gambled on chiplets to be the future for CPUs and paid off hard. TSMC gambled on EUV it worked (fueled by capital from Apple)
Intel got leadfrogged on CPU architecture and foundry. Something that never happened to them before. During the Netburst era, Intel had their foundry side to fall back on despite architecture's shortcomings.
Intel is trying to play catch-up since Zen. Their ace is getting the whole heterogeneous computing down and getting on board with chiplets

>>105946324
shit one was replaced by amd rma and that one has worked as expected, so you can stuff your You problem back into your ass.

>>105933364 (OP)
>How did AMD go from an amusing annoyance to Nvidia to completely being a laughing stock in the GPU market?

I find it a bit strange how long intel stuck to making inferior cpus on their own fabs instead of using tsmc to make actually competitive products.

Ofc they have higher margins that way, but the decline was happening for several years on desktop gaming cpus and then the crash happened and intel is looking to be bought out by someone.

How did their board not go "lets reduce the risk of crashing out in the future by having lower margins now, instead of later on just so we stay competitive

>>105933610
Cyrix

>>105933364 (OP)
>tsmc
>intel 7 a massive failure
>chiplets
it's shrimple, really

>>105951550
>stuff
>problem back into your ass.
Strange choice of words for a straight man. just saying.

>>105940375
>>105942006
Yields from other fabs are not as good as TSMC iinm. Of course TSMC are smart enough to not raise their wafer costs until it is more profitable to use a lesser yield fab

>>105933364 (OP)
Because intel fucked up and sold defective chips.

>>105933364 (OP)
By relying on jeets and other diversity... at Intel.

>>105954761
kys racist

>>105952543
ATI was a serious threat kiddo back in the day which started with Radeon 8xxx (Despite its driver issuers). Nvidia was only able to turn it around with ATI fumbling with Terascale a.k.a (HD 2xxx-5xxx) and Nvidia make a large break with Tesla (the architecture not the brand) a.k.a 8xxx-9xxx family. Nvidia kept building on its momentum until it pass clear away with Maxwell. We have been here since.

>>105952601
>How did their board not go "lets reduce the risk of crashing out in the future by having lower margins now, instead of later on just so we stay competitive
The line must always go up, short term loss of profits is unacceptable. Future long term gains are not important, give my fat check right now!

>>105954898
SAAR

>>105933416
Steam is biased they have a section for intel cpu speed but no section for AMD cpu speed

>>105933364 (OP)
they ran so out of money that they could only afford engineers, and as such they started making useful products instead of marketable ones

they're currently on the other side of the bell curve now tho, AMD is getting fat. they spend like 100x more on R&D now than when they developed zen, yet radeon and zen are making less progress at longer cadence

>>105954898
We are going to ethnically cleanse India at the end of this century...

... by we, I mean Pakistan.

>>105933416
Steam survey is heavily skewed towards prebuilts (which is why the 50/60 series from nvidia are so goddamn popular even though they're dogshit).

>>105959647
>India
>Pakistan
you are the same

>>105959647
>We are going to ethnically cleanse India at the end of this century
Do you know why?

>>105933364 (OP)
people wanted more cores and they delivered, simple as that

>>105933416

To everybody that hates e-cores (aka cuck cores), amd is going the same way as intel with P-cores, E-cores and some sort of LP-E-cores. With intel the P and E cores come from two different design teams, but are looking to unify them into a single design that can be synthesized with different parameters (beyond cache sizes).

AMD is currently winning because of several things going their way at the same time.
#1 they executed. They had a good idea with zen. Lisa Su has done a great job.

The rest are intel fuckups.

#2 intel botched 10nm terribly. They tried to make too many changes at once. They moved from copper to cobalt interconnects on at least some metal layers. They also tried to push density reductions beyond what they could do with multi-patterning (i.e. not EUV).

#3 intel had design flows that were not compatible with anybody else. That means they couldn't just up and move their designs easily into an external fab. I understand this point least well, but intel is apparently changing how they design chips to move chip(let) manufacturing around at will. For example Nova-lake is going to have its main cpu tile fabbed on 18AP for mobile, but tsmc n2 for desktop.

#4 management hell at intel resulting in the company not being able to function and adapt where they are not in a market-leading position. Intel has and has had incredible technology, they just can't seem to bring it to market in price and performance competitive ways. Their product development has been like moon-shots (ex ponte-vecchio) that were basically abandoned. If you want to break into new markets (like gpu compute), you can't do that, you need to be sure that your products are capable and competitive on price / performance then sustain a roadmap and build adoption. This is the path nvidia and amd have taken with their compute accelerators.

>>105933364 (OP)
For no reason at all.

>>105960654
>not being able to function and adapt where they are not in a market-leading position.

Another example of this is AVX-512. Intel invented it, it is actually in the p-core designs, but is disabled for product segmentation (so you would have to buy Xeon). We are 2 -almost 3- generations beyond amd introducing it into Zen on desktop and server, intel: crickets. They could have done it even if e-cores don't have it by pinning cores that call avx512 instructions to p-cores on an invalid instruction exception.

Another example is the compute accelerators built into sapphire rapids. You were supposed to enable these post sale by spending extra $, but nobody was going to do that - they would just buy epyc, and in xeon6, they are just there to use. Years lost getting people to start coding to them. The hubris was staggering on this one.

>>105960747
>pinning cores
*pinning threads

>>105933416
i have three computers in my house running intel chips.
two of them are optiplexes i got for $100 each, neither has a GPU. they're in my kitchen and bedroom for watching youtube videos.

one is my old laptop, sitting in a cabinet in my bathroom, which i use as a literal shitposting rig.

they all have steam on them despite not being able to run anything newer than skyrim.
not because i run games on them, but because i can stream games to them from my actual gaming rigs.

i have three gaming rigs. they all run ryzen chips. 2700, 5700G and 5800X3D.

>>105933364 (OP)
>AMD makes dual core CPU in 2005
>be intel panic make dual core the next year.
>it was not that hard
>an year later double the cores and make quad core
>one would think it will take you maybe 2-3 years for a six core and 5 years for 8 core cpu on you main socket.
>nope make quad cores on you main socket for 10 year.
>but it gets even better you move one pin on the same socket just to force people to get new motherboard for no reason and the same quad core that is 3 to 16% faster.
>then after 10 years of quad cores comes the competition offering 6 and 8 cores on the price of your duo and quad cores
>and they promise to keep the socket alive for 10 year - surprise they did - people with 10 years old motherboards can enjoy 16 core cpus with over 50% higher single core and 300% multi-core performance.
>at the same time you shit 3 sockets with 0 upgrade path
>low end motherboards can't even even drive you high end cpus without shitting the bed and restarting with overheating vrm even mid tier mb have problem because you CPUs boost to 230W.
>and this +230w boost - it makes you cpus degrade overtime
It is a complete mystery.

>>105933364 (OP)
AMD made a good product. and at the time 2018-+ intel cpus were not getting any faster and were shown to be vulnerable and the patches slowed them down. so it's a no brainer. ain't letting anymore jewish cpus in my house

>>105933453
>14gh gen has high failure rate
This is the biggest part of it, intel is dead to me. I don't even care that it got fixed, I don't know if it did. I just don't want to even risk it when AMD is proven and cheaper. I used intel my whole life but I have no loyalty to companies who fuck over their customers. Computer parts are too expensive to risk it. Same reason I'll never buy asrock for at least 5 more years. No point in taking a chance when there are safer better bets.