| If AMD had been extremely conservative then they would not have wasted 12 Billion+ US dollars on Share Buybacks in 2021/2022 and on through to today and invested that money in growing their Client PC/Laptop market share. But Lisa Su and Management team wants a more IBM like AMD that goes after the higher margin business and consumer is just there to uptake the Zen generational CCDs samples that do not make the binning grade to be Branded Eoyc/TR Pro and command the higher markups. So all of AMD's Zen generational CCDs are produced on the same diffusion lines at TSMC and then the Zen CCDs go into AMD's CCD Binning operation where the CCDs with the best electrical characteristics(Operate Stably at the lowest voltage and use the least about of power/generate the lowest amount of heat in doing so) get taken aside for Server/HPC and Epyc branding and markups. While AMD takes the CCD samples with the not so performant electrical characteristics(Needs more Voltage to operate stably and uses more power/generates more heat) and that gets binned down for Ryzen/Consumer Branding and the not so good markups. And so for AMD those Zen CCDs, that do not make the Epyc/TR Pro binning grade, are costs recovered via the consumer market and that's why AMD has a consumer market segment to begin with to unload any non performant for server/HPC/Pro Workstation market CCD samples. Has anyone read the recent article on the Epyc 4004 processors that are for the AM5 platform but for small business usage and the Overclockers getting a hold of that and they have found a way to unlock a CPU clocked locked part that's not supposed to have that ability! And look at the high Overclocks achieved on those not for consumer market Epyc 4004 parts that are the better binned Epyc Grade Silicon and thus actually can be overclocked higher because that's from the cream of the Zen Generational CCD samples. And so for AMD they have to have some minimal Ryzen/consumer market to uptake the non performant for Professional Market usage CCD samples that are a little more power hungry as they need higher voltage to operate stably compared to the Epyc Binned CCD samples. The server binned CCDs are for clients that favor low power usage and performance per watt over higher clocks at the cost of power efficiency. The laptop market is a bit different save AMD's Dragon Range laptop parts currently that are just the same desktop Ryzen CCD based processors that are BGA packaged for laptops instead of LGA packaged for socket based MBs! But eventually most of AMD's laptop parts may not be monolithic die based parts and become more like the CCD based desktop parts where the cream of the CCD samples never makes its way into consumer based SKUs. Most of AMD's current lineup of Mobile APUs are monolithic APU DIE based and so have no Epyc equivalent like parts to get the better Die samples and leaving the less performant samples for consumers. AMD also has terrible ROCm/HIP support for AMD's Consumer iGPUs and dGPUs and so it's nearly impossible for any Integrated GPU Accelerated Blender 3D Cycles rendering on Ryzen Integrated Graphics or on most consumer Radeon dGPUs, save some limited SKUs of the most recent Radeon dGPU generation parts! And so Intel's consumer hardware via Intel's OneAPI and Level-0 GPU compute API has better support for creative applications on Linux and Windows for Intel's consumer iGPU and dGPU SKUs. And AMD's support for iGPU/dGPU compute workloads has always been a sore point among end users while Nvidia's support for Blender 3D and other Professional graphics applications on Nvidia's consumer parts has always been better. And really for Apple's M series processors on Apple silicon Apple makes sure that its iGPUs can be used for Blender 3D's iGPU accelerated cycles rendering. So if one is looking to do creative workloads other than gaming AMD's not the first choice there as Intel/Nvidia/Apple have the better support for iGPU/dGPU compute workloads on their respective consumer processors. |
Such cheap servers are the right choice for small businesses and for individuals, for whom the great increases in the prices of the server CPUs that have happened starting in 2017, since the launch of the Skylake Server CPUs, have made them either unaffordable or not worthwhile.
Because Intel cripples all their consumer CPUs, they cannot really compete with AMD for this purpose, except for some very restricted applications, e.g. for a server which does nothing else but code compilation, or for a server which does nothing else but hosting a Web server. Even for such applications, buying one of the very few motherboards for Intel that support ECC memory is likely to increase the price well over the AMD solution.