That is because Apple is snagging all the sales.The only thing that I can think of is that the Workstation market isn't very large - compared to server and desktop/laptop. So Workstation CPUs are a lower priority - plus they still take a fair bit of time to validate. It's not like, oh sorry we just corrupted your 700GB AutoCAD file, everything is fine.
As already mentioned - that would push up the price of desktop boards, and they are already expensive.I am surprised AMD didn’t go quad channel with Ryzen.
Really? For Pro software aside from Video/Photo and Sound editing?That is because Apple is snagging all the sales.
This is a false assumption for a couple reasons:As already mentioned - that would push up the price of desktop boards, and they are already expensive.
Yes, Apple has been dominating or overtaking nearly every pro workload out there except the Windows only ones. At least in all the niches I know of.Really? For Pro software aside from Video/Photo and Sound editing?
Well, that's an option 4x single channel. Surprised that Quad channel isn't more expensive - are DRAM clocks limited compared to enthusiast boards?As it is, we now have a bunch of boards with 4 DIMM slots, but 2 of them may or may not work, and if they do, likely at lower than advertised speeds.
No idea, Zen 1 was pretty limited in terms of speeds. I bought a DDR4 3200 kit and it worked out of the box with XMP. In theory, because it is still 1DPC, it should still support high speeds.Well, that's an option 4x single channel. Surprised that Quad channel isn't more expensive - are DRAM clocks limited compared to enthusiast boards?
I'm not sure only "marketing" would block it (not like AMD excelled at that part anyway). We already have seen AMD target a lot of different markets with the fewest actually different chips. So from that point of view adding yet another option should be no issue. It however does become an issue if it is deemed to weaken the overall branding, to bind significant software investment (in the case of mixing Zen4 and Zen4c CCDs), to use up stock volume in the channels for lower margin units, and whatever other more or less valid excuses different management levels could come up.
Zen3 with 16C had no problem with only DDR4 3200-3600Mhz.
Now, we can buy 8000MHz memory, that's 122-150% more.
If I calculated:
100% for Zen3.
129% for Zen4 (+29%)
174% for Zen5 (+35%)
Now, If I added 50% more cores, I would end up with 261%.
I would need 8350-9400MHz memory.
Small servers that would officially build upon the AM5 platform would be nice indeed. A too low margin niche for AMD though I guess.The easiest way to market it would be as a micro server chip, with 2x Zen 5c based CCDs. Also, possibly with just 1 CCD.
It would not interfere or contradict anything. The CPU would work very well in this environment.
Of course, marketing heats would probably explode in what to call it... Easiest thing would be just to give it a different suffix, just as 8950 MS (or something like similar), but the marketing would probably have to burn a couple of thousand man hours to come up with this...
Small servers that would officially build upon the AM5 platform would be nice indeed. A too low margin niche for AMD though I guess.
Small servers that would officially build upon the AM5 platform would be nice indeed. A too low margin niche for AMD though I guess.
I'm well aware of these offers by board manufacturers. ASRock Rack as well. Point is that AMD itself doesn't target small servers at all.
Right there. And there are others.
Edit: at least this one also
Fixed it for you.There are two kinds of people who "need" morecoresST performance:
a) Those who have an actualmany-coreST workload to run. These folks simply buy or rentmany-coreST computers or computer clusters.
b) Those who don't actually care about workloads which benefit frommany coresST performance. These folks are telling us how certain computer platforms need a certain increase ofcore countST performance. How about they tell us about it in a dedicated thread? It's not particular to Zen 5.
Thats BS. The entire DC community needs more cores. Its only gamers and a few niche users (software that can only use one core or a very few cores) that need ST performance above all.Fixed it for you.
I am beyond certain many if not most engineering loads need a combination of both. Engineering workloads are vast and diverse.Its only gamers and a few niche users (software that can only use one core or a very few cores) that need ST performance above all.
Its only gamers and a few niche users (software that can only use one core or a very few cores) that need ST performance above all.
The same CPUs are used in mainstream, gaming, office, and workstation PCs. All on AM5.If anything it's the complete opposite.
And gaming is what drives DIY.
The same CPUs are used in mainstream, gaming, office, and workstation PCs. All on AM5.
Yes. And a mix of those actually.Dont think you ll find some 7950X/7900X or even 7700/7700X in office PCs, most likely some APU with 4C/8T at best.
and what does gaming have to do with desktop ? Yes, its a PART of desktop, but there are others. Some want office, some want mini workstations, some participate in DC or things that require a lot of cores. Why do you think DIY owns gaming ? Consoles ? handhelds ? Gaming is all over, but they don't own DIY.If anything it's the complete opposite.
And gaming is what drives DIY.
Also, just because there would be 24C/32C CPUs does not mean everyone has to buy them.
If anything it's the complete opposite.
And gaming is what drives DIY.
This is not true. Higher core counts matter for literally everything except gaming. Find me a game that runs better on the upcoming 96 core Threadripper than on a 7800X3D and we will talk. Single core performance benefits ALL tasks. Multicore performance benefits productivity primarily.If anything it's the complete opposite.
Lots of single threaded or lightly threaded apps, which does not benefit from extra cores.Higher core counts matter for literally everything except gaming.