What's going on with all the new motherboards & cpu's , why the downgrade?

[lou4asus]

Hi,
In the past, upgrading normally meant, better, faster, and more, more, more than the previous.
For some reason, this time the specs seem to go down.

For example
My old system is:
CPU i9 10980xe (48 lanes )

Motherboard
Asus prime x299 Memory 256 gigs MAX
48 lanes max
--------------------
New system to buy:
i9 13900k (28 lanes)
Motherboard ASUS ROG Strix Z790- Memory 128 Max
28 Lanes
----------------------------

Please forgive me for my ignorance, my ability is limited to just assembling new components.
But even limited to that, I do know that 256 gigs of ram are better than 128 and 48 lanes (seems) better than 28.

In my work, I use 80% of my 128 gigs installed, so I really don't need any more than that, however I was expecting the max ram to at least stay the same maxing out at 256 gigs of ram capability. I was surprised to see most i9 13900k boards max out at 128.

Why are they limiting and why not 48 lanes ??

Thanks!

[Murph_9000]

X299 is HEDT, but both AMD & Intel have abandoned HEDT for now. AMD more or less won the HEDT war with Threadripper 3000, Intel gave up, then AMD also exited the HEDT segment with 5000 series (not sure what will happen with 7000 series, as it's only just hit production). There are currently no HEDT processors or platforms in the current generations.

It's unfair to compare Z790 and X299; the fair comparison would be Z490 and Z790, where there's a small increase in PCIe lanes. Unfortunately both AMD and Intel seem to be sticking to the limit of 128GB RAM for mainstream desktops, something I had rather hoped would rise to 256GB with this generation; but they want to continue to force people across to the professional workstation segment that's more expensive and profitable than the former HEDT segment. They just don't want to allow enthusiasts to have that much IO or memory, at present, unless they are super rich. PCIe gen 4 & 5 have partially compensated for the limited lanes, as you now have 2x or 4x the bandwidth in addition to a few more lanes, but you are kinda stuck if you actually need lots of cards.

Board vendors can get a little creative with how they use the lanes, but only a little. Your 10980XE was the last gasp of HEDT from Intel, you now have to go to Xeon workstation if you want bigger from them.

[lou4asus]

So that's interesting in many ways. I'm assuming you mean that the cpu "XE" is the HEDT vs the " K" in terms of performance ratings.
Interesting because I originally had the 6 core i7 8700K and upgraded to the i9 10980xe. My expectations on that particular upgrade were optimistic, since I was going from a 6 core to an 18 core.

I immediately ran real world benchmarks after I built it and recorded my simulation & render times from the 8700K vs 10980xe.
I was soooo disappointed because the 8700K simulated a fluid simulation scene in 8 minutes and the 10980xe did it in 7 minutes, 10 seconds. (Similar with all other tests too)

Fast forward 2 years, I am looking to upgrade now from the 10980xe to the i9 13900k (or ks )
Hoping that my results won't be as horrible as the previous build.


Ha, so I take the "HEDT" title with some hesitation, since the K vs xe had similar performance. (in my case)
So frustrating for people like me that just want to believe that 18 cores are 3X faster than 6 cores, or that 5.8 ghz should be way faster than 3.2ghz.
That is until we run our real world tests and hear Intel faintly laughing in the far distance.

Thanks for the info!

[Murph_9000]

Yes, the X and XE processors were the HEDT offering. Despite the very similar marketing names, the HEDT processors were a completely different family from the mainstream processors. They were cut down Xeon chips (big Xeon, not the Xeon branded mainstream chips; Intel's marketing confusion goes both ways).

Some workloads scale well with the number of cores, others not so much. Your software just wasn't able to take full advantage of those cores for a single task, as the 10980XE has does have higher theoretical raw performance (compared to a 10900K). Moving up to a 13900K should be a good overall improvement for you with the type of task that doesn't scale well with core count; it's the same family as the 8700K/10900K, with higher clocks, more cache, faster memory and IO, and improved IPC (instructions per clock).

With a suitably scalable workload, the 10980XE was significantly faster than the 10900K (and the 8700K). E.g. Cinebench R23 multi-core scores (pulled off the web) are about 15,000 for the 10900K and 24,000 for the 10980XE. Other workloads could actually be faster on the 10900K due to higher clock and the iGPU, despite it being the slower chip in raw total performance.

I'm using the 10900K for comparison purposes above to keep it in the same generation as the 10980XE.

[WHO_FARTED?]

also too, i don't know what your using your computer for, just gaming or for productivity? like murph was saying hedt and main stream k are really 2 different tools. not that you can't do both on either, they just really have 2 separate purposes. which may be why you didn't notice a great improvement in whatever your doing.