Review: Core i9 10980XE processor (Intel 18-core galore)

Cheap... not really.
But indeed a good upgrade for someone who had the 10-core version on X299 during the age of 7000 series, near double the cores for $1000 and that's it !

For anyone else... NO, NO, NO and NO. Building a new PC with X299 today is pure insanity.

---
A bit side story...

I kept looking for an upgrade to my old 6800K system on X99 but unfortunately all CPUs on that platform are still expensive (6900K, 6950X, or Xeons).
Guess I'll have to wait until new EPYC Rome replaces all datacenter installations en masse and maybe I can snag a cheap 18-core Xeon for X99...
- Xeon E5-2699 v3 (18c)
- Xeon E5-2698 v3 (16c)
- Xeon E5-2697 v3 (14c)

These would make a nice CPU upgrade for the already old X99 platform, assuming I can snag one on the 2nd hand market somewhere...

 

Can you post the newegg link? I dont see it there.
nm... https://www.newegg.com/intel-core-i9-10th-gen-core-i9-10980xe/p/N82E16819118112

 

Hardware Unboxed has the entire Casket Lake X lineup... from an unknown source:



They get beaten at all price points, basically.

 

its like intel loves us for only a 1000

 

Ouch, that was harsh.

 

@CyberSparky
I know, still like to post for ppl that dont know better
or that come by search engine and no idea that some
here would still say "best ever" while their unit is literally on fire in the background

 

You dont know how PCIe works, Gen 4 x4 only equals to gen 3.0 x8 in Bandwidth, but if you connect Gen 3.0 x16 GPU to Gen 4.0 x8 slot, the card will only run at x8 speed, because it needs electric connection across all 16 lanes.
Your theory only valid in some future where every device is Gen 4, then a gen 4 GPU can use 8 lanes and be as fast as gen 3 16 lanes, but right now the only gen 4 devices we have is some ssds.
Everything else is Gen 3, i mean Capture cards, GPUs and RAID cards [things editors and most workstations use]

 

Forbes so far was the only realistic Reviewer showing us how it performs Stock vs 4.7Ghz [light overclock] vs 3950x and 3950x overclocked
This CPU at 4.7Ghz is unbeatable

Just checked, plenty on Xeon E5-2699 v3 eBay

 

Forbes... wow, you must have been fishing far and wide to find the results you wanted .

 

Okay Chief or is it Chef? Nicely put

 

Since I'm like 90% gaming and 10% productivity stuff, I can hardly see any reason to upgrade from my overclocked i9-7900X. I run my games in 2560x1440 on my 2080Ti so at least until graphics card can speed up and put more demand on the CPU at higher resolution it doesn't seem like its worth the $$ to upgrade yet. Maybe have to wait and see where AMD and Intel fall when both have PCIe 4.0 and graphics cards that can use it.

 

Their tech reviews are sourced from this guy https://twitter.com/antonyleather
Also, how hard can you shill for AyMD? Because you dont like their Truthful results you hate them? Facts are facts, deal with it.
Intel CAN be overclocked, everyone knows it, 4.7Ghz is enough to lead the pack.
Personally i think ill got with 3960x, the platform has future and I dont know how many years it will take for Intel to release a CPU that outright beats 3960x without overclocking.
The only thing that scares me personally is value, AMD CPU lose value like old prostitute, the 1350$ I Paid for the CPU will turn into 700$ next year, I hope this new generation keeps value But unlike Intel, AMD wants their old CPUs to lose value, so people buy new ones.
I really want to give AyMD a try, the CPU is out of stock, if by the time it comes in-stock Intel wont have it for sale for their employees [50% discount for all CPUs if you work for Intel] ill keep the 24c AyMD
If not then, 500$ for the 18 core + 500$ for one oft he best mobos = 1K for CPU/Mobo set
My only problem is that right now i have 64GB of AMD GSKILL RAM, Neo something, when i got it 2 months ago, their QVL said its compatible iwth my z370 Gamer 7 and it works full speed, and im sure it will work with TR3 too, but x299 is unknown...

Seriously? How old are you? Because When i see all these Intel Dead, Cascet lake etc AyMD fanboy shilling I laugh, like none of you have an idea about Intel.
Unlike AMD, Intel is one of the biggest corporations on planet that has its hands into most business on planet earth, every otehr Corporation has something made by Intel and Intel makes more then just CPUs.
The amount of money Intel makes is more then most 3rd world and even second world countries budgets, the amount of properties like buildings, factories, land, companies etc they own is an immeasurable value [Im sure someone measured, but i dont have this number], i know its a lot, intels owns property, buildings, otehr companies in most developed and even not that developed countries.
AyMD has ZERO fabs, Intel has 15 Wafer fabs in 10 locations and 13 assembly/test sites.
Last time Intel released a number of how many Coders work for them to optimize code for Intel CPUs [they help 3rd parties for free], its MORE then WHOLE AMD stuff.
On top of all of the above, Intel has hundreds of contracts for supercomputers and workstations [post sale support] that sighed for years forward, 10-20 years into future and im sure even longer and there are supercomputers build as we speak.
If you think that Intel can fall, you have to be either Trolling or be AMD teenage fanboy.
I mean almost every WEEK there is info that Intel is OVERsold, with their 15 Fabs, they making less products than they can sell.

Now tell me again, how Intel going to flop and die..

 

@MegaFalloutFan
physical connection 16x and bandwidth speed 16x are not the same.
reducing the slot speed to 4x worked with the 16x 3.0 card (1080 as well as the 1080ti/2080ti from a friend),
and produced slightly better results for benchmarks/games as expected.
multiple tests from different sites have proven that as well, no matter if reducing 16x to 8x,
and/or going with reduction from 3.0 to 2.0.

 

-->Careful there, you DO NOT EVER want zero competition!! Do you forget what Intel did to us for so many years, applying a very deliberate and calculated level of 'innovation' - amounting to practically zero, while absolutely ripping off ALL customers.
-->We need AMD to profit very well from Zen2 and Zen3 for the next 2-3 years, Intel to extract the finger from its ass ass start real innovation, then BOTH companies (the more the better) to REMAIN competitive. At the same time...
-->We NEED this to happen to Ngreedia too - let's hope all the best for rDNA2..., !!

Competition, in this context, is the primary driver for innovation. Without that, profit will always stifle innovation.

 

Don't forget ARM.
Their CPUs are becoming quite strong, and they are catching up in GPU department as well (at least on lower end).

And the Chinese are progressing quite fast with their CPU/GPU designs...
I think in the next decade we'll have a 4 or even 5-way battle between silicon giants.

 

Then dont try AMD processors, they are not worth your time.

 

There is absolutely no doubt Intel will come back from this, just like they have before with C2D and then X58 (which was probably their best platform to date). They have the R&D budget and are flush with cash to sustain themselves over several years of AMD domination. That said, there's no doubt AMD have completely blindsided them, and no one in their right mind can possibly defend their current lineup given all the independent benchmarks we have been presented with.

 

I agree Intel will come back from this, but I don't at all think this is going to be C2D all over again.
For one thing, Intel's progress with 10nm has been delayed by several years, and they already seem to be giving up on that and moving smaller since by the time they complete 10nm, it'll be already be obsolete. To my understanding, architectures need to be optimized per-node, so, every month Intel delays something <=10nm, the more time they're buying AMD to release something better.

But here's the more important issue we have to accept: x86 as we know it is at a dead end. We're practically at the limits of silicon transistors, and despite how vastly different Intel and AMD have their architectures, they both have roughly the same IPC; the real-world performance differences are actually quite minimal, and can be narrowed down to things like cache. We can keep adding more cores, but most applications don't benefit from that and never will. Whatever Intel supersedes their Core i# series with, the architecture is going to have a very minimal per-core IPC improvement. I'm sure it'll have a vastly superior HT and it'll have more cores for a lower price, but otherwise I don't see how it can handily outperform Zen without making compromises.
In theory, there's quite a lot that could be done to x86 to improve it, either in terms of efficiency or raw performance. But the problem is backward compatibility. Pretty much the reason x86 is the most used architecture is because of binary compatibility. And it goes deeper than that: software runs better depending on the design of the pipelines. Bulldozer wasn't anywhere near as bad of an architecture as people made it out to be, but the problem is you had to write something for that architecture in order to tap into its true potential. Nobody was going to do that for AMD. So, even though Intel has the money and resources to create the best architecture x86 could ever be, they won't do that because that means current software is likely to take a performance hit, and that is not how you sell a product. Regressions for the sake of a better future is never acceptable to consumer products.

 

Yet AMD is capable of cramming 64 cores with SMT in one single CPU, and also make each one of them faster than equivalent Intel cores.

No x86/x64 is not dead.
Monolithic chips are. And until Intel finally gets that, they will fall more and more behind.

 

None of that disproves anything I just said.

I didn't say x86 is dead, I said it's at a dead end. I basically hinted before that x86 isn't going anywhere. Intel is capable of outdoing AMD, but not by much. I think we all know Intel's next major architecture isn't going to be monolithic, because it just isn't cost effective when trying to pump out a lot of cores. But all that's going to do is just get them more cores at more affordable prices. It isn't going to give them a per-core IPC advantage, just as it didn't do for AMD.
So my point remains:
Individual x86 cores don't have much room to improve. Silicon transistors are almost as small as they can get. Adding more cores isn't going to make everything run faster and it is unrealistic to expect everything to adapt. x86 isn't going to die off in the foreseeable future, but its progress will stagnate very soon (which is what I meant by "dead end"). AMD's chiplet approach breathed new life into x86, but only for high-end applications. It didn't really do anything for the average person other than lower prices.
But, perhaps it doesn't need to get any better. More and more workloads are being pushed to GPUs, and quantum computers will soon be handling the heavy tasks that traditional binary processors just aren't fit for. I see x86 to be like the internal combustion engine: decently well refined and pretty much as good as it's ever going to get, but still found everywhere and not effectively obsoleted for the majority of people.