It’s a Cascade of 14nm CPUs: AnandTech’s Intel Core i9-10980XE Reviewby Dr. Ian Cutress on November 25, 2019 9:00 AM EST
The most profitable process node in the history of Intel has been its 14nm process. Since 2014, the company has been pumping out CPUs built on a variety of configurations of 14nm – slowly optimizing for power and frequency. We used to call these variants 14+ and 14++, but as the next process node isn’t yet ready, rather than draw attention to a soon-to-be 6-year old process, Intel just calls it all ‘14nm class’. The latest launch on 14nm is Intel’s new Cascade Lake-X processors: high-end desktop hardware that gives a slight frequency improvement over Skylake-X from 2017 but it also has the first round of hardware mitigations. Today we’re testing the best CPU of the new list, the Core i9-10980XE.
The Ups and Downs of Intel’s High-End Strategy
Way back in June 2017, Intel first launched its Skylake-X high-end desktop processors. The Core i7-7900X was a 10-core processor built using the smallest silicon die from Intel’s enterprise processor range. It was on sale for $999, a noticeable drop from the $1729 pricing of the 10-core in the previous generation, and fit into a market where AMD had just started to launch its 8-core Ryzen processors for half this price. The benefits over AMD at the time, as explained in our review, came down to new vector extensions, more PCIe lanes, more memory channels, and a higher rate of instruction throughput, all equating to more performance – if the cost didn’t frighten you away.
AMD quickly launched 16-core processors and then 32-core processors into the high-end desktop market, turning most of the areas in which Intel had been winning into wins for AMD. The 16-core 1950X/2950X and the 32-core 2990WX were able to stifle the usefulness of Intel’s 10-core offerings by being much more competitively priced. In response, Intel moved up another step in its enterprise CPU silicon, and started offering up to 18 cores to the high-end desktop market, first with the Core i9-7980XE at $1979, and then the Core i9-9980XE at the same price but with a small clock increase.
For 2019, both companies have kicked it up a gear. AMD now offers for its mainstream platform 16 cores built on TSMC’s 7nm process with the Ryzen 9 3950X, which has a recommended price of $749. It also has a fundamental performance per clock advantage, as well as a higher frequency than Intel's HEDT parts. This now means that Intel’s 18-core CPU, at $1979, competes against AMD’s 16-core CPU at half the price and with better efficiency.
Today’s Launch: Cascade Lake-X and the Core i9-10980XE
In order to be competitive, Intel is doing the only thing it can do, based on what it has in its arsenal: the new 18-core Core i9-10980XE that comes out today is going to have a tray price of $979. The new Cascade Lake-X processor, based on the same silicon as Intel's already-launched Cascade Lake generation of Xeon processors, comes with many of the same features introduced for those parts. In particular, this means the new Intel HEDT chips come with hardware protections for the first round of Spectre/Meltdown security patches. Intel is launching a range of processors, from 10-core all the way up to 18-core.
The Core i9-10980XE is an 18-core processor that has a base frequency of 3.0 GHz (same as the 9980XE) and a turbo frequency of 4.6 GHz (+100 MHz higher than the 9980XE) and a turbo max frequency of 4.8 GHz (+100 MHz higher than 9980XE). It can support up to 256 GB of DDR4-2933 with a quad-channel design, and has a 165W TDP.
|Intel Cascade Lake-X|
|Core i9-10980XE||18C / 36T||3.0||3.8||4.6||4.8||165 W||$979|
|Core i9-10940X||14C / 28T||3.3||4.1||4.6||4.8||165 W||$784|
|Core i9-10920X||12C / 24T||3.5||4.3||4.6||4.8||165 W||$689|
|Core i9-10900X||10C / 20T||3.7||4.3||4.5||4.7||165 W||$590|
|Skylake-X (previous generation)|
|Core i9-9980XE||18C / 36T||3.0||4.5||4.7||165 W||$1979|
|Core i9-9940X||14C / 28C||3.3||4.5||165 W||$1387|
|Core i9-9920X||12C / 24T||3.5||4.5||165 W||$1189|
|Core i9-9900X||10C / 20T||3.5||4.5||165 W||$989|
If we compare the top parts from AMD and Intel, we get an interesting differential.
|Intel vs AMD
|Core i9-10980XE||AnandTech||Ryzen 9 3950X|
|18 / 36||Cores / Threads||16 / 32|
|3.0 GHz||Base Frequency||3.5 GHz|
|4.6 / 4.8 GHz||Turbo Frequency||4.7 GHz|
|18 MB||L2 Cache||8 MB|
|24.75 MB||L3 Cache||64 MB|
|256 GB||DRAM Capacity||128 MB|
|165 W||TDP||105 W|
|$979 (1ku)||Price||$749 (MSRP)|
What we have here are two processors that are technically in different markets: AMD is making the ‘high-end desktop market’ for its processors go beyond $749, while Intel’s HEDT market is now from $569 to $979. This means that Intel does have an advantage in this price range for memory controllers and PCIe lanes. It is worth noting that Intel is not launching a 16-core processor in this family, to compete directly with AMD’s 16-core. The official reason is that Intel doesn’t see a need to insert a product between the 10940X and the 10980XE in that price range; however as most people have gathered, not having a direct competition product on core count saves Intel some expected embarrassment in performance comparisons.
With that being said, AMD is also launching its newest HEDT processors today as well. The AMD Threadripper 3960X (24-core) and AMD Ryzen Threadripper 3970X (32-core) are (just) derivative designs of their enterprise processors, but signify that Intel has nothing to compete in this 24-core and above space.
|Intel vs AMD
|18 / 36||Cores / Threads||24 / 48||32 / 64|
|3.0 GHz||Base Frequency||3.8 GHz||3.5 GHz|
|4.6 / 4.8 GHz||Turbo Frequency||4.5 GHz||4.7 GHz|
|18 MB||L2 Cache||12 MB||16 MB|
|24.75 MB||L3 Cache||128 MB||128 MB|
|256 GB||DRAM Capacity||512 GB||512 GB|
|165 W||TDP||280 W||280 W|
If we were to compare the 10980XE to the 3960X/3970X, it wouldn’t necessarily be a fair fight, with the AMD processors costing a good chunk more. But comparing the 10980XE to the 3950X is comparing a mainstream processor against HEDT, so the mainstream CPU automatically loses on most memory bound and PCIe bound tasks.
If we put up a price list for the updated product families, it shows the following:
|$2000+||28/56||Xeon W-3175X ($2999)|
|TR 3970X ($1999)||32/64||$1750-$1999|
|TR 3960X ($1399)||24/48||$1250-$1499|
|$900-$999||18/36||Core i9-10980XE ($979)|
|Ryzen 9 3950X ($749)||16/32||$700-$799||14/28||Core i9-10940X ($784)|
|$600-$699||12/24||Core i9-10920X ($689)|
|$550-$599||10/20||Core i9-10900X ($590)|
|$500-$549||8/16||Core i9-9900KS ($513)|
|Ryzen 9 3900X ($499)||12/24||$450-$499||8/16||Core i9-9900K/F ($488)|
|Ryzen 7 3800X ($399)||8/16||$350-$399||8/8||Core i7-9700K/F ($374)|
|Ryzen 7 3700X ($329)||8/16||$300-$349|
|$250-$299||6/6||Core i5-9600K ($262)|
|Ryzen 5 3600X ($249)||6/12||$200-$249|
|Ryzen 5 3600 ($199)||6/12||Below $200||4/4||Core i3-9350K ($173)|
|*Intel quotes OEM/tray pricing. Retail pricing will sometimes be $20-$50 higher.|
Keep an eye on all our benchmarks, just to see where everyone ends up.
Post Your CommentPlease log in or sign up to comment.
View All Comments
Thanny - Wednesday, November 27, 2019 - linkZen does not support AVX-512 instructions. At all.
AVX-512 is not simply AVX-256 (AKA AVX2) scaled up.
Something to consider is that AVX-512 forces Intel chips to run at much slower clock speeds, so if you're mixing workloads, using AVX-512 instructions could easily cause overall performance to drop. It's only in an artificial benchmark situation where it has such a huge advantage.
Everett F Sargent - Monday, November 25, 2019 - linkObviously, AMD just caught up with Intel's 256-bit AVX2, prior to Ryzen 3 AMD only had 128-bit AVX2 AFAIK. It was the only reason I bought into a cheap Ryzen 3700X Desktop (under $600US complete and prebuilt). To get the same level of AVX support, bitwise.
I've been using Intel's Fortran compiler since 1983 (back then it was on a DEC VAX).
So I only do math modeling at 64-bits like forever (going back to 1975), So I am very excited that AVX-512 is now under $1KUS. An immediate 2X speed boost over AVX2 (at least for the stuff I'm doing now).
rahvin - Monday, November 25, 2019 - linkI'd be curious how much the AVX512 is used by people. It seems to be a highly tailored for only big math operations which kinda limits it's practical usage to science/engineering. In addition the power use of the module was massive in the last article I read, to the point that the main CPU throttled when the AVX512 was engaged for more than a few seconds.
I'd be really curious what percentage of people buying HEDT are using it, or if it's just a niche feature for science/engineering.
TEAMSWITCHER - Tuesday, November 26, 2019 - linkIf you don't need AVX512 you probably don't need or even want a desktop computer. Not when you can get an 8-core/16-thread MacBook Pro. Desktops are mostly built for show and playing games. Most real work is getting done on laptops.
Everett F Sargent - Tuesday, November 26, 2019 - linkLOL, that's so 2019.
Where I am from it's smartwatches all the way down.
Queue Four Yorkshiremen.
AIV - Tuesday, November 26, 2019 - linkVideo processing and image processing can also benefit from AVX512. Many AI algorithms can benefit from AVX512. Problem for Intel is that in many cases where AVX512 gives good speedup, GPU would be even better choice. Also software support for AVX512 is lacking.
Everett F Sargent - Tuesday, November 26, 2019 - linkNot so!
It compiles and runs on both Intel and AMD. Full AVX-512 support on AVX-512 hardware.
You have to go full Volta to get true FP64, otherwise desktop GPU's are real FP64 dogs!
AIV - Wednesday, November 27, 2019 - linkThere are tools and compilers for software developers, but not so much end user software actually use them. FP64 is mostly required only in science/engineering category. Image/video/ai processing is usually just fine with lower precision. I'd add that also GPUs only have small (<=32GB) RAM while intel/amd CPUs can have hundreds of GB or more. Some datasets do not fit into a GPU. AVX512 still has its niche, but it's getting smaller.
thetrashcanisfull - Monday, November 25, 2019 - linkI asked about this a couple of months ago. Apparently the 3DPM2 code uses a lot of 64b integer multiplies; the AVX2 instruction set doesn't include packed 64b integer mul instructions - those were added with AVX512, along with some other integer and bit manipulation stuff. This means that any CPU without AVX512 is stuck using scalar 64b muls, which on modern microarchitectures only have a throughput of 1/clock. IIRC the Skylake-X core and derivatives have two pipes capable of packed 64b muls, for a total throughput of 16/clock.
I do wish AnandTech would make this a little more clear in their articles though; it is not at all obvious that the 3DPM2 is more of a mixed FP/Integer workload, which is not something I would normally expect from a scientific simulation.
I also think that the testing methodology on this benchmark is a little odd - each algorithm is run for 20 seconds, with a 10 second pause in between? I would expect simulations to run quite a bit longer than that, and the nature of turbo on CPUs means that steady-state and burst performance might diverge significantly.
Dolda2000 - Monday, November 25, 2019 - linkThanks a lot, that does explain much.