Friday, October 25th 2019
Intel Core i9-10980XE "Cascade Lake-X" Benchmarked
One of the first reviews of Intel's new flagship HEDT processor, the Core i9-10980XE, just hit the web. Lab501.ro got their hands on a freshly minted i9-10980XE and put it through their test bench. Based on the "Cascade Lake-X" silicon, the i9-10980XE offers almost identical IPC to "Skylake-X," but succeeds the older generation with AI-accelerating DLBoost instruction-set, an improved multi-core boosting algorithm, higher clock speeds, and most importantly, a doubling in price-performance achieved by cutting the cores-per-Dollar metric by half, across the board.
Armed with 18 cores, the i9-10980XE is ahead of the 12-core Ryzen 9 3900X in rendering and simulation tests, although not by much (for a chip that has 50% more cores). This is probably attributed to the competing AMD chip being able to sustain higher all-core boost clock speeds. In tests that not only scale with cores, but are also hungry for memory bandwidth, such as 7-zip and Media, Intel extends its lead thanks to its quad-channel memory interface that's able to feed its cores with datasets faster.
As we move to gaming and gaming-related 3D benchmarks, we see the i9-10980XE only marginally ahead of the 3900X in the 3DMark Physics test. This lends credibility to the report where the unreleased 16-core 3950X was seen beating the i9-10980XE in this particular test. With gaming still being the forte of mainstream-desktop processors with lower core counts and higher clock-speeds, we see the likes of the i9-9900K racing ahead on account of significantly higher speeds while having sufficient muscle to handle games. Find more interesting results in the Lab501 review here.
Source:
Lab501
Armed with 18 cores, the i9-10980XE is ahead of the 12-core Ryzen 9 3900X in rendering and simulation tests, although not by much (for a chip that has 50% more cores). This is probably attributed to the competing AMD chip being able to sustain higher all-core boost clock speeds. In tests that not only scale with cores, but are also hungry for memory bandwidth, such as 7-zip and Media, Intel extends its lead thanks to its quad-channel memory interface that's able to feed its cores with datasets faster.
81 Comments on Intel Core i9-10980XE "Cascade Lake-X" Benchmarked
Behold, the same damn thing, but for AMD.
so it's going to be even boring benchmarks then 2080Super ;-).
As to everything else, if you’re gaming at 1440p or 4K - you do not notice the difference. If anything, my current X399 build seems to have more consistent FPS (“smoother”) at 4K vs my previous AM4 or Z270 builds. SIngle threaded performance is a terrible metric to build a system around FYI - the tasks and applications that are truly single-threaded are mostly the stuff of benchmarks now. ITX isn’t realistic with X399 due to the size of the socket. Windows optimizations / lack thereof had much to do with NUMA vs UMA and this has been resolved for Zen 2.
Overclocking is where I have to disagree most. Zen overclocking has more to do with building a custom cooling solution, tweaking PBO options / wattage / scalar, and many hours of memory adjustment and testing. To do it well is harder than bumping vcore and cranking the multiplier on a mainstream intel board - this is especially true regarding the memory.
BTW I'm using my Optane SSD 900P AIC with a X570. All client Optane devices are standard NVMe SSDs including those H10 hybrids. It's hilarious to think a standard NVMe SSD will not work on AMD platform.
Only server part like Optane DC Persistence Memory are limit to Xeon platform and anyway they are not compatible with desktop or HEDT platform and not even Xeon E or Xeon W.
It works day one as those drives are just NVMe SSDs.
But then i was like
If these results are true, intel… sorry i mean Shintel is gonna have a hard year a head of them on the desktop marked. Intel HEDT is a dying breed as it is right now.
Well this just confirms for my needs Ryzen 9 3950X is a better choise for me. I want the best possible gaming performance mixed with high core count for multi tasking perpose and other core needy stuff and there it seems so far Ryzen 9 3950X doing a better job based on the 3900X gaming results. AMD wins my money this round.
9700K is 8 cores / 8 threads
apparently more threads was more beneficial for such benchmarks
PS....now I see a 9900K is in there too which is 8 cores / 16 threads....OK so 8700K beating 9900K is a bit perplexing. However, not perplexing in my initial response between 8700K and 9700K
8700k is 6 cores so ring bus latency is 20% lower by average. Unless 8700k is limited by its 6 cores throutput, its always better than a Intel 8 core CPU. Same goes for 7700k, if game does not overwhelm a quad core then 7700k is slightly faster than a 8700k.
Next gen Intel high end 10 cores only make this situation worse.
The problem for Intel right now is more core == less gaming performance.
This is their fault in architecture, AMD does not affected by this. Each CCX is a 7700k equivalence and can enjoy low latency. For old games 1CCX is enough so Ryzen will act like a 7700k. For newer games they can group threads into group of 4 and make them run on different CCX to minimize cross CCX latency. And AMD is trying to improve that by adding a larger cache to the CCX.
and no thanks (jk) for worrying me about waiting for comet lake. the yet to be released 10 core / 20 thread Intel CPU. I've been holding out for the new Z490 series chipsets and 10 core / 20 thread CPU. Thinking it would be just that much better for gaming. Though, I use my workstation as an audio workstation running Pro Tools, but also game on it just as much. So I'm trying to upgrade into a system that still gives me the best of both worlds.