Features
Here are the features Intel lists for the new 109xx series Core i9 CPU’s. These are taken from Intel’s documentation directly and apply to the entire Cascade Lake-X lineup. The only difference between these CPUs are core counts and clock speeds.
Intel® Turbo Boost Technology 2.0: Dynamically increases the processor frequency up to 4.6GHz when applications demand more performance. Speed when you need it, energy efficiency when you don’t.
Intel® Turbo Boost Max Technology 3.0: Now identifies the four best performing cores to provide increased single- and dual-core performance up to 4.8 GHz, quad-core performance up to 4.7 GHz. Natively supported with the latest Microsoft Windows* 10 releases, Linux* distributions based on kernels since 2017.
Intel® Deep Learning Boost: A new technology unique to Intel architecture designed to accelerate artificial intelligence for faster image tagging, image enhancement, and marker-less motion tracking.
Intel® Hyper-Threading Technology: Allows each processor core to work on two tasks at the same time for up to 36 total independent tasks (threads) providing parallel processing capability for better multi-tasking with threaded applications.
Intel® Smart Cache: 24.75 MB of shared cache allows faster access to your data by enabling dynamic and efficient allocation of the cache to match the needs of each core, which can help significantly reduce latency to frequently used data and improving performance.
CPU Overclocking Enabled (with Intel® X299 Chipset) Fully unlocked core multiplier, base clock, and memory ratios as well as options to modify power settings and per core overclocking enable ultimate flexibility for overclocking.
Solder Thermal Interface Material: Delivering improved thermal conductivity between the CPU die and the integrated heat spreader for improved overclocking capability.
Integrated Memory Controller: Supports 4 channels of DDR4-2933 memory. Support for memory based on the Intel® Extreme Memory Profile (Intel® XMP) specification.
PCI Express* 3.0 Interface: Supports up to 8 GT/s for fast access to peripheral devices and networking with up to 48 lanes configurable, per x16 port, as 1×16, 2×8, or 1×8 and 2×4 depending on the motherboard design.
Chipset/Motherboard Compatibility: Compatible with the Intel® X299 chipset with the latest BIOS and drivers.
What does this all mean?
So, where the rubber meets the road, what is the result of this? A few things. The feature list speaks for itself, but there are a few points I’ll expand on. However, the biggest improvement above all else are higher clock speeds. TDP’s haven’t changed and the feature set is pretty much the same. All the CPU’s in the Cascade Lake-X family have four additional PCIe lanes for a total of 48 via the CPU.
Intel also increased its T-junction temperature ceiling. This rating increased by two degrees to 86c. This should at least contribute to higher overclocking potential vs. the previous iteration of these CPUs. This by itself wouldn’t make a huge difference, but Intel does tout these CPUs as clocking quite high which I’ll talk about later. The 9980XE used a soldered heat spreader and the 7980XE didn’t. While Intel lists it here, it was an improvement it had already made earlier.
You can also now use a maximum of 256GB of RAM with the 109xx series CPUs. This is one much-needed improvement as we now have regular mainstream hardware which can now be configured with up to 128GB of RAM.
Deep Learning
Intel’s Deep Learning Boost feature is something I’ll talk about briefly. It’s best to get this straight from the horse’s mouth:
“A new set of embedded processor technologies designed to accelerate AI deep learning use cases. It extends Intel AVX-512 with a new Vector Neural Network Instruction (VNNI) that significantly increases deep learning inference performance over previous generations.”
The feature by itself isn’t exactly new either. It’s more like “new’ish.” The feature was first made available on the Cascade Lake server parts which have been out since Q2 of this year. However, anytime new instructions are added into the mix, it takes time for software to make use of it. As a result, people may not be able to do much with this today but over time that will change. It is important to note that AVX-512 is something that only Intel supports right now on the desktop. AMD is still only using AVX/AVX2 in its Ryzen 3000 series and third-generation Threadripper.
Deep Learning benchmarks have often been run on GPU’s like the RTX 2080Ti and RTX Titan. GPU’s are often the way to go in this application due to efficiency and lower cost. Intel did recently beat NVIDIA in Deep Learning benchmarks using a dual-processor Xeon Platinum 9282 system. The only reason I mention this is because the Core i9 10980XE is a far cry from the Xeon Platinum 9282. That said, there are and will be applications that can leverage these instructions. Right now, AVX-512 workloads are one area where Intel is the only game in town.
Boost Clocks
In case you aren’t familiar with them, I’ll talk about Intel’s Intel’s Turbo Boost 2.0 and Turbo Boost Max Technology 3.0. These are features that need to be touched on. Like all modern CPUs, the Core i9 10980XE has a base frequency and then a turbo boost clock value. This is something that we’ve seen on both Intel and AMD CPU’s over the years. The Core i9 10980XE has a base frequency of 3.0GHz and a Turbo Boost 2.0 clock speed of 4.6GHz. Like AMD’s CPU’s, these higher frequencies are something you will only ever really see when you are running single-threaded applications and the system is for the most part otherwise idle. When using multithreaded workloads, the clock speeds will generally go beyond the base frequency but, will fall well short of the maximum frequency advertised.
One thing that comes to mind is the difference between AMD and Intel’s boosting methods. People have been very upset with AMD over the fact that its advertised boost clocks can only be achieved by one or two cores out of however many a given CPU model has. With Intel CPU’s, all of the cores are capable of reaching the advertised boost clocks. Thus, each of the 18 cores present on the 10980XE should be capable of achieving 4.6GHz were you to isolate any one of them and clock it to that speed. This is where Intel’s Turbo Boost Max Technology 3.0 comes into play.
This feature takes the best four cores on the processor and clocks them slightly higher. These cores are all capable of achieving a 4.8GHz clock speed. So, for single-threaded and lightly threaded workloads such as gaming or general tasks, you could see clocks around 4.8GHz. HEDT processors in the past have fallen well short of the mainstream CPU clocks and still do somewhat, but the 10980XE is pretty close to say a 9900K. Close enough, that you wouldn’t or shouldn’t have to sacrifice too much in the way of single-threaded performance just to get the I/O and memory bandwidth offered by the HEDT platform. I’ll come back to this point a bit later so keep it in mind.