Intel has confirmed that they are going to be bringing their first Multi-Chip Package deal design Xeon CPUs to the market within the type of Cascade Lake Superior Efficiency or Cascade Lake-AP for brief. We’ve been listening to reviews in regards to the new collection for some time now however Intel has now formally revealed the brand new Xeon lineup and in addition talked about some early efficiency numbers.
Intel Declares Cascade Lake Superior Efficiency Xeon Processors – Extra Cores, Extra Reminiscence Lanes, Quicker Clock Speeds, Optimized Cache and Spectre/Meltdown Mitigations Included
Intel isn’t speaking a lot in regards to the Cascade Lake Superior Efficiency specs as these will probably be offered in the course of the Supercomputing 2018 in early November.
Associated Intel Z399 Chipset Motherboards To Change X299 Platform, Alleges Rumor – Most of 22 Cores on The LGA 2066 Socket
Right this moment, Intel is sharing the preliminary particulars together with some efficiency numbers versus their very own flagship Xeon, the 8180 Platinum and AMD’s EPYC 7601. Intel is stating that the brand new Xeon class is constructed upon 20 years of the Xeon innovation. A number of the options within the new lineup embody:
- Management Efficiency
- Optimized Cache Hierarchy
- Increased Frequencies
- Safety Mitigations
- Intel Deep Studying Increase (VNNI)
- Optimized Framework and Libraries
We stay extremely centered on delivering a variety of workload-optimized options that finest meet our prospects’ system necessities. The addition of Cascade Lake superior efficiency CPUs and Xeon E-2100 processors to our Intel Xeon processor lineup as soon as once more demonstrates our dedication to delivering performance-optimized options to a variety of consumers.
– Lisa Spelman, vice chairman and basic supervisor of Intel Xeon merchandise and knowledge middle advertising
Associated Intel Formally Confirms Cascade Lake Superior Efficiency Processors, Makes use of MCP With Up To 48 Cores, 12 Channel Reminiscence Help – Reveals First Efficiency Numbers Versus AMD EPYC
It seems to be like the brand new Xeon processors will probably be totally rounded up, not solely that includes the above-mentioned applied sciences but additionally that includes much more cores and reminiscence channels. Intel is revealing that the Xeon Cascade Lake Superior Efficiency processors will home as much as 48 cores and 96 threads on a single bundle which will probably be using a Multi-Chip Package deal (MCP).
Intel’s Xeon processors presently home as much as 28 cores and 56 threads so the brand new AP collection processors could be a superb enhance in efficiency, simply from the upper variety of cores. Along with that, the chip will characteristic assist for 12 DDR4 reminiscence channels, an enormous bump from the 6 channel reminiscence that will probably be featured on the Cascade Lake-SP Xeons.
The efficiency numbers have been taken from a 2 socket server, that includes two 48 core chips which imply a complete of 96 cores and 192 threads. The 12 channel reminiscence will give a complete of 24 DIMM slots and contemplating you’ll be able to fill that up with some actually dense ECC reminiscence, we will probably be as much as Three TB of reminiscence assist. Along with ECC reminiscence, the brand new Xeons may also assist the Optane DC Persistent reminiscence, that includes capacities of as much as 512 GB. 24 of those within the 2S server would give a mind-boggling 12 Terabytes of system reminiscence. This is able to ship an unprecedented quantity of reminiscence bandwidth which is just attainable via a CPU within the class of Cascade Lake-AP.
We will additionally inform that the platform is of a approach totally different design in comparison with Purley which homes the Skylake-SP and would home the upcoming Cascade Lake-SP Xeons. We will in all probability be a distinct socket which was rumored to be the BGA 5903 and therefore, a distinct platform will probably be proven.
Now coming to the official efficiency numbers, Intel claims that the 2S Cascade Lake-AP server is:
- 3.4x Quicker Than AMD EPYC 7601 (2S) in Linpack
- 1.2x Quicker Than Intel Xeon Scalable 8180 (2S) in Linpack
- 1.3x Quicker Than AMD EPYC 7601 (2S) in Stream Triad
- 1.83x Quicker Than Intel Xeon Scalable 8180 (2S) in Stream Triad
- 17.0x Pictures Per Second Versus Intel Xeon Scalable in AI/DL Inference
Following are the testing methodologies used for testing as talked about within the presentation footnotes:
Efficiency outcomes are based mostly on testing or projections as of 6/2017 to 10/3/2018 (Stream Triad), 7/31/2018 to 10/3/2018 (LINPACK) and seven/11/2017 to 10/7/2018 (DL Inference) and will not mirror all publicly accessible safety updates.
LINPACK: AMD EPYC 7601: Supermicro AS-2023US-TR4 with 2 AMD EPYC 7601 (2.2GHz, 32 core) processors, SMT OFF, Turbo ON, BIOS ver 1.1a, 4/26/2018, microcode: 0x8001227, 16x32GB DDR4-2666, 1 SSD, Ubuntu 18.04.1 LTS (4.17.0-041700-generic Retpoline), Excessive-Efficiency Linpack v2.2, compiled with Intel(R) Parallel Studio XE 2018 for Linux, Intel MPI model 220.127.116.11, AMD BLIS ver 0.4.0, Benchmark Config: Nb=232, N=168960, P=4, Q=4, Rating = 1095GFs, examined by Intel as of July 31, 2018. in comparison with 1-node, 2-socket 48-core Cascade Lake Superior Efficiency processor projections by Intel as of 10/3/2018.
Stream Triad: 1-node, 2-socket AMD EPYC 7601, examined by AMD as of June 2017 in comparison with 1-node, 2-socket 48-core Cascade Lake Superior Efficiency processor projections by Intel as of 10/3/2018. DL Inference:
Platform: 2S Intel Xeon Platinum 8180 CPU 2.50GHz (28 cores), HT disabled, turbo disabled, scaling governor set to “efficiency” by way of intel_pstate driver, 384GB DDR4-2666 ECC RAM. CentOS Linux launch 7.3.1611 (Core), Linux kernel 3.10.0-514.10.2.el7.x86_64. SSD: Intel SSD DC S3700 Collection (800GB, 2.5in SATA 6Gb/s, 25nm, MLC).
Lastly, Intel is saying that the Xeon Cascade Lake Superior Efficiency processors will probably be launching within the first half of 2019 which additionally follows the sooner rumors we had heard.
Intel confirmed their different Xeon households some time in the past too which would come with:
- Cascade Lake is a future Intel Xeon Scalable processor based mostly on 14nm know-how that may introduce Intel Optane DC persistent reminiscence and a set of recent AI options known as Intel DL Increase. This embedded AI accelerator will velocity deep studying inference workloads, with an anticipated 11 occasions quicker picture recognition than the present technology Intel Xeon Scalable processors after they launched in July 2017. Cascade Lake is focused to start delivery late this 12 months.
- Cooper Lake is a future Intel Xeon Scalable processor that’s based mostly on 14nm know-how. Cooper Lake will introduce a brand new technology platform with important efficiency enhancements, new I/O options, new Intel DL Increase capabilities (Bfloat16) that enhance AI/deep studying coaching efficiency, and extra Intel Optane DC persistent reminiscence improvements. Cooper Lake is focused for 2019 shipments.
- Ice Lake is a future Intel Xeon Scalable processor based mostly on 10nm know-how that shares a standard platform with Cooper Lake and is deliberate as a quick follow-on focused for 2020 shipments.
It is going to be attention-grabbing to see how Intel packs the brand new Cascade Lake-AP processors contemplating their HCC dies to scale as much as 28 cores. We could also be a totally unique die designed for the AP collection processors however I assume we now have to attend until the Supercomputing 2018 convention to seek out out extra. Until then, tell us your ideas under in regards to the Cascade Lake-AP processors and the way you assume they’ll evaluate towards AMD EPYC processors, particularly when 7nm EPYC chips are going to launch subsequent 12 months.
Along with all the Cascade Lake Superior Efficiency information, the Xeon E-2100 (1S) CPUs additionally hit basic market availability at this time. They have been introduced some time again and at the moment are accessible by way of Intel and their distributor companions across the globe, that includes as much as 6 cores, 12 threads, and quicker reminiscence assist.
Intel Xeon SP Households:
|Household Branding||Skylake-SP||Cascade Lake-SP||Cascade Lake-AP||Cooper Lake||Ice Lake-SP||Ice Lake-AP|
|Course of Node||14nm+||14nm++||14nm++||14nm++||10nm+||10nm+|
|Platform Identify||Intel Purley||Intel Purley||TBA||Intel Whitley||Intel Whitley||TBA|
|MCP (Multi-Chip Package deal) SKUs||No||No||Sure||No||No||Sure|
|Socket||LGA 3647||LGA 3647||BGA 5903||LGA 4189||LGA 4189||TBA|
|Max Core Depend||Up To 28||Up To 28||Up To 48||Up To 28||TBA||TBA|
|Max Thread Depend||Up To 56||Up To 56||Up To 56||Up To 56||TBA||TBA|
|Max L3 Cache||38.5 MB L3||38.5 MB L3||TBA||TBA||TBA||TBA|
|Reminiscence Help||DDR4-2666 6-Channel||DDR4-2933 6-Channel||DDR4 2933 12-Channel||TBA||8-Channel||TBA|
|TDP Vary||140W-205W||165W-205W||TBA||TBA||Up To 230W||TBA|
|Competitors||AMD EPYC Naples 14nm||AMD EPYC Rome 7nm||AMD EPYC Rome 7nm||AMD EPYC Rome 7nm||AMD EPYC Milan 7nm+||AMD EPYC Milan 7nm+|
Share Tweet Submit