Intel Xeon Scalable Processor Family (Skylake-SP) Launch Coverage Central

4
Intel Skylake SP Wafer And Logo
Intel Skylake SP Wafer And Logo

At STH, we have been working on the Intel Xeon Scalable Processor Family launch for months. We also tend to keep our articles in a single-page format. Since this is perhaps the biggest mainstream architecture change in the data center in well over five years, there is simply too much to cover. As a result, we are creating this Intel Xeon Scalable Processor Family (formerly codenamed Skylake-SP) coverage central. Here you can find the latest in STH coverage.

Intel Xeon Scalable Processor Family (Skylake-SP) Overview and Architecture

The following articles discuss the Intel Xeon Scalable Processor Family (Skylake-SP) at a general level. These include pieces on branding, SKU lists, chipsets, and overall architecture details.

STH Skylake-SP Launch Coverage

STH Intel Xeon Scalable Processor Video Coverage

STH Skylake-SP Pre-Launch Coverage

Intel Xeon Scalable Processor Family (Skylake-SP) Benchmarks

We have a number of SKUs that we will be releasing benchmark data for over the next few weeks. Here is the benchmark data you are looking for all validated in our STH/ DemoEval lab.

  • Dual and Quad Socket Intel Xeon Platinum Cache Latency and Memory Bandwidth Performance
  • Quad Intel Xeon Platinum 8180 Benchmarks
  • Quad Intel Xeon Platinum 8176 Benchmarks
  • Dual Intel Xeon Platinum 8176 Benchmarks
  • Dual Intel Xeon Gold 6150 Benchmarks
  • Dual Intel Xeon Gold 6140 Benchmarks
  • Dual Intel Xeon Silver 4110 Benchmarks
  • Dual Intel Xeon Silver 4108 Benchmarks
  • Dual Intel Xeon Bronze 3106 Benchmarks
  • Dual Intel Xeon Bronze 3104 Benchmarks

More to come. As you may be able to infer, we do have a methodology we are following as we work through the SKU stack.

Note: We did pull AMD EPYC benchmarks just before launch. We have seen numbers move on production systems and ours is set to get a new motherboard refresh, new RAM, and more importantly new firmware this week. Performance has changed with new firmware and we wanted to use a production system.

Intel Xeon Scalable Processor Family (Skylake-SP) Hardware Vendor Systems and Product Family Releases

Here we will post information on the new hardware we are seeing at launch. We do expect Intel Xeon Scalable Processor Family systems to launch over the next few months and will cover those separately as those releases happen.

Beyond the above areas, head over to the STH Forums Processor section to ask questions. You can also follow us on RSS, Twitter, Facebook and YouTube to follow the latest coverage.

4 COMMENTS

  1. Just one request for testing: Please try to use a relatively recent Linux distro! Phoronix has done tests on the Skylake X series parts (that are actually the same silicon as LCC Xeons) and there are numerous workloads where upgrading even from Ubuntu 16.04 to 17.04 produce major performance improvements since these chips are using a pretty different architecture from the old Broadwells.

    I’d personally steer clear of Ubuntu and use a recent Fedora spin (Fedora 26 is practically launched), but if you want to use Ubuntu, at least try to use a recent version.

  2. Don – We actually have results coming for both Ubuntu 14.04 (for comparing to old data sets) and Ubuntu 17.04 for comparing new processors among each other.

    Ed – Johan did a great piece but the AMD system has a smaller PSU (better efficiency) than the Intel system. Also, the AMD reference system is not a commercial product you can buy so it is a bit hard to use as an efficiency part. We will be investigating but Johan’s initial findings are directionally correct from what we have seen over the past few months.

  3. Maybe I’m missing something here but where are the high frequency, single socket parts like the E5-16xx series that are great for workstations or low cost servers that need more memory than the E3 parts can support?

LEAVE A REPLY

Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.