Four Intel Xeon Platinum 8180 processors delivered 112 cores and 224 threads in ServeTheHome’s July 2017 test—but the headline is throughput, not a universal speed advantage. The system pulled as much as 1,336W in the lab and was estimated to cost about $50,000 or more, making workload fit, power, and software licensing central to its value.
What was tested
ServeTheHome tested four Xeon Platinum 8180 CPUs in Intel’s S4PR1SY2B four-socket platform. Each processor was specified at a 2.5GHz base clock, up to 3.7GHz turbo, and 38.5MB of L3 cache. The system therefore had 112 physical cores and 224 threads. It used 768GB of memory across 24 32GB DDR4-2666 DIMMs, an Intel DC S3700 400GB OS SSD, and Ubuntu 14.04, Ubuntu 17.04, and CentOS 7.2. ServeTheHome’s July 11, 2017 benchmark report describes the configuration and results.
How the benchmark mix worked
The test suite used ServeTheHome’s Linux-Bench scripts, selected as a cross-platform baseline and built with gcc. The Linux kernel test compiled version 4.4.2 with make using every thread and reported compiles per hour. Other tests targeted distinct kinds of work:
- c-ray 1.1: highly parallel ray tracing.
- 7-Zip: compression and decompression throughput.
- NAMD: molecular modeling.
- Sysbench CPU: processor throughput.
- OpenSSL: signing and verification.
- UnixBench: Dhrystone and Whetstone tests.
- Redis: set/get requests against a single instance.
- Intel Memory Latency Checker: an initial look at inter-socket latency.
This is a suite for observing server throughput across varied workloads, not a direct measure of responsiveness in lightly threaded desktop applications.
#1 Best Overall
What the results indicate
Strong scaling in parallel workloads
The clearest case for the quad 8180 is work that can keep many cores busy. Kernel compilation, ray tracing, compression, molecular modeling, and several of the other workloads can benefit from parallel execution, memory bandwidth, or both. ServeTheHome reported that the system beat the fastest quad Broadwell system it had tested, the Xeon E7-8890 V4, in kernel compilation, and described a generational improvement in 7-Zip. These are the author’s qualitative comparisons; exact per-test values should be read from the original charts rather than inferred from those descriptions.
OpenSSL results have an accelerator caveat
The reported OpenSSL tests ran without an Intel QuickAssist PCH. QuickAssist support could affect results in deployments using compatible hardware and software, so the benchmark should not be treated as a comparison of every possible accelerated configuration.
Rank #2
- Universal Product Code: This processor comes with UPC 735858340267 for easy identification and verification
- Product Weight Specification: The unit weighs approximately 2.000 lbs, making it suitable for standard server installations
- High-Performance Processing Power: Features 26 cores running at 2.00GHz base frequency for demanding enterprise workloads and data center applications
- Large Cache Memory: Equipped with 35.75MB of cache memory to enhance processing efficiency and reduce latency for complex computational tasks
- Server-Grade Processor: Intel Xeon Platinum 8164 model SR3BB designed for professional server environments and high-performance computing systems
Inter-socket latency is a configuration-specific result
The initial Memory Latency Checker result showed better idle inter-socket latency than ServeTheHome’s AMD EPYC 7601 system equipped with DDR4-2400. That is a comparison between particular tested systems at a point in time, not a general ranking of the two processor architectures or all memory configurations.
Why the later Cinebench result needs caution
In a July 14, 2017 follow-up, Patrick Kennedy reported Cinebench R15 scores above 9,300 and later above 11,000 using release drivers. He also observed roughly 15% run-to-run variation and said the 224-thread Windows scheduler was struggling; dual-socket 8180 runs were more consistent. He noted that the default render scene was not complex enough. The score is therefore sensitive to workload and software scheduling, rather than a stable all-purpose measure of quad-socket performance. The Cinebench R15 follow-up provides that context.
Rank #3
- PART NUMBER: CD8069504201101
- CPU SERIES: 2ND GEN INTEL XEON SCALABLE ( PLATINUM 8000 SERIES )
- CPU FREQUENCY: 2.40GHZ
- OEM TRAY PROCESSOR
- COOLING DEVICE: NOT INCLUDED - PROCESSOR ONLY
Power and cost shape the value case
ServeTheHome observed maximum system power of 1,336W on its 208V lab racks. Its estimated 2017 list price was around $50,000 and up. Those figures make power delivery, cooling, rack density, and acquisition cost part of the performance decision—not afterthoughts.
The strongest economic rationale is consolidation: replacing several servers with one high-throughput system can matter when applications scale across cores and the workload benefits from fewer systems. The counterweight is licensing. Software licensed per socket, core, or another metric can change the economics substantially; a high core count is not automatically a lower-cost way to run an application. The 2017 estimate is historical, not a current quote or statement of availability.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Who should care about this benchmark
- Potentially relevant: organizations evaluating highly parallel compute, HPC, rendering, or server consolidation workloads that can use many cores and have a clear power and licensing model.
- Less informative: readers choosing hardware for lightly threaded tasks, where this suite’s aggregate throughput emphasis does not establish a comparable everyday speed advantage.
- Requires workload-specific validation: environments sensitive to NUMA placement, inter-socket communication, accelerators, or scheduler behavior. The test’s single-instance Redis and initial latency results do not settle performance for every deployment.
The results document what one four-socket configuration did in a 2017 test environment. They do not establish current product availability, pricing, or how a different application and software stack will scale.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
