AMD EPYC 4584PX and 4484PX processors, using 3D V-Cache Stacking to Double L3 Cache to 128 MB

AMD EPYC 4004 Series Processors

Data centers, supercomputers, hyperscalers, and large companies require performance and scalability. However, the AMD EPYC 4004 Series Processors target small businesses and dedicated hosting providers seeking economical entry-level server workload solutions.

These processors provide the speed, scalability, and reliability users expect from AMD EPYC while having low core counts, Thermal Design Power (TDP) as low as 65 watts, and affordable prices. This blog discusses workloads in this market segment and AMD EPYC 4004 processors’ performance benefits.

With boost rates up to 5.7 GHz, configurations ranging from 4 to 16 “Zen 4” cores over up to 2 Core Complex Dies (CCDs), and from 8 to 32 threads with Simultaneous Multi-Threading (SMT) enabled, AMD EPYC 4004 Series Processors are a reliable option. Every AMD EPYC 4004 processor includes Gen 3 Infinity Fabric architecture, which supports up to 32 Gbps of die-to-die bandwidth, up to 192GB of DDR5-5200 RAM with ECC enabled, and up to 28 PCIe Gen 5 lanes from the processor, with additional lanes available based on system vendor design specifications.

AMD EPYC 4584PX

Technical Details

Each CCD offers up to 32 MB of shared L3 cache, for a total of up to 64 MB per processor. Packages including all of these are offered at low Thermal Design Power (TDP) levels of 65 to 170 watts.

The tried-and-true AM5 socket is used by all AMD EPYC 4004 variants, providing flexible deployment choices for a range of computing requirements. AMD 3D V-Cache die stacking technology is used by the 12-core AMD EPYC 4484PX and the 16-core AMD EPYC 4584PX, doubling the maximum L3 cache to 128 MB per unit.

Only to the degree that a feature improves efficiency and performance is it beneficial. Small companies and hosting providers need strong systems that can handle demanding workloads while keeping acquisition and running costs under control. In addition to the simplified memory and I/O capabilities that discussed earlier, servers with high-performance AMD EPYC 4004 CPUs provide attractive cost-to-performance ratios across critical customer applications.

By comparing the performance and possible cost savings of 16-core 4th Gen AMD EPYC 4004 processors to those of the competition, let’s take a deeper look at the outstanding performance and value these processors provide to the market.

Utilization Examples

Broadly speaking, the AMD EPYC 4004 CPU is versatile, performant, and economical for a variety of computing workloads, from compute-intensive jobs to common business applications. Among the instances are:

General computing: Workloads including web serving, DNS administration, file sharing, printing, email hosting, messaging, CRM, and enterprise resource planning (ERP) are well handled by AMD EPYC 4004 CPUs.

Web serving and e-commerce: Applications requiring scalability and dependability, such as web serving, are especially well-suited for AMD EPYC 4004 CPUs.

Applications requiring a lot of computation: AMD EPYC 4004 processors with 16 cores and 32 SMT threads speed up compilation and can handle demanding applications.

Gaming: Even the most demanding games run very well because to the powerful “Zen4” CPUs.

Processor Price

Performance and price must be balanced by small and medium-sized enterprises, especially when choosing processors for server construction. In this context, processor costs are an important factor to take into account. Here, it will use the following to demonstrate how much more affordable AMD EPYC 4004 CPUs are than those of their rivals.

AMD EPYC 4584PX 16-core: $699, or around $43.69 per core

Intel Xeon E-2388G 8-core processor: $606 (about $75.75) per core

Intel Xeon E-2488 8-core processor: $606 (about $75.75) per core

Put otherwise, the cost of an AMD EPYC 4584PX CPU core is just around 58% that of an Intel Xeon core. These costs highlight the comparative pricing and leading performance capabilities of AMD EPYC 4004 processors, which makes them an appealing choice for small and medium-sized enterprises trying to maximize their server infrastructure expenditures.

Fundamental Leadership in Workload

Comparing a single-socket 8-core Intel Xeon E-2488 system to a single-socket 16-core AMD EPYC 4584PX system, illustrates the ~1.73x SPECrate 2017_int_rate_base performance uplift achieved by the latter. A performance gain of around 1.50x is also achieved using the identical AMD EPYC 4584PX and Intel Xeon E-2488 CPU.

Advantage of Power Efficiency

Small- to medium-sized businesses and huge data centers struggle with energy costs. The SPECpower_ssj 2008 benchmark standardizes energy efficiency evaluation of volume server-class computers.

The power efficiency of 4th-generation AMD EPYC 4004 CPUs leads SPECpower_ssj 2008. A 16-core AMD EPYC 4584PX system with ~1.81x more energy efficiency than an Intel system. Once again, the AMD EPYC 4004 CPU offers a noteworthy increase in performance per CPU dollar of around 1.57 times.

Java Server Side

The SPECjbb 2015 benchmark simulates a business IT environment that manages online activities, data mining jobs, and point-of-sale transactions in order to assess server-side Java programs. This benchmark is significant to JVM suppliers, hardware manufacturers, Java developers, researchers, and academics because of how widely used Java is. Max-jOPS, which measures maximum throughput without stringent response time restrictions, and critical-jOPS, which measures maximum throughput with reaction time limits, are the performance metrics used by SPECjbb 2015.

A single-socket 16-core AMD EPYC 4584PX system achieves ~2.59x the performance of the same Intel processor on the SPECjbb 2015 composite critical jOPS metric at a performance/CPU dollar result of up to ~2.24x, and ~2.02x the performance of a single-socket 8-core Intel Xeon E-2388G system on the SPECjbb 2015 composite max jOPS metric.

Processing Transactions for Small and Medium-Sized Enterprises

Online transaction processing (OLTP) benchmark TPC Benchmark C describes a set of functional criteria that are common to all transaction processing systems, independent of operating system or hardware. The TPROC-C workload was developed and generated using the HammerDB benchmark tool.Image Credit To AMD

Because the results of this open-source workload do not adhere to the TPC-C Benchmark Standard, they cannot be compared to published TPC-C results. Instead, they are generated from the TPC-C Benchmark Standard. On the other hand, HammerDB TPROC-C is a useful tool for quickly evaluating the performance of database systems, contrasting databases, and system optimization.

A 16-core AMD EPYC 4584PX single-socket system is shown in Figure, delivering about 1.50x MySQL TPROC-C TPM performance and approximately 1.30x performance/CPU dollar compared to the identical Intel processor.

Processing of Media

Image Credit To AMD

The increased demand for high-quality video material has made media processing an increasingly typical edge activity. A flexible multimedia framework, FFmpeg may be used to encode, decode, transcode, stream, filter, and play back video files in a variety of historical and contemporary formats and standards. In comparison to the same Intel system, a single-socket, 16-core AMD EPYC 4584PX system can achieve average FFmpeg encode speed-ups of ~2.13x (8 jobs @ 2 threads per job), ~2.25x (4 jobs @ 4 threads per job), and ~2.45x (2 threads @ 8 cores per job) at a processor cost that is only about 15% higher.