Crucial lessons discovered
Micron DDR5 128GB memory with 5th Gen AMD EPYC CPUs with 128 cores offer a powerful solution for AI and database infrastructures. This method effectively handles the computational difficulty of large model sizes and large datasets, which are characteristics of AI, machine learning, data analytics, and IMDB applications.| SVM 1.3x Improves AI/ML support vector machine (SVM) performance by 1.3x.1.3x higher bandwidth use due to higher memory clock speeds on Micron 128GB RDIMMs, along with increased core counts on 5th Gen AMD EPYC processors. | IMDB REDIS 1.2x Improves IMDB Redis performance by 1.2x for a set:get ratio of 1:10.30% Delivers 30% better average latency and 60% better p99 latency. | SAP SD 201k users Improves SAP Sales and Distribution (SAP SD) benchmark. With 30% higher memory capacity and 30% higher clock, SAP SD scores higher than previous six-socket performance score. |
- Combining 5th Gen AMD EPYC CPUs (codenamed Turin) with Micron 128GB DDR5 RDIMMs.
- In contrast to 4th Gen AMD EPYC CPUs with 96 cores (codenamed Genoa) and Micron 96GB DDR5 RDIMMs.
Micron presents the results of benchmark testing for AI/ML support vector machines (SVM), SAP SD, and Redis in-memory databases (IMDB) in this blog. The hardware configuration of these systems contrasted with the following:
- Turin-coded 5th generation AMD EPYC CPUs with Micron DDR5 128GB DIMMs
- Micron DDR5 96GB DIMMs and 4th Gen AMD EPYC CPUs (codenamed Genoa)
Testing shows that Micron DDR5 RDIMMs with higher bandwidth (up to 8000 MT/s) and capacity (128GB) enhance SVM, SAP SD, and Redis IMDB performance.
System configuration and hardware
The details of the system architecture are shown in the table below. In this blog, two systems, A and B, were contrasted. System B has 128GB and 5th Gen AMD EPYC CPUs (96 cores), whereas System A had Micron 96GB DDR5 DIMMs. A 12GB/core configuration is supported by both systems; the 128-core CPU has 128GB of memory, while the 96-core CPU has 96GB spread across 12 memory channels.System A | System B | |
|---|---|---|
| Hardware | 4th Gen AMD EPYC processors (codenamed Genoa) | 5th Gen AMD EPYC processors (codenamed Turin) |
| Memory | Micron 96GB DDR5 4800 MT/sDual rank, 12 channels | Micron 128GB DDR5 6400 MT/sDual rank, 12 channels |
| CPU | Dual-socket AMD EPYC 9654 (96-core) | Dual-socket AMD EPYC (128-core) |
| Storage (for SVM) | Micron 9400 8TB (3) | Micron 9400 8TB (3) |
AI/ML Support Vector Machine
A common machine learning method for preparing datasets for various cloud-based data science applications is support vector machines (SVM). In testing, it used the Intel Hi-Bench architecture and the SparkML engine to handle a 2TB dataset.Quicker time of execution
In terms of SVM execution time, system B performed 30% better than system A. The main causes of this are the additional cores in system B's processor, the larger capacity and bandwidth provided by the 128GB memory modules, and the effective utilization of bandwidth.Increased use of bandwidth
Data show that SVM on system B needs 1.3 times more bandwidth than system A because to the higher memory rate (6400 MT/s vs. 4800 MT/s) and the additional Zen5 cores made feasible by the 5th Gen AMD EPYC processors with 128 cores.Because system B has a bigger capacity (128GB vs. 96GB), which lowers storage input/output, the SVM can store more data in memory. maintained a constant memory capacity of 12GB per core for both configurations. This approach enabled us to distinguish between the effects of increased processing capacity and quicker clock speed (memory) in comparison to the baseline configuration (system A).
Redis
Redis is a rapid in-memory database (IMBD) that may be used to store and retrieve data for applications that need low latency. Memtier benchmarks Redis with multiple set:get operations to mimic a multithreaded and multiclient execution architecture.Redis runs 1.2 times faster on system B (128GB and 128 cores). Furthermore, the same combination lowers average latency by 30% and p99 latency by 60%. The larger capacity and bandwidth of the Micron DDR5 128GB DIMMs compared to previous AMD EPYC CPU generations can be better utilized by higher core counts, such as the 5th Gen's 128 cores. With the extra cores, enterprise data centers can simply serve more users with the improved throughput.
SAP Distribution and Sales (SAP SD)
Systems Application and Products (SAP) is a commonly used enterprise resource planning (ERP) software suite. It is made up of several smaller components that are part of the SAP ecosystem. The component that houses all of the operations and processes for SAP Sales and Distribution (SAP SD) was benchmarked against the Dell PowerEdge R6725 server and equipped with Micron DDR5 128GB RDIMMs and 5th Gen AMD EPYC processors, setting a new performance world record of 201,000 users for the SAP SD benchmark on a two-socket system.The top six-socket score is surpassed by that. The higher number of benchmark users indicates that there may be a performance benefit to combining Micron memory with 5th Gen AMD EPYC CPUs on Dell PowerEdge servers for database use cases.
0 Comments