AMD EPYC Turin delivers better performance/power efficiency than AmpereOne
The AMD EPYC 9965 processor outperforms the AmpereOne A192-32X in high-performance computing benchmarks, benefiting from superior memory bandwidth and Zen 5 architecture, despite some variability in performance across tests.
Read original article
The AMD EPYC 9965 "Turin Dense" processor has demonstrated superior performance and power efficiency compared to the AmpereOne A192-32X ARM CPU in various high-performance computing (HPC) benchmarks. Despite a suboptimal configuration due to DDR5 memory throttling, the EPYC 9965 consistently outperformed the AmpereOne in most tests, benefiting from its greater memory bandwidth and the advantages of the Zen 5 architecture. The EPYC 9965 supports 12-channel DDR5-6000 memory, while the AmpereOne is limited to 8 channels with DDR5-5200 memory. Although the AmpereOne A192-32X excelled in the synthetic Coremark benchmark, the EPYC 9965's overall performance in HPC workloads, such as AMG, was significantly better. The LULESH hydrodynamics benchmark was one of the few instances where the EPYC 9965 did not lead, indicating some variability in performance across different tests. Overall, the EPYC 9965 is positioned as a strong contender in the server processor market, particularly for applications requiring high computational power and efficiency.
- AMD EPYC 9965 outperforms AmpereOne A192-32X in most HPC benchmarks.
- EPYC 9965 features 12-channel DDR5-6000 memory, enhancing its performance.
- AmpereOne A192-32X shows better results only in the Coremark benchmark.
- The performance advantage of EPYC 9965 is attributed to its Zen 5 architecture.
- The EPYC 9965's performance remains strong even under suboptimal conditions.
Related
AMD Ryzen AI 9 HX 370: Linux Benchmarks
The AMD Ryzen AI 9 HX 370, part of the Zen 5 generation, shows promising performance and efficiency in benchmarks, with ongoing improvements for Linux support and future applications in thin clients and IoT.
AMD Ryzen 9 9950X Review: Zen 5 at Full Power
The AMD Ryzen 9 9950X, priced at $650, excels in multi-threaded tasks but underperforms in gaming compared to competitors. It features AVX-512 support and is compatible with AM5 motherboards.
Zen5's AVX512 Teardown and More (Without Redacted Content)
AMD's Zen5 architecture enhances AVX512 capabilities with full 512-bit execution paths, but faces memory bandwidth limitations affecting high-performance computing. IPC improvements vary, with some workloads achieving up to 98% gains.
AmpereOne at Hot Chips 2024: Maximizing Density
Ampere Computing introduced its AmpereOne server CPU at Hot Chips 2024, featuring a custom core design, high branch prediction accuracy, and advanced memory handling for optimized performance in cloud-native applications.
AmpereOne Scaling from 32 to 192 Cores, Core-for-Core Benchmarks vs. Altra Max
The AmpereOne A192-32X processor scales from 32 to 192 cores, maintains 3.2GHz clock speed, has high idle power consumption over 100 Watts, and lacks lower core count options in its lineup.
6 commentsFor example, take 7-Zip Compression 22.01. The CPU Power Consumption Monitor chart states:
AmpereOne: Average 278.72W EPYC: Average 311.64W
But the fine print under that same chart states:
AmpereOne: 6968J per run EPYC: 14439J per run
By the Joules per run numbers, AmpereOne is far more power efficient than EPYC, requiring only less than half of the energy to complete a run.
In that case, how could the average power of EPYC to be only 11.8% higher than that of AmpereOne? For this benchmark EPYC is 14.2% faster than AmpereOne, and if the average power numbers are correct, the EPYC should have slightly lower Joules per run than AmpereOne.
That is not the only anomaly. For example, the CPU Power Consumption Monitor chart for John the Ripper 2023.03.14 also does not make sense.
Given that most (all?) cutting-edge chips use TSMC nowadays, can you really have an apples-to-Apples comparison if the chips being compared aren't on the same process node?
Unless you're comparing price/performance, since nowadays there's no guarantee that a process shrink will get you significantly cheaper transistors (RIP, Dr. Moore).
Do these EPYCs usually go this low when idling? I ask because im considering getting one but it would idle more than 50% of the time, or would waiting for 5c make more sense?
I find 19 Watts surprisingly low. I know that the mainboard and peripherals would consume more, but my system running a 5950x, which im planning to upgrade to an EPYC, idles at around 130 Watts.
But top of the line ARM machines are really hard to get a hold of. We need an OpenAI for ARM ;)
RISC-V is.
Related
AMD Ryzen AI 9 HX 370: Linux Benchmarks
The AMD Ryzen AI 9 HX 370, part of the Zen 5 generation, shows promising performance and efficiency in benchmarks, with ongoing improvements for Linux support and future applications in thin clients and IoT.
AMD Ryzen 9 9950X Review: Zen 5 at Full Power
The AMD Ryzen 9 9950X, priced at $650, excels in multi-threaded tasks but underperforms in gaming compared to competitors. It features AVX-512 support and is compatible with AM5 motherboards.
Zen5's AVX512 Teardown and More (Without Redacted Content)
AMD's Zen5 architecture enhances AVX512 capabilities with full 512-bit execution paths, but faces memory bandwidth limitations affecting high-performance computing. IPC improvements vary, with some workloads achieving up to 98% gains.
AmpereOne at Hot Chips 2024: Maximizing Density
Ampere Computing introduced its AmpereOne server CPU at Hot Chips 2024, featuring a custom core design, high branch prediction accuracy, and advanced memory handling for optimized performance in cloud-native applications.
AmpereOne Scaling from 32 to 192 Cores, Core-for-Core Benchmarks vs. Altra Max
The AmpereOne A192-32X processor scales from 32 to 192 cores, maintains 3.2GHz clock speed, has high idle power consumption over 100 Watts, and lacks lower core count options in its lineup.