HPE ProLiant Compute Gen12 achieves multiple world records in AI inference benchmarks
If you're thinking of server modernization, consider starting with HPE Compute ProLiant Gen12. Recently setting multiple world records in AI inference benchmarks, its superiority is undeniable. The DL384 Gen12 and DL380a Gen12 servers achieved #1 rankings across over 23 MLPerf™ Benchmark models - quite a stride in server performance. The Gen12 delivered 13 new world records on MLPerf Inference: Datacenter v5.0, thanks to the NVIDIA® GH200 NVL2 accelerator that powers advanced, high-quality image models and high-accuracy AI reasoning models. Notably, the ProLiant DL384 Gen12 server achieved a groundbreaking feat - a cool 2X increase in performance over single-socket setups, while almost doubling AI performance in various scenarios. It's a significant step toward reimagining server performance in data centers. And let's not forget the stellar performance of the Gen12 Server in offline high-batch situation inference applications - it ranked as a top performer in 4 benchmarks. HPE Compute Gen12 is shaping up to be a key player in server modernization, providing phenomenal performance. Shouldn't you be a part of this transformation too? Take the first step towards transforming your business with us! Contact us today to get started.
What are the key performance achievements of HPE ProLiant Compute Gen12?
HPE ProLiant Compute Gen12 servers have achieved #1 rankings across more than 23 MLPerf™ Benchmark models, with the DL384 Gen12 delivering 13 new world records on the MLPerf Inference: Datacenter v5.0. Notably, the DL384 Gen12 server with dual-socket NVIDIA GH200 Grace Hopper Superchip 144 GB has shown to deliver 2x the performance of a single-socket setup, indicating significant advancements in AI inference capabilities.
How does the HPE ProLiant Compute DL380a Gen12 perform in benchmarks?
The HPE ProLiant Compute DL380a Gen12 server has demonstrated superior performance in several benchmarks, including image classification with ResNet50, object detection with Retinanet, and LLM summarization with both GPT-J 99 and GPT-J 99.9. It stands out as a top performer on four benchmarks, showcasing its capabilities in various AI applications.
What AI models are supported by HPE ProLiant Compute Gen12 servers?
HPE ProLiant Compute Gen12 servers support a range of advanced AI models, including Stable Diffusion XL (SDXL) for image generation, Llama2-70B-99 for enhanced accuracy in AI reasoning tasks, and Llama2-70B-99.9 for ultra-high precision in complex applications. These models leverage the servers' high-performance capabilities to deliver efficient and reliable AI inference.
HPE ProLiant Compute Gen12 achieves multiple world records in AI inference benchmarks
published by Strictly Tech, LLC
Strictly Tech is a full line IT Solution Provider headquartered in Fort Lauderdale.
We help customers with Client Computing , Datacenter Infrastructure, Cloud Computing and many other areas.
We are eager to add value.
A few simple questions to consider;
- Does your current IT provider have local 500,000 square foot warehouses around the state stocked with 100,000 products within a 1 or 2-day FedEx ground shipment from you?
- Do they offer free FedEx ground shipping on those products?
- Do they have 20 of those warehouses Nationwide?
- Do they hold the highest level of sales and warranty certification with HP, HPE, Lenovo, Dell, Cisco, APC, Microsoft, etc.?
- Are they Minority Owned and Certified?
- Do they offer pre deployment services?
- Image Build and Imaging Service
- No Charge Pre-Sale Engineering Support
If you answered NO to any of these questions, let's discuss how Strictly Tech can save your organization time and budget money when procuring or leasing IT products and services.
Watch this brief 1-Minute Animated Video about how StrictlyTech can help your firm
Stephen Rudner
Senior Client Executive
StrictlyTech
2300 NW 55th Court, Suite 110 Fort Lauderdale, FL 33309
D: 954.666.6215 | M: 561.572.1489 | srudner@strictlytech.com
https://strictlytech.com/
Sign in with LinkedIn