NeuRealityleader in artificial intelligence technology, announced extraordinary results in terms of performance of solutions available on the market NR1-S™ AI inference device, which significantly reduces the costs and energy consumption of AI data centers, offering a much-needed solution to the growing concerns about high AI expenses and energy consumption. As governments, environmental groups and enterprises raise alarms about AI’s unsustainable energy consumption and prohibitive costs, NeuReality’s breakthrough comes at a critical time with the rapid development of generative AI. NR1-S is a responsible and affordable option for the 65% of global and 75% of U.S. businesses and governments currently struggling to adopt AI.
NR1-S does not compete with GPUs or other AI accelerators, but rather enhances and complements their performance. NeuReality’s published results compare the NR1-S inference appliance paired with Qualcomm® Cloud AI 100 Ultra and Pro accelerators to conventional CPU-centric inference servers powered by Nvidia® H100 or L40S GPUs. NR1-S achieves dramatically improved savings and energy efficiency in AI data centers for popular AI applications compared to the CPU-centric systems currently used by enormous cloud service providers (hyperscalers), server OEMs and manufacturers such as Nvidia.
Key performance benefits of the NR1-S
According to tech blog shared on Medium this morning, NeuReality’s real-world performance results show the following improvements:
- Huge savings: Combined with AI 100 Ultra, NR1-S can save up to 90% on costs for various types of AI data such as image, audio and text. These are the key building blocks of generative AI, including enormous language models, mix of experts (MoE), search-assisted generation (RAG), and multimodality.
- Critical energy efficiency: In addition to capital expenditure (CAPEX) savings in AI operate cases, NR1-S demonstrates up to 15 times better energy efficiency compared to conventional CPU-centric systems, further reducing operational expenditure (OPEX).
- Optimal operate of the AI accelerator: Unlike conventional CPU-centric systems, NR1-S ensures 100% utilization of integrated AI accelerators without the performance drops or latency seen in today’s CPU-centric systems.
Critically impacting ever-evolving real-world AI applications
Performance data included key metrics such as AI queries per dollar, queries per watt, and total cost per 1 million queries (both CAPEX and OPEX). Data zone for Natural Language Processing (NLP), Automatic Speech Recognition (ASR) and Computer Vision (CV) commonly used in medical imaging, fraud detection, call centers, online assistants and much more:
- Cost-effectiveness: One ASR test shows that the NR1-S reduces the cost of processing 1 million seconds of audio from 43 cents to just 5 cents, making voice bots and other audio-based NLP applications cheaper and able to handle more intelligence on demand.
- Energy saving: Power consumption was also measured in the tests, with ASR showing seven seconds of audio processing per watt for the NR1-S compared to 0.7 seconds for conventional CPU-centric systems. This translates into a 10-fold escalate in efficiency in relation to the energy consumed.
- Linear scalability: NR1-S delivers consistent performance regardless of the number of AI accelerators deployed, enabling customers to efficiently scale their AI infrastructure up or down with zero performance loss. This ensures maximum return on investment without the diminishing returns typically caused by adding more GPUs or other accelerators to CPU-centric servers.
NR1-S offers a practical solution for companies and governments looking to implement artificial intelligence without breaking the bank or overloading power grids. It supports a variety of AI applications commonly used in the financial services, healthcare, biotechnology, entertainment, content creation, government, public safety and transportation sectors.
These real-world performance results provide a welcome remedy to the energy crisis facing AI infrastructure providers and next-generation hyperscale supercomputers. “While increasingly faster GPUs are driving innovation for new AI capabilities, the current systems that support them are also taking us further away from most companies’ budget and carbon reduction goals,” said Ilan Avital, director of research and development NeuReality. “Our NR1-S is designed to reverse this trend, enabling sustainable AI development without sacrificing performance.”