In a new blog posted on NVIDIA Develope, the company reiterates its strengths and how it itself used its AI capabilities to design its most powerful GPU to date, the Hopper H100. NVIDIA GPUs are primarily designed using state-of-the-art EDA (electronic design automation) tools, but with the help of AI that leverages the PrefixRL approach help, using deep reinforcement learning to optimize parallel prefix circuits, the company was able to design smaller, faster and more energy-efficient chips while delivering better performance.

Arithmetic circuits in computer chips are constructed using networks of logic gates (such as NAND, NOR and XOR) and wires. The ideal circuit should have the following characteristics.

● Small: Smaller area so that more circuits can be mounted on the chip.

● Fast: lower latency to improve chip performance.

● Consume less power: lower power consumption of the chip.

NVIDIA has designed nearly 13,000 AI-assisted circuits using this approach, reducing their area by 25% compared to equally fast and functionally identical EDA tools. But PrefixRL was mentioned as a very computationally demanding task, and for each GPU physically simulated, it required 256 CPUs and over 32,000 GPU hours. To remove this bottleneck, NVIDIA developed Raptor, an in-house distributed reinforcement learning platform that specifically leverages NVIDIA hardware for this industrial reinforcement learning.

Raptor has several features that improve scalability and training speed, such as job scheduling, custom networks, and GPU-aware data structures. In the context of PrefixRL, Raptor enables mixed job allocation across CPU, GPU, and Spot instances.

The networks in this reinforcement learning application are diverse and benefit from the following.

● Raptor’s ability to switch between NCCLs for peer-to-peer transfers to transfer model parameters directly from the learner GPU to the inference GPU.

● Redis for asynchronous and smaller messages, such as rewards or statistics.

● A JIT-compiled RPC for handling high-volume and low-latency requests, such as uploading experience data.

NVIDIA concluded that applying AI to real-world circuit design problems could lead to better GPU designs in the future. The full paper is here, and you can visit the developer blog here for more information.

The post NVIDIA uses AI to design and develop GPUs – Latest Hopper already has 13,000 circuit instances appeared first on TechGoing.

There is also a gap is the AI unit, NVIDIA in the RTX 20 series of the Turing architecture on the introduction of Tensor Core, used to accelerate AI computing, DLSS 1.0 relies on AI acceleration, in the implementation of AI-related operations in the performance of more advantageous.

AMD’s RDNA and RDNA2 architectures do not have a dedicated AI unit and instruction set, but that may change in the RDNA3 architecture (codenamed GFX11) at the end of this year. The latest revelation shows that AMD is providing WMMA (Wave Matrix Multiply-Accumulate) instruction support for the architecture, which can perform multiplication and addition.

AMD supported the MFMA instruction set in its CDNA2 architecture for computing cards last year, and the WMA instruction set is more flexible and supports data formats that are closer to AI needs.

After this upgrade, the RDNA3 architecture should be able to narrow the gap with NVIDIA’s GPU architecture in terms of AI and DLSS. Coupled with AMD’s previously announced 50% improvement in energy efficiency, this year’s RX 7000 graphics card is still worth looking forward to, and is expected to be on par with N cards in terms of performance, energy efficiency, and technology.

The post After taking on DLSS, AMD graphics cards target NVIDIA’s unique AI acceleration appeared first on TechGoing.