What is NVIDIA Blackwell? All about the GPU architecture

NVIDIA’s Blackwell microarchitecture is setting new standards in GPU technology, bringing significant performance improvements and higher efficiency to various areas of use. NVIDIA Blackwell offers modern technologies that are specially optimised for complex calculations.

What is NVIDIA Blackwell?

NVIDIA Blackwell is modern microarchitecture that was launched in 2024 and named after the mathematician David Blackwell. It was specially developed for use in high-performance graphics processors. Blackwell is the successor to Hopper GPU architecture and delivers significant improvements in performance and higher energy efficiency.

Blackwell was optimised for use in artificial intelligence (in particular generative AI and large language models), machine learning, scientific calculations and gaming. The architecture is based on advanced manufacturing technologies and uses modern chiplet design to achieve higher performance. A special feature of Blackwell is the improved memory architecture that enables faster data processing. It also offers optimised support for ray tracing and more efficient use of Tensor Cores for AI applications. NVIDIA prioritised optimised scalability in the design, which makes the architecture suitable for data centres as well as high-end consumer products.

What new features does NVIDIA Blackwell have?

Compared to its predecessor Hopper, Blackwell GPUs offer a number of technical innovations. One of the most important is the introduction of an advanced chiplet design that enables increased scalability and efficiency. The design makes it possible to combine several smaller chips rather than using a single monolithic chip, which significantly improves performance. Additionally, Tensor Cores were optimised for accelerating deep learning models more efficiently.

Ray tracing performance was also improved, which enables more realistic lighting and shadow calculation for games and graphics applications. The memory architecture was also modified. With the new generation of HBM (High Bandwidth Memory), NVIDIA Blackwell can reach much higher memory bandwidths. Energy efficiency was also improved using new manufacturing technologies and better cooling mechanisms. And finally, NVIDIA Blackwell also supports interfaces such as the next generation of NVlink and PCIe 5.0, which enable faster communication between different GPUs and CPUs.

What are the main areas of use for NVIDIA Blackwell?

NVIDIA Blackwell architecture is designed for a variety of high-performance applications:

• Artificial intelligence and machine learning: Optimised Tensor Cores enable Blackwell to train and run very large AI models. Blackwell GPUs are equipped with Confidential Computing, which uses hardware-based security features to protect confidential data and AI models from unauthorised access. That makes it ideal for companies that work with generative AI, neural networks and autonomous driving and want to increase both performance and security.
• Data centres and cloud computing: Blackwell GPUs are specially optimised for use in data centres and offer high computing power with relatively little energy use. That makes them suitable for big data analysis, simulations and scientific calculations (e.g. weather forecasts and climate modelling).
• Gaming and 3D graphics: Thanks to more efficient AI acceleration, game developers can better combine ray tracing with classic rasterisation technologies to create more realistic scenes with optimised performance. That makes Blackwell suitable for high-end gaming computers and VR applications.

What are the advantages and disadvantages of NVIDIA Blackwell?

Advantages of NVIDIA Blackwell

NVIDIA Blackwell microarchitecture has a number of significant advantages that make it a great choice for professional applications and gaming. One of the biggest advantages is the substantially increased computing power achieved with the new chiplet design. Rather than incorporating a monolithic chip, Blackwell uses a modular structure that enables better scalability and more efficient manufacturing. That means that high-performance GPUs can be produced more flexibly and optimised for different areas of use.

Another advantage of Blackwell GPUs is improved energy efficiency. The use of advanced manufacturing technology allows for a reduction in energy use in relation to computing power. That’s particularly important for data centres that require high performance for the lowest possible energy consumption.

In addition, Blackwell GPUs are equipped with Tensor Cores, which were specially developed for machine learning and AI applications. They make it possible to quickly train and run complex models, making Blackwell an ideal solution for companies and research institutions.

Gamers also benefit from these improvements, since improved ray tracing technology enables a more realistic representation of light and shadow in modern games. New memory technologies also ensure faster data processing and reduced latency.

Disadvantages of NVIDIA Blackwell

Despite all those advances, there are still several challenges and potential downsides to Blackwell GPUs. One of the most obvious is the high price, especially for professional models and server solutions. Its advanced technology makes Blackwell an expensive investment that doesn’t necessarily make sense for every use case. Although energy efficiency was improved, the power usage involved in high-performance computing is significant, especially when it comes to multi-GPU setups.

Another potential hurdle is the need for specific software optimisations. To get maximum performance from Blackwell-based GPUs, developers need to adapt their software accordingly. That can present a challenge in professional workflows if existing programs aren’t fully optimised for the new architecture.

Finally, Blackwell GPUs may not be widely available at first, as the demand for high-performance GPUs increases and new technologies are often initially produced in small quantities.

Advantages and disadvantages at a glance