Monday, August 25th 2025
NVIDIA "Blackwell Ultra" Brings PCIe Gen 6 and 1.5x NVFP4 Performance Boost
NVIDIA's long-anticipated "Blackwell Ultra" is the final form of the Blackwell family before the transition to "Rubin." However, this silicon doesn't seem to share much resemblance with its predecessors, especially in terms of I/O and performance. On its blog, NVIDIA detailed the silicon design of its latest creation and all the accompanying features like enhanced software support and optimizations. One of the most striking aspects of the Blackwell Ultra, designed for AI servers, is its use of PCIe Gen 6, whereas the consumer Blackwell and regular server Blackwell use PCIe Gen 5. Made using the TSMC 4NP node, the massive Ultra chip features 208 billion transistors, which is 2.6 times more than the last generation Hopper, based solely on raw transistor count. This comes with a 1,400 W TDP, which means a massive cooling system is a must.
When it comes to performance, Blackwell Ultra delivers approximately 1.5 times denser NVFP4 compute compared to Blackwell, resulting in higher tokens-per-second on inference and improved throughput for large-batch training. The chip pairs 160 SMs across two reticle dies via NVIDIA's NV-HBI link, bringing a 10 TB/s die-to-die fabric, 288 GB of HBM3E at up to 8 TB/s bandwidth, and fifth-generation Tensor Cores tuned for NVFP4. Attention-layer throughput benefits from the doubled performance of special function units (SFUs) for transcendental operations, reducing softmax latency and enhancing reasoning responsiveness.At system scale, NVIDIA positions the GB300 NVL72 as the prime build: liquid-cooled racks built from Grace Blackwell Ultra Superchips reach exascale-class dense FP4 and offer substantial gains in throughput-per-megawatt versus prior HGX platforms. Software changes are equally important; CUDA remains fully supported, while frameworks such as TensorRT-LLM, Dynamo scheduling, and precision-aware model optimizers leverage NVFP4 and the dual-die topology to increase utilization and reduce the cost per token. Enterprise features, such as advanced MIG partitioning, TEE-I/O confidential computing, hardware decompression, and expanded NVLink fabrics, round out the platform for multi-tenant AI factories. For datacenter operators seeking density and efficiency, Blackwell Ultra is designed for larger on-chip memory footprints, tighter host I/O with PCIe Gen 6, and software-hardware co-design that prioritizes inference scale and operational stability.
Source:
NVIDIA
When it comes to performance, Blackwell Ultra delivers approximately 1.5 times denser NVFP4 compute compared to Blackwell, resulting in higher tokens-per-second on inference and improved throughput for large-batch training. The chip pairs 160 SMs across two reticle dies via NVIDIA's NV-HBI link, bringing a 10 TB/s die-to-die fabric, 288 GB of HBM3E at up to 8 TB/s bandwidth, and fifth-generation Tensor Cores tuned for NVFP4. Attention-layer throughput benefits from the doubled performance of special function units (SFUs) for transcendental operations, reducing softmax latency and enhancing reasoning responsiveness.At system scale, NVIDIA positions the GB300 NVL72 as the prime build: liquid-cooled racks built from Grace Blackwell Ultra Superchips reach exascale-class dense FP4 and offer substantial gains in throughput-per-megawatt versus prior HGX platforms. Software changes are equally important; CUDA remains fully supported, while frameworks such as TensorRT-LLM, Dynamo scheduling, and precision-aware model optimizers leverage NVFP4 and the dual-die topology to increase utilization and reduce the cost per token. Enterprise features, such as advanced MIG partitioning, TEE-I/O confidential computing, hardware decompression, and expanded NVLink fabrics, round out the platform for multi-tenant AI factories. For datacenter operators seeking density and efficiency, Blackwell Ultra is designed for larger on-chip memory footprints, tighter host I/O with PCIe Gen 6, and software-hardware co-design that prioritizes inference scale and operational stability.



Comments on NVIDIA "Blackwell Ultra" Brings PCIe Gen 6 and 1.5x NVFP4 Performance Boost
There are no comments yet.