GB200
NVIDIA's Grace Blackwell Superchip based on the Blackwell architecture, the flagship product targeting data-center inference workloads. It integrates a Grace CPU and a B200 GPU interconnected via NVLink-C2C.
Timeline (last 90 days)
CoreWeave's publicly listed price for the older GB200 is $10.5 per GPU per hour
According to CoreWeave's comparison chart, Rubin holds approximately a 3x advantage over the GB200 at the slow end, climbing to more than 7x at the fast end
The attention forward pass of sigmoid attention improves by 7.4%, but the full GPU step improves by only 0.3%.
On the NVIDIA GB200 platform, full training step throughput improves by 8.0% for routed-expert SwiGLU tasks, the highest gain among the four tasks.
On the NVIDIA GB200 platform, full training step throughput improves by 2.7% for dense SiLU tasks and by 2.9% for tanh soft-capping attention.
Jalapeño has a rated power consumption of 700W with measured sustained power consumption not exceeding 550W; by comparison, the GB200 is rated at 1200W and the GB300 at 1400W.
In the gpt-oss 120B test, Jalapeño achieved a peak throughput of approximately 85,448 per watt compared to GB200's 44,960, roughly 1.9×; end-to-end latency was reduced from 1.80 seconds to 1.03 seconds; and minimum TBT dropped from 1.87 ms to 0.69 ms.
The appendix of this public benchmark only compared the GB200 and GB300, and did not include the upcoming Vera Rubin, which is about to enter mass production and shipment.
The throughput level for running Kimi K3 on the GB200 platform is 450 tokens/s.
The GB300 is the latest iteration of NVIDIA's Grace Blackwell architecture, with the previous generation being the GB200.
5 more timeline events
All Facts (14)
GB200中的'G'代表Grace CPU，'B'代表Blackwell GPU
95%UnverifiedIn the gpt-oss 120B test, Jalapeño achieved a peak throughput of approximately 85,448 per watt compared to GB200's 44,960, roughly 1.9×; end-to-end latency was reduced from 1.80 seconds to 1.03 seconds; and minimum TBT dropped from 1.87 ms to 0.69 ms.
60%UnverifiedCoreWeave's publicly listed price for the older GB200 is $10.5 per GPU per hour
50%UnverifiedAccording to CoreWeave's comparison chart, Rubin holds approximately a 3x advantage over the GB200 at the slow end, climbing to more than 7x at the fast end
50%UnverifiedOn the NVIDIA GB200 platform, full training step throughput improves by 2.7% for dense SiLU tasks and by 2.9% for tanh soft-capping attention.
50%UnverifiedThe attention forward pass of sigmoid attention improves by 7.4%, but the full GPU step improves by only 0.3%.
50%UnverifiedOn the NVIDIA GB200 platform, full training step throughput improves by 8.0% for routed-expert SwiGLU tasks, the highest gain among the four tasks.
50%UnverifiedJalapeño has a rated power consumption of 700W with measured sustained power consumption not exceeding 550W; by comparison, the GB200 is rated at 1200W and the GB300 at 1400W.
50%UnverifiedThe GB300 is the latest iteration of NVIDIA's Grace Blackwell architecture, with the previous generation being the GB200.
50%UnverifiedNVIDIA has launched DGX systems and rack-level solutions based on the GB200, selling complete AI computing infrastructure rather than individual chips
50%UnverifiedCompared to the H100, the GB200 achieves a 30× improvement in energy efficiency and a 25× reduction in latency for AI inference tasks.
50%UnverifiedThe Blackwell architecture's B200/GB200 contains 208 billion transistors, uses a dual-chip package design, and achieves an inter-chip interconnect bandwidth of 10 TB/s.
50%UnverifiedThe Blackwell architecture released in 2024 (represented by the B200/GB200) has a transistor count of 208 billion, uses a dual-chip packaging design, and is merged into a single logical GPU via 10 TB/s chip-to-chip interconnect.
50%UnverifiedNvidia's Blackwell architecture released in 2024 (with representative products B200/GB200) features 208 billion transistors and adopts second-generation Transformer Engine and NVLink 5.0 interconnect technology.
50%