Introduction to Google’s Ironwood Pods
Google’s Ironwood pods are a significant improvement over previous chips, with a staggering 42.5 Exaflops of inference computing. Each chip has a peak throughput of 4,614 TFLOPs, which Google claims is a substantial improvement over previous chips.
Technical Specifications
Google has also boosted memory for the new TPUs, with each chip sporting 192GB, which is six times more than Google’s last-gen Trillium TPU. The memory bandwidth has also increased to 7.2 Tbps, a 4.5x improvement.
Comparison with Previous Chips
There are numerous ways to measure AI throughput, making it difficult to compare chips. Google is using FP8 precision as its benchmark for the new TPU, but it’s comparing it to some systems, like the El Capitan supercomputer, that don’t support FP8 in hardware. So you should take its claim that Ironwood "pods" are 24 times faster than comparable segments of the world’s most powerful supercomputer with a grain of salt.
TPU v6 Hardware
Google’s TPU v6 hardware is also conspicuously absent from the comparison chart above. The company says Ironwood is twice as powerful per watt compared to that chip, though. According to a spokesperson, Ironwood is best thought of as a successor to v5p, while TPU v6 (Trillium) was a follow-up to the less powerful TPU v5e. Google opted not to show the lower-specced chips on this chart, but for the record, Trillium was capable of hitting about 918 TFLOPS at FP8 precision.
Impact on AI Ecosystem
While the provided benchmarks are a bit odd, Ironwood is clearly a big improvement for Google’s AI ecosystem. It’s faster and more efficient than previous TPUs by a considerable margin, and Google’s existing infrastructure has enabled rapid improvements to LLMs and simulated reasoning. Google’s market-leading Gemini 2.5 model is running on last-gen TPUs right now, and Google says the higher inference speed and efficiency of Ironwood sets the stage for more breakthroughs in the coming year.
Update
Updated April 9 with more detail on how Ironwood compares to Trillium (TPUv6).
Conclusion
In conclusion, Google’s Ironwood pods are a significant improvement over previous chips, with a substantial increase in inference computing, memory, and memory bandwidth. While the comparison with previous chips is a bit tricky, Ironwood is clearly a big improvement for Google’s AI ecosystem, and its faster and more efficient performance sets the stage for more breakthroughs in the coming year.
FAQs
- Q: What is the peak throughput of each Ironwood chip?
A: Each chip has a peak throughput of 4,614 TFLOPs. - Q: How much memory does each Ironwood chip have?
A: Each chip has 192GB of memory, which is six times more than Google’s last-gen Trillium TPU. - Q: What is the memory bandwidth of Ironwood chips?
A: The memory bandwidth has increased to 7.2 Tbps, a 4.5x improvement. - Q: How does Ironwood compare to Trillium (TPUv6)?
A: Ironwood is twice as powerful per watt compared to Trillium (TPUv6), and is best thought of as a successor to v5p.
