Cerebras Disaggregated Inference
emschwartz
59 points
14 comments
November 18, 2026
Related Discussions
Found 5 related stories in 146.8ms across 13,334 title embeddings via pgvector HNSW
- AMD and Cerebras Launch AI Inference Solution rbanffy · 20 pts · July 24, 2026 · 68% similar
- Cerebras Deploys Trillion-Parameter AI Models gmays · 77 pts · October 25, 2026 · 67% similar
- Cerebras CS-4 sunils34 · 177 pts · August 19, 2026 · 62% similar
- Qwen 3.8 27B available on Cerebras at 1500 tokens/s altertable · 509 pts · September 03, 2026 · 54% similar
- Data Processing Incidents on Inference Providers sdko · 76 pts · November 22, 2026 · 53% similar
Discussion Highlights (4 comments)
augment_me
Have they solved the issue yet where they need to keep running ghost tokens to keep the board hot to prevent it from cracking, which causes massive energy draw in idle, and reduces lifetime by 3 times the estimate? https://culperresearch.com/wp-content/uploads/2026/10/Culper...
aabdi
5x throughput more than b4, same 10x tokens/s as compared to standard. looks like disagg/wide ep is applicable to cerebras chips as well. kind of amazing. https://www.cerebras.ai/blog/how-we-deploy-trillion-paramete...
hephaes7us
It sounds like this might mean they could do real prefix caching? Cerebras would be much more interesting to me if cache hits resulted in a commensurate savings on my end.
speedping
Cross-platform prefill/decode split is an incredible feat, kudos to them. I wonder how much time is spent transporting kv cache, and if the gains would be even higher for LLMs with ultra-compressed kv cache like the latest deepseek.