FP4
A 4-bit floating-point format, native in Blackwell-class silicon, pushing inference throughput and density further.
Current numbers
~2.9 EF FP4AMD Helios (72× MI455X) rack-scale FP4 (1.4 EF FP8), 31 TB HBM4, ORW double-wide
216 GBHBM3E per Microsoft Maia 200 @ 7 TB/s; >10 PFLOPS FP4 / >5 PFLOPS FP8, 750W, +30% perf/$ (Microsoft claim)
~1.44 EFLOPSGB200 NVL72 rack FP4 with sparsity; dense and other-precision figures are lower (the datasheet-reading trap)