|
gptkbp:instanceOf
|
gptkb:microprocessor
|
|
gptkbp:additionalCores
|
4 assistant cores
|
|
gptkbp:architecture
|
gptkb:ARMv8.2-A
|
|
gptkbp:connects
|
Tofu D
|
|
gptkbp:coreCount
|
48
|
|
gptkbp:countryOfOrigin
|
gptkb:Japan
|
|
gptkbp:floatingPointUnit
|
2.7 TFLOPS (double precision)
|
|
gptkbp:heldBy
|
gptkb:ARMv8.2-A_SVE
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
Fujitsu A64FX processor
|
|
gptkbp:instructionSet
|
gptkb:SVE
|
|
gptkbp:L2Cache
|
none
64 KB per core
8 MB per 12 cores
|
|
gptkbp:manufacturer
|
gptkb:Fujitsu
|
|
gptkbp:memoryBusWidth
|
1 TB/s
|
|
gptkbp:memoryType
|
gptkb:HBM2
|
|
gptkbp:notableFor
|
used in the world's fastest supercomputer (Fugaku, June 2020)
first processor to implement ARM SVE
|
|
gptkbp:processNode
|
7 nm
|
|
gptkbp:releaseYear
|
2019
|
|
gptkbp:socket
|
custom LGA
|
|
gptkbp:speed
|
up to 2.2 GHz
|
|
gptkbp:supports
|
gptkb:Scalable_Vector_Extension_(SVE)
|
|
gptkbp:supportsECCMemory
|
yes
|
|
gptkbp:targetMarket
|
high-performance computing
|
|
gptkbp:TDP
|
150 W
|
|
gptkbp:type
|
multi-chip module
|
|
gptkbp:usedIn
|
gptkb:Fugaku_supercomputer
|
|
gptkbp:bfsParent
|
gptkb:AArch64_architecture
|
|
gptkbp:bfsLayer
|
7
|