From WikiChip
Difference between revisions of "cambricon/mlu"
< cambricon

(mlu)
 
(Models)
 
(2 intermediate revisions by the same user not shown)
Line 26: Line 26:
 
}}
 
}}
 
'''Machine Learning Unit''' ('''MLU''') is a family of [[neural processors]] designed by [[Cambricon]].
 
'''Machine Learning Unit''' ('''MLU''') is a family of [[neural processors]] designed by [[Cambricon]].
 +
 +
== Overview ==
 +
Announced in late 2017, the MLU family of [[neural processors]] designed for cloud-based workloads for both inference and training. In contrast to Cambricon's mobile and edge computing [[IP cores]], those processors have a higher power envelope and are designed for much higher performance.
 +
 +
== Models ==
 +
{{empty section}}
 +
 +
 +
* {{\|MLU100}}
 +
* {{\|MLU200}}
 +
 +
== See also ==
 +
* Google's {{google|TPU}}

Latest revision as of 01:08, 27 May 2018

MLU
Developer Cambricon
Manufacturer TSMC
Type Neural Processors
Introduction Nov 7, 2017 (announced)
May 3, 2018 (launch)
ISA MLU
Word size 64 bit
8 octets
16 nibbles
Process 16 nm
0.016 μm
1.6e-5 mm
Technology CMOS
Clock 1,000 MHz-1,300 MHz

Machine Learning Unit (MLU) is a family of neural processors designed by Cambricon.

Overview[edit]

Announced in late 2017, the MLU family of neural processors designed for cloud-based workloads for both inference and training. In contrast to Cambricon's mobile and edge computing IP cores, those processors have a higher power envelope and are designed for much higher performance.

Models[edit]

New text document.svg This section is empty; you can help add the missing info by editing this page.


See also[edit]

designerCambricon +
first announcedNovember 7, 2017 +
first launchedMay 3, 2018 +
full page namecambricon/mlu +
instance ofintegrated circuit family +
instruction set architectureMLU +
main designerCambricon +
manufacturerTSMC +
nameMLU +
process16 nm (0.016 μm, 1.6e-5 mm) +
technologyCMOS +
word size64 bit (8 octets, 16 nibbles) +