From WikiChip
Difference between revisions of "habana/hl"
< habana

(HL-x)
 
Line 22: Line 22:
  
 
== Members ==
 
== Members ==
 +
=== Inference ===
 +
==== HL-1000 ====
 +
{{main|habana/microarchitectures/goya|l1=Goya}}[[File:goya card.png|right|thumb|HL-100/102 PCIe Card]]
 +
The '''HL-1000''' is a first-generation [[neural processor]] for inference. The HL-1000 is fabricated on [[TSMC]] [[16-nanometer]] process based on the {{habana|Goya|l=arch}} microarchitecture. This chip ships in a number of products in a [[PCIe cards]] form factor.
 +
 +
* '''Proc''' [[16 nm]]
 +
* '''Mem'''' Dual-channel DDR4-2667
 +
* '''I/O''' x16 PCIe Gen 4.0
 +
 +
{| class="wikitable
 +
|-
 +
! Model !! TDP !! Memory !! Type
 +
|-
 +
| '''HL-100''' || 200 W || 4/8/16 GiB || PCIe Card; Dual-slot, passive cooling
 +
|-
 +
| '''HL-102''' || 200 W || 4/8/16 GiB || PCIe Card; Dual-slot, active cooling
 +
|}
 +
=== Training ===
 +
==== HL-2000 ====
 +
{{main|habana/microarchitectures/gaudi|l1=Gaudi}}
 +
[[File:habana gaudi oam board.JPG|right|thumb|HL-205 OAM board]]
 +
[[File:habana hls-1.jpg|right|thumb|Habana HLS-1 system with eight HL-205s]]
 +
The '''HL-2000''' is a first-generation [[neural processor]] for training. The HL-1000 is fabricated on [[TSMC]] [[16-nanometer]] process based on the {{habana|Gaudi|l=arch}} microarchitecture. This chip ships in a number of products in both a [[PCIe card]] and an [[OCP OAM]] form factors.
 +
 +
* '''Proc''' [[16 nm]]
 +
* '''Mem'''' [[HBM2]] (1 TB/s)
 +
* '''I/O''' x16 PCIe Gen 4.0, 8-10x 10 GbE
 +
 +
{| class="wikitable
 +
|-
 +
! Model !! TDP !! Memory !! Scale-Out !! Type
 +
|-
 +
| '''HL-200''' || 200 W || 32 GiB || 8x100 GbE || PCIe Card; passive cooling
 +
|-
 +
| '''HL-202''' || 200 W || 32 GiB || 8x100 GbE || PCIe Card; active cooling
 +
|-
 +
| '''HL-205''' || 300 W || 32 GiB || 10x100 GbE || [[OCP-OAM]]
 +
|}
  
 
== See also ==
 
== See also ==
 
* {{nervana|NNP}}
 
* {{nervana|NNP}}

Revision as of 02:46, 15 December 2019

HL Series
Developer Habana
Manufacturer TSMC
Type NPU
Introduction 2017 (announced)
2018 (launch)
µarch Goya, Gaudi
Process 16 nm
0.016 μm
1.6e-5 mm
Technology CMOS

HL-x is a series of neural processors designed by Habana for the data center market.

Overview

HL-x is a series of neural processors designed by Habana for both training and inference in the data center. There are two distinct products based on two separate microarchitectures - one for inference and one for training. Both designs share common unified architecture, simplifying software development.

Members

Inference

HL-1000

Main article: Goya
HL-100/102 PCIe Card

The HL-1000 is a first-generation neural processor for inference. The HL-1000 is fabricated on TSMC 16-nanometer process based on the Goya microarchitecture. This chip ships in a number of products in a PCIe cards form factor.

  • Proc 16 nm
  • Mem' Dual-channel DDR4-2667
  • I/O x16 PCIe Gen 4.0
Model TDP Memory Type
HL-100 200 W 4/8/16 GiB PCIe Card; Dual-slot, passive cooling
HL-102 200 W 4/8/16 GiB PCIe Card; Dual-slot, active cooling

Training

HL-2000

Main article: Gaudi
HL-205 OAM board
Habana HLS-1 system with eight HL-205s

The HL-2000 is a first-generation neural processor for training. The HL-1000 is fabricated on TSMC 16-nanometer process based on the Gaudi microarchitecture. This chip ships in a number of products in both a PCIe card and an OCP OAM form factors.

  • Proc 16 nm
  • Mem' HBM2 (1 TB/s)
  • I/O x16 PCIe Gen 4.0, 8-10x 10 GbE
Model TDP Memory Scale-Out Type
HL-200 200 W 32 GiB 8x100 GbE PCIe Card; passive cooling
HL-202 200 W 32 GiB 8x100 GbE PCIe Card; active cooling
HL-205 300 W 32 GiB 10x100 GbE OCP-OAM

See also

Retrieved from "https://en.wikichip.org/w/index.php?title=habana/hl&oldid=94884"
Facts about "HL Series - Habana"
RDF feed
designerHabana +
first announced2017 +
first launched2018 +
full page namehabana/hl +
instance ofintegrated circuit family +
main designerHabana +
manufacturerTSMC +
microarchitectureGoya + and Gaudi +
nameHL Series +
process16 nm (0.016 μm, 1.6e-5 mm) +
technologyCMOS +