From WikiChip
Difference between revisions of "princeton/piton"
< princeton

(Update name of OpenPiton page)
 
(9 intermediate revisions by 3 users not shown)
Line 1: Line 1:
 
{{princeton title|Piton}}
 
{{princeton title|Piton}}
−
{{mpu
+
{{chip
 
| name                = Piton
 
| name                = Piton
 
| no image            =  
 
| no image            =  
Line 47: Line 47:
 
| max memory addr    =  
 
| max memory addr    =  
  
−
| electrical          = Yes
+
 
 
| power              =  
 
| power              =  
 
| v core              = 0.9 V
 
| v core              = 0.9 V
Line 79: Line 79:
 
| socket 0 type      = QFP
 
| socket 0 type      = QFP
 
}}
 
}}
−
'''Piton''' is a {{arch|64}} [[many-core microprocessor]] developed by [[Princeton]]'s Parallel Computing Group and announced in August of 2016. The [[massively parallel processor array|MPPA]] chip contains 25 modified [[OpenSPARC T1]] cores (an implementation of {{sparc|V9|SPARC V9}}). The chip, which was manufactured on [[IBM]]'s [[32 nm|32 nm SOI process]], operates at 1 GHz. The chip was presented On August 32 2016 at the [[Hot Chips]] 28.
+
'''Piton''' is a {{arch|64}} [[many-core microprocessor]] developed by [[Princeton]]'s Parallel Computing Group and announced in August of 2016. The [[massively parallel processor array|MPPA]] chip contains 25 modified [[OpenSPARC T1]] cores (an implementation of {{sparc|V9|SPARC V9}}). The chip, which was manufactured on [[IBM]]'s [[32 nm|32 nm SOI process]], operates at 1 GHz. The chip was presented in August 2016 at the [[Hot Chips]] 28.
  
 
== Architecture ==
 
== Architecture ==
−
The chip is designed as a [[massively parallel processor array]], with 25 cores ("tiles") arranged as a grid 5 by 5. Each core is a modified [[OpenSPARC T1]] which implements {{sparc|V9|SPARC V9}} capable of booting a standard [[operating system|OS]]. Piton implements a 64-bit [[network on chip]] (NoC) interconnect made of 3 physical networks operating at 1 cycle/hop latency.
+
The chip is designed as a [[massively parallel processor array]], with 25 cores ("tiles") arranged as a 2D grid of 5 by 5. Each core is a modified [[OpenSPARC T1]] which implements {{sparc|V9|SPARC V9}} capable of booting a standard [[operating system|OS]]. Piton implements a 64-bit [[network on chip]] (NoC) interconnect made of 3 physical networks operating with a 1 cycle/hop latency.
  
 
[[File:piton layout.svg|650px]]
 
[[File:piton layout.svg|650px]]
Line 97: Line 97:
 
|l1i desc=
 
|l1i desc=
 
|l1i extra=
 
|l1i extra=
−
|l1d cache=200 KB
+
|l1d cache=200 KiB
−
|l1d break=25x8 KB
+
|l1d break=25x8 KiB
 
|l1d desc=4-way set associative
 
|l1d desc=4-way set associative
 
|l1d extra=(write-back, per tile)
 
|l1d extra=(write-back, per tile)
−
|l2 cache=1.6 MB
+
|l2 cache=1.5 MiB
−
|l2 break=25x64 KB
+
|l2 break=25x64 KiB
 
|l2 desc=4-way set associative
 
|l2 desc=4-way set associative
 
|l2 extra=(per tile)
 
|l2 extra=(per tile)
Line 112: Line 112:
  
 
== Network On-chip (NoC) ==
 
== Network On-chip (NoC) ==
−
Piton implements 3 physical [networks on chip]] (NoC) that provide all the communication between the [[#tiles|tiles]], deliver I/O and memory traffic, and pass inter-core [[interrupt]]s. Piton implements 3 physical neworks - each consisting of two 64-bit uni-directional links (one in each direction). Packets routing implements [[dimension-order routing]]. Each packet reserves 29 bits for core address, allowing a theoretical network size 500 million core. The three NoCs have descending priorities - i.e. NoC3 has the highest priority, followed by NoC2, followed by NoC1 with the lowest priority.
+
Piton implements 3 physical [[networks on chip]] (NoC) that provide all the communication between the [[#tiles|tiles]], deliver I/O and memory traffic, and pass inter-core [[interrupt]]s. Piton implements 3 physical neworks - each consisting of two 64-bit uni-directional links (one in each direction). Packets routing implements [[dimension-order routing]]. Each packet reserves 29 bits for core address, allowing a theoretical network size of 500 million cores. The three NoCs have descending priorities - i.e. NoC3 has the highest priority, followed by NoC2, followed by NoC1 with the lowest priority.
  
 
== Documents ==
 
== Documents ==
−
* Balkind J, McKeown M, Fu Y, Nguyen T, Zhou Y, Lavrov A, Shahrad M, Fuchs A, Payne S, Liang X, Matl M. [https://parallel.princeton.edu/papers/openpiton-asplos16.pdf OpenPiton: An open source manycore research framework]. In Proceedings of the Twenty-First International Conference on Architectural Support for Programming Languages and Operating Systems 2016 Mar 25 (pp. 217-232). ACM.
+
* Jonathan Balkind, Michael McKeown, Yaosheng Fu, Tri Nguyen, Yanqi Zhou, Alexey Lavrov, Mohammad Shahrad, Adi Fuchs, Samuel Payne, Xiaohua Liang, Matthew Matl, David Wentzlaff. [https://parallel.princeton.edu/papers/openpiton-asplos16.pdf OpenPiton: An open source manycore research framework]. In Proceedings of the Twenty-First International Conference on Architectural Support for Programming Languages and Operating Systems 2016 Mar 25 (pp. 217-232). ACM.
 
* Michael McKeown, Yaosheng Fu, Tri Nguyen, Yanqi Zhou, Jonathan Balkind, Alexey Lavrov, Mohammad Shahrad, Samuel Payne, and David Wentzlaf. [http://parallel.princeton.edu/piton/piton_hotchips28_talk.pdf Piton: A 25-core Academic Manycore Research Processor]
 
* Michael McKeown, Yaosheng Fu, Tri Nguyen, Yanqi Zhou, Jonathan Balkind, Alexey Lavrov, Mohammad Shahrad, Samuel Payne, and David Wentzlaf. [http://parallel.princeton.edu/piton/piton_hotchips28_talk.pdf Piton: A 25-core Academic Manycore Research Processor]
−
* OpenPit [http://parallel.princeton.edu/openpiton/docs/micro_arch.pdf OpenPiton Microarchitecture Specification]. Hot Chips 28. August 23, 2016.
+
* OpenPiton [http://parallel.princeton.edu/openpiton/docs/micro_arch.pdf OpenPiton Microarchitecture Specification]. Hot Chips 28. August 23, 2016.
  
 
== External links ==
 
== External links ==
 
* [http://parallel.princeton.edu/piton/ Piton]
 
* [http://parallel.princeton.edu/piton/ Piton]
 
* [http://www.openpiton.org Open Piton]
 
* [http://www.openpiton.org Open Piton]
 +
 +
{{DEFAULTSORT: Piton}}

Latest revision as of 14:41, 5 July 2018

Edit Values
Piton
Princeton piton.png
Piton face
General Info
DesignerPrinceton
ManufacturerIBM
MarketServer
IntroductionAugust 23, 2016 (announced)
General Specs
Frequency1000 MHz
Bus speed350 MHz
Bus rate2,800 MT/s
Clock multiplier2.85
Microarchitecture
Process32 nm
Transistors460,000,000
TechnologyCMOS
Die36 mm²
6 mm × 6 mm
Word Size64 bit
Cores25
Threads50
Multiprocessing
Max SMP20,000-Way (Multiprocessor)
Electrical
Vcore0.9 V

Piton is a 64-bit many-core microprocessor developed by Princeton's Parallel Computing Group and announced in August of 2016. The MPPA chip contains 25 modified OpenSPARC T1 cores (an implementation of SPARC V9). The chip, which was manufactured on IBM's 32 nm SOI process, operates at 1 GHz. The chip was presented in August 2016 at the Hot Chips 28.

Architecture[edit]

The chip is designed as a massively parallel processor array, with 25 cores ("tiles") arranged as a 2D grid of 5 by 5. Each core is a modified OpenSPARC T1 which implements SPARC V9 capable of booting a standard OS. Piton implements a 64-bit network on chip (NoC) interconnect made of 3 physical networks operating with a 1 cycle/hop latency.

piton layout.svg

Tiles[edit]

Piton is made of an array of tiles in a grid of 5x5. Each tile is composed of a modified OpenSPARC T1 core (+L1$), an L1.5 cache, L2 cache, a floating-point unit (FPU), a CPU Cache-Crossbar (CCX) arbiter, and three network on chip (NoC) routers.

Cache[edit]

Pitons uses a distributed write-back L2$ model that implements a directory-based MESI protocol - adhering to OpenSPARC's total store order (TSO) model. Each tile contains 64 KB slice of the L2 cache and an attached cache directory. The L2 cache is inclusive of both the L1.5$ and the L1$. Note that the L1.5 is called as such because the OpenSPARC T1 core already implements an 8 KB L1d$ and 16 KB L1i$ which are tightly coupled with the pipeline and was thus only modified to work in a scaleable multi-core environment. The L1.5$ acts as a middleman between the OpenSPARC T1's crossbar protocol and the Piton's NoC. It relays requests and replies to and from the core through CCX.

Cache Info [Edit Values]
L1D$ 200 KiB
204,800 B
0.195 MiB
25x8 KiB 4-way set associative (write-back, per tile)
L2$ 1.5 MiB
1,536 KiB
1,572,864 B
0.00146 GiB
25x64 KiB 4-way set associative (per tile)

Network On-chip (NoC)[edit]

Piton implements 3 physical networks on chip (NoC) that provide all the communication between the tiles, deliver I/O and memory traffic, and pass inter-core interrupts. Piton implements 3 physical neworks - each consisting of two 64-bit uni-directional links (one in each direction). Packets routing implements dimension-order routing. Each packet reserves 29 bits for core address, allowing a theoretical network size of 500 million cores. The three NoCs have descending priorities - i.e. NoC3 has the highest priority, followed by NoC2, followed by NoC1 with the lowest priority.

Documents[edit]

External links[edit]


Retrieved from "https://en.wikichip.org/w/index.php?title=princeton/piton&oldid=79908"
Facts about "Piton - Princeton"
RDF feed
base frequency1,000 MHz (1 GHz, 1,000,000 kHz) +
bus rate2,800 MT/s (2.8 GT/s, 2,800,000 kT/s) +
bus speed350 MHz (0.35 GHz, 350,000 kHz) +
clock multiplier2.85 +
core count25 +
core voltage0.9 V (9 dV, 90 cV, 900 mV) +
designerPrinceton +
die area36 mm² (0.0558 in², 0.36 cm², 36,000,000 µm²) +
die length6 mm (0.6 cm, 0.236 in, 6,000 µm) +
die width6 mm (0.6 cm, 0.236 in, 6,000 µm) +
first announcedAugust 23, 2016 +
full page nameprinceton/piton +
instance ofmicroprocessor +
l1d$ description4-way set associative +
l1d$ size200 KiB (204,800 B, 0.195 MiB) +
l2$ description4-way set associative +
l2$ size1.5 MiB (1,536 KiB, 1,572,864 B, 0.00146 GiB) +
ldateAugust 23, 2016 +
main imageFile:Princeton piton.png +
main image captionPiton face +
manufacturerIBM +
market segmentServer +
max cpu count20,000 +
namePiton +
process32 nm (0.032 μm, 3.2e-5 mm) +
smp max ways20,000 +
technologyCMOS +
thread count50 +
transistor count460,000,000 +
word size64 bit (8 octets, 16 nibbles) +