High-speed, multi-input, multi-output control using GPU processing in the HBT-EP tokamak

N. Rath; J. Bialek; P. J. Byrne; B. Debono; J. P. Levesque; B. Li; M. E. Mauel; D. A. Maurer; G. A. Navratil; D. Shiraki

doi:10.1016/j.fusengdes.2012.04.003

High-speed, multi-input, multi-output control using GPU processing in the HBT-EP tokamak

N. Rath, J. Bialek, P. J. Byrne, B. Debono, J. P. Levesque, B. Li, M. E. Mauel, D. A. Maurer, G. A. Navratil, D. Shiraki

Research output: Contribution to journal › Article › peer-review

23 Scopus citations

Abstract

We report on the design of a new plasma control system for the HBT-EP tokamak that utilizes a graphical processing unit (GPU) to magnetically control the 3D perturbed equilibrium state [1] of the plasma. The control system achieves cycle times of 5 μs and I/O latencies below 10 μs for up to 96 inputs and 64 outputs. The number of state variables is in the same order. To handle the resulting computational complexity under the given time constraints, the control algorithms are designed for massively parallel processing. The necessary hardware resources are provided by an NVIDIA Tesla M2050 GPU, offering a total of 448 computing cores running at 1.3 GHz each. A new control architecture allows control input from magnetic diagnostics to be pushed directly into GPU memory by a D-TACQ ACQ196 digitizer, and control output to be pulled directly from GPU memory by two D-TACQ AO32 analog output modules. By using peer-to-peer PCI express connections, this technique completely eliminates the use of host RAM and central processing unit (CPU) from the control cycle, permitting single-digit microsecond latencies on a standard Linux host system without any real-time extensions.

Original language	English
Pages (from-to)	1895-1899
Number of pages	5
Journal	Fusion Engineering and Design
Volume	87
Issue number	12
DOIs	https://doi.org/10.1016/j.fusengdes.2012.04.003
State	Published - Dec 2012
Externally published	Yes

Funding

The authors would like to thank D-TACQ and PathScale for their assistance with implementing the direct communication channels between GPU and digitizer/output boards. This work was supported by U.S. Department of Energy (DOE) grant DE-FG02-86ER53222 .

Keywords

GPU
High speed
MIMO
Perturbed equilibria
Plasma control
Tokamak

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1016/j.fusengdes.2012.04.003

Cite this

@article{2bcbefb630b2484a80831a09dec19f21,

title = "High-speed, multi-input, multi-output control using GPU processing in the HBT-EP tokamak",

abstract = "We report on the design of a new plasma control system for the HBT-EP tokamak that utilizes a graphical processing unit (GPU) to magnetically control the 3D perturbed equilibrium state [1] of the plasma. The control system achieves cycle times of 5 μs and I/O latencies below 10 μs for up to 96 inputs and 64 outputs. The number of state variables is in the same order. To handle the resulting computational complexity under the given time constraints, the control algorithms are designed for massively parallel processing. The necessary hardware resources are provided by an NVIDIA Tesla M2050 GPU, offering a total of 448 computing cores running at 1.3 GHz each. A new control architecture allows control input from magnetic diagnostics to be pushed directly into GPU memory by a D-TACQ ACQ196 digitizer, and control output to be pulled directly from GPU memory by two D-TACQ AO32 analog output modules. By using peer-to-peer PCI express connections, this technique completely eliminates the use of host RAM and central processing unit (CPU) from the control cycle, permitting single-digit microsecond latencies on a standard Linux host system without any real-time extensions.",

keywords = "GPU, High speed, MIMO, Perturbed equilibria, Plasma control, Tokamak",

author = "N. Rath and J. Bialek and Byrne, {P. J.} and B. Debono and Levesque, {J. P.} and B. Li and Mauel, {M. E.} and Maurer, {D. A.} and Navratil, {G. A.} and D. Shiraki",

year = "2012",

month = dec,

doi = "10.1016/j.fusengdes.2012.04.003",

language = "English",

volume = "87",

pages = "1895--1899",

journal = "Fusion Engineering and Design",

issn = "0920-3796",

publisher = "Elsevier",

number = "12",

}

TY - JOUR

T1 - High-speed, multi-input, multi-output control using GPU processing in the HBT-EP tokamak

AU - Rath, N.

AU - Bialek, J.

AU - Byrne, P. J.

AU - Debono, B.

AU - Levesque, J. P.

AU - Li, B.

AU - Mauel, M. E.

AU - Maurer, D. A.

AU - Navratil, G. A.

AU - Shiraki, D.

PY - 2012/12

Y1 - 2012/12

N2 - We report on the design of a new plasma control system for the HBT-EP tokamak that utilizes a graphical processing unit (GPU) to magnetically control the 3D perturbed equilibrium state [1] of the plasma. The control system achieves cycle times of 5 μs and I/O latencies below 10 μs for up to 96 inputs and 64 outputs. The number of state variables is in the same order. To handle the resulting computational complexity under the given time constraints, the control algorithms are designed for massively parallel processing. The necessary hardware resources are provided by an NVIDIA Tesla M2050 GPU, offering a total of 448 computing cores running at 1.3 GHz each. A new control architecture allows control input from magnetic diagnostics to be pushed directly into GPU memory by a D-TACQ ACQ196 digitizer, and control output to be pulled directly from GPU memory by two D-TACQ AO32 analog output modules. By using peer-to-peer PCI express connections, this technique completely eliminates the use of host RAM and central processing unit (CPU) from the control cycle, permitting single-digit microsecond latencies on a standard Linux host system without any real-time extensions.

AB - We report on the design of a new plasma control system for the HBT-EP tokamak that utilizes a graphical processing unit (GPU) to magnetically control the 3D perturbed equilibrium state [1] of the plasma. The control system achieves cycle times of 5 μs and I/O latencies below 10 μs for up to 96 inputs and 64 outputs. The number of state variables is in the same order. To handle the resulting computational complexity under the given time constraints, the control algorithms are designed for massively parallel processing. The necessary hardware resources are provided by an NVIDIA Tesla M2050 GPU, offering a total of 448 computing cores running at 1.3 GHz each. A new control architecture allows control input from magnetic diagnostics to be pushed directly into GPU memory by a D-TACQ ACQ196 digitizer, and control output to be pulled directly from GPU memory by two D-TACQ AO32 analog output modules. By using peer-to-peer PCI express connections, this technique completely eliminates the use of host RAM and central processing unit (CPU) from the control cycle, permitting single-digit microsecond latencies on a standard Linux host system without any real-time extensions.

KW - GPU

KW - High speed

KW - MIMO

KW - Perturbed equilibria

KW - Plasma control

KW - Tokamak

UR - http://www.scopus.com/inward/record.url?scp=84870731159&partnerID=8YFLogxK

U2 - 10.1016/j.fusengdes.2012.04.003

DO - 10.1016/j.fusengdes.2012.04.003

M3 - Article

AN - SCOPUS:84870731159

SN - 0920-3796

VL - 87

SP - 1895

EP - 1899

JO - Fusion Engineering and Design

JF - Fusion Engineering and Design

IS - 12

ER -

High-speed, multi-input, multi-output control using GPU processing in the HBT-EP tokamak

Abstract

Funding

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this