Print
Univis
Search
 
FAU-Logo
Techn. Fakultät Willkommen am Institut für Informatik FAU-Logo

Ranking

Problem size: 256 | 512 | 1024
Rank Algorithm Description qrmse* Error Hist.† PSNR‡ Time✢Performance∗
[HU] [dB] [s]
1
GPU
Elmer
Submitter: Sam Hawker
Institution: Nikon Metrology
RabbitCT dataset version: 2
Date of submission: 2016-07-21

Elmer is a multi-GPU implementation using CUDA 5.5
0.1688.290.3100.00%
2
GPU
Elmer (high accuracy)
Submitter: Sam Hawker
Institution: Nikon Metrology
RabbitCT dataset version: 2
Date of submission: 2016-07-21

High accuracy version of Elmer avoiding the fixed-point bilinear interpolation hardware
0.0598.470.389.47%
3
GPU
AccuRabbit
Submitter: Oleg Konings,Tobias Funk ,Scott Hsieh,Paul Kahn
Institution: Triple Ring Technologies
RabbitCT dataset version: 2
Date of submission: 2016-07-30

We have developed a fast accurate 2 GPU implementation of the back projection algorithm for output volumes of sizes 256^3, 512^3, 1024^3 and 2048^3
0.03102.180.644.74%
4
GPU
Thumper
Submitter: Timo Zinßer
Institution: Siemens AG
RabbitCT dataset version: 2
Date of submission: 2013-06-11

Thumper is a CUDA-based back-projection implementation.
0.1787.390.643.59%
5
GPU
RapidRabbit
Submitter: Eric Papenhausen, Ziyi Zheng
Institution: Stony Brook University
RabbitCT dataset version: 2
Date of submission: 2013-06-04

A CUDA 3.0 based back projection implementation using a variety of optimization techniques.
0.1688.290.735.88%
6
GPU
CERA on GTX 680
Submitter: Matthias Elter
Institution: Siemens AG
RabbitCT dataset version: 2
Date of submission: 2013-02-15

The CUDA 5.0 based CERA back-projection implementation (extended with CUDA streams) running on a NVIDIA GTX 680.
0.1887.380.929.31%
7
GPU
GAPBckPrj
Submitter: Giovanni Di Domenico
Institution: University of Ferrara - Italy
RabbitCT dataset version: 2
Date of submission: 2016-07-29

GAPBckPrj is a CUDA 8.0 back-projection implementation.
0.1688.271.026.01%
8
GPU
CERA on GTX 670
Submitter: Matthias Elter
Institution: Siemens AG
RabbitCT dataset version: 1
Date of submission: 2012-06-11

A CUDA 4.2 based back-projection implementation running on a NVIDIA GTX 670 GPU.
2.8463.173.47.73%
9
GPU
CERA on GTX 570
Submitter: Matthias Elter
Institution: Siemens AG
RabbitCT dataset version: 1
Date of submission: 2012-06-11

A CUDA 4.1 based back-projection implementation running on a NVIDIA GTX 570 GPU.
2.8463.173.96.78%
10
GPU
LionEatsRabbit
Submitter: Wolfgang Wein
Institution: White Lion Technologies AG
RabbitCT dataset version: 1
Date of submission: 2011-08-03

The back-projection component of our rapid iterative reconstruction implemented in OpenGL / GLSL.
2.1665.545.94.52%
11
CPU
fastrabbitEX
Submitter: Jan Treibig
Institution: RRZE (FAU Erlangen)
RabbitCT dataset version: 1
Date of submission: 2011-07-21

OpenMP parallel implementation with SSE assembly kernel. A variant with higher precision is also available, see details.
1.6867.737.43.56%
12
GPU
CLeopatra
Submitter: Sebastian Schuberth
Institution: Zuse Institute Berlin, Department Visualization and Data Analysis
RabbitCT dataset version: 1
Date of submission: 2011-01-19

This is an OpenCL-based implementation in order to evaluate its performance against CUDA, and its scalability across multiple GPUs / CPUs.
2.8663.1210.52.52%
13
GPU
ConradCL
Submitter: Andreas Maier
Institution: Pattern Recognition Lab, Friedrich-Alexander University Erlangen-Nuremberg
RabbitCT dataset version: 2
Date of submission: 2013-09-06

OpenCL Backprojector implemented in CONRAD.
0.1688.2812.42.14%
14
GPU
SpeedyGonzales
Submitter: Christopher Rohkohl
Institution: Pattern Recognition Lab, Department of Computer Science, Friedrich-Alexander University Erlangen-Nuremberg, Germany
RabbitCT dataset version: 1
Date of submission: 2009-06-26

SpeedyGonzales is a CUDA 2.1 implementation. It uses the nice capabilities of the cuda texture-based interpolation.
2.8463.1714.71.80%
15
CPU
fastrabbit
Submitter: Jan Treibig
Institution: RRZE (FAU Erlangen)
RabbitCT dataset version: 1
Date of submission: 2011-06-07

OpenMP parallel implementation with SSE assembly kernel.
2.5064.2915.01.77%
16
GPU
RoadRunner
Submitter: Christian Siegl
Institution: Pattern Recognition Lab, Department of Computer Science, Friedrich-Alexander University Erlangen-Nuremberg, Germany
RabbitCT dataset version: 1
Date of submission: 2011-05-23

OpenCL implementation using the CUDA 3.1 framework (OpenCL 1.1). Uses texture-based interpolation and coalesced memory access.
2.8463.1716.81.58%
17
CPU
TomatoSalad (formerly LolaTBBSSE)
Submitter: Hannes G. Hofmann
Institution: Pattern Recognition Lab, FAU Erlangen
RabbitCT dataset version: 1
Date of submission: 2009-06-16

LolaBunny multi-threaded and vectorized using Intel's Threading Building Blocks (TBB) and SSE.
0.03101.81518.90.05%
18
CPU
LolaSSE
Submitter: Hannes G. Hofmann
Institution: Pattern Recognition Lab, FAU Erlangen
RabbitCT dataset version: 1
Date of submission: 2009-06-16

LolaBunny vectorized using SSE.
0.03101.81840.70.03%
19
CPU
LolaTBB
Submitter: Hannes G. Hofmann
Institution: Pattern Recognition Lab, FAU Erlangen
RabbitCT dataset version: 1
Date of submission: 2009-06-16

LolaBunny multi-threaded using Intel's Threading Building Blocks (TBB).
0.02106.651,287.50.02%
20
CPU
LolaOMP
Submitter: Hannes G. Hofmann
Institution: Pattern Recognition Lab, FAU Erlangen
RabbitCT dataset version: 1
Date of submission: 2009-06-16

LolaBunny multi-threaded using OpenMP.
0.00∞1,345.40.02%
21
CPU
LolaBunny
Submitter: RabbitCT Team
Institution: Pattern Recognition Lab, Department of Computer Science, Friedrich-Alexander University Erlangen-Nuremberg, Germany
RabbitCT dataset version: 1
Date of submission: 2009-06-25

This is the official RabbitCT reference implementation.
0.00∞2,509.50.01%
Notes:
*Mean squared error. qrmse = sum(error2) / Nvoxels.
†The error histogram shows the distribution of the absolute errors. Bins 0–9 count errors between (bin, (bin+1)] HU, bin 10 counts all errors > 10 HU. Figures show relative amount: y = count / Nvoxels * 100.
‡Peak SNR. PSNR = 10.0 * log10(4095.0*4095.0 / qrmse2).
✢Runtime for backprojection of all views, measured in seconds.
∗Performance relative to fastest implementation for this problem size. perf = tavg,fastest / tavg
versionBetween RabbitCT dataset version 1 and 2 we have changed the reference volume due to an error in the first version. Learn more
whereNvoxels = 5123 (on this page) and Nviews = 496 for the RabbitCT data set.

Please take the time to participate and become one of the famous RabbitCT backprojectors.