TY - GEN
T1 - Hardware implementation of micropolygon rasterization with motion and defocus blur
AU - Brunhaver, J. S.
AU - Fatahalian, K.
AU - Hanrahan, P.
N1 - Publisher Copyright:
© The Eurographics Association 2010.
PY - 2010/6/25
Y1 - 2010/6/25
N2 - Current GPUs rasterize micropolygons (polygons approximately one pixel in size) inefficiently. Additionally, they do not natively support triangle rasterization with jittered sampling, defocus, or motion blur. We perform a microarchitectural study of fixed-function micropolygon rasterization using custom circuits. We present three rasterization designs: the first optimized for triangle micropolygons that are not blurred, a second for stochastic rasterization of micropolygons with motion and defocus blur, and third that is a hybrid combination of the two. Our designs achieve high area and power efficiency by using low-precision operations and rasterizing pairs of adjacent triangles in parallel. We demonstrate optimized designs synthesized in a 45 nm process showing that a micropolygon rasterization unit with a throughput of 3 billion micropolygons per second would consume 2.9 W and occupy 4.1 mm2 which is 0.77% of the die area of a GeForce GTX 480 GPU. Categories and Subject Descriptors (according to ACM CCS): I.3.1 [Computer Graphics]: Graphics processors-I.3.7 Three-Dimensional Graphics and Realism.
AB - Current GPUs rasterize micropolygons (polygons approximately one pixel in size) inefficiently. Additionally, they do not natively support triangle rasterization with jittered sampling, defocus, or motion blur. We perform a microarchitectural study of fixed-function micropolygon rasterization using custom circuits. We present three rasterization designs: the first optimized for triangle micropolygons that are not blurred, a second for stochastic rasterization of micropolygons with motion and defocus blur, and third that is a hybrid combination of the two. Our designs achieve high area and power efficiency by using low-precision operations and rasterizing pairs of adjacent triangles in parallel. We demonstrate optimized designs synthesized in a 45 nm process showing that a micropolygon rasterization unit with a throughput of 3 billion micropolygons per second would consume 2.9 W and occupy 4.1 mm2 which is 0.77% of the die area of a GeForce GTX 480 GPU. Categories and Subject Descriptors (according to ACM CCS): I.3.1 [Computer Graphics]: Graphics processors-I.3.7 Three-Dimensional Graphics and Realism.
UR - http://www.scopus.com/inward/record.url?scp=84874186913&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84874186913&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:84874186913
SN - 9783905674262
T3 - High-Performance Graphics - ACM SIGGRAPH / Eurographics Symposium Proceedings, HPG
SP - 1
EP - 9
BT - High-Performance Graphics 2010 - ACM SIGGRAPH / Eurographics Symposium Proceedings, HPG 2010
A2 - Laine, Samuli
A2 - Hunt, Warren
A2 - Doggett, Michael
PB - Association for Computing Machinery
T2 - 2nd ACM SIGGRAPH / Eurographics on High-Performance Graphics, HPG 2010
Y2 - 25 June 2010 through 27 June 2010
ER -