Using mobile GPU for general-purpose computing &amp;#x2013; a case study of face recognition on smartphones

Cheng, Kwang-Ting; Wang, Yi-Chu

doi:10.1109/vdat.2011.5783575

Cited by 36 publications

(13 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Many authors are attempting to increase power eciency of heterogeneous architectures by dividing workloads between the heterogeneous processing elements [3,9,12,14,21]. These target mobile SoCs such as the Tegra 2 [3,21], Samsung S4, Samsung Note II, Google Nexus 7 and Tegra 250 [14], Tegra 3 [9] and Texas Instruments' OMAP 3530 platform [12]. A typical application area is SIFT [9,14], but Huang and Lai [9] also experiment with BLAS benchmarks, mobile face recognition [3,21] and face tracking [12].…”

Section: Load Balancingmentioning

confidence: 99%

“…These target mobile SoCs such as the Tegra 2 [3,21], Samsung S4, Samsung Note II, Google Nexus 7 and Tegra 250 [14], Tegra 3 [9] and Texas Instruments' OMAP 3530 platform [12]. A typical application area is SIFT [9,14], but Huang and Lai [9] also experiment with BLAS benchmarks, mobile face recognition [3,21] and face tracking [12]. The common approach in these studies is to ooad certain computation blocks entirely to the on-board GPU using the OpenGL ES graphics library.…”

Section: Load Balancingmentioning

confidence: 99%

“…GPU-ooaded computational blocks for SIFT is for example gaussian scale space [14] or convolution, dierence of gaussians, octave gradient and key description generation [9,21]. In their face recognition system for the Tegra 2, Cheng and Wang ooad the Gabor wavelet to the mobile GPU [3]. Lopez et.…”

Section: Load Balancingmentioning

confidence: 99%

“…Authors who ooad work to GPUs [3,8,9,12,14,21] report performance speedups and increased energy eciency when both the CPU and GPU are used. However, as explicitly noted by Wang and Cheng [21], the increase in energy eciency is probably a result of reduced execution time.…”

Section: Measuring Energy-eciencymentioning

confidence: 99%

“…al. [10] in Section 2.2, and means that authors who quantitatively measure energy eciency, for example in total runtime energy [3,9,12,21] or Performance Per Watt (PPW) [7,10,14], observe increases in power eciency which is potentially caused only by reduced energy consumption from idle power components. Therefore, interpreting the result that hybrid processing saves energy is dangerous.…”

Section: Measuring Energy-eciencymentioning

confidence: 99%

See 4 more Smart Citations

Load Balancing of Multimedia Workloads for Energy Efficiency on the Tegra K1 Multicore Architecture

Stokke

Stensland

Griwodz

et al. 2017

Proceedings of the 8th ACM on Multimedia Systems Conference

View full text Add to dashboard Cite

Energy eciency is a timely topic for modern mobile computing. Reducing the energy consumption of devices not only increases their battery lifetime, but also reduces the risk of hardware failure. Many researchers strive to understand the relationship between software activity and hardware power usage. A recurring strategy for saving power is to reduce operating frequencies. It is widely acknowledged that standard frequency scaling algorithms generally overreact to changes in hardware utilisation. More recent and original eorts attempt to balance software workloads on heterogeneous multicore architectures, such as the Tegra K1, which includes a quad-core CPU and a CUDA-capable GPU. However, it is not known whether it is possible to utilise these processor elements in parallel to save energy. Research into these types of systems are unfortunately often evaluated with the Performance Per Watt (PPW) metric, which is an unaccurate method because it ignores constant power usage from idle components. We show that this metric can end up increase energy usage on the Tegra K1, and give a false impression of how such systems consume energy. In reality, we show that it is much harder to save energy by balancing workloads between the heterogeneous cores of the Tegra K1, where we demonstrate only a 5% energy saving by ooading 10% DCT workload from the GPU to the CPU. Signicantly more energy can be saved (up to 50 %) using the appropriate processor for dierent workloads. CCS CONCEPTS • Computer systems organization → Heterogeneous (hybrid) systems; • Applied computing → Electronics; • Human-centered computing → Mobile devices;

show abstract