Guohao Xu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Guohao Xu has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
- Xiaoyang Zeng
Fudan University
4 shared publications - Yibo Fan
Fudan University
4 shared publications - Shiyan Yi
State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, China
3 shared publications - Lingfei Lu
State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, China
2 shared publications - Yudi Qiu
State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, China
2 shared publications - Zhijian Hao
Xidian University
2 shared publications
Latest Publications
- 11 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 74 (10) , 3419-3432
GATe: Efficient Graph Attention Network Acceleration With Near-Memory Processing
Graph Attention Network (GAT) has gained widespread adoption thanks to its exceptional performance in processing non-Euclidean graphs. The critical components of a GAT model involve aggregation and attention, which cause numerous main-memory access, occupying significant inference time. Recently, much research has proposed near-memory processing (NMP) architectures to accelerate aggregation. However, graph attention requires additional operations distinct from aggregation, making previous NMP architectures less suitable for supporting GAT, as they typically target aggregation-only workloads. In this paper, we propose GATe, a practical and efficient GAT accelerator with NMP architecture. To the best of our knowledge, this is the first time that accelerates both attention and aggregation computation on DIMM. We unify feature vector access to eliminate the two repetitive memory accesses to source nodes caused by the sequential phase-by-phase execution of attention and aggregation. Next, we refine the computation flow to reduce data dependencies in concatenation and softmax, which lowers on-chip memory usage and communication overhead. Additionally, we introduce a novel sharding method that enhances data reusability of high-degree nodes. Experiments show that GATe achieves substantial speedup of GAT attention and aggregation phases up to 6.77${\boldsymbol\times}$ and 2.46${\boldsymbol\times}$ , with average to 3.69${\boldsymbol\times}$ and 2.24${\boldsymbol\times}$ , respectively, compared to state-of-the-art NMP works GNNear and GraNDe.- 11 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 74 (10) , 3419-3432
GATe: Efficient Graph Attention Network Acceleration With Near-Memory Processing
Graph Attention Network (GAT) has gained widespread adoption thanks to its exceptional performance in processing non-Euclidean graphs. The critical components of a GAT model involve aggregation and attention, which cause numerous main-memory access, occupying significant inference time. Recently, much research has proposed near-memory processing (NMP) architectures to accelerate aggregation. However, graph attention requires additional operations distinct from aggregation, making previous NMP architectures less suitable for supporting GAT, as they typically target aggregation-only workloads. In this paper, we propose GATe, a practical and efficient GAT accelerator with NMP architecture. To the best of our knowledge, this is the first time that accelerates both attention and aggregation computation on DIMM. We unify feature vector access to eliminate the two repetitive memory accesses to source nodes caused by the sequential phase-by-phase execution of attention and aggregation. Next, we refine the computation flow to reduce data dependencies in concatenation and softmax, which lowers on-chip memory usage and communication overhead. Additionally, we introduce a novel sharding method that enhances data reusability of high-degree nodes. Experiments show that GATe achieves substantial speedup of GAT attention and aggregation phases up to 6.77${\boldsymbol\times}$ and 2.46${\boldsymbol\times}$ , with average to 3.69${\boldsymbol\times}$ and 2.24${\boldsymbol\times}$ , respectively, compared to state-of-the-art NMP works GNNear and GraNDe.- 21 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 34 (11) , 11256-11270
A High Compression Efficiency Hardware Encoder for Intra and Inter Coding With 4K@30fps Throughput
The promotion of the HEVC standard has significantly alleviated the burden of network transmission and video storage. However, its inherent complexity and data dependencies pose a significant challenge in achieving high compression efficiency hardware encoder. To tackle this challenge, we propose several hardware-oriented algorithms and achieve a hardware encoder supporting both intra and inter coding. In terms of algorithms, our optimizations focus on intra mode decision, motion estimation (ME), rate estimation, and merge mode estimation. These optimizations reduce the computational complexity and address the data dependencies within and between encoder modules while maintaining an acceptable compression efficiency. As for hardware, we propose an encoder architecture that supports not only 35 intra prediction modes but also ME with an extensive search range of [±64, ±64]. The uniform 4×4 engine, 2-D data reuse, and timing schedule for intra and inter coding are presented in this architecture to optimize the hardware resource consumption and throughput. Compared with HM 15.0, the proposed hardware-oriented algorithms lead to a 1.88% and 14.57% increase in BD-Rate under the configurations of all intra and low delay P, respectively. Notably, the BD-Rate outperforms all existing hardware encoders supporting 4K resolution. In a GF 28nm fabrication process, the hardware design achieves a clock frequency of 550MHz, supporting 4K@30fps throughput with a hardware gate count of 3154K and memory usage of 1.02MB, and the proposed architecture demonstrates substantial advantages in terms of area, throughput, and power compared to other studies.
- 21 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 34 (11) , 11256-11270
A High Compression Efficiency Hardware Encoder for Intra and Inter Coding With 4K@30fps Throughput
The promotion of the HEVC standard has significantly alleviated the burden of network transmission and video storage. However, its inherent complexity and data dependencies pose a significant challenge in achieving high compression efficiency hardware encoder. To tackle this challenge, we propose several hardware-oriented algorithms and achieve a hardware encoder supporting both intra and inter coding. In terms of algorithms, our optimizations focus on intra mode decision, motion estimation (ME), rate estimation, and merge mode estimation. These optimizations reduce the computational complexity and address the data dependencies within and between encoder modules while maintaining an acceptable compression efficiency. As for hardware, we propose an encoder architecture that supports not only 35 intra prediction modes but also ME with an extensive search range of [±64, ±64]. The uniform 4×4 engine, 2-D data reuse, and timing schedule for intra and inter coding are presented in this architecture to optimize the hardware resource consumption and throughput. Compared with HM 15.0, the proposed hardware-oriented algorithms lead to a 1.88% and 14.57% increase in BD-Rate under the configurations of all intra and low delay P, respectively. Notably, the BD-Rate outperforms all existing hardware encoders supporting 4K resolution. In a GF 28nm fabrication process, the hardware design achieves a clock frequency of 550MHz, supporting 4K@30fps throughput with a hardware gate count of 3154K and memory usage of 1.02MB, and the proposed architecture demonstrates substantial advantages in terms of area, throughput, and power compared to other studies.
- 10 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Computer Architecture Letters
- Vol. 23 (1) , 87-90
GATe: Streamlining Memory Access and Communication to Accelerate Graph Attention Network With Near-Memory Processing
Graph Attention Network (GAT) has gained widespread adoption thanks to its exceptional performance. The critical components of a GAT model involve aggregation and attention, which cause numerous main-memory access. Recently, much research has proposed near-memory processing (NMP) architectures to accelerate aggregation. However, graph attention requires additional operations distinct from aggregation, making previous NMP architectures less suitable for supporting GAT. In this paper, we propose GATe, a practical and efficient GAT acc e lerator with NMP architecture. To the best of our knowledge, this is the first time that accelerates both attention and aggregation computation on DIMM. In the attention and aggregation phases, we unify feature vector access to reduce repetitive memory accesses and refine the computation flow to reduce communication. Furthermore, we introduce a novel sharding method that enhances the data reusability. Experiments show that our work achieves substantial speedup of up to 6.77× and 2.46×, respectively, compared to state-of-the-art NMP works GNNear and GraNDe.
- 10 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Computer Architecture Letters
- Vol. 23 (1) , 87-90
GATe: Streamlining Memory Access and Communication to Accelerate Graph Attention Network With Near-Memory Processing
Graph Attention Network (GAT) has gained widespread adoption thanks to its exceptional performance. The critical components of a GAT model involve aggregation and attention, which cause numerous main-memory access. Recently, much research has proposed near-memory processing (NMP) architectures to accelerate aggregation. However, graph attention requires additional operations distinct from aggregation, making previous NMP architectures less suitable for supporting GAT. In this paper, we propose GATe, a practical and efficient GAT acc e lerator with NMP architecture. To the best of our knowledge, this is the first time that accelerates both attention and aggregation computation on DIMM. In the attention and aggregation phases, we unify feature vector access to reduce repetitive memory accesses and refine the computation flow to reduce communication. Furthermore, we introduce a novel sharding method that enhances the data reusability. Experiments show that our work achieves substantial speedup of up to 6.77× and 2.46×, respectively, compared to state-of-the-art NMP works GNNear and GraNDe.
- 26 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Broadcasting
- Vol. 70 (2) , 693-707
Fast Transform Kernel Selection Based on Frequency Matching and Probability Model for AV1
As a fundamental component of video coding, transform coding concentrates the energy scattered in the spatial domain onto the upper-left region of the frequency domain. This concentration contributes significantly to Rate-Distortion performance improvement when combined with quantization and entropy coding. To better adapt the dynamic characteristics of image content, Alliance for Open Media Video 1 (AV1) introduces multiple transform kernels, which brings substantial coding performance benefits, albeit at the cost of considerably computational complexity. In this paper, we propose a fast transform kernel selection algorithm for AV1 based on frequency matching and probability model to effectively accelerate the coding process with an acceptable level of performance loss. Firstly, the concept of Frequency Matching Factor (FMF) based on cosine similarity is defined for the first time to describe the similarity between the residual block and the primary frequency basis image of the transform kernel. Statistical results demonstrate a clear distribution relationship between FMFs and normalized Rate-Distortion optimization costs (nRDOC). Then, leveraging these distribution characteristics, we establish Gaussian normal probability model of nRDOC for each FMF by characterizing the parameters of the normal model as functions of FMFs, enhancing the normal model’s accuracy and coding performance. Finally, based on the derived normal models, we design a fast selection algorithm with scalability and hardware-friendliness to skip the non-promising transform kernels. Experimental results show that the performance loss of the proposed fast algorithm is 1.15% when 57.66% of the transform kernels are skipped, resulting in a saving of 20.09% encoding time, which is superior to other fast algorithms found in the literature and competitive with the pruning algorithm based on the neural network in the AV1 reference software.
- 26 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Broadcasting
- Vol. 70 (2) , 693-707
Fast Transform Kernel Selection Based on Frequency Matching and Probability Model for AV1
As a fundamental component of video coding, transform coding concentrates the energy scattered in the spatial domain onto the upper-left region of the frequency domain. This concentration contributes significantly to Rate-Distortion performance improvement when combined with quantization and entropy coding. To better adapt the dynamic characteristics of image content, Alliance for Open Media Video 1 (AV1) introduces multiple transform kernels, which brings substantial coding performance benefits, albeit at the cost of considerably computational complexity. In this paper, we propose a fast transform kernel selection algorithm for AV1 based on frequency matching and probability model to effectively accelerate the coding process with an acceptable level of performance loss. Firstly, the concept of Frequency Matching Factor (FMF) based on cosine similarity is defined for the first time to describe the similarity between the residual block and the primary frequency basis image of the transform kernel. Statistical results demonstrate a clear distribution relationship between FMFs and normalized Rate-Distortion optimization costs (nRDOC). Then, leveraging these distribution characteristics, we establish Gaussian normal probability model of nRDOC for each FMF by characterizing the parameters of the normal model as functions of FMFs, enhancing the normal model’s accuracy and coding performance. Finally, based on the derived normal models, we design a fast selection algorithm with scalability and hardware-friendliness to skip the non-promising transform kernels. Experimental results show that the performance loss of the proposed fast algorithm is 1.15% when 57.66% of the transform kernels are skipped, resulting in a saving of 20.09% encoding time, which is superior to other fast algorithms found in the literature and competitive with the pruning algorithm based on the neural network in the AV1 reference software.