Comparison of MLPerf™ Inference v1.1 Results of Dell EMC PowerEdge R7525 Servers with NVIDIA GPUs
Mon, 04 Apr 2022 11:08:17 -0000
|Read Time: 0 minutes
Abstract
This blog showcases the MLPerf Inference v1.1 performance results of Dell EMC PowerEdge R7525 servers configured with NVIDIA A100 40 GB GPUs or with NVIDIA A30 GPUs. We compare the cost of a system with both types of GPUs to help you choose the best configuration for your AI inference workloads.
Introduction
MLPerf Inference v1.1 falls under the benchmarks and metrics category from MLCommons™ and serves as the industry standard for machine learning (ML) inference performance. The MLPerf benchmarking suite measures the performance of ML workloads consistently and fairly. The MLPerf Inference benchmark measures how fast a system can perform ML inference by using a pretrained model in various deployment scenarios. For a comprehensive understanding of MLPerf Inference, see this blog.
Test bed details
The systems under test (SUT) include:
- PowerEdge R7525 server that is configured with three NVIDIA A100 PCIe 40 GB (250 W, 40 GB passive, double wide, full height GPU) GPUs. All references to the PowerEdge R7525 server with A100 GPUs assume that the configuration includes three NVIDIA A100 GPUs.
- PowerEdge R7525 server that is configured with three NVIDIA A30 (165 W, 24 GB passive, double wide, full height GPU with cable) GPUs. All references to the PowerEdge R7525 server with A30 GPUs assume that the configuration includes three NVIDIA A100 GPUs.
The following figure shows the PowerEdge R7525 server:
Both systems run TensorRT, which is a library designed and developed for improved performance in inference on NVIDIA GPUs. For more information about TensorRT, see the NVIDIA documentation.
SUT configuration
The following table shows the MLPerf system configurations for the SUTs:
Table 1: SUT configuration
Platform | PowerEdge R7525 with 3 A100 PCIe 40 GB GPUs | PowerEdge R7525 with 3 A30 GPUs |
MLPerf System ID | R7525_A100-PCIE-40GBx3_TRT | R7525_A30x3_TRT |
Operating system | CentOS 8.2.2004 | |
Memory | 512 GB | 1 TB |
GPU | NVIDIA A100-PCIE-40GB | NVIDIA A30 |
GPU count | 3
| |
Software stack | TensorRT 8.0.2 CUDA 11.3 cuDNN 8.2.1 GPU Driver 470.42.01 DALI 0.31.0
| |
MLPerf Inference v1.1 results per model
ResNet 50
ResNet50 is a 50-layer deep convolution neural network that is used for many computer vision applications. This neural network can address vanishing gradients using the concept of skip connections by allowing gradients to move through layers in the network. For an introduction to ResNet, see Deep Residual Learning for Image Recognition.
We conducted four tests on this model across the two SUTs: two in the Offline scenario and two in the Server scenario. The following figure shows our ResNet50 results. The performance of the PowerEdge R7525 server with A30 GPUs across both scenarios is approximately 50 percent higher than the PowerEdge R7525 server with A100 GPUs.
Figure 1: ResNet50 results on a PowerEdge R7525 server with A100 GPUs and a PowerEdge R7525 server with A30 GPUs
BERT
Bidirectional Encoder Representations from Transformers (BERT) is a state-of-the-art language representational model. In essence, BERT is a stack of Transformer encoders. The Transformer architecture is fast because it can process words simultaneously, and the context of words can be learned from both directions simultaneously. BERT can be used for neural machine translation, question answering, sentiment analysis, and text summarization, all of which require language understanding. BERT is trained in two phases: pretrain in which the model understands language and context, and fine-tuning in which BERT learns specific tasks such as questioning and answering. For an in-depth understanding, see BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding from Google AI Language.
For this model, we conducted eight tests across our systems in which we considered the default and high accuracy modes in both the Server and Offline scenarios. In the default mode, the PowerEdge R7525 server with A100 GPUs performed 69 percent better than the PowerEdge R7525 server with A30 GPUs in the Offline scenario and 99 percent better in the Server scenario. The high accuracy mode provided similar results in which the PowerEdge R7525 server with A100 GPUs performed 72 percent better than the PowerEdge R7525 server with A30 GPUs in the Offline scenario and 96 percent better in the Server scenario. In the following figure, bert-99 refers to the default accuracy target, whereas bert-99.9 refers to the high accuracy target.
Figure 2: BERT results on a PowerEdge R7525 with A100 GPUs and a PowerEdge R7525 with A30 GPUs
SSD-ResNet34
ResNet34 is an encoder on top of Single Shot Multibox Detector (SSD) that is used to improve performance and reduce training time. As the full form suggests, the SSD is a single stage objection detection model that is known for speed. For an in-depth understanding, see Small Object Detection using Context and Attention.
For this model, we conducted four tests across both of our systems. In the Offline scenario, the PowerEdge R7525 server with A100 GPUs outperformed the PowerEdge R7525 server with A30 GPUs by 74 percent. Similarly, in the Server scenario, the PowerEdge R7525 server with A100 GPUs performed 78 percent better than the PowerEdge R7525 server with A30 GPUs.
Figure 3: SSD-ResNet34 results on a PowerEdge R7525 server with A100 GPUs and a PowerEdge R7525 server with A30 GPUs
DLRM
DLRM, an open-source Deep Learning Recommendation Model, is available on Facebook’s PyTorch platform. The model is composed of compute-dominated multilayer perceptrons (MLPs) and relies on data parallelism to improve performance. When predicting click percentage for certain items, for example, it is aligned with randomized Las Vegas algorithms in which resources (time and memory) are used freely but the results are always correct. DLRM uses collaborative filtering and predicative analysis-based approaches to process large amounts of data. For more information about DLRM, see Deep Learning Recommendation Model for Personalization and Recommendation Systems.
For this model, we conducted eight tests across both of our systems. For the PowerEdge R7525 server with A100 GPUs, we notice a tight range with a lower and upper bound of 764,569 and 768,806 result samples per second, respectively. Also, the results produced across the default and high accuracy tests are the same for their respective systems. The initial numbers from the PowerEdge R7525 server with A30 GPUs were slightly below expectations. After the submission deadline, our team was able to extract additional performance, particularly in the Server scenario. The numbers for the PowerEdge R7525 server with A30 GPUs shown in the following figure are not the same as the numbers published on the MLCommons website. However, these numbers are valid and pass all the required compliance tests. The PowerEdge R7525 server with A30 GPUs behaved like the PowerEdge R7525 server with A100 GPUs in that the Server scenario results are slightly lower than the Offline results. The tuned numbers provided the best per card performance among all A30 GPU submissions.
Figure 4: DLRM results on a PowerEdge R7525 server with A100 GPUs and a PowerEdge R7525 server with A30 GPUs
RNNT
Recurrent Neural Network (RNNT) is a type of neural network in which outputs are recycled as inputs for the current step. By using one-hot encoding and memory, RNNT can remember information through time that might be useful in time series prediction. This model uses a squashing function to learn to predict the next potential word or step to take. The result of the squashing function is always between –1 and 1, which allows neural networks to remain nonlinear and thus effective as the same values are passed through the neural network.
For this model, we conducted four tests across both of our systems. In the Offline scenario, the PowerEdge R7525 server with A100 GPUs outperformed the PowerEdge R7525 server with A30 GPUs by 80 percent. In the Server scenario, the PowerEdge R7525 server with A100 GPUs excelled by performing 199 percent better than the PowerEdge R7525 server with A30 GPUs.
Figure 5: RNNT results on a PowerEdge R7525 server with A100 GPUs and a PowerEdge R7525 server with A30 GPUs
3D U-Net
3D U-Net is an elegant improvement to the sliding window approach of convolution neural networks (CNNs) in which fewer training images can be used and more precise segmentations can be yielded. In brief, an input image goes through a contraction and expansion path (in a U shape architecture with skip connections) and becomes a segmentation map output. This segmentation map provides class labels for what is inside the image. For a deeper understanding of 3D U-Net's architecture, see U-Net: Convolutional Networks for Biomedical.
Across the two systems, we conducted Offline scenario tests for the default and high accuracy modes. The default and high accuracy modes yielded the same results across the two systems. Across the two systems, the PowerEdge R7525 server with A100 GPUs performed 75 percent better than the PowerEdge R7525 server with A30 GPUs.
Figure 6: 3D U-Net results on a PowerEdge R7525 server with A100 GPUs and a PowerEdge R7525 server with A30 GPUs
Cost Considerations
When placing an order for the PowerEdge R7525 Rack Server on the Dell Technologies website, customers are guided through the purchasing process with suggestions and requirements for their specific rack server. The PowerEdge R7525 server with three NVIDIA Ampere A100 GPUs is 1.423 times more expensive than the PowerEdge R7525 server with three NVIDIA Ampere A30 GPUs. The price difference between the two configurations is due to the powerful GPU itself. Also, the PowerEdge R7525 server with A100 GPUs requires higher performance fans and a more powerful thermal configuration. Despite the additional options required for the PowerEdge R7525 server with A100 GPUs, understanding the throughput performance (queries per second (QPS) in the Server mode and samples per second in the Offline mode) per dollar provides valuable insight into achievable performance per dollar spent.
The following figure shows the relative performance of the two systems per dollar. If we divide the performance achieved on a system for a particular benchmark by the total cost of the system, we determine the achievable throughput per dollar spent on the system. The higher the throughput per dollar amount indicates that greater performance can be extracted from the system per dollar spent.
Figure 7: Relative QPS per cost of a PowerEdge R7525 server with A100 GPUs and a PowerEdge R7525 server with A30 GPUs
In the figure, the orange line shows the normalized data of the throughput per cost of the PowerEdge R7525 server with A30 GPUs. The blue bars indicate the relative achievable performance of the PowerEdge R7525 server with A100 GPUs. For most of the benchmarks, we see an acceptable range of performance on both systems. However, the PowerEdge R7525 server with A100 GPUs unconditionally outperformed the PowerEdge R7525 server with A30 GPUs in the DLRM Server default and high accuracy modes as well as in the RNNT Server mode. Both systems perform well per dollar spent.
Note: We compiled the cost data in this section from the PowerEdge R7525 Rack Server page on the Dell Technologies website on September 7, 2021. The data might be subject to change.
Conclusion
The blog provides a detailed comparison of performance between the Dell EMC PowerEdge R7525 server configured with three A100s and the Dell EMC PowerEdge R7525 server configured with three A30 GPUs. If your ML workload focuses on inferencing, the PowerEdge R7525 server configured with A100s might suit your needs well. However, if you are looking for a system that not only performs well, but also is more cost-effective, the PowerEdge R7525 server configured with A30 GPUs will suit those needs. Both systems performed well and are a great investment based on your ML workload requirements.
Next Steps
In future blogs, we plan to describe:
- How to run MLPerf Inference v1.1
- The PowerEdge R750xa server as a platform for inference v1.1
- The DSS8440 server as a platform for inference v1.1
- The PowerEdge R725 server as a platform for inference v1.1
- The PowerEdge XE8545 server as a platform for inference v1.1
- Comparison of inference v1.0 performance with inference v1.1 performance
Related Blog Posts
MLPerf™ Inference v2.0 Edge Workloads Powered by Dell PowerEdge Servers
Fri, 06 May 2022 19:54:11 -0000
|Read Time: 0 minutes
Abstract
Dell Technologies recently submitted results to the MLPerf Inference v2.0 benchmark suite. This blog examines the results of two specialty edge servers: the Dell PowerEdge XE2420 server with the NVIDIA T4 Tensor Core GPU and the Dell PowerEdge XR12 server with the NVIDIA A2 Tensor Core GPU.
Introduction
It is 6:00 am on a Saturday morning. You drag yourself out of bed, splash water on your face, brush your hair, and head to your dimly lit kitchen for a bite to eat before your morning run. Today, you have decided to explore a new part of the neighborhood because your dog’s nose needs new bushes to sniff. As you wait for your bagel to toast, you ask your voice assistant “what’s the weather like?” Within a couple of seconds, you know that you need to grab an extra layer because there is a slight chance of rain. Edge computing has saved your morning run.
Although this use case is covered in the MLPerf Mobile benchmarks, the data discussed in this blog is from the MLPerf Inference benchmark that has been run on Dell servers.
Edge computing is computing that takes place at the “edge of networks.” Edge of networks refers to where devices such as phones, tablets, laptops, smart speakers, and even industrial robots can access the rest of the network. In this case, smart speakers can perform speech-to-text recognition to offload processing that ordinarily must be accomplished in the cloud. This offloading not only improves response time but also decreases the amount of sensitive data that is sent and stored in the cloud. The scope for edge computing expands far beyond voice assistants with use cases including autonomous vehicles, 5G mobile computing, smart cities, security, and more.
The Dell PowerEdge XE2420 and PowerEdge XR 12 servers are designed for edge computing workloads. The design criteria is based on real life scenarios such as extreme heat, dust, and vibration from factory floors, for example. However, despite these servers not being physically located in a data center, server reliability and performance are not compromised.
PowerEdge XE2420 server
The PowerEdge XE2420 server is a specialty edge server that delivers high performance in harsh environments. This server is designed for demanding edge applications such as streaming analytics, manufacturing logistics, 5G cell processing, and other AI applications. It is a short-depth, dense, dual-socket, 2U server that can handle great environmental stress on its electrical and physical components. Also, this server is ideal for low-latency and large-storage edge applications because it supports 16x DDR4 RDIMM/LR-DIMM (12 DIMMs are balanced) up to 2993 MT/s. Importantly, this server can support the following GPU/Flash PCI card configurations:
- Up to 2 x PCIe x16, up to 300 W passive FHFL cards (for example, NVIDIA V100/s or NVIDIA RTX6000)
- Up to 4 x PCIe x8; 75 W passive (for example, NVIDIA T4 GPU)
- Up to 2 x FE1 storage expansion cards (up to 20 x M.2 drives on each)
The following figures show the PowerEdge XE2420 server (source):
Figure 1: Front view of the PowerEdge XE2420 server
Figure 2: Rear view of the PowerEdge XE2420 server
PowerEdge XR12 server
The PowerEdge XR12 server is part of a line of rugged servers that deliver high performance and reliability in extreme conditions. This server is a marine-compliant, single-socket 2U server that offers boosted services for the edge. It includes one CPU that has up to 36 x86 cores, support for accelerators, DDR4, PCIe 4.0, persistent memory and up to six drives. Also, the PowerEdge XR12 server offers 3rd Generation Intel Xeon Scalable Processors.
The following figures show the PowerEdge XR12 server (source):
Figure 3: Front view of the PowerEdge XR12 server
Figure 4: Rear view of the PowerEdge XR12 server
Performance discussion
The following figure shows the comparison of the ResNet 50 Offline performance of various server and GPU configurations, including:
- PowerEdge XE8545 server with the 80 GB A100 Multi-Instance GPU (MIG) with seven instances of the one compute instance of the 10gb memory profile
- PowerEdge XR12 server with the A2 GPU
- PowerEdge XE2420 server with the T4 and A30 GPU
Figure 5: MLPerf Inference ResNet 50 Offline performance
ResNet 50 falls under the computer vision category of applications because it includes image classification, object detection, and object classification detection workloads.
The MIG numbers are per card and have been divided by 28 because of the four physical GPU cards in the systems multiplied by second instances of the MIG profile. The non-MIG numbers are also per card.
For the ResNet 50 benchmark, the PowerEdge XE2420 server with the T4 GPU showed more than double the performance of the PowerEdge XR12 server with the A2 GPU. The PowerEdge XE8545 server with the A100 MIG showed competitive performance when compared to the PowerEdge XE2420 server with the T4 GPU. The performance delta of 12.8 percent favors the PowerEdge XE2420 system. However, the PowerEdge XE2420 server with A30 GPU card takes the top spot in this comparison as it shows almost triple the performance over the PowerEdge XE2420 server with the T4 GPU.
The following figure shows a comparison of the SSD-ResNet 34 Offline performance of the PowerEdge XE8545 server with the A100 MIG and the PowerEdge XE2420 server with the A30 GPU.
Figure 6: MLPerf Inference SSD-ResNet 34 Offline performance
The SSD-ResNet 34 model falls under the computer vision category because it performs object detection. The PowerEdge XE2420 server with the A30 GPU card performed more than three times better than the PowerEdge XE8545 server with the A100 MIG.
The following figure shows a comparison of the Recurrent Neural Network Transducers (RNNT) Offline performance of the PowerEdge XR12 server with the A2 GPU and the PowerEdge XE2420 server with the T4 GPU:
Figure 7: MLPerf Inference RNNT Offline performance
The RNNT model falls under the speech recognition category, which can be used for applications such as automatic closed captioning in YouTube videos and voice commands on smartphones. However, for speech recognition workloads, the PowerEdge XE2420 server with the T4 GPU and the PowerEdge XR12 server with the A2 GPU are closer in terms of performance. There is only a 32 percent performance delta.
The following figure shows a comparison of the BERT Offline performance of default and high accuracy runs of the PowerEdge XR12 server with the A2 GPU and the PowerEdge XE2420 server with the A30 GPU:
Figure 8: MLPerf Inference BERT Offline performance
BERT is a state-of-the-art, language-representational model for Natural Language Processing applications such as sentiment analysis. Although the PowerEdge XE2420 server with the A30 GPU shows significant performance gains, the PowerEdge XR12 server with the A2 GPU exceeds when considering achieved performance based on the money spent.
The following figure shows a comparison of the Deep Learning Recommendation Model (DLRM) Offline performance for the PowerEdge XE2420 server with the T4 GPU and the PowerEdge XR12 server with the A2 GPU:
Figure 9: MLPerf Inference DLRM Offline performance
DLRM uses collaborative filtering and predicative analysis-based approaches to make recommendations, based on the dataset provided. Recommender systems are extremely important in search, online shopping, and online social networks. The performance of the PowerEdge XE2420 T4 in the offline mode was 40 percent better than the PowerEdge XR12 server with the A2 GPU.
Despite the higher performance from the PowerEdge XE2420 server with the T4 GPU, the PowerEdge XR12 server with the A2 GPU is an excellent option for edge-related workloads. The A2 GPU is designed for high performance at the edge and consumes less power than the T4 GPU for similar workloads. Also, the A2 GPU is the more cost-effective option.
Power Discussion
It is important to budget power consumption for the critical load in a data center. The critical load includes components such as servers, routers, storage devices, and security devices. For the MLPerf Inference v2.0 submission, Dell Technologies submitted power numbers for the PowerEdge XR12 server with the A2 GPU. Figures 8 through 11 showcase the performance and power results achieved on the PowerEdge XR12 system. The blue bars are the performance results, and the green bars are the system power results. For all power submissions with the A2 GPU, Dell Technologies took the Number One claim for performance per watt for the ResNet 50, RNNT, BERT, and DLRM benchmarks.
Figure 10: MLPerf Inference v2.0 ResNet 50 power results on the Dell PowerEdge XR12 server
Figure 11: MLPerf Inference v2.0 RNNT power results on the Dell PowerEdge XR12 server
Figure 12: MLPerf Inference v2.0 BERT power results on the Dell PowerEdge XR12 server
Figure 13: MLPerf Inference v2.0 DLRM power results on the Dell PowerEdge XR12 server
Note: During our submission to MLPerf Inference v2.0 including power numbers, the PowerEdge XR12 server was not tuned for optimal performance per watt score. These results reflect the performance-optimized power consumption numbers of the server.
Conclusion
This blog takes a closer look at Dell Technologies’ MLPerf Inference v2.0 edge-related submissions. Readers can compare performance results between the Dell PowerEdge XE2420 server with the T4 GPU and the Dell PowerEdge XR12 server with the A2 GPU with other systems with different accelerators. This comparison helps readers make informed decisions about ML workloads on the edge. Performance, power consumption, and cost are the important factors to consider when planning any ML workload. Both the PowerEdge XR12 and XE2420 servers are excellent choices for Deep Learning workloads on the edge.
Appendix
SUT configuration
The following table describes the System Under Test (SUT) configurations from MLPerf Inference v2.0 submissions:
Table 1: MLPerf Inference v2.0 system configuration of the PowerEdge XE2420 and XR12 servers
Platform | PowerEdge XE2420 1x T4, TensorRT | PowerEdge XR12 1x A2, TensorRT | PowerEdge XR12 1x A2, MaxQ, TensorRT | PowerEdge XE2420 2x A30, TensorRT |
MLPerf system ID | XE2420_T4x1_edge_TRT | XR12_edge_A2x1_TRT | XR12_A2x1_TRT_MaxQ | XE2420_A30x2_TRT |
Operating system | CentOS 8.2.2004 | Ubuntu 20.04.4 | ||
CPU | Intel Xeon Gold 6238 CPU @ 2.10 GHz | Intel Xeon Gold 6312U CPU @ 2.40 GHz | Intel Xeon Gold 6252N CPU @ 2.30 GHz | |
Memory | 256 GB | 1 TB | ||
GPU | NVIDIA T4 | NVIDIA A2 | NVIDIA A30 | |
GPU form factor | PCIe | |||
GPU count | 1 | 2 | ||
Software stack | TensorRT 8.4.0 CUDA 11.6 cuDNN 8.3.2 Driver 510.47.03 DALI 0.31.0 | |||
Table 2: MLPerf Inference v1.1 system configuration of the PowerEdge XE8545 server
Platform | PowerEdge XE8545 4x A100-SXM-80GB-7x1g.10gb, TensorRT, Triton |
MLPerf system ID | XE8545_A100-SXM-80GB-MIG_28x1g.10gb_TRT_Triton |
Operating system | Ubuntu 20.04.2 |
CPU | AMD EPYC 7763 |
Memory | 1 TB |
GPU | NVIDIA A100-SXM-80GB (7x1g.10gb MIG) |
GPU form factor | SXM |
GPU count | 4 |
Software stack | TensorRT 8.0.2 CUDA 11.3 cuDNN 8.2.1 Driver 470.57.02 DALI 0.31.0 |
Performance of the Dell PowerEdge R750xa Server for MLPerf™ Inference v2.0
Thu, 21 Apr 2022 18:20:33 -0000
|Read Time: 0 minutes
Abstract
Dell Technologies recently submitted results to the MLPerf Inference v2.0 benchmark suite. The results provide information about the performance of Dell servers. This blog takes a closer look at the Dell PowerEdge R750xa server and its performance for MLPerf Inference v1.1 and v2.0.
We compare the v1.1 results with the v2.0 results. We show the performance difference between the software stack versions. We also use the PowerEdge R750xa server to demonstrate that the v1.1 results from all systems can be referenced for planning an ML workload on systems that are not available for MLPerf Inference v2.0.
PowerEdge R750xa server
Built with state-of-the-art components, the PowerEdge R750xa server is ideal for artificial intelligence (AI), machine learning (ML), and deep learning (DL) workloads. The PowerEdge R750xa server is the GPU-optimized version of the PowerEdge R750 server. It supports accelerators as 4 x 300 W DW or 6 x 75 W SW. The GPUs are placed in the front of the PowerEdge R750xa server allowing for better airflow management. It has up to eight available PCIe Gen4 slots and supports up to eight NVMe SSDs.
The following figures show the PowerEdge R750xa server (source):
Figure 1: Front view of the PowerEdge R750xa server
Figure 2: Rear view of the PowerEdge R750xa server
Figure 3: Top view of the PowerEdge R750xa server
Configuration comparison
The following table describes the software stack configurations from the two rounds of submission for the closed data center division:
Table 1: MLPerf Inference v1.1 and v2.0 software stacks
NVIDIA component | v1.1 software stack | v2.0 software stack |
TensorRT | 8.0.2 | 8.4.0 |
CUDA | 11.3 | 11.6 |
cuDNN | 8.2.1 | 8.3.2 |
GPU driver | 470.42.01 | 510.39.01 |
DALI | 0.30.0 | 0.31.0 |
Triton | 21.07 | 22.01 |
Although the software has been updated across the two rounds of submission, performance is consistent, if not better, for the v2.0 submission. For MLPerf Inference v2.0, Triton performance results can be extrapolated from MLPerf Inference v1.1 except for the 3D U-Net benchmark, which is due to a v2.0 dataset change.
The following table describes the System Under Test (SUT) configurations from MLPerf Inference v1.1 and v2.0 of data center inference submissions:
Table 2: MLPerf Inference v1.1 and v2.0 system configuration of the PowerEdge R750xa server
Component | v1.1 system configuration | v2.0 system configuration |
Platform | R750xa 4x A100-PCIE-80GB, TensorRT | R750xa 4xA100 TensorRT |
MLPerf system ID | R750xa_A100-PCIE-80GBx4_TRT | R750xa_A100_PCIE_80GBx4_TRT |
Operating system | CentOS 8.2 | |
CPU | Intel Xeon Gold 6338 CPU @ 2.00 GHz | |
Memory | 1 TB | |
GPU | NVIDIA A100-PCIE-80GB | |
GPU form factor | PCIe | |
GPU count | 4 | |
Software stack | TensorRT 8.0.2 | TensorRT 8.4.0 CUDA 11.6 cuDNN 8.3.2 Driver 510.39.01 DALI 0.31.0 |
In the v1.1 round of submission, Dell Technologies submitted four different configurations on the PowerEdge R750xa server. Although the GPU count of four was maintained, Dell Technologies submitted the 40 GB and the 80 GB versions of the NVIDIA A100 GPU. Additionally, Dell Technologies submitted Multi-Instance GPU (MIG) numbers using 28 instances of the one compute instance of the 10gb memory profile on the 80 GB A100 GPU. Furthermore, Dell Technologies submitted power numbers (MaxQ is a performance and power submission) for the 40 GB version of the A100 GPU and submitted with the Triton server on the 80 GB version of the A100 GPU. A discussion about the v1.1 submission by Dell Technologies can be found in this blog.
Performance comparison of the PowerEdge R70xa server for MLPerf Inference v2.0 and v1.1
ResNet 50
ReNet50 is a 50-layer deep convolution neural network that is made up of 48 convolution layers along with a single max pool and average pool layer. This model is used for computer vision applications including image classification, object detection, and object classification. For the ResNet 50 benchmark, the performance numbers from the v2.0 submission match and outperform in the server and offline scenarios respectively when compared to the v1.1 round of submission. As shown in the following figure, the v2.0 submission results are within 0.02 percent in the server scenario and outperform the previous round by 1 percent in the offline scenario:
Figure 4: MLPerf Inference v2.0 compared to v1.1 ResNet 50 per card results on the PowerEdge R750xa server
BERT
Bidirectional Encoder Representation from Transformers (BERT) is a state-of-the-art language representational model for Natural Language Processing applications. This benchmark performs the SQuAD question answering task. The BERT benchmark consists of default and high accuracy modes for the offline and server scenarios. In the v2.0 round of submission, the PowerEdge R750xa server matched and slightly outperformed its performance from the previous round. In the default BERT server and offline scenarios, the extracted performance is within 0.06 and 2.33 percent respectively. In the high accuracy BERT server and offline scenarios, the extracted performance is within 0.14 and 1.25 percent respectively.
Figure 5: MLPerf Inference v2.0 compared to v1.1 BERT per card results on the PowerEdge R750xa server
SSD-ResNet 34
The SSD-ResNet 34 model falls under the computer vision category. This benchmark performs object detection. For the SSD-ResNet 34 benchmark, the results produced in the v2.0 round of submission are within 0.14 percent for the server scenario and show a 1 percent improvement in the offline scenario.
Figure 6: MLPerf Inference v2.0 compared to v1.1 SSD-ResNet 34 per card results on the PowerEdge R750xa server
DLRM
Deep Learning Recommendation Model (DLRM) is an effective benchmark for understanding workload requirements for building recommender systems. This model uses collaborative filtering and predicative analysis-based approaches to process large amounts of data. The DLRM benchmark consists of default and high accuracy modes, both containing the server and offline scenarios. For the server scenario in both the default and high accuracy modes, the v2.0 submissions results are within 0.003 percent. For the offline scenario across both modes, the PowerEdge R750xa server showed a 2.62 percent performance gain.
Figure 7: MLPerf Inference v2.0 compared to v1.1 DLRM per card results on the PowerEdge R750xa server
RNNT
The Recurrent Neural Network Transducers (RNNT) model falls under the speech recognition category. This benchmark accepts raw audio samples and produces the corresponding character transcription. For the RNNT benchmark, the PowerEdge R750xa server maintained similar performance behavior within 0.04 percent in the server mode and showing 1.46 percent performance gains in the offline mode.
Figure 8: MLPerf Inference v2.0 compared to v1.1 RNNT per card results on the PowerEdge R750xa server
3D U-Net
The 3D U-Net performance numbers have changed in terms of scale and are not comparable in a bar graph because of a change to the dataset. The new dataset for this model is the Kitts 2019 Kidney Tumor Segmentation set. However, the PowerEdge R750xa server yielded Number One results among the PCIe form factor systems that were submitted. This model falls under the computer vision category, but it specifically deals with medical image data.
Results summary
Figure 1 through Figure 8 show the consistent performance of the PowerEdge R750xa server across both rounds of submission.
The following figure shows that in the offline scenarios for the benchmarks there is a small but noticeable performance improvement:
Figure 9: Performance improvement in percentage of the PowerEdge R750xa server across MLPerf Inference v2.0 and v1.1
The small percentage delta in the server scenarios can be a result of noise and are consistent with the previous round of submission.
Conclusion
This blog confirms the consistent performance of the Dell PowerEdge R750xa server across the MLPerf Inference v1.1 and MLPerf Inference v2.0 submissions. Because an identical system from round v1.1 performed at a consistent level for MLPerf Inference v2.0, we see that the software stack upgrades had minimal impact on performance. Therefore, the optimal results from the v1.1 round of submission can be used for making informed decisions about server performance for a specific ML workload. Because Dell Technologies submitted a diverse set of configurations in the v1.1 round of submission, customers can take advantage of many results.