NVIDIA GPU GN* instances provide powerful computing capabilities to help you process a large number of concurrent computing tasks in real time. NVIDIA GPU instances are not only suitable for general computing scenarios such as deep learning and scientific computing, but also graphic and image processing such as 3D rendering and video encoding and decoding. Tencent Cloud GPU Cloud Computing (GCC) provides a fast, stable, and elastic computing service and allows you to manage NVIDIA GPU instances the same way you manage CVM instances.

Overview

NVIDIA GPU instances are categorized into rendering and computing types:

• Rendering: suitable for 3D rendering, video encoding and decoding, computer-aided design (CAD), etc.
• Computing: suitable for deep learning, scientific computing, computer-aided engineering (CAE), etc.

GCC provides the following types of NVIDIA GPU instances:

Type	NVIDIA GPU instance	GPU type	GPU performance	Availability zone
Computing	GN10X or GN10Xp	Tesla V100 NVLink 32GB	15.7 TFLOPS of single-precision floating point performance 7.8 TFLOPS of double-precision floating point performance 125 TFLOPS of deep learning accelerator performance through Tensor cores 300 GB/s NVLink	GN10X: Guangzhou, Shanghai, Nanjing, Beijing, Chengdu, Chongqing, Singapore, and Silicon Valley GN10Xp: Guangzhou, Shanghai, Nanjing, Beijing, Chengdu, and Chongqing
	GN8	Tesla P40	12 TFLOPS of single-precision floating point performance 47 TOPS for INT8	Hong Kong (China), Guangzhou, Shanghai, Beijing, Chengdu, Chongqing, and Silicon Valley
	GN7	Tesla T4	8.1 TFLOPS of single-precision floating point performance 130 TOPS for INT8 260 TOPS for INT4	Guangzhou, Shanghai, Nanjing, Beijing, Chengdu, Chongqing, Singapore, and Silicon Valley
	GN6 or GN6S	Tesla P4	5.5 TFLOPS of single-precision floating point performance 22 TOPS for INT8	GN6: Chengdu GN6S: Guangzhou, Shanghai, and Beijing
	GN2	Tesla M40	7 TFLOPS of single-precision floating point performance with GPU Boost 0.2 TFLOPS of double-precision floating point performance	Guangzhou, Shanghai, and Beijing
Rendering	GN7vw	Tesla T4	8.1 TFLOPS of single-precision floating point performance 130 TOPS for INT8 260 TOPS for INT4	-

Note：

Availability zone: accurate to the city level. For more information, please see the instance configuration information below.

Recommended NVIDIA GPU instances

Tencent Cloud provides NVIDIA GPU instances to meet business needs in different scenarios. Refer to the following tables to select an NVIDIA GPU instance as needed.

The table below lists recommended NVIDIA GPU instances provided by GCC. A tick (✓) indicates that the NVIDIA GPU instance supports the corresponding feature. A pentagram (★) indicates that the NVIDIA GPU instance is recommended.

Feature\Instance	GN2	GN6 or GN6S	GN7	GN8	GN10X or GN10Xp
Graphic and image processing	-	✓	✓	✓	✓
Video encoding and decoding	✓	✓	★	✓	✓
Deep learning training	✓	✓	✓	★	★
Deep learning inference	✓	★	★	★	✓
Scientific computing	✓	-	-	-	★

Note：

GN2 supports video encoding and decoding in H.264 format, but not H.265. For more information, please see Video Encode and Decode GPU Support Matrix.

Except for GN2, other types of NVIDIA GN* instances allow you to install a GRID driver to process graphics and images. To use the GRID driver, you must purchase a license.

Video encoding and decoding

Recommended instance: GN7. GN7 uses T4 GPU. It features high performance and the lowest cost for transcoding a single video, making it suitable for video encoding and decoding.

Deep learning training

Recommended instance: GN8, GN10X, or GN10Xp. GN8 and GN10X use P40 or V100 mid- to high-end GPU. They feature powerful single-precision floating-point computing capabilities and large on-board memories, making them ideal for deep learning training.

Deep learning inference

Recommended instance: GN6, GN6S, GN7, or GN8. These instances use P4, T4, or P40 GPU. They feature INT8 computing capabilities and cost-efficiency, making them suitable for mass deployment.

Scientific computing

Recommended instance: GN10X or GN10Xp. GN10X and GN10Xp use V100 GPU. They feature powerful double-precision floating-point computing capabilities and provide optimal acceleration for computational science and engineering applications.

Note：

• These recommendations are for reference only. Please select an appropriate instance based on your needs.
• To use NVIDIA GPU instances for general computing tasks, you must install the Tesla driver and CUDA toolkit. For more information, please see Installing NVIDIA Driver and Installing CUDA Toolkit.
• To use NVIDIA GPU instances for 3D rendering tasks such as high-performance graphic processing and video encoding and decoding, you must install a GRID driver and configure a license server.

Service options

• NVIDIA GPU instances can be launched in Virtual Private Cloud (VPC).
• NVIDIA GPU instances support services such as Cloud Load Balancer (CLB) with no additional management and OPS costs. Private network traffic is free of charge.

GN10X and GN10Xp for computing tasks

NVIDIA GPU instances GN10X and GN10Xp support not only general GPU computing tasks such as deep learning and scientific computing, but also graphic and image processing tasks such as 3D rendering and video encoding and decoding.

Application scenarios

GN10X and GN10Xp feature powerful double-precision floating-point computing capabilities. They are applicable to the following scenarios:

• Large-scale deep learning training and inference as well as scientific computing scenarios, such as:
  • Deep learning
  • High-performance database
  • Computational fluid dynamics
  • Computational finance
  • Seismic analysis
  • Molecular modeling
  • Genomics and others
• Graphic and image processing scenarios, such as:
  • Graphic and image processing
  • Video encoding and decoding
  • Graph database

Hardware specifications

• CPU: GN10X is configured with an Intel^® Xeon^® Gold 6133 CPU with a clock rate of 2.5 GHz. GN10Xp is configured with an Intel^® Xeon^® Platinum 8255C CPU with a clock rate of 2.5 GHz.
• GPU: NVIDIA^® Tesla^® V100 NVLink 32GB, providing 15.7 TFLOPS of single-precision floating point performance, 7.8 TFLOPS of double-precision floating point performance, 125 TFLOPS of deep learning accelerator performance with Tensor cores, and 300 GB/s NVLink.
• Memory: DDR4, providing memory bandwidth up to 2,666 MT/s.
• Storage: Select the appropriate CBS cloud disk type. To expand the cloud disk capacity, create and mount an elastic cloud disk.
• Network: network optimization is enabled by default. The network performance of an instance depends on its specification. You can purchase public network bandwidth as needed.

GN10X and GN10Xp instances provide the following configurations:

Model	GPU (NVIDIA Tesla V100 NVLink 32GB)	GPU memory (HBM2)	vCPU	Memory (DDR4)	Private network bandwidth	Packet forwarding rate (PPS)	Number of queues	Availability zone
GN10X.2XLARGE40	1	1 * 32 GB	8 cores	40 GB	4 Gbps	800,000	2	Guangzhou Zone 3, Guangzhou Zone 4, Shanghai Zone 2, Shanghai Zone 3, Nanjing Zone 1, Beijing Zone 4, Beijing Zone 5, Chengdu Zone 1, Chongqing Zone 1, Singapore Zone 1, and Silicon Valley Zone 2
GN10X.9XLARGE160	4	4 * 32 GB	36 cores	160 GB	13 Gbps	2.5 million	9
GN10X.18XLARGE320	8	8 * 32 GB	72 cores	320 GB	25 Gbps	4.9 million	16
GN10X.4XLARGE80	2	2 * 32 GB	18 cores	80 GB	7 Gbps	1.5 million	4	Guangzhou Zone 3, Guangzhou Zone 4, Nanjing Zone 1, Chengdu Zone 1, and Chongqing Zone 1
GN10X.MEDIUM10	1/4	8 GB vGPU	2 cores	10 GB	1 Gbps	200,000	2	-
GN10X.LARGE20	1/2	16 GB vGPU	4 cores	20 GB	2 Gbps	400,000	2
GN10Xp.2XLARGE40	1	1 * 32 GB	10 cores	40 GB	4 Gbps	800,000	10	Guangzhou Zone 3, Guangzhou Zone 4, Shanghai Zone 2, Nanjing Zone 1, Beijing Zone 5, Chengdu Zone 1, and Chongqing Zone 1
GN10Xp.5XLARGE80	2	2 * 32 GB	20 cores	80 GB	7 Gbps	1.5 million	16
GN10Xp.10XLARGE160	4	4 * 32 GB	40 cores	160 GB	13 Gbps	2.5 million	16
GN10Xp.20XLARGE320	8	8 * 32 GB	80 cores	320 GB	25 Gbps	4.9 million	16

Note：

vGPU: GN10X instance cluster provides vGPU-based instances, which are currently in beta. The vGPU type is vComputeServer, which only supports CUDA APIs.

GN8 for computing tasks

NVIDIA GPU instance GN8 supports not only general GPU computing tasks such as deep learning, but also graphic and image processing tasks such as 3D rendering and video encoding and decoding.

Application scenarios

GN8 is applicable to the following scenarios:

• Deep learning training and inference scenarios, such as:
  • AI inference with high throughput
  • Deep learning
• Graphic and image processing scenarios, such as:
  • Graphic and image processing
  • Video encoding and decoding
  • Graph database

Hardware specifications

• CPU: Intel^® Xeon^® E5-2680 v4 CPU, 2.4 GHz.
• GPU: NVIDIA^® Tesla^® P40, providing 12 TFLOPS of single-precision floating point performance and 47 TOPS for INT8.
• Memory: DDR4, providing memory bandwidth up to 2,666 MT/s.
• Storage: Select the appropriate CBS cloud disk type. To expand the cloud disk capacity, create and mount an elastic cloud disk.
• Network: network optimization is enabled by default. The network performance of an instance depends on its specification. You can purchase public network bandwidth as needed.

GN8 instance provides the following configurations:

Model	GPU (NVIDIA Tesla P40)	GPU memory (GDDR5)	vCPU	Memory (DDR4)	Private network bandwidth	Packet forwarding rate (PPS)	Number of queues	Availability zone
GN8.LARGE56	1	24 GB	6 cores	56 GB	1.5 Gbps	450,000	6	Hong Kong Zone 2, Guangzhou Zone 3, Shanghai Zone 3, Beijing Zone 2, Beijing Zone 4, Chengdu Zone 1, Chongqing Zone 1, and Silicon Valley Zone 1
GN8.3XLARGE112	2	48 GB	14 cores	112 GB	2.5 Gbps	500,000	8
GN8.7XLARGE224	4	96 GB	28 cores	224 GB	5 Gbps	700,000	8
GN8.14XLARGE448	8	192 GB	56 cores	448 GB	10 Gbps	700,000	8

GN7 for computing tasks

NVIDIA GPU instance GN7 supports not only general GPU computing tasks such as deep learning, but also graphic and image processing tasks such as 3D rendering and video encoding and decoding.

Application scenarios

GN7 is cost-effective and applicable to the following scenarios:

• Deep learning inference and small-scale training scenarios, such as:
  • AI inference for mass deployment
  • Small-scale deep learning training
• Graphic and image processing scenarios, such as:
  • Graphic and image processing
  • Video encoding and decoding
  • Graph database

Hardware specifications

• CPU: Intel^® Xeon^® Platinum 8255C CPU, 2.5 GHz.
• GPU: NVIDIA^® Tesla^® T4, providing 8.1 TFLOPS of single-precision floating point performance, 130 TOPS for INT8, and 260 TOPS for INT4.
• Memory: DDR4, providing memory bandwidth up to 2,666 MT/s.
• Storage: Select the appropriate CBS cloud disk type. To expand the cloud disk capacity, create and mount an elastic cloud disk.
• Network: network optimization is enabled by default. The network performance of an instance depends on its specification. You can purchase public network bandwidth as needed.

GN7 instance provides the following configurations:

Model	GPU (NVIDIA Tesla T4)	GPU memory (GDDR6)	vCPU	Memory (DDR4)	Private network bandwidth	Packet forwarding rate (PPS)	Number of queues	Availability zone
GN7.LARGE20	1/4	4 GB vGPU	4 cores	20 GB	2 Gbps	500,000	4	Guangzhou Zone 3, Guangzhou Zone 4, Shanghai Zone 2, Shanghai Zone 4, Nanjing Zone 1, Nanjing Zone 2, Beijing Zone 3, Beijing Zone 5, Chengdu Zone 1, Chongqing Zone 1, and Silicon Valley Zone 2
GN7.2XLARGE40	1/2	8 GB vGPU	10 cores	40 GB	4 Gbps	700,000	10
GN7.2XLARGE32	1	1 * 16 GB	8 cores	32 GB	7 Gbps	600,000	8	Guangzhou Zone 3, Guangzhou Zone 4, Shanghai Zone 2, Shanghai Zone 4, Nanjing Zone 1, Nanjing Zone 2, Beijing Zone 3, Beijing Zone 5, Chengdu Zone 1, Chongqing Zone 1, Singapore Zone 1, and Silicon Valley Zone 2
GN7.5XLARGE80	1	1 * 16 GB	20 cores	80 GB	7 Gbps	1.4 million	16
GN7.8XLARGE128	1	1 * 16 GB	32 cores	128 GB	7 Gbps	2.4 million	16
GN7.10XLARGE160	2	2 * 16 GB	40 cores	160 GB	13 Gbps	2.8 million	16
GN7.20XLARGE320	4	4 * 16 GB	80 cores	320 GB	25 Gbps	5.6 million	16

Note：

vGPU: GN7 instance cluster provides vGPU-based instances. The vGPU type is vComputeServer, which only supports CUDA APIs.

GN6 and GN6S for computing tasks

NVIDIA GPU instances GN6 and GN6S support not only general GPU computing tasks such as deep learning, but also graphic and image processing tasks such as 3D rendering and video encoding and decoding.

Application scenarios

GN6 and GN6S are cost-effective and applicable to the following scenarios:

• Deep learning inference and small-scale training scenarios, such as:
  • AI inference for mass deployment
  • Small-scale deep learning training
• Graphic and image processing scenarios, such as:
  • Graphic and image processing
  • Video encoding and decoding
  • Graph database

Hardware specifications

• CPU: GN6 is configured with an Intel^® Xeon^® E5-2680 v4 CPU with a clock rate of 2.4 GHz. GN6S is configured with an Intel^® Xeon^® Silver 4110 CPU with a clock rate of 2.1 GHz.
• GPU: NVIDIA^® Tesla^® P4, providing 5.5 TFLOPS of single-precision floating point performance and 22 TOPS for INT8.
• Memory: DDR4, providing memory bandwidth up to 2,666 MT/s.
• Storage: Select the appropriate CBS cloud disk type. To expand the cloud disk capacity, create and mount an elastic cloud disk.
• Network: network optimization is enabled by default. The network performance of an instance depends on its specification. You can purchase public network bandwidth as needed.

GN6 and GN6S instances provide the following configurations:

Model	GPU (NVIDIA Tesla P4)	GPU memory (GDDR5)	vCPU	Memory (DDR4)	Private network bandwidth	Packet forwarding rate (PPS)	Number of queues	Availability zone
GN6.7XLARGE48	1	8 GB	28 cores	48 GB	5 Gbps	1.2 million	16	Chengdu Zone 1
GN6.14XLARGE96	2	16 GB	56 cores	96 GB	10 Gbps	1.2 million	16
GN6S.LARGE20	1	8 GB	4 cores	20 GB	7 Gbps	500,000	2	Guangzhou Zone 3, Shanghai Zone 2, Shanghai Zone 3, Shanghai Zone 4, Beijing Zone 4, and Beijing Zone 5
GN6S.2XLARGE40	2	16 GB	8 cores	40 GB	13 Gbps	800,000	2

GN2 for computing tasks

NVIDIA GPU instance GN2 supports not only general GPU computing tasks such as deep learning and scientific computing, but also graphic and image processing tasks such as video encoding and decoding.

Application scenarios

GN2 is applicable to deep learning training and inference as well as scientific computing scenarios, such as:

• Deep learning
• High-performance database
• Computational fluid dynamics
• Computational finance
• Seismic analysis
• Molecular modeling
• Genomics and others

GN2 is also applicable to graphic and image processing scenarios. For example, GN2 supports video encoding and decoding in H.264 format, but not H.265. For more information, please see Video Encode and Decode GPU Support Matrix.

Hardware specifications

• CPU: Intel^® Xeon^® E5-2680 v4 (Broadwell), 2.4 GHz.
• GPU: NVIDIA^® Tesla^® M40, providing 7 TFLOPS of single-precision floating point performance with GPU Boost and 0.2 TFLOPS of double-precision floating point performance.
• Memory: DDR4, providing memory bandwidth up to 2,666 MT/s.
• Storage: local SSD disk. Currently, this instance type does not support cloud disks.
• Network: network optimization is enabled by default. The network performance of an instance depends on its specification. You can purchase public network bandwidth as needed.

GN2 instance provides the following configurations:

Model	GPU (NVIDIA Tesla M40)	GPU memory (GDDR5)	vCPU	Memory (DDR4)	Private network bandwidth	Packet forwarding rate (PPS)	Number of queues	Availability zone
GN2.7XLARGE48	1	24 GB	28 cores	48 GB	5 Gbps	400,000	8	Guangzhou Zone 3, Beijing Zone 2, and Shanghai Zone 2
GN2.14XLARGE96	2	48 GB	56 cores	96 GB	10 Gbps	700,000	8
GN2.7XLARGE56	1	24 GB	28 cores	56 GB	5 Gbps	400,000	8	Guangzhou Zone 3, Beijing Zone 2, and Shanghai Zone 2
GN2.14XLARGE112	2	48 GB	56 cores	112 GB	10 Gbps	700,000	8

Application scenarios

GN7vw is applicable to graphic and image processing scenarios, such as:

• Graphic and image processing
• Video encoding and decoding
• Graph database

Hardware specifications

• CPU: Intel^® Xeon^® Platinum 8255C CPU, 2.5 GHz.
• GPU: NVIDIA^® Tesla^® T4, providing 8.1 TFLOPS of single-precision floating point performance, 130 TOPS for INT8, and 260 TOPS for INT4.
• Memory: DDR4, providing memory bandwidth up to 2,666 MT/s.
• Storage: Select the appropriate CBS cloud disk type. To expand the cloud disk capacity, create and mount an elastic cloud disk.
• Network: network optimization is enabled by default. The network performance of an instance depends on its specification. You can purchase public network bandwidth as needed.