admin 管理员组文章数量: 887021
2024年4月13日发(作者:高性能mysql第四版和第三版)
NVIDIA CONNECTX-7
400G ETHERNET
SMART ACCELERATION FOR CLOUD,
DATA-CENTER AND EDGE
ACCELERATED NETWORKING AND SECURITY FOR THE
MOST ADVANCED CLOUD AND AI WORKLOADS
The NVIDIA
®
ConnectX
®
-7 SmartNIC is optimized to deliver accelerated networking for
modern cloud, artificial intelligence, and traditional enterprise workloads. ConnectX-7
provides a broad set of software-defined, hardware accelerated networking, storage,
and security capabilities which enable organizations to modernize and secure their IT
infrastructures.
Extending the tradition of NVIDIA’s industry leading innovation for networking,
ConnectX-7, is available in 1, 2, or 4-port configurations and delivers up to 400Gb/s of
bandwidth. With features such as NVIDIA ASAP
2
- Accelerated Switching and Packet
Processing
®
, advanced RoCE, NVIDIA GPUDirect
®
Storage, and in-line hardware
acceleration for TLS/IPsec/MACsec encryption/decryption, ConnectX-7 empowers agile
and high-performance solutions from edge to core data centers to clouds, all while
enhancing network security and reducing the total cost of ownership.
Available in PCIe card and OCP3.0 form factors, ConnectX-7 empowers solutions for
cloud, hyperscale, and enterprise networking.
PRODUCT SPECIFICATIONS
Maximum Total
Bandwidth
Supported Ethernet
Speeds
Number of
Network Ports
Network Interface
Technologies
Host Interface
Cards Form Factors
Network Interfaces
400GbE
10/25/40/50/100/
200/400GbE
1/2/4
NRZ (10/25G) / PAM4
(50/100G)
PCIe Gen5.0 x16/ x32
PCIe FHHL/ HHHL,
OCP3.0 SFF
SFP56, QSFP56,
QSFP56-DD,
QSFP112, SFP112
Accelerate Software-Defined
Networking
NVIDIA ASAP
2
technology
accelerates software-defined
networking, delivering line-rate
performance with no CPU penalty.
Provide Security from
Edge to Core
Hardware engines in ConnectX-7
offload and accelerate security, with
in-line encryption/decryption of TLS,
IPsec, and MACsec.
Enhance Storage Performance
ConnectX-7 enables high-
performance storage and data access
with RoCE and GPUDirect Storage and
accelerates NVMe-oF over both RoCE
and TCP.
Enable Precision Timing
ConnectX-7 provides extremely
accurate time synchronization for
data-center applications and timing-
sensitive infrastructures.
NVIDIA CONNECTX-7 | DATASHEET | APR21
FEATURES
Network Interface
> Up to 4 network ports supporting NRZ,
PAM4 (50G and 100G), in various ports
configurations:
> 1 x 10/25/40/50/100/200/400GbE
> 2 x 10/25/40/50/100/200/400GbE
> 4 x 10/25/40/50/100/200GbE
> Up to 400Gb/s total bandwidth
> ASAP
2
accelerations/actions: counters,
QoS, NAT, aging, mirroring, sampling,
flow tag
> Hairpin flow for full hardware offload
> Highly-scalable number of classifications
and actions
> Application access to hardware statistics
> Application access to crypto offloads
> PTP-based packet pacing
> Time-based SDN acceleration (ASAP
2
)
Storage Accelerations
> NVMe
™
over Fabrics (NVMe-oF) storage
target offloads
> NVMe-oF
™
over TCP/RoCE acceleration
> Storage protocols: iSER, NFSoRDMA, SMB
Direct, NVMe-oF
™
, and more
Host Interface
> 32 lanes of PCIe Gen 5.0, compatible with
PCIe Gen 2/3/4
> Integrated PCI switch
> NVIDIA Multi-Host
™
(up to 8 hosts)
and NVIDIA Socket Direct
™
> MSI/MSI-X mechanisms
> Advanced PCIe capabilities
Cyber Security
> IInline hardware IPsec encryption
and decryption
> AES-GCM 128/256-bit key
> IPsec over RoCE
> Inline hardware TLS encryption
and decryption
> AES-GCM 128/256-bit key
> Inline hardware MACsec encryption and
decryption
> AES-GCM 128/256-bit key
> AES-GCM-XPN 128/256-bit key
> Data-at-rest AES-XTS encryption
and decryption
> AES-XTS 256/512-bit key
> Platform security
> Secure boot with hardware root-of-trust
> Secure firmware update
> On-board flash encryption
HPC / AI
> All-to-All engine
> NVIDIA GPUDirect
> NVIDIA GPUDirect Storage
Management and Control
> SMBus 2.0
> Network Controller Sideband
Interface (NC-SI)
> NC-SI, MCTP over SMBus and MCTP
over PCIe - Baseboard Management
Controller interface
> PLDM for Monitor and Control DSP0248
> PLDM for Firmware Update DSP026
> I2C interface for device control
and configuration
> General Purpose I/O pins
> SPI interface to flash
> JTAG IEEE 1149.1 and
> IEEE 1149.6
Networking
> RoCE, Zero Touch RoCE
> ASAP² - Accelerated Switch and Packet
> Processing
®
for SDN and VNF acceleration
> Single Root I/O Virtualization (SR-IOV)
> VirtIO acceleration
> Overlay network acceleration:
VXLAN, GENEVE, NVGRE
> Programmable flexible parser:
user-defined classification
> Connection tracking (L4 firewall)
> Flow mirroring, sampling and statistics
> Header rewrite
> Hierarchical QoS
> Stateless TCP offloads
Advanced Timing and
Synchronization
> Advanced PTP
> IEEE 1588v2 (any profile)
> Meets G.8273.2 Class C standard
> PTP hardware clock (PHC) (UTC format)
> 12 nanosecond accuracy
> Line-rate hardware timestamp
(UTC format)
> SyncE
> Meets G.8262.1 (eEEC)
> Configurable PPS In and configurable
PPS Out
> Time-triggered scheduling
Remote Boot
> Remote boot over Ethernet
> Remote boot over iSCSI
> UEFI support for x86 and Arm servers
> PXE boot
VNF Acceleration
> Hardware offload programmable pipeline:
> Packet classification on network layers
L2 to L4 and tunneled traffic such as GTP
and VXLAN
> Packet dispatching to multiple cores
> Multi-threaded API for concurrent update
of offloaded rules
Form Factors and Options
> PCIe HHHL/FHHL
> OCP 3.0 SFF
To learn more about the NVIDIA ConnectX SmartNICs visit /en-us/networking/ethernet-adapters/
© 2021 NVIDIA Corporation. All rights reserved. NVIDIA, the NVIDIA logo, ConnectX, ConnectX-7, GPUDirect, Multi-Host, Socket Direct, and ASAP
2
- Accelerated Switch and Packet Processing are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and other countries.
Other company and product names may be trademarks of the respective companies with which they are associated. MAY21.
版权声明:本文标题:NVIDIA ConnectX-7 商品说明书 内容由网友自发贡献,该文观点仅代表作者本人, 转载请联系作者并注明出处:http://www.freenas.com.cn/jishu/1713005812h627434.html, 本站仅提供信息存储空间服务,不拥有所有权,不承担相关法律责任。如发现本站有涉嫌抄袭侵权/违法违规的内容,一经查实,本站将立刻删除。
发表评论