Among the common RAID levels of servers, which RAID level has the best write performance? (Multiple choice)

• A.RAID60
• B.RAID0
• C.RAID1E
• D.RAID60

Comment: *

Name: *

Email: *

Verification: *

Use the preset models of the deep learning platform or use the provided Notebook development environment to develop your own deep learning models, create and start training tasks, and then import the trained models into the model warehouse. Which steps of the Huawei deep learning system belong to the above process? (Multiple choice)

• A.Data annotation
• B.Model training
• C.Model verification
• D.Model storage

Comment: *

Name: *

Email: *

Verification: *

Which of the following is correct about the characteristics of heterogeneous computing? (Multiple choice)

• A.Heterogeneous computing is to develop a series of software and hardware standards so that different types of computing devices can share the process and results of computing.
• B.GPU and CPU can work together to form a future computing system. The advantage of heterogeneous computing is computing acceleration and energy saving. Cooperative work can form a heterogeneous computing system. The advantage of heterogeneous computing is computing acceleration and energy saving.
• C.The purpose of heterogeneous computing is to solve the problem of insufficient CPU computing power.
• D.Common heterogeneous systems are CPU+GPU, GPU+FPGA, CPU+ special chip.

Comment: *

Name: *

Email: *

Verification: *

GPU is widely used in embedded systems. Which of the following descriptions about GPU is wrong in mobile phones, personal computers, workstations and video game solutions?

• A.GPU is a general-purpose processor, which must take into account calculation and control. Its processor system is SISD (Single Instruction single Data) type.
• B.GPU is composed of thousands of smaller and more efficient cores, which are designed to handle multiple tasks at the same time.
• C.GPU is mainly good at parallel computing similar to image processing, and its processor system is SIMD (Single Instruction Multiple Data) type
• D.GPU is very efficient for image and graphics processing. This is because the GPU is designed as a very high parallel architecture, which makes it more advantageous than general-purpose processor CPUs in parallel processing algorithms for large data blocks.

Comment: *

Name: *

Email: *

Verification: *

Huawei Kunpeng's full-stack cloud solution includes four levels: physical infrastructure, resource pool, cloud service, and operation management platform from the bottom up. In the ECS elastic cloud hosting service, which of the following is a feature that the ECS ARM virtual machine cannot support?

• A.In- band and out-of-band information query
• B.GPU pass-through
• C.CPU, memory, interrupt hardware assisted virtualization.
• D.Offline adjustment of virtual devices (CPU, memory, network card, disk)

Comment: *

Name: *

Email: *

Verification: *