• By quade
• 15 July 2024

New Apple AI and HPC Development Tools Will Not Use Nvidia Products 

In surprising news, the new Mac developer tools and their upcoming Apple AI will not use Nvidia products or components. Instead, the system will leverage Apple’s homegrown chips. This is the Xcode 16 released in the recent WWDC and has AI integrated into its system which is supposed to make coding and programming easier.

This will limit HPC users’ ability to use parallel programming frameworks with Intel or Nvidia. But this can also affect how well this system can be integrated with other ecosystems.

What Can the New Apple AI Do?

The current line of Apple PCs uses silicone, GPUs, and AI chips that are locally made in their companies instead of using other systems. Before that, they relied on the x86 chips and GPUs from AMD and Nvidia. 

While this prevents the risks of hackers and other security threats, it comes at the cost of limiting the environment that can be used to create AI applications or code. However, the WWDC event revealed Apple’s plans to continue developing machine-learning models that use the same CoreML format. This system will leverage homegrown CPUs, GPUs, and neural processors.

However, tools like CoreML can convert PyTorch models to work with Apple AI’s existing hardware. Aside from that, other tools like Jax, TensorFlow, and MLX can also be used.

Apple meanwhile reveals that their Apple AI will be trained with the Google Tensor Processing Unit. This will include new features like the Swift Programming language and a Code Complete ability can predict code. Last is an extra feature called Swift Assisst which answers questions about coding.

Apple Drops Nvidia and Intel

With Apple no longer supporting providers like Intel and Nvidia, they said they will not waste time with MacOS. Intel has already pulled out its support with MacOS in its newest version of the OneAPI programming network.

Meanwhile, Nvidia discontinued support for Apple products, especially in AI and HPC CUDA programming tools several years ago. CUDA is a critical tool for AI development when using Nvidia products. A system like this can help Nvidia lock down customers to use only their hardware and software. This means that developers will need to use either Linux or Windows if they want to use Nvidia GPUs to develop applications.

One reason for Apple’s refusal to use these systems is that they want to focus on power efficiency and have their systems handle that. In contrast, the Nvidia systems run power-intensive training and inference to train its LLM.

Instead, Apple has used a Private Compute Cloud housed in Google’s data centers. They have also started to use their own AI framework called Metal which can be used for GPUs and is almost entirely exclusive to Apple’s in-house GPU.

However, there are still some ways for Mac users to use the Apple AI and Nvidia’s GPUs as long as the latter is hosted in the cloud. this is because Cloud providers have working environments without any ties to existing operating systems. 

Does the Apple AI Developments Affect BPOs?

These changes to Apple AI can have a big impact on BPOs. Different companies and Global BPO providers use different systems. this can include Apple, Windows, or Linux. Having easy integration is important to get good clients.

If things like Apple won’t allow Nvidia products to be used, this can cause problems with integration and make coordination between two groups difficult. 

On the other hand, this can also have some benefits for BPO IT services. Because these tools are developing independently of each other, they will be more complete as each ecosystem will want to include everything to entice more users.