Well-known chip design firm Imagination Technologies has raised a huge amount of funding to get started and develop native technology to meet AI demands.
Imagination is responsible for the PowerVR GPU architecture, whose technology still powers 12 billion devices, and is currently planning a major comeback.
It would be fair to classify Imagination Technologies as the “NVIDIA” of the late 20th century. The company is known for its PowerVR GPUs, which have seen widespread adoption in both desktop and mobile markets, with the likes of Apple being key partners of the GPU manufacturer. However, things started to go wrong for the company in 2017 when Apple decided to dump the company in favor of its own custom GPU solution. Imagination Technologies was then acquired by Chinese investment firm Canyon Bridge, and the company has since gained considerable popularity in China.
Imagination Technologies is poised to make significant strides in the edge AI space, having undergone a major shift in leadership and worked from the ground up to advance its GPU IP technology. Separately, the company received $100 million from Fortress Investment Group in an attempt to revive the operator, this time with a primary focus on AI.
Imagination believes that its GPU IP is already sophisticated enough when it comes to power efficiency and data movement, so to gain an advantage in the AI boom it needs to focus on the software side.
“There are two ways to approach AI,” Mamutla says. “Either you take something like an NPU and make it more general purpose, sacrificing some PPA, or you start with an inherently flexible GPU and increase the PPA by adding dense, low-precision compute capabilities.”
– Tim Mamutola, Chief of Innovation and Engineering, EE Times
Mamutla said we've already reached the peak in terms of what manufacturers can integrate into their GPUs, and the only way to stay competitive in this space is to build a robust AI software ecosystem, adding that the breakthrough here would be for Imagination to achieve some mechanism that would allow it to compete with NVIDIA's CUDA software stack, perhaps by offering a platform that allows developers to run their hardware with any software stack, hinting at something like ZLUDA.
The key problem we're trying to solve is to develop something that can compete with (Nvidia's) CUDA. Developers just want to run (their applications) on the platform, they just want it to be translated and it to work. Performance aside, it just needs to work. The fundamental problem we had with our AI accelerator was just getting it to work in the first place.
Imagination Technologies is known for its impressive portfolio of AI products. Notable products include the PowerVR Series 9XE and their Neural Network Accelerator, designed specifically for neural network computations. The company plans to release next-generation products, but for now, they are primarily focused on improving their software stack and watching how market demand changes.
Whether Imagination can make a comeback here remains to be seen, but the overhaul of both its plans and leadership indicates the company is determined to make a breakthrough in the market. We'll just have to wait and see what they have in store for us.
News source: EE Times
Share this story
