According to industry publication Graphic Speak, Samsung brought on Dr. Chien-Ping Lu for his experience in the GPU space.
Lu’s previous experience includes work on Nvidia’s integrated graphics project nForce. He also lead the team that helped MediaTek with its own in-house GPU project. Unfortunately, MediaTek reportedly cancelled the project near completion.
Samsung scooped up Lu for its own GPU project, dubbed SGPU.
The South Korean electronics giant’s Austin, Texas and San Jose, California research labs are currently working on the SGPU project as well.
Interestingly, reports surfaced in June 2018 that both labs were hiring for a GPU project. Lu’s work on the SGPU project could be the catalyst for those hirings.
Graphic Speak reports that Samsung’s GPU efforts started at its Suwon facilities in South Korea.
Researchers there were investigating system architectures similar to ‘very long instruction word’ or VLIW GPU architecture.
VLIW GPUs allows programs to specify instructions to be executed in parallel. This runs contrary to typical CPU architecture, which allows programs to specify instructions to execute only in a sequence.
VLIW GPUs can process large amounts of data very quickly, which is great for graphics processing. Additionally, VLIW GPUs could have applications in artificial intelligence and virtual reality.
Despite all this, it’s unclear where we’ll see these new GPUs. According to the reports from June, Samsung considered in-house GPUs for low-tier devices.
However, Lu’s GPU work is reportedly showing better than expected performance. If the SGPU project goes well for Samsung, we could see Samsung GPUs in phones, TVs and more.
Additionally, these GPUs could prove to be a boon for Samsung’s AI and VR teams too.