Skip to content
TheChatArchiveThe Chat Archive
Human-written demonstration (OpenAssistant volunteers)· OpenAssistant (human-written)Hugging Face OpenAssistant Conversations (oasst2)

How has the number of transistors in GPUs increased over the years since the first…

In OpenAssistant data, “assistant” turns were written by human volunteers as demonstrations. They are not model outputs.
  1. User

    How has the number of transistors in GPUs increased over the years since the first commercial GPU?

  2. Assistant

    When talking about the transistor count in GPUs, it is important to differentiate between the processors and their memory. Typically, the transistor count in random-access memory (RAM) used in GPUs is much higher than in their processors.

    The "NEC µPD7220", released 1981, is one of the firsts graphic display processors and able to draw basic shapes, like lines, circles, arcs and text to a display. It consisted of arounc 40 thousand transistors and was manuctured using a 6µm process. The "Sony GPU", wich was manufactured for the Playstation 1 (1994) already had a transistor count of around a million, similar to the first Nvidia Card "NV1", released 1995. Progress did not slow down and by 1998 the "PowerVR2 CLX2" cracked the 10 million transistor mark. This card powered the Dreamcast and Sega Naomi consoles. Jumping into the 2010s, AMD and NVIDIA are the only two big competitors left on the GPU market. By this time, the transistor count has already surpassed 1 billion.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.