• Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    AI
    Announcement

    Build-A-Chip-Bench tests AI on end-to-end chip design

    A researcher introducing the benchmark says the top model scored 16.6%, while meeting physical requirements remains a major challenge.

    William WangWW
    Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)T(
    2 Sources, 1h ago, first seen 1h ago

    TLDR

    A researcher introducing Build-A-Chip-Bench says ChipAgents built it with NVIDIA to test whether AI can take a chip design from specification to verified GDS. The top model scored 16.6%, he says. Models can sometimes produce functionally correct RTL, but reliably meeting physical requirements remains a major challenge.

    Combined views

    2.4K

    2 Sources, first seen 1h ago

    50 likes4 comments11 saves2 reposts

    Combined views

    2.4K

    2 Sources, first seen 1h ago

    50 likes4 comments11 saves2 reposts
    Image: @WilliamWangNLP on X

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Featured Source

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    2 Sources

    William Wang@WilliamWangNLPAs many of you know, over the past two years I’ve shifted my research from foundation models and NLP into AI-native chip design. What I’ve found: the problems feel 10x harder than software engineering. Writing functionally correct RTL is only the beginning. As models saturate on benchmarks like CVDP and ChipBench, I keep coming back to one question: how far can AI go toward designing a complete chip on its own? Today, we’re thrilled to introduce Build-A-Chip-Bench, the industry’s first AI benchmark for end-to-end chip design, built by ChipAgents in collaboration with @NVIDIA. It tests whether AI can take a design from specification all the way to verified GDS. The top model scored just 16.6%. Today’s models can produce functionally correct RTL in some cases, but reliably meeting physical requirements remains a major challenge. We still have a long way to go, and that’s what makes this research so exciting to me. Proud of our research team and grateful to our industry partners for helping us give the community an honest measure of where we are and how much work lies ahead.1h
    Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTexthese are really high scores actually, seeing as these models are 2 tiers behind the closed frontier. Also however OpenAI made Jalapeno, they can clearly accelerate chip designs without end-to-end autonomy. Nvidia should sweat bullets and double down on supply chain thuggery1h
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI

    2 Sources

    William Wang@WilliamWangNLPAs many of you know, over the past two years I’ve shifted my research from foundation models and NLP into AI-native chip design. What I’ve found: the problems feel 10x harder than software engineering. Writing functionally correct RTL is only the beginning. As models saturate on benchmarks like CVDP and ChipBench, I keep coming back to one question: how far can AI go toward designing a complete chip on its own? Today, we’re thrilled to introduce Build-A-Chip-Bench, the industry’s first AI benchmark for end-to-end chip design, built by ChipAgents in collaboration with @NVIDIA. It tests whether AI can take a design from specification all the way to verified GDS. The top model scored just 16.6%. Today’s models can produce functionally correct RTL in some cases, but reliably meeting physical requirements remains a major challenge. We still have a long way to go, and that’s what makes this research so exciting to me. Proud of our research team and grateful to our industry partners for helping us give the community an honest measure of where we are and how much work lies ahead.1h
    Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTexthese are really high scores actually, seeing as these models are 2 tiers behind the closed frontier. Also however OpenAI made Jalapeno, they can clearly accelerate chip designs without end-to-end autonomy. Nvidia should sweat bullets and double down on supply chain thuggery1h
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet