InkCore AI: MLPerf Wins, Sparse Computing Leads the Era of Large Model Applications

cartridgeartificial intelligence (AI)Once again, it scored a huge victory in the MLPerf Reasoning Test, winning all three GPT-J Large Model Reasoning titles in one fell swoop. This victory not only solidifies its place in theAIThe status of the chip field also paves the way for the application of sparse computing techniques in the era of large models.

InkCore Artificial Intelligence's S30 Compute Card has once again reached the pinnacle of MLPerf by excelling in GPT-J large model inference performance on single, 4 and 8 cards. Previously, they have won MLPerf twice in a row with consistent and strong performance.

On the 6-billion-parameter GPT-J model, InkCore achieves an amazing performance advantage of up to 1.8 times that of the 4nm-processed NVIDIA H100 through its original dual-sparse algorithm and hardware co-design. this magnificent performance once again proves the importance of sparse computing technology in the era of large models.

The emergence of sparse computing technology means a profound change in large model applications. It is expected to reduce the amount of model computation, solve the problem of sharp increase in the demand for computing resources and energy consumption, and provide new possibilities for the development of large models. InkCore has already released AI computing platforms that support a wide range of large models with impressive high performance and efficiency.

The industry agrees that the innovation of sparse computing technology, such as InkCore, will lead the development of big model applications and penetrate them into various industries.MLPerf's achievement is not only an affirmation of InkCore's product strength, but also a clear evidence that sparse computing has a revolutionary impact in big model computing. The victory of InkCore AI signifies a promising future for big model applications, in which sparse computing will play a key role.