Stability AI, the developer of the Stable Diffusion model; conducted some important tests for its artificial intelligence. Stable Diffusion published a performance report testing Diffusion 3 with different hardware accelerators. As part of the comparison, Intel Gaudi2 and NVIDIA H100 chips faced each other.
Intel Gaudi2 vs NVIDIA H100 vs A100
Stability AI introduces Stable Diffusion 3 Intel Gaudi2 96GB, NVIDIA H100 80GB Hopper GPU and A100 80GB Tested it on Ampere GPU. In a test with 16 images per device, it was reported that Gaudi2 was able to produce images 56 percent faster than the H100 GPU.
In numerical terms, it was reported that Gaudi2 reached 927 images/second and H100 reached 595 images/second. Additionally, A100 ranked third with 381 images/second.
iQOO Z9, which attracts attention with its screen, was introduced! Here are its features and price
iQOO Z9, which has been awaited for a while, was introduced in India. The middle class model attracts attention with its display capabilities. Here are the details…
As cluster sizes grew, the performance gap became wider. A cluster of 256 Gaudi2 chips achieved 12,654 images per second (49.4 images per second per device). In comparison, the A100 cluster reached 3,992 images per second (15.6 images per second).
However, Stability AI TensorRT technology He said that thanks to this, the A100 chips are better in performance. It was stated that there was a 40 percent performance difference compared to Gaudi2.
Stability AI plans to use the Intel Gaudi2 chip on the Stability Cloud platform in the coming period. Performance tests also show that NVIDIA, which leads the industry, is facing a formidable rival.
So how did you find these tests? You can share your opinions with us in the Comments section below.