Survey Reveals AI Professionals Considering Switching From NVIDIA To AMD, Cite Instinct MI300X Performance & Cost

wccftech.com:

A recently conducted survey has revealed that a large chunk of AI professionals are looking to switch from NVIDIA to AMD Instinct MI300X GPUs.

AMD's Instinct MI300X May Be The Gateway To The Firm's Entrance In The Ongoing AI "Gold Rush" As It Attracts Huge Interest

Jeff Tatarchuk, who works at TensorWave, recently surveyed 82 engineers and AI professionals, claiming that around 50% of them expressed their confidence in utilizing AMD Instinct MI300X GPU, considering that it offers a better price-to-performance ratio, along with having extensive availability compared to counterparts such as NVIDIA H100s. Apart from that, Jeff also says that TensorWave will employ the MI300X AI accelerators, which is another promising news for Team Red, considering that their Instinct lineup, in general, hasn't seen the adoption level compared to others like NVIDIA.

Related Story AMD Raises FHD Monitor Requirements To 144 Hz For Freesync Tag, 200Hz FHD For Premium Pro

For a quick rundown on the MI300X Instinct AI GPU, it is designed solely on the CDNA 3 architecture, and a lot of stuff is going on. It hosts a mix of 5nm and 6nm IPs, combining to deliver up to 153 Billion transistors (MI300X). Memory is another area where a huge upgrade has been witnessed, with the MI300X boasting 50% more HBM3 capacity than its predecessor, the MI250X (128 GB). Here is how it compares to NVIDIA's H100:

• 2.4X Higher Memory Capacity
• 1.6X Higher Memory Bandwidth
• 1.3X FP8 TFLOPS
• 1.3X FP16 TFLOPS
• Up To 20% Faster Vs H100 (Llama 2 70B) In 1v1 Comparison
• Up To 20% Faster Vs H100 (FlashAttention 2) in 1v1 Comparison
• Up To 40% Faster Vs H100 (Llama 2 70B) in 8v8 Server
• Up To 60% Faster Vs H100 (Bloom 176B) In 8v8 Server

We recently reported how AMD's flagship Instinct AI accelerator has been causing "headaches" for market competitors. This is because not only are the performance gains with the MI300X huge, but AMD has timed its release perfectly, as NVIDIA is currently stuck with the "weight" of order