Intel Arc B580 vs. AMD Radeon RX 9000 vs. NVIDIA RTX 50 Series For Llama.cpp Vulkan Performance

Written by Michael Larabel in Display Drivers on 8 December 2025 at 10:50 AM EST. Page 2 of 5. 11 Comments.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: Llama-3.1-Tulu-3-8B-Q8_0, Test: Text Generation 128. RTX 5090 was the fastest.

While the Intel open-source graphics drivers for Battlemage matured a lot over 2025, the Arc B580 itself remains much slower than the likes of the RX 9060 XT and RTX 5060 Ti. It will be interesting if the rumored Arc B770 will be able to at least tango with the Radeon RX 9060 XT.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: Llama-3.1-Tulu-3-8B-Q8_0, Test: Text Generation 128. RTX 5090 was the fastest.

The Arc B580 with the latest Xe kernel driver wasn't exposing the GPU power consumption but when it came to AMD RDNA4 vs. Blackwell, the NVIDIA RTX 50 graphics cards were delivering the best performnace-per-Watt during this testing with Llama 3.1.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: Llama-3.1-Tulu-3-8B-Q8_0, Test: Text Generation 128. RTX 5090 was the fastest.

Even at $249 USD, the Arc B580 was in last place for the performance-per-dollar in this benchmark.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: Llama-3.1-Tulu-3-8B-Q8_0, Test: Prompt Processing 512. RTX 5090 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: Llama-3.1-Tulu-3-8B-Q8_0, Test: Prompt Processing 1024. RTX 5090 was the fastest.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: Llama-3.1-Tulu-3-8B-Q8_0, Test: Prompt Processing 2048. RTX 5090 was the fastest.

With prompt processing as well the Arc B580 was a fraction of the speed of the Radeon RX 9060 XT on the latest Linux drivers.

👁 Llama.cpp benchmark with settings of Backend: Vulkan, Model: Llama-3.1-Tulu-3-8B-Q8_0, Test: Prompt Processing 2048. RTX 5090 was the fastest.

For the Llama 3.1 8B prompt processing the NVIDIA RTX 50 graphics cards continued delivering much greater performance-per-Watt than the RDNA4 graphics cards. The RTX 5060 Ti and RTX 5070 at current pricing levels were delivering the best value.

11 Comments - Next Page

Page: 1 2 3 4 5 Next Page

URL: https://www.phoronix.com/review/llama-cpp-vulkan-eoy2025/2

⇱ Intel Arc B580 vs. AMD Radeon RX 9000 vs. NVIDIA RTX 50 Series For Llama.cpp Vulkan Performance - Phoronix

Intel Arc B580 vs. AMD Radeon RX 9000 vs. NVIDIA RTX 50 Series For Llama.cpp Vulkan Performance