Graphics Cards Linux Reviews & Articles

There have been 398 Linux hardware reviews and benchmark articles on Phoronix for graphics cards. Separately, check out our news section for related product news.

Following last week's AMD Radeon RX 9060 XT Linux graphics/gaming review for launch day, today's article is providing an initial look at the GPU compute performance for this new RDNA4-powered ~$349 graphics card on Linux with ROCm 6.4.1.

Last month I began the much anticipated AMD Strix Halo Linux benchmarking at Phoronix by testing the top-end Ryzen AI Max+ PRO 395 that features 16 cores / 32 threads and the very impressive Radeon 8060S integrated graphics. Coming in one step below that flagship Strix Halo SoC is the Ryzen AI Max (PRO) 390 with Radeon 8050S graphics. Coming out today on Phoronix - coincidentally timed for the 21st birthday of Phoronix.com - is the first benchmarks of the Ryzen AI Max PRO 390 along with the Radeon 8050S graphics.

Ahead of the AMD Radeon RX 9060 XT graphics card hitting retailers tomorrow, today the review embargo lifts on this latest addition to the RDNA4 family. Here are the initial Linux graphics performance benchmarks for this new $349 graphics card compared to other AMD Radeon graphics cards as well as the NVIDIA GeForce and Intel Arc competition.

This month I have been running many Linux benchmarks of the HP ZBook Ultra G1a with the very exciting Ryzen AI MAX+ PRO 395 Strix Halo SoC featuring the powerful Radeon 8060S graphics. While there were the very promising OpenGL and Vulkan benchmarks shown so far from AMD Strix Halo on Linux -- including the very compelling performance compared to Microsoft Windows 11 -- many are interested in the ROCm compute aspects for Strix Halo. Here are some of the first benchmarks of the GPU compute performance for the Ryzen AI MAX+ PRO with the new ROCm 6.4.1 release compared to Strix Point as well as Intel Xe2 Lunar Lake on their Compute Runtime stack.

Intel is using Computex 2025 to showcase their new Arc Pro B-Series graphics cards that will be available in Q3 for professional use-cases as well as focusing on AI inference workstations and edge computing workloads. Plus they are noting some significant improvements coming to their Linux software stack.

As shown in today's article the AMD Ryzen AI Max+ PRO 395 Linux performance is incredible with its 16 Zen 5 cores delivering staggering laptop / mobile workstation performance with a 55 Watt default TDP. But that's only half the magic of Strix Halo, with the other aspect being the very capable integrated RDNA 3.5 graphics with unified memory support. Given this being an equally interesting topic for Linux users considering a Strix Halo laptop or desktop, this article is centered around the integrated Radeon 8060S graphics support and performance under Linux.

Earlier this week the NVIDIA GeForce RTX 5060 Ti launched and there were launch-day Linux CUDA/OpenCL compute benchmarks on Phoronix. But for the Linux gaming performance tests we were waiting on a new supported driver release, which happened to be on launch day with the NVIDIA 575.51.02 Linux beta. Now that the gaming-ready Linux driver is available for the GeForce RTX 5060 series, here are some initial benchmarks of the PNY GeForce RTX 5060 Ti 16GB up against other NVIDIA GeForce and AMD Radeon graphics cards using the newly-released Ubuntu 25.04.

Yesterday NVIDIA announced the GeForce RTX 5060 "Blackwell" graphics cards as their new, most affordable offering of the RTX 50 series. While the $299 GeForce RTX 5060 isn't shipping until next month, today the GeForce RTX 5060 Ti 8GB goes on sale for $379 USD and the GeForce RTX 5060 Ti 16GB is also hitting Internet retailers today and starting out at $429 USD. I've been testing out the GeForce RTX 5060 Ti 16GB the past several days under Linux and have initial GPU compute benchmarks to share today.

Earlier this month for launch-day there were NVIDIA GeForce RTX 5070 Linux GPU compute benchmarks. The graphics/gaming benchmarks of the GeForce RTX 5070 on Linux were held up by waiting for a new R570 Linux driver release with proper support for this new Blackwell graphics card. Last week that new Linux driver arrived in the form of the NVIDIA 570.133.07 Linux build. That new NVIDIA Linux driver is working out great with the GeForce RTX 5070 Founder's Edition and in this article are some initial Linux gaming/graphics performance benchmarks for that new graphics card competing with the AMD Radeon RX 9070 series.

AMD recently allowed me some time with their AMD Accelerator Cloud (AAC) leveraging multiple Instinct MI300X accelerators. During this brief opportunity to try out their latest software advancements with the Instinct MI300X and the ROCm compute stack, one of the most striking takeaways was their documentation improvements compared to previous forays into ROCm+Instinct compute. In addition, AMD is now offering more robust container options for easier Instinct compute deployments with more software options available and being more regularly updated.

Last week AMD formally announced the Radeon RX 9070 series graphics cards that will begin shipping tomorrow at $549 for the Radeon RX 9070 and $599 for the RX 9070 XT. Today the review embargo is lifted so we can now share Linux performance benchmarks and more details on the open-source Linux driver support for these first AMD RDNA4 graphics cards.

In addition to the Radeon RX 9070 series Linux gaming/graphics benchmarks with today's embargo lift, I've also spent some time working on some GPU compute benchmarks for these first RDNA4 graphics cards. Here is a look at some initial GPU compute benchmarks of the Radeon RX 9070 and Radeon RX 9070 XT graphics cards, mostly on cross-vendor OpenCL benchmarks, and a few words on the GPU compute stack support for the Radeon RX 9070.

Ahead of NVIDIA GeForce RTX 5070 graphics cards seeing retail availability tomorrow, today the review embargo expires on the GeForce RTX 5070 Founders Edition graphics cards. I've been testing out the GeForce RTX 5070 under Linux and today have a number of GPU compute benchmarks to share.

The embargo is over! We finally can share details on the exciting Radeon RX 9070 series graphics cards powered by RDNA4 that will be available from Internet retailers next week.

Complementing the recent Linux GPU benchmarks of the NVIDIA GeForce RTX 5080 and GeForce RTX 5090 looking at both the Linux / Steam Play gaming performance as well as GPU compute and other areas, in today's testing is a wide multi-generation look seeing how the NVIDIA GeForce performance has evolved going back to the GeForce GTX 980 Maxwell GPUs up through the newest GeForce RTX 5080/5090 graphics cards.

Over the past week I have published a number of GeForce RTX 5090 Linux compute benchmarks as well as the GeForce RTX 5080 on Linux. With that early NVIDIA R570 Linux driver build as part of the CUDA 12.8 package I was asked to wait on Linux gaming benchmarks until the proper RTX 50 Linux driver is released. Well, it was released this morning with the NVIDIA 570.86.16 Linux beta availability and have in turn been pushing the GeForce RTX 5080 and RTX 5090 through a number of Linux gaming/graphics benchmarks.

Last week was the review embargo lift on the NVIDIA GeForce RTX 5090 graphics card while today the review embargo lifts on the GeForce RTX 5080, both of which graphics cards are officially available in the retail channel tomorrow (30 January). Due to waiting on the official NVIDIA R570 Linux driver release that is recommended for the RTX 50 series Linux gaming, today's tests at Phoronix are looking at the GeForce RTX 5080 GPU compute performance.

In beginning the NVIDIA Blackwell Linux testing with the GeForce RTX 5090 compute performance, besides all the CUDA/OpenCL/OptiX benchmarks delivered last week a number of readers asked about AI performance and in particular the Llama.cpp performance with the RTX 5090 flagship graphics card. Here are some initial benchmarks looking at the GeForce RTX 5090 performance in Llama.cpp compared to prior RTX 40 and RTX 30 graphics cards.

While there have been a lot of GeForce RTX 5090 Windows gaming benchmarks since the review embargo lift yesterday, for those more fascinated by this high-end Blackwell desktop graphics card for its GPU compute potential on Linux, this article is for you. Up today are my very initial GPU compute benchmarks for the GeForce RTX 5090 Founders Edition on Linux with NVIDIA graphics card comparisons across the prior RTX 20, RTX, 30, and RTX 40 series too.

Earlier this month at CES was the announcement of the GeForce RTX 50 "Blackwell" series. Among the first of these consumer Blackwell GPUs is the GeForce RTX 5090 flagship graphics card that is set to retail for $1999 USD. The NVIDIA GeForce RTX 5090 Founder's Edition graphics card arrived at Phoronix a few days ago to begin Linux testing.

Last month when Intel formally introduced Battlemage graphics their initial products in the B-Series were the B570 and B580 graphics cards. The B580 went on sale in December and we've been busy testing the B580 on Linux since while today the embargo expires on the Arc B570 with those graphics cards going on sale this morning. Here is a first look at the Intel Arc B570 graphics and compute performance under Linux with their latest open-source drivers.

Yesterday I shared the Linux gaming performance and OpenCL / GPU compute performance for the new Intel Arc B580 Battlemage graphics card. Today the focus is a look at how well the Linux workstation graphics performance for Battlemage is looking relative to the existing Alchemist hardware with the Intel Arc A-Series graphics cards.

Last week Intel announced the Arc B-Series Battlemage graphics cards as their first Xe2 discrete GPUs. Ahead of the Arc B580 graphics card hitting Internet retailers tomorrow, today the review embargo lifts on the Intel Arc B580. Here is what to expect from the Linux driver support at launch for the Intel Battlemage graphics cards and how the Arc B580 is performing for Linux gaming/graphics workloads.

Now that the initial Intel Arc B580 graphics card gaming review on Linux is out, for productivity-minded users you may be more curious about the GPU compute potential... Here are some initial OpenCL and Level Zero results for the Intel Arc B580 graphics card compared to the Intel Arc A-Series and the AMD Radeon and NVIDIA GeForce competition under Linux.

With the recent release of Blender 4.3 for this leading open-source 3D modeling software, I've been carrying out some fresh NVIDIA vs. AMD GPU benchmarks for accelerated rendering across several different popular benchmark scenes.

Succeeding the Intel Arc Graphics discrete graphics cards that launched two years ago as the DG2/Alchemist series, the next-gen Battlemage graphics cards are being announced today. The embargo lifts today on the new Intel Arc B-Series graphics cards with initial availability next week. Like the prior generation Intel graphics and as discussed already in many Phoronix articles, Battlemage is still treated to fully open-source graphics driver support on Linux.

Now that Linux 6.12 has a fix for the Lunar Lake performance with the ASUS Zenbook I have been using for my Core Ultra 200V series Linux testing as well as there recently being an updated Intel Compute Runtime with Lunar Lake fixes, I have been working on some fresh Lunar Lake Xe2 graphics benchmarks using the very latest upstream open-source code. In today's article is exploring how the Xe2 Lunar Lake graphics is performing for OpenCL / GPU compute relative to the prior Meteor Lake Arc Graphics that were already a nice step-up over earlier Intel integrated graphics.

While I have been very eager to test out the Core Ultra 200V Lunar Lake series on Linux in part due to the new Xe2 integrated graphics, after several days of pushing a new Lunar Lake laptop on Linux the results have been very disappointing. Besides needing a very leading-edge software stack to enjoy the Xe2 accelerated graphics out-of-the-box, the performance currently is poor. It's a fraction of the Windows performance and currently falls behind the Meteor Lake graphics performance and in turn also being well behind the AMD RDNA3.5 competition with the Ryzen AI 300 series laptops.

NVIDIA recently sent over their RTX 2000 Ada Generation and RTX 4000 Ada Generation graphics cards suited for designers, engineers, and creative professionals. In my testing the past several weeks these professional graphics cards have been working out with NVIDIA's Linux driver stack -- including their open-source kernel modules now the default with the R555 driver series and later. While there is that previous article looking at how their open-source kernel drivers are at parity to the former proprietary kernel modules, today's article is looking at how the NVIDIA RTX 2000/4000 Ada Generation performance stacks up against the AMD Radeon Pro competition.

While the upcoming AMD Ryzen 9000 series desktop processors continue to make use of RDNA2 graphics, with the Ryzen AI 300 series shipping today in notebooks there are RDNA3.5 graphics being introduced alongside the Zen 5 CPU cores and upgraded Ryzen AI XDNA2 NPU. While just an evolution of RDNA3, the initial benchmarks of RDNA3.5 graphics with the Ryzen AI 9 HX 370 are looking rather promising for both the raw graphics performance as well as the power efficiency. The Radeon 890M RDNA3.5 graphics are working on Linux when using a new enough software stack.