Enlarge / AMD's Radeon RX 7800 XT. (credit: Andrew Cunningham)

Nearly a year ago, Nvidia kicked off this GPU generation with its GeForce RTX 4090 . The 4090 offers unparalleled performance but at an unparalleled price of $1,600 (prices have not fallen). It's not for everybody, but it's a nice halo card that shows what the Ada Lovelace architecture is capable of. Fine, I guess.

The RTX 4080 soon followed, along with AMD's Radeon RX 7900 XTX and XT . These cards also generally offered better performance than anything you could get from a previous-generation GPU, but at still-too-high-for-most-people prices that ranged from between $900 and $1,200 (though all of those prices have fallen by a bit). Fine, I guess.

By the time we got the 4070 Ti launch in May, we were getting down to the level of performance that had been available from previous-generation cards. These GPUs offered a decent generational jump over their predecessors (the 4070 Ti performs kind of like a 3090, and the 4070 performs kind of like a 3080). But those cards also got big price bumps that took them closer to the pricing levels of the last-gen cards they performed like. Fine, I guess.

Enlarge / The RTX 4060 Ti Founders Edition. (credit: Andrew Cunningham)

Last week, AMD announced what are probably the last major GPU launches of this generation of graphics cards: the $449 Radeon RX 7700 XT and $499 Radeon RX 7800 XT . AMD's pricing and performance numbers pit the cards against Nvidia's GeForce RTX 4060 Ti (specifically the $499 16GB version) and the $599 RTX 4070 .

AMD's pricing is aggressive enough that Nvidia is quietly cutting the prices of some 16GB RTX 4060 Ti cards to $449, to match the RX 7700 XT. The announcement about the $50 reduction was buried toward the bottom of an email that Nvidia sent to GPU reviewers ahead of AMD's launch next week; it also drew attention to Nvidia-specific features like DLSS upscaling and frame generation, which compete with AMD's GPU-agnostic FSR , plus recent DLSS improvements that improve ray-tracing performance.

"Finally, as a reminder, market prices can vary from the original launch MSRPs," Nvidia's Brian Burke wrote. "Today, GeForce RTX 4070 is widely available at $599, and GeForce RTX 4060 Ti 16GB is now available at $449. Both of these GPUs are great upgrade choices for gamers seeking their next GPU for the upcoming 2 to 3 years."

Enlarge / The specs of AMD's Radeon RX 7700 XT and 7800 XT. (credit: AMD)

AMD has been slower than Nvidia to fill out its next-generation GPU lineup, and for months there has been a huge gap between the Radeon RX 7900 XT (currently retailing between $750 and $850) and the Radeon RX 7600 (holding steady at $270ish). Today, the company is finally filling in that gap with the new Radeon RX 7800 XT and RX 7700 XT, both advertised as 1440p graphics cards and available starting at $449 and $499, respectively. Both cards will be available on September 6. And most Radeon RX 6000 and RX 7000 GPUs sold between now and September 30 will come with a free copy of Bethesda's upcoming " Skyrim in space " title, Starfield.

AMD kept the prices of both cards under wraps while pre-briefing members of the press about the announcement, which is unusual but not hard to explain. AMD's RX 7600 launch was spoiled a bit by Nvidia, which preempted the 7600's announcement by offering a more powerful GeForce RTX 4060 at the same $299 price that AMD had planned for the 7600. This prompted AMD to cut the 7600's price to $269 before it was even announced; we'll have to wait and see if Nvidia chooses to change its prices in response to the new Radeon cards' launch.

The full lineup of RX 7000-series graphics cards. AMD pictures a reference version of the 7700 XT, though it won't be selling one. (credit: AMD)

The RX 7700 XT and 7800 XT are based on the same RDNA 3 graphics architecture as the other 7000-series GPUs, which means a more efficient manufacturing process than the RX 6000 series, DisplayPort 2.1 support, and hardware acceleration for encoding with the AV1 video codec, which promises game streamers either higher-quality video at the same bitrate as older codecs or the same quality with a lower bitrate. AMD compared the 7800 XT and 7700 XT favorably to Nvidia's $600 upper-midrange RTX 4070 and the $500 16GB version of the RTX 4060 Ti .

Enlarge / AMD's FSR 3 will compete with Nvidia's proprietary DLSS Frame Generation feature starting in September. (credit: AMD)

Even if you're not interested in buying one of the new Radeon graphics cards AMD announced today , the company still has some software-related announcements of interest to anyone who plays games on their PC. And that includes not just owners of older AMD GPUs but people who use Nvidia GeForce or Intel Arc cards, too.

First, AMD is finally ready to reveal more details about FidelityFX Super Resolution version 3, the latest major update to the company's open source upsampling technology. A competitor to Nvidia's proprietary Deep Learning Super Sampling (DLSS) and Intel's GPU-agnostic but nascent XeSS, all of these technologies attempt to generate a high-resolution image by rendering a lower-resolution image, blowing it up and filling in the gaps algorithmically to approximate what a natively rendered image would have looked like.

What GPUs support FSR 3?

Last year, FSR 2.0 went a long way toward making the technology more competitive with DLSS while also working on a wider range of graphics hardware from AMD, Nvidia, and Intel. Contrary to some prior speculation, FSR 3 will continue to support a wide range of old and new GPUs from all three major GPU companies. AMD has confirmed to us that the following graphics hardware should all support FSR 3:

Enlarge / PNY's take on the basic $299 version of the Nvidia GeForce RTX 4060. (credit: Andrew Cunningham)

Nvidia's GeForce 1060, 2060, and 3060 graphics cards are some of the most widely used GPUs in all of PC gaming. Four of Steam's top five GPUs are 60-series cards, and the only one that isn't is an even lower-end GTX 1650.

All of this is to say that, despite all the fanfare for high-end products like the RTX 4090, the new GeForce RTX 4060 is Nvidia's most important Ada Lovelace-based GPU. History suggests that it will become a baseline for game developers to aim for and the go-to recommendation for most entry-level-to-mainstream PC gaming builds.

The RTX 4060, which launches this week starting at $299, is mostly up to the task. It's faster and considerably more power efficient than the 3060 it replaces, and it doesn't come with the same generation-over-generation price hike as the higher-end Lovelace GPUs. It's also a solid value compared to the 4060 Ti, typically delivering between 80 and 90 percent of the 4060 Ti's performance for 75 percent of the money.

Enlarge / Has any game been more associated with proof of concept than SuperTuxKart? It's the "Hello World" of 3D racing. (credit: Asahi Linux)

For the brave people running Linux on Apple Silicon, their patience has paid off. GPU drivers that provide desktop hardware acceleration are now available in Asahi Linux , unleashing more of the M-series chips’ power.

It has taken roughly two years to reach this alpha-stage OpenGL driver, but the foundational groundwork should result in faster progress ahead, writes project leads Alyssa Rosenzweig and Asahi Lina. In the meantime, the drivers are “good enough to run a smooth desktop experience and some games.”

Older games that don't require Vulkan support, like the original Quake trilogy, should work (within alpha-release limits) on Asahi Linux on M-series Macs.

The drivers offer non-conformance-tested OpenGL 2.1 and OpenGL ES 2.0 support for all M-series Apple devices. That’s enough for desktop environments and older games running at 60 frames per second at 4K. But the next target is Vulkan support . OpenGL work is being done “with Vulkan in mind,” Lina writes, but some OpenGL support was needed to get desktops working first. There's a lot more you can read about the interplay between OpenGL, Vulkan, and Zink in Asahi's blog post .

For a while now, Asahi Linux has been making do with software-rendered desktops, but M-series chips are fast enough that they feel almost native (and sometimes faster than other desktops on ARM hardware). And while the Asahi project is relatively new , some core bits of Apple's silicon are backward compatible with known and supported devices, like the original iPhone. And Asahi's work is intended to move upstream, helping other distributions get up and running on Apple's hardware.

The team of developers includes three core members—Rosenzweig, Lina, and Dougall Johnson—plus Ella Stanforth, who works on Vulkan drivers and future reuse. The developers note that their work stands "on the shoulders of FOSS giants." That includes the NIR backend, the Direct Rendering Manager in the Linux kernel, and the Gallium3D API inside the open source Mesa drivers, which themselves build on 30 years of OpenGL work.

Installing the new drivers requires running a bleeding-edge kernel, Mesa drivers, and a Wayland-based desktop. The team welcomes bug reports, but not of the "this specific app isn't working" variety. Their blog post details how and where to submit reports about certain kinds of GPU-specific issues.

Enlarge / The Cerebras Andromeda, a 13.5 million core AI supercomputer. (credit: Cerebras )

On Monday, Cerebras Systems unveiled its 13.5 million core Andromeda AI supercomputer for deep learning, reports Reuters. According Cerebras, Andromeda delivers over one 1 exaflop (1 quintillion operations per second) of AI computational power at 16-bit half precision.

The Andromeda is itself a cluster of 16 Cerebras C-2 computers linked together. Each CS-2 contains one Wafer Scale Engine chip (often called "WSE-2"), which is currently the largest silicon chip ever made, at about 8.5-inches square and packed with 2.6 trillion transistors organized into 850,000 cores.

Cerebras built Andromeda at a data center in Santa Clara, California, for $35 million. It's tuned for applications like large language models and has already been in use for academic and commercial work. "Andromeda delivers near-perfect scaling via simple data parallelism across GPT-class large language models, including GPT-3, GPT-J and GPT-NeoX," writes Cerebras in a press release.

Enlarge / The Nvidia RTX 4090 founders edition. If you can't tell, those lines are drawn on, though the heft of this $1,599 product might convince you that they're a reflection of real-world motion blur upon opening this massive box. (credit: Sam Machkovech)

The Nvidia RTX 4090 makes me laugh.

Part of that is due to its size. When a standalone GPU is as large as a modern video gaming console—it's nearly identical in total volume to the Xbox Series S and more than double the size of a Nintendo Switch—it's hard not to laugh incredulously at the thing. None of Nvidia's highest-end "reference" GPUs, previously branded as "Titan" models, have ever been so massive, and things only get more ludicrous when you move beyond Nvidia's "Founders Edition" and check out AIB options from third-party partners. (We haven't tested any models other than the 4090 FE yet.)

• 

  A gallery of size comparisons. 4090 vs. Xbox Series X gamepad. [credit: Sam Machkovech ]

After figuring out how to safely mount and run power to the RTX 4090, however, the laughs become decidedly different. You're going to consistently laugh with , not at , the RTX 4090, either in joy or excited disbelief.