At this year's Gamescom event, two forthcoming games heavily feature ray tracing as the means for producing the best possible graphics. Star Wars Outlaws and Indiana Jones and the Great Circle both have graphics features, either exclusively developed by Nvidia or are so demanding that only high-end GeForce RTX cards can really handle it.

Nvidia was the first GPU vendor to bring real-time ray tracing hardware to the gaming masses at another Gamescom in 2018, with its GeForce RTX 20-series graphics cards. But since then, AMD and Intel have both followed suit, and all new gaming PCs and consoles are more or less capable of ray tracing. Most PC gamers are well aware that Nvidia's GPUs can do ray tracing faster than the competition but three games show that Nvidia is doing its best to push the technology to new heights—and very much in its favour.

Take the recently launched Black Myth: Wukong. Its graphics are spectacular but to get the very best visuals, you need to enable a setting called Full Ray Tracing. The game runs on Unreal Engine 5 and it uses Lumen ray-traced global illumination by default. However, 'full ray tracing' is a path tracing algorithm.

If you're a hardware enthusiast, you'll already know about path tracing but even if you don't, you may well have already come across it in Cyberpunk 2077 or more recently, a mod for Doom 2. Path tracing is actually a whole bunch of different algorithms but the general idea is that it's a means to let developers push the amount of ray tracing taking place, without utterly tanking the performance. It's still very demanding but the results are super impressive.

At Gamescom 2024, it was announced that the forthcoming Indiana Jones and the Great Circle game will also have path tracing—sorry, Full Ray Tracing. Now, there's no indication whatsoever that the option will only work on GeForce RTX cards but given how demanding it is, you can be pretty sure that just the likes of the RTX 4080 Super or RTX 4090 will be able to run at a decent frame rate.

Mind you, Black Myth: Wukong runs quite well with path—darn it—Full Ray Tracing enabled on mid-range RTX graphics cards, as long as you keep the resolution down and use lots of upscaling and frame generation. In theory, AMD and Intel GPUs should run it too but my own testing showed that the performance hit is far too big to make it worth using.

And then there's Star Wars Outlaws, which is due for release August 30. That doesn't have Full Path Tracing (I'm going to call it FRT from now on) but it does feature an Nvidia technology: RTX Direct Illumination or RTXDI, for short. 

Now, any GPU that supports DirectX Ray Tracing 1.0 (aka DXR) or better can run this algorithm but, while I'm not allowed to give any specifics, I'll give you just one guess as to which GPU tech runs it the best, in terms of raw speed and lack of bugs.

I personally spend quite a bit of my free time messing about with DXR coding and using Lumen in Unreal Engine 5, so it's clear that I'm somewhat biased when it comes to having an opinion on the use of ray tracing in games. But if more and more games start to feature FRT as being the only way to get the best possible graphics or utilise more of Nvidia's APIs for ray tracing, I am a little worried that the whole thing is slowly becoming a one horse race.

Not that ray tracing is a race, of course, but let's be honest here—the technology is going to be used increasingly more in games, especially the big-budget ones, and consoles are only going to become more capable in this aspect.

The latter is the best counterpoint to Nvidia's dominance in the ray tracing industry as consoles are powered by AMD chips and developers that make games exclusively for that platform only need to think about how RDNA RT units handle it all.

But in the world of PC gaming, it's pretty much RTX all the way, be it hardware or software, right now. That's great news if you're an Nvidia shareholder, less so if you have an Arc or Radeon graphics card.

With regards to AMD's desktop PC graphics cards, it's expected that RNDA 4 will have improved ray tracing units but at the same time, it's looking unlikely that there will be any high-end models sporting these chips.

As for Intel's next-gen Battlemage GPUs—well, with the senior management hell-bent on saving big chunks of money, there's a chance that the hardware is fine but the software support won't be as good.

Don't get me wrong, Nvidia has made some pretty amazing stuff for PC gaming, be it hardware or comprehensive software kits. But I rather hope that hope that game developers just use Black Myth: Wukong as inspiration for what modern graphics can look like, rather than being the best way to implement ray tracing.

Another day, another delay. This time it looks like it's Nvidia's Blackwell AI GPUs that may have fallen victim to last minute delay-itus, as reports now suggest they'll be launching in the first half of 2025 instead.

While engineering samples for the B100 and B200 GPUs have already been delivered, it seems enterprise customers including Microsoft, Meta and xAI might have to wait for their orders. According to Tom's Hardware, two anonymous sources who worked on the chips first released news of the pushback, which was later allegedly corroborated with an anonymous Microsoft source by Bloomberg.

Previously, it was thought that Nvidia was aiming for an end of 2024 release. With multiple high-profile customers placing huge orders for Nvidia's latest mega-GPUs, and prices estimated to run at $70,000 per "superchip", significant delays are not likely to be well received.

While these sources have pointed at design flaws as a reason for the pushback, Dylan Patel, Chief Analyst at independent research company SemiAnalysis, has posted a tweet telling a different story. 

According to an update sent out to SemiAnalysis' clients on July 22, CoWoS-L (Chip on Wafer on Substrate) ramping issues are primarily to blame, with TSMC's AP3 packaging fab needing to be shut down to convert from CoWoS-S production to CoWoS-L.

This process is not expected to be completed until the second quarter of next year. The tweet also mentions rumours about overheating and firmware issues, but characterises them as "overblown". While the update mentions a variety of "teasing pains to be fixed", SemiAnalysis seems to think that they are not the primary cause of the delays.

Whatever the cause, any significant delay in shipping could potentially shake confidence in Nvidia's ability to deliver its costly AI-processing GPUs en masse and on time.

That being said, Nvidia are unlikely to be worried that any significant delay will result in lost customers. While AMD has its own competing AI GPU, the Instinct MI300, AMD CEO Dr. Lisa Su has recently indicated that it's currently selling as many as it can make, and that "the overall supply chain is tight and will remain tight through 2025."

Given Nvidia's dominance in the sector and the reported potential performance of the Blackwell series AI GPUs, it's unlikely any major customer will switch over to other hardware providers in the face of a potential delay, especially if the top candidates are in the same short supply boat.

Once again it seems, TSMC holds the key to the world's chipmaking capabilities, and if it needs to reconfigure to keep up with demand, the rest of the world's top tech companies may just have to stand in line and wait. Or perhaps there really are design issues with Nvidia's latest AI hardware holding things back.

Still, at least it doesn't seem to be a mistake at the printers, ey?

Is three years too early to discontinue one of the most popular GPUs in recent times? According to a report, Nvidia is getting ready to distribute the final batches of its GA106 graphics processor, the chip that powers the GeForce RTX 3060, and if that's the case, it would mean the jolly green giant has officially closed the doors on the GPU that's been top of the Steam hardware charts for ages.

Launched in early 2021, the GeForce RTX 3060 stood out for all kinds of reasons. At the time, global supplies of graphics cards were extremely poor, thanks to the effects of the global pandemic and cryptominers snapping every GPU around the globe.

With a launch MSRP of $329, the RTX 3060 was the cheapest Ampere-powered model at that point, even though it was loaded with 12 GB of VRAM. However, it wasn't an instant success as its performance wasn't brilliant enough, especially when compared to the RTX 3060 Ti.

But hey, it was a GPU and at that time, it was more than enough. Over the following years, though, the RTX 3060 rapidly gained in popularity and it's been topping the Steam hardware survey charts for a long time.

But all good things must come to an end and a report by Board Channels (via Videocardz) states that Nvidia has stated the model is now officially discontinued, and that the last few batches of GA106 GPUs will be it.

Its successor, the RTX 4060, has got a long way to go before it becomes the most popular GPU with Steam users, if it ever even does. That's because the aforementioned RTX 3060 Ti, the RTX 3070, and the Turing-powered GTX 1650 are all considerably more popular than the RTX 4060.

While its launch MSRP of $299 was less than the RTX 3060's, the fact that it only has 8 GB of VRAM didn't go down well with many reviewers and PC gamers.

At face value, the RTX 4060 is a really good 1080p graphics card, especially when you factor in the full DLSS 3.5 support and the superb NVENC video encoders. It's just that it feels far more like an RTX '4050' rather than 60-class GPU and had Nvidia marketed it as such, and priced it accordingly, the reception would have been far better.

Imagine if the original RTX 4070 had been sold as an RTX 4060—with a price tag of $329, it would have sold out instantly.

Fortunately, the RTX 3060 will actually be around a little while longer, as it's still in stock with plenty of retailers, although its continued popularity means prices are somewhat higher than they perhaps should be. They'll probably take a big dip once Nvidia launches the RTX 5060 but that won't happen until January 2025 at the very earliest.

If you're looking for a really good GPU that's under $300, you're far better off getting a Radeon RX 6750 XT. Over-priced at launch, they're much cheaper now and at 1080p, it'll give the RTX 4060 Ti a good run for its money, as long as you don't enable ray tracing in games.

Pour one out for EVGA graphics cards. Back in September 2022, the company announced it would make no more, bringing to a close a run of rather performant and well-cooled GPUs, many of which were capable of some serious overclocking.

One of the figures behind some of those super-tweakable designs was Vince "Kingpin" Lucido, an extreme overclocker known for his all-black design ethos and his work with EVGA (via Videocardz). After EVGA's exit, however, Lucido did seem like he was open to offers from other vendors. 

Now that period is over, after exploring collaborations with some other big-name companies he's now working with PNY on some of its latest designs.

Gamers Nexus' Steve Burke got the chance to take a tour of the lab Lucido calls "Kingpin studios" in Taipei, a futuristic building that looks like a gigantic workshop of PC gaming dreams. In amongst his impressive array of hardware (and some fancy-looking electric bikes), Lucido explains why he chose to work with PNY over others.

"They're keen to dive into extreme overclocking. Asus, MSI, Galax, the other companies, they already do it. Too many cooks in the kitchen, right?"

"There's a huge hole right now in the enthusiast market from EVGA being gone, and it doesn't really seem like the other vendors are too keen to fill it"

That's not too surprising, I suppose. While overclocking used to be practically mainstream, it's been a while since we've seen an OC'ed card that delivers much more than a couple of percent performance improvement over a standard model. 

What with GPUs being able to effectively boost themselves to high-levels with relatively standard coolers, performance gains in recent years don't really seem to make the extra expense of an overclocked card worth it for most.

If anyone's capable of putting together a card with a real performance advantage though, it's probably Lucido. And given some predicted upcoming GPU launches (I'm looking at you, Nvidia), perhaps he may have had the chance to get his hands on something next-generation?

"I heard, y'know, the vendors...maybe this year, there's a new card coming" Lucido says, coyly.

"I spoke with several of the board partners, and they're all like, 5080, end of year" says Steve. "Definitely", Lucido replies, with a nod.

Definitely, that's what they're saying, or definitely, as in "that's correct", who's to say. Anyway, whether it's working on the upcoming GPUs or merely tinkering with the current models (and I know where my money would be), Lucido certainly seems busy testing out new GPU overclocking software. Although as he points out, in his signature style, the first thing he did with his PNY RTX 4090 was turn off the RGB.

Matte black and highly overclocked GPUs, cooked up in an honest-to-goodness secret laboratory? Living the dream right there. A little RGB is fine though, right? Me, I like a little dash of red. Horses for courses, I suppose.

Many PC enthusiasts don't like the fact that the best top-end GPUs use lots of power and try all kinds of ways of reducing the consumption, such as undervolting, frame rate caps, or lowering the max power limit. Graphics card vendor PowerColor is experimenting with a slightly different approach by using an NPU to manage the power usage in games, without impacting performance, in a system called Edge AI.

A demonstration of the work in progress was on display at PowerColor's Computex stand. While we didn't get a chance to see it in action (there was a huge amount to try and see at the event), tech site IT Home and X user Harukaze5719 managed to grab some pictures of the setup and see it running on two computers running Final Fantasy XV.

PowerColor's engineers hotwired an external NPU to an AMD graphics card, in one of the computers, and programmed it to manage the GPU's power consumption while rendering. At the moment, there's no indication of exactly what's going on behind the scenes but NPUs (neural processing units) are specialised processors for handling the math operations involved in AI routines.

What I suspect is going on is that the NPU is running a neural network that takes metrics such as the GPU's load, voltage, and temperature, as well as aspects from the game being rendered, and alters the GP voltage in such a way that the power consumption is significantly reduced on average.

In the Final Fantasy XV demonstration, the PC without an NPU ran the game at 118 fps, with the graphics card using 338 W to achieve this. The other setup, with the NPU-GPU combo, was hitting 107 fps at 261 W of power. That's a 23% reduction in energy consumption for a 9% drop in frame rate.

PowerColor's demo stand actually claims that Edge AI improves performance but it obviously forgot that if you're going to showcase a new bit of technology, then you kind of want to check that it does what you're saying it does before you wave it about in public. But even with that minor marketing boo-boo, the whole concept of Edge AI looks like it has quite a bit of potential.

Reducing the power consumption of a graphics card has multiple benefits—less heat is dissipated into your gaming room, the whole PC uses less electricity, and the peripheral components on the graphics card will last longer. All that seems worth the relatively small drop in performance.

At the moment, Edge AI requires an external NPU to be wired to various points on a graphics card but if it's only monitoring voltages and temperatures, then an internal NPU could be used to do the same thing. The majority of these are embedded in laptop-based chips, such as AMD's new Ryzen AI series, Intel's Core Ultra range, and Qualcomm's Snapdragon X—processors that are rarely going to be used with a discrete graphics card.

The neural network that Edge AI runs could, in theory, be done on a GPU but they're not really designed to do such things on as little power as possible, unlike NPUs.

That 77 W decrease in GPU power seen in the Final Fantasy demonstration would probably be far smaller if the routines were GPU-accelerated instead (and the fps drop would likely be larger, too).

I don't think PowerColor is planning on releasing a graphics card with an NPU on the circuit board, as that would eat into the profit margins. Instead, I suspect it's preparing Edge AI to be ready for when NPUs are routinely embedded in desktop CPUs from every vendor and if that's the case, then it's one of the few uses of AI that I'd genuinely look forward to seeing in action.

And if it's a success for PowerColor, you can bet your last dollar that every other graphics card manufacturer will want to replicate. As long as all these systems are optional to use, then that would be a positive step forward for the GPU industry as a whole.

We expect Nvidia's next-generation flagship graphics card to be a powerful beast. In terms of pixel-pushing power, it surely will be. But aside from being fast, will it end up as the big, hot, power guzzling behemoth that we're expecting?

Perhaps not, according to reliable leakster and Kimi Raikkonen superfan Kopite7kimi. They suggest that Nvidia's hypothetically named RTX 5090 could actually end up with a dual -slot, twin-fan cooler. That would fly in the face of current thinking, when even 250-300W cards are commonly equipped with very large triple fan coolers that can take up over three slots. 

Now it may be that Nvidia is able to extract a lot of performance from the Blackwell silicon without having to push 450 watts through it. That's the rated power consumption of the RTX 4090, which features a triple slot cooler.

But, if Nvidia could shrink the RTX 5090 cooler down to two slots, I can't imagine it being a 450W card. If so, that opens up new lines of speculation. Would it mean the RTX 5090 will not use the high end GB202 chip? Or if it does, are an unusually high number of shader units disabled?

It's all speculation for now, but I would contend that the rumor of a dual slot cooler could simply be an unintentional mis-statement. Manufacturers are often generous when reporting a card's width. A dual slot card could be 2.5 or 2.7 slots in width, even though it takes up three in reality, as the extra fraction means an adjacent PCIe slot cannot be used.

We know Nvidia has tested a very large quad-slot cooler with a perpendicular mounting. It looks like the kind of cooler that could handle a ludicrous 600W TDP. 

Whatever form Nvidia's cooler takes, partner cards are certain to push the envelope. There are several quad slot cards on the market already, and Asus even has a five-slot RTX 4080 complete with Noctua fans. We can expect more of the same when it comes to next generation behemoths.

From what we can gather, the launch of Nvidia's next-gen Blackwell-series graphics cards is not imminent. There are plenty of rumors, but with many months remaining, take any rumors with a grain of salt for now.

What is certain is that the RTX 5090 will be a fast card. But how fast? We look forward to finding out.

Sony's PS4 classic, Ghost of Tsushima, has finally made its way to the humble PC. In terms of hardware support, porting wizards Nixxes has tried its best to incorporate all the latest GPU features, such as upscaling, frame generation, and ultra-wide monitor support. But there's one thing that Sony hasn't checked properly: does the game run properly on Intel Ark GPUs?

Spoiler: nope.

A lack of early code means I've been working on testing the performance of Ghost of Tsushima across a range of CPUs and GPUs once the game released this week. So far things have been going very smoothly. The game has a wealth of options to help it run well on a broad range of PC gaming hardware.

For the most part, both AMD and Nvidia graphics cards have worked well in the game, although it must be said one particular model has been prone to crashing—you'll have to wait for the final report to find out which one, though! Exciting teaser, eh?

However, my testing of Intel's Arc cards has delivered results and scenes which are... well... odd, to say the least. No freezes, total PC crashes or anything like that, and even the performance has been sort of okay. Not really, it's actually been rubbish (it's really not good), but the most entertaining bit has been how the Alchemist GPU has coped with the game's rendering duties.

Harking back to the golden age of graphics cards, the Arc A770 I've been testing on Intel's latest drivers has taken the ghost part in the game's name quite literally. Okay, missing upper arms aren't exactly Exorcist levels of fright but it's nothing compared to the cloak and arm clothing—these follow you around, suspended mid-air, like some kind of poltergeist.

Nixxes other recent PC port, Horizon Forbidden West, didn't have any such issues with the Arc A770. While I do not doubt that Intel will eventually release a set of drivers that fix these bugs (or Nixxes will, via a game patch), the one thing that this all shows is that Arc cards weren't tested properly for the port of Ghost of Tsushima. And that's despite the game fully supporting Intel's XeSS upscaler.

Now you might think I'm being unduly harsh on Sony/Nixxes or Intel here, but I haven't seen this kind of bug in a PC game, backed by a huge publisher and from experienced developers, for a very long time. At least not since another Sony port from last year, anyway. Joel's eyebrows, anyone? Aside from those caterpillars, these days, it's normally performance issues or outright game-ending crashes.

Displaced meshes and non-rendered parts are hallmarks of an alpha or beta game release and certainly shouldn't be there in a full version, even accounting for wonky drivers. I've tried multiple graphics options to fix the rendering bugs, but none of them worked. Even rolling back to an older driver didn't help.

At least I can play Ghost of Alchemist Tsushima, on an A770—upper arms and spooky cloak action notwithstanding. Now if I can get a certain popular and powerful chiplet-y graphics card to run the game longer than three seconds before crashing to the desktop, I'll be a happy chap.