Afrika zal er in de toekomst veel groter uitzien op kaarten

The way our hobby talks about player counts is all sorts of messed up. Steam concurrents are a useful metric for gauging a game's general popularity, but they paint an incomplete picture. They're a single data point, yet have become the flimsy scaffolding on which the internet builds elaborate headcanon about a game's quality, viability, or deadness. Sometimes dogpiling is so successful, and the resulting perception so difficult to fight, that simply saying a game is dead is enough to help kill it.

It's a trend that annoys me to no end, so it's refreshing to see a developer speaking openly about concurrents—both about how many it considers "healthy," and why it's normal for the number to rise and fall. It's been a recurring topic for Wardogs developer Bulkhead, whose closed beta surpassed 200,000 concurrents today on Steam. Responding to a TimTheTatman clip on X, CEO Joe Brammer spoke on the milsim's unexpected popularity, and its inevitable (but healthy) decline.

"Wardogs will continue to lose players for months as the hype dies down," Brammer wrote. "Doesn't make it a bad game or even a problem, we planned for 5k players concurrent! It'll maintain a healthy playerbase and we'll continue to build the 1.0 version of the game then... it'll boom again :) And so the cycle continues :)"

It's a cycle many of us are familiar with, but rarely acknowledge. I tend to build a rotation of multiplayer favorites that I regularly return to, either to check out an update or scratch an itch. It is, I think, a pretty healthy way to enjoy videogames, but the rise of live service "forever" games, as well as high-profile failures, have proliferated the idea that videogames are those species of shark that will die if they ever stop swimming.

NachrichtenVon der Hitze in die Halle: Was beim Trinken im Tennis wirklich zähltVon der Hitze bei den US Open bis zur europäischen Hallensaison: Die Bedingungen ändern sich, die Bedeutung der richtigen Flüssigkeitsversorgung bleibt. Was der Körper über den ...Öffnen →

Much like the majority of videogames before 2016, Bulkhead does not need Wardogs to be a consistent, top-charting megahit to continue making it. In a video published last month outlining "10 reasons not to buy Wardogs," game director Howard Philpott expanded on Bulkhead's expectations:

"We expect our initial days after launch to have a lot more players than expected, but we can tell you for a fact that this is completely normal. Those players will fall away over the next coming months. Player numbers will drop off, and journalists and influencers on the fringe of the community will try to ragebait your enjoyment out of the game for social media clicks. This doesn't mean the game is dying. We're thinking longer term than a flash in the plan. Our goal is to settle into a sustainable, close-knit community. If we're pulling in 3,000 to 5,000 daily concurrent players who genuinely love the game, it'll allow new players to buy in anytime and join an active community."

If we put our hands together on this one, we can all be normal about a multiplayer shooter. Wardogs is out September 10, but pre-ordering grants access to the closed beta ending Sunday.

Another day, another DLSS 5 story that actually kinda doesn't involve Nvidia itself doing anything, because it seems now AMD GPU owners can get in on the act. A new homebrew app has been released on Github that will allow you to run the DLSS Neural Rendering feature on a Radeon GPU. I've tried it and it works, but damn does it hurt the poor Radeon silicon.

From having DLSS 5 .dll files plucked from an early access version of NBA 2K27, we've had modders working hard to create surprisingly polished ways to essentially get the tech working in practically any game, even if it doesn't natively support traditional DLSS itself.

I've been playing around with it a lot this past week, even to the point of jamming it into Football Manager 26 just to see what would happen. And, y'know what, I kinda like the result so I'm keeping it. I wasn't expecting that, and I also wasn't expecting this technology, created by Nvidia, to appear so quickly on AMD GPUs. Historically we've found that where Nvidia goes, so AMD follows, at least in the GPU feature space, but it seems like AMD might not actually have to do much this time. Well, apart from actually make the silicon a little more tensor-friendly.

Because I have spent the afternoon trying to get the new app from user danielblnc on Github to work on my Radeon RX 9070 XT in Cyberpunk 2077. Essentially you download the DLSS-NR-on-AMD file, drop it into the relevant folder (the bin/x64 folder in the Cyberpunk 2077 directory in Steam, for example) and run it. So long as you have the necessary nvngx_dlssnr.dll file that's been doing the rounds installed in the same place it ought to work.

It didn't.

Initially Windows flagged it as containing a virus, but I did some checking around with VirusTotal and seemingly the Defender flag is a machine learning false positive. Probably. But even then I still struggled to have it running happily.

I am not entirely sure what I did wrong, but when I enabled the feature in-game all I got was a black output where the 3D world ought to be, even though the HUD and game kept on running. After some file deletion, redownloading, verifying of integrity (mine and Cyberpunk's), and a rollback of my AMD Adrenalin drivers, I finally got it working. Kinda. I mean it's still kinda crashy, and I can only now launch Cyberpunk 2077 directly from the .exe file rather than from Steam, but it works.

But my poor GPU wished I hadn't. Now, we know that DLSS 5 is a hell of a resource hog on an RTX 50-series GPU that it's designed for, with a touted 50 - 60% performance hit officially noted by Nvidia, so I should have known it would trouble the AMD GPU.

DLSS NR on Nvidia

DLSS 5 was designed for GeForce GPUs, so is a little more capable, but still a massive resource hog.

DLSS NR Off

DLSS NR On

Still… running at 4K RT Ultra settings is tough enough, and I was getting just 61 fps on the RX 9070 XT even with FSR 4 set to Balanced. And then flicking the DLSS Neural Rendering toggle to 'on' made that figure drop to just 9 fps.

And I thought the 73% drop Andy found with NBA 2K27 was rough.

DLSS NR on AMD

AMD GPUs running DLSS 5 looks great, but runs horribly.

NachrichtenHuss Valley codes (October 2026) and how to redeem them for free Gems, Boosts, and moreCodes for Huss Valley will give you Gems, Boosts, and cosmetics while you do your best to run away from or catch your opponentsÖffnen →

DLSS NR Off

DLSS NR On

Now it's worth noting that the fact it is running at all on AMD GPUs is impressive on its own, but it's not going to make any Radeon graphics card owners any happier about it when they're getting unplayable frame rates out of their new ill-gotten pretties. An 85% performance drop is not sustainable. Nice for screenshots though, eh?

The issue is that AMD's RDNA 4 doesn't have the level of dedicated silicon for the sort of matrix calculations the feature is running… in the same way that RTX 20-series chips don't.

The Turing architecture does have Tensor Cores, but they're far weaker compared to those in the RTX 40- or RTX 50-series, and the shared resources AMD plucks from the standard RDNA 4 Compute Units for these sorts of calculations aren't as robust either. Hence the weaker performance.

Will they be able to squeeze more performance out of this feature on AMD chips? I don't know, but YouTuber Ancient Gameplays noted the app in a recent video, and states that performance has gone up 12% since the first version of the DLSS-NR-on-AMD software first hit Github, so maybe there's a chance. But, realistically, I think there's a hard hardware limit in there which is going to stop DLSS Neural Rendering from being a realistic part of AMD's armoury in this generation.

It's nearly impossible to nail down any one feature or mechanic that defines a Grand Theft Auto game, but the two specific suggestions from the title are thievery and driving. My memories of the early GTA games from my childhood, Vice City and San Andreas, are largely comprised of stealing stuff, cruising around listening to 2Pac, and mowing down innocent citizens, a horde of cop cars blaring sirens in the rearview. Driving is essential to a GTA game, so it makes sense that it's a huge focus in the community leading up to GTA 6.

What does it actually feel like in the hands? Does it have the bulk and heft of vehicular traversal in GTA 4? I remember that one being a little clunky to handle, something that was dramatically improved in GTA 5, which I've been playing a lot in anticipation of GTA 6. According to Rockstar co-studio head Rob Nelson, driving in GTA 6 is something of a hybrid of the two, combining the best of each.

Asked by TGG if GTA 6's cars are as "weighty" as GTA 4's, Nelson says:

"Only in the best way possible. The team has worked hard to maintain some of the best parts of the weightiness and feel of GTA 4, but have the accessibility of GTA 5. It has all been built up from scratch on this hardware, and they have much more control tuning drivetrain, tires, suspension, steering, all of that is getting applied to the physics model."

As no one outside of Rockstar and publisher Take-Two have had the chance to go hands-on with GTA 6, we don't have objective, unbiased impressions of any of the game's mechanics. Of course the guy in charge of Rockstar is going to say it's the best of both worlds; we'll just have to wait until November to find out if he's right.

GTA 6 launches November 19 for PS5 and Xbox Series X.

GTA 6 isn't flashy for the sake of it – I'm impressed at how additive every new detail feels to gameplay