PC hardware, technology, reviews and analysis from Steve and Tim
Melbourne, Australia
Joined April 2015
- Tweets17.5K
- Following74
- Followers773K
- Likes15.9K
Radeon 6000 series owners, in your opinion did AMD take too long to support FSR 4.1 on your RDNA 2 GPU?
80%Yes
20%No
1,576 votes • Final resultsTim Sweeney getting upset at Steam's AI disclosure is an endless stream of funny content.
He claims it's bad for developers, but what he actually means is it's bad for Epic Games. Because Epic are literally building UE6 around generative AI, which they say will play a "central role"
If generative AI is good, and developers are proud to use it, there shouldn't be an issue disclosing it. But the problem is gamers don't agree, and some actively avoid games using AI. This in turn will mean avoiding UE6 games using their fancy gen AI integrations, which hurts Epic's business model.
It's another one of those things where companies are not on the same page as gamers. Gamers want engine developers to solve problems they are having (eg poor performance, traversal stutter, visual artefacts). They don't want those problems to remain unsolved while Epic makes it easier to fill storefronts with low quality AI-made games.
Transparency is always good for consumers and shouldn't be sacrificed. Ever. Why hide the use of a feature if you think the feature is good?
If it's a "hate mob" issue, maybe time to reassess the messaging. Prove that AI use is good, instead of hiding it. Because if it's actually good, it should be easy to explain and show why. The work speaks for itself with these things.
If Epic are ever wondering why gamers love Steam and avoided the Epic Games Store for so long, it's because Valve continue to deliver what gamers want. Being hostile to gamers doesn't work.
Based on our previous results, if Nvidia halves the runtime for DLSS 5, it would reduce the average performance hit from 55% to 38% when enabling it
This is based off a 120 FPS frame rate before turning on DLSS 5, a 1440p resolution, and an average of results across all RTX 50 models. It's about what you'd see on an RTX 5070, slightly lower hit on higher models, larger hit on slower models.
Look at how much they've toned down DLSS 5 on the faces in Hogwarts Legacy for the final release. Barely does anything now to faces
Original reveal: youtube.com/watch?v=t50AdVMu…
Final game: nvidia.com/en-us/geforce/com…
So no effective change. 5090s have been priced like RTX PRO cards for months
NVIDIA reportedly discontinuing RTX 5090, GB202 GPUs to be reserved for RTX PRO series
🔗videocardz.com/newz/nvidia-r…
Hardware Unboxed retweeted
We're still burning in our OLED monitor, here's the results after 8000 hours youtube.com/watch?v=ERrp0TjP…
Hardware Unboxed retweeted
It's funny that a video like this had to be made in the big 2026, but I'm glad Steve did as there was a lot of "selective information" being shared around this topic, and some I would even label as disinformation.
youtube.com/watch?v=1sC8eVVL…
Hardware Unboxed retweeted
Nothing has changed here since we first covered it 5 years ago. There’s also no point talking about the RTX 5090, no one is pairing what was meant to be a $2000 US graphics card with anything but a high-end CPU.
5 years ago when I investigated this issue it was clear that it really only affected those with low-end CPUs (mostly old low-end CPUs) who were upgrading to a modern mid-range GPU.
This is because the CPU overhead issue is only an issue when the CPU is maxed out, you’re not going to max out a Core i3-13700K in the vast majority of games with a GeForce GPU. The 5800X3D and 7900X will also be fast enough to avoid the issue for the vast majority of games/configurations.
In reality this is a low-end CPU concern, so real-world configurations would be looking more at 5060 class GPUs, maybe the 5070, but certainly not the 5080 and 5090.
Sure you can create scenarios with a 9800X3D where the 9070 XT can beat the RTX 5090, but those are rare and performance of both configurations is in the hundreds of frames per second range.
I just came across a channel called "TrashBench" and it's very cool, the videos are fun, I really enjoyed this one where he strapped a CPU AIO onto his GPU... just coz youtube.com/watch?v=YyjrMwQz…
This was also great for a 5 min video! youtube.com/watch?v=wW-RkIIt…
The worst thing to come out of this whole 12VHPWR cable disaster are the guys blaming people for not monitoring their power cable, like that's some kind of embarrassing skill issue, or in this case 'unprofessional'.
These people need to get a grip, you've lost the plot when blaming people for their shitty 12VHPWR cable melting.
That said if you'd like to learn more about why we weren't using certain monitoring tools all is explained in our latest podcast, I've timestamped it at 11:54 for you: youtu.be/yrPHWhtZNF0?si=dy5s…
Techwando has now added links to the videos he discussed, he had planned on doing it, just a slip up which is fair enough. I didn’t mean to make the “clickbait” comment seem like an attack (if I did), and Techwando seems like a genuinely good guy, so to avoid any headaches for him I have deleted the original post. But I will still link to the video here so you guys can check it out.
I got a lot of good feedback from the original post and it sounds like the vast majority of you agree, not clickbait, so that’s good :) youtube.com/watch?v=c8udgihj…
Driver notes by the way, you can't make this stuff up: amd.com/en/resources/support…
A couple of people suggested that FSR FG frame pacing is fixed if you force Vsync on. This is not true, frame pacing is broken regardless of Vsync settings. Here are some examples of FSR FG with Vsync forced on in the driver (and in the case of Crimson Desert, also enabled in the game itself) youtube.com/watch?v=rXk67jHw…
