AMD's FidelityFX Super Resolution leverages standard upscaling techniques and appears to offer worse image quality than DLSS 2.0

[Redaktion]

Screen captures from Godfall - one of the first games to support FSR - suggest that image quality with AMD's new upscaling technique might not be substantially better than conventional resolution scaling, while offering worse performance, bringing back memories of DLSS 1.0

https://www.notebookcheck.net/AMD-s-FidelityFX-Super-Resolution-leverages-standard-upscaling-techniques-and-appears-to-offer-worse-image-quality-than-DLSS-2-0.543263.0.html

[TruthIsThere]

This and RT coming to Sammy's/AMD's mobile SoCs just shows how really clueless and how stupid Su really is and she's only CEO of AMD due to affirmative action and publicity.

[Rinnnnnn]

This and RT coming to Sammy's/AMD's mobile SoCs just shows how really clueless and how stupid Su really is and she's only CEO of AMD due to affirmative action and publicity.

I would have agree with you if AMD was losing sales, but the results are positive. It kept on increasing. In addition, AMD was nearing the brink of bankruptcy several years ago, so she is doing something right.

Meanwhile, Intel's keynote was absolutely laughable, where they talked a whole lot of...nothing. Almost as if looking at soulless NPCs talking during their presentation xD

[RinzImpulse]

This and RT coming to Sammy's/AMD's mobile SoCs just shows how really clueless and how stupid Su really is and she's only CEO of AMD due to affirmative action and publicity.

You pascal user will get this thing for free without changing GPU, LMAO

[TruthIsThere]

This and RT coming to Sammy's/AMD's mobile SoCs just shows how really clueless and how stupid Su really is and she's only CEO of AMD due to affirmative action and publicity.

I would have agree with you if AMD was losing sales, but the results are positive. It kept on increasing. In addition, AMD was nearing the brink of bankruptcy several years ago, so she is doing something right.

Meanwhile, Intel's keynote was absolutely laughable, where they talked a whole lot of...nothing. Almost as if looking at soulless NPCs talking during their presentation xD

AMD was at rock bottom before butch Su entry at AND. The only direction from the floor is up, you know?!

So, anyone knowledgeable executive with a decent background would've turned AMD around from its disaster BullDozer shenanigans.

What have she innovated?

AMD new about the competitors DLSS and RT prowess since its release 2018 (more likely a lot sooner due to industry intel) and how did she answer to that a minmum of two-years later, by not having their own HARDWARE solution or anything when AMD released LITTLE Navi! 

[Bill]

AMD was at rock bottom before butch Su entry at AND. The only direction from the floor is up, you know?!

So, anyone knowledgeable executive with a decent background would've turned AMD around from its disaster BullDozer shenanigans. 

No, rock bottom, and by far the most likely option, was bankruptcy.

1.) AMD had a tiny fraction of the R&D budget of Intel and were bleeding money.
2.) They hadn't produced a competitive CPU architecture in well over a decade
3.) They needed ~60% improvement in IPC just to match Intel
4.) AMD engineers themselves lacked belief they could compete with Intel, the company had already abandoned the high end CPU market and was on a path towards complete obsolescence

What AMD did would be equivalent to Blackberry becoming the 2nd or 3rd biggest player in the smartphone space. Or 3dfx coming back to destroy Nvidia after Voodoo 5.

[Vuyo Ncube]

"FSR isn't as good as DLSS 2.0"

WE KNOW. FSR will be sprinkled with a sharpening filter to deal with the GPU drought while AMD is likely working on its DL upscaler.

[TruthIsThere]

"FSR isn't as good as DLSS 2.0"

WE KNOW. FSR will be sprinkled with a sharpening filter to deal with the GPU drought

Still an inferior solutions. FSR is worse than checkerboard upscaling it seems and adding a sharpening filter would just add awful artifacts. This weak solution you can't have your cake and eat too. Either suffer from smoothness or sharpness, that's the reality for algorithmic rendering vs ML/DL/AI upscaling.

[_MT_]

No, rock bottom, and by far the most likely option, was bankruptcy.

The problem is that clean-sheet CPU design takes years. The project could have easily started before she became CEO. Who knows how much input she had on it. All we know for certain is that she didn't kill it. She also helped AMD by refocusing on non-PC markets which brought much needed money. Perhaps some trimming, streamlining, reorganization, rationalization. But I'm not sure how much we can credit her for Ryzen. Unless someone somewhere explained her involvement. Bulldozer was really an exercise in poor product planning. It was designed for workloads that practically didn't exist in consumer sphere.

[Bullet4564]

Anyone who knows about Resolution Scaling already knows AMDFidelityFX Super Resolution using software based Resolution Algorithms vs DLSS 2.0 that uses Machine Learning an Artificial Intelligence to upscale it's games! AI will always be better then Humans when putting together anything!

[_MT_]

AI will always be better then Humans when putting together anything!

Actually, AI is trying to do the thing a human would - understand what is in the picture. While traditional methods can't do that. You are trying to produce detail that simply isn't there.

[Spunjji]

Garbage in, garbage out... but I'm referring to the WCCF Tech article this was based on, where they specifically took the examples rendered with the GTX 1060 and not the 6800 (which are noticeably better) and based their whole conclusion on that, then referred to it as a "bait and switch" when it was pointed out that AMD won't optimise the algorithm for Pascal GPUs. Why would they?!