Nvidia GeForce RTX 5070 vs AMD Radeon RX 9070 in DLSS 4 vs FSR 4

[Scorpus]

Nvidia's GeForce RTX 5070 and AMD's Radeon RX 9070 are going head to head at the same price. We benchmark both with DLSS 4 and FSR 4 to see which mid-range GPU delivers the stronger gaming experience.

Read the full article here.

Nvidia GeForce RTX 5070 vs AMD Radeon RX 9070 in DLSS 4 vs FSR 4

 

[gamerk2]

I used to be someone who would spend premium for maximum image quality, but we've *long* gotten to the point where I can't tell much of a difference anymore. I just go 1440p High and call it a day. If I can enable any other bells and whistles I do, but more often then not it's just not worth the hassle anymore.

 

[messy]

I always fall on the side of Radeon simply due to Linux support. I know that's not everyone's cup of tea but built-in mesa support is just so nice

 

[GoldenGoat]

This would be a perfect article for a site called GameSpot, but on TechSpot I want to know about more than just gaming performance. It's 2025, tech fans use their GPU for more than just gaming.

 

[j4ni1]

9070 beats the 5070 across the board, except in two games, one of which is margin of error and still would advice to buy the 5070. Typical Tim Schiesser.
If the benchmars show anything than that the two cards are a tier level apart.

 

[montrealer88]

This would be a perfect article for a site called GameSpot, but on TechSpot I want to know about more than just gaming performance. It's 2025, tech fans use their GPU for more than just gaming.

The article focuses more on comparing GPU upscaling technologies than on comparing GPUs themselves.or at least, that's how I interpreted the author's intent.

 

[James321]

Pity AMD price their cards stupidly I mean $50 difference between the non [XT] and the XT?

Then you would be lucky to find anyone of these cards at MSRP anyways.

 

[James321]

The article focuses more on comparing GPU upscaling technologies than on comparing GPUs themselves.or at least, that's how I interpreted the author's intent.

The comparison is out of whack then, what is the point we all know the faster card will be faster with upscaling enabled as well. Maybe image quality should have been taken in consideration then?

Bit of a pointless exercise comparing upscaling without image quality.

 

[Kirby1]

I have been very happy with me 7900XT. The RT support and upscaling is not the best, but the raster is so strong that in most games I either don't need upscaling, or only need to use quality mode.

 

[Thatsdisgusting]

I used to be someone who would spend premium for maximum image quality, but we've *long* gotten to the point where I can't tell much of a difference anymore. I just go 1440p High and call it a day. If I can enable any other bells and whistles I do, but more often then not it's just not worth the hassle anymore.

CP77 Path traced wanna have a word on this...

 

[daffy duck]

This would be a perfect article for a site called GameSpot, but on TechSpot I want to know about more than just gaming performance. It's 2025, tech fans use their GPU for more than just gaming.

Yes, I've complained about the same thing. Can we grow up yet.

 

[Axiarus]

Pity AMD price their cards stupidly I mean $50 difference between the non [XT] and the XT?

Then you would be lucky to find anyone of these cards at MSRP anyways.

XFX 9070XT is 679.99 right now on Amazon. Thats $80 from MSRP. With tariffs I would say thats a decent price.

 

[ddferrari]

9070 beats the 5070 across the board, except in two games, one of which is margin of error and still would advice to buy the 5070. Typical Tim Schiesser.
If the benchmars show anything than that the two cards are a tier level apart.

He didn't advise to buy the 5070.

His first sentence of the last paragraph is:
"In the end, neither card lands as the obvious slam dunk choice"

 

[EkmaNaff]

Great article. I would have liked to have seen the difference between the Average FPS and the 1% Low. I may be wrong on this but I feel that a sudden jump from the Average FPS to the 1% Low might cause a jitter in the game. If so, I'd prefer the card with the less chance of a jitter than necessarily the Higher FPS.

Case in point, years ago I bought an AMD card (don't remember exactly which one now) and kept having jitters during gameplay. I took it back and traded it in for the Nvidia equivalent (even though it was considered slightly lower FPS). Couldn't have been happier.

While I realize that the software (ie; drivers) for the card plays a huge part of the overall ability of a card, at that time Nvidia still had the better, more polished software. This was even while using an AMD CPU...with no configuration changes between the 2 cards.

Has anyone else noticed the same issues?

 

[Alfatawi Mendel]

Nvidia drivers seem to have declined in quality over the past few years, whereas AMD's have gotten better.
As the article refers to.
I'm sure Tim would've mentioned problems by either card within games, if there were any.

 

[Alfatawi Mendel]

"Right now both GPUs retail for $550, which puts the Radeon at an 11% advantage in cost per frame. That still falls just short of the 15% threshold we typically look for when recommending Radeon over GeForce. However, context matters."
Good to know that AMD needs a 15% lead in order for Techspot to recommend it over Nvidia. And 'context' is ALWAYS what the author decides it is. Makes a mockery of the supposedly 'rigorous testing regime'.

 

[Julio Franco]

The comparison is out of whack then, what is the point we all know the faster card will be faster with upscaling enabled as well. Maybe image quality should have been taken in consideration then?

Bit of a pointless exercise comparing upscaling without image quality.

We have tested and compared both of these GPUs at native first. Then we have done several round ups of DLSS (2,3,4, features) vs FSR (2,3,4) image quality. And now we're publishing more follow ups dedicated to upscaling performance versus, especially because FSR 4 is more comparable to DLSS 4 in terms of image quality, whereas before FSR was definitely lower quality.

On every newer piece we link to and refer to previous testing/findings, but we can't repeat the same things over and over.

"Right now both GPUs retail for $550, which puts the Radeon at an 11% advantage in cost per frame. That still falls just short of the 15% threshold we typically look for when recommending Radeon over GeForce. However, context matters."
Good to know that AMD needs a 15% lead in order for Techspot to recommend it over Nvidia. And 'context' is ALWAYS what the author decides it is. Makes a mockery of the supposedly 'rigorous testing regime'.

It's not the first time Tim says so. Instead of making up facts, we are giving you the numbers (native, upscaling, and IQ samples) and explaining our reasoning straight. He believes GeForce gets an edge because DLSS 4 game support is about twice as strong as FSR 4 (list will change but it's like 175 vs 80 titles at the moment).

I'm willing to bet that 15% will get revised soon because until the previous generation Radeon was worse in ray tracing, upscaling quality and number of supported games (and local AI processing/CUDA, important, but that's not gaming). That gap has shortened this past year.

But again, that's what he thinks but you have the benchmarks, and you can vote with your wallet.

 

[loki1944]

I never us upscaling; it looks like total *** vs native.

 

[loki1944]

Nvidia drivers seem to have declined in quality over the past few years, whereas AMD's have gotten better.
As the article refers to.
I'm sure Tim would've mentioned problems by either card within games, if there were any.

Nah; AMD drivers still have tons of issues just like Nvidia.

 

[TempleOrion]

Nah; AMD drivers still have tons of issues just like Nvidia.

Another baseless assertion without evidence.

 

[mbk34]

It's 2025, tech fans use their GPU for more than just gaming.

I'm genuinely curious, is it mostly image and video manipulation or are there a whole range of applications that use GPU's for processing?

 

[NightAntilli]

The 15% difference to be able to recommend AMD is an absolute joke. With these two cards, if the prices are near MSRP and you go for the 5070 instead of the 9070 for gaming, you're either ignorant, misinformed, biased or a fanboy.

 

[GoldenGoat]

I'm genuinely curious, is it mostly image and video manipulation or are there a whole range of applications that use GPU's for processing?

Yes, editing, also analyzing large data sets, and of course AI. AI performance is the one I am most interested in seeing benchmarks on and is rapidly evolving. I like to run open weights LLMs and use ComfyUI.

 

[mbk34]

Yes, editing, also analyzing large data sets, and of course AI. AI performance is the one I am most interested in seeing benchmarks on and is rapidly evolving. I like to run open weights LLMs and use ComfyUI.

Perhaps TS should do a series of articles on what these sort of tools offer and how to run them with GPU's. Maybe even an intro into CUDA (or ROCm) programming.

 

[Jarenien]

"There are also extra features Nvidia offers like CUDA, AI tooling, and multi frame generation, but for pure gaming, these are the main considerations."

MFG IS PURE GAMING. Geezus f-in c.... While people continue to not use it when measuring how good a card is is just laughable. If you put MFG on the table, it blows the crap out of the donkey *** AMD card, lmao...