That depends on the workload, for heavier RT/PT games such as Black Myth or Indiana Jones the difference is near 100%.Right now RT core per core, Nvidia is about ~40% better with RDNA4.
Aaaaand here comes the backtrack.
Yes but he already got his clicks. Still Shameful.Aaaaand here comes the backtrack.
Yes but he already got his clicks. Still Shameful.
Going by a single game isn't good enough.That depends on the workload, for heavier RT/PT games such as Black Myth or Indiana Jones the difference is near 100%.
There's no backtracking.Aaaaand here comes the backtrack.
They are earning a fuck ton and I'm not asking for some crazy performance likeBojji that is disappointed that tb Pro isn't a 5070. I'm asking for a little of decency in a product where they usually eat costs and still make a fuck ton of money. Or else: disappear. There will be better options if the hardware is not subsidised.
I was worried that the amount of RAM was going to be minimal next gen but seeing all of these rumours about a starving bandwidth at this point I'd gladly take 16GB of whatever they can muster that is fast enough.
Why is 9700 XT slightly worse than 7900 XTX in fully pathtraced games? Doesn't look good for PS6 pathtraced games IMO.Going by a single game isn't good enough.
Here is a proper test result.
The 9070XT (64 RT cores, 304W) vs the 5070TI (70 RT cores, 300W) is the closest comparison we can get, which isn't double the RT performance.
Radeon RX 9070 XT (ASUS TUF OC) review
![]()
![]()
7900 XTX has 96 RT cores vs 64 RT cores in the 9070 XT.Why is 9700 XT slightly worse than 7900 XTX in fully pathtraced games? Doesn't look good for PS6 pathtraced games IMO.
I'm not going by a single game. I listed two (although even one would be better than 3D mark of all things), but there are even more if you want, Doom The Dark Ages is ~80% faster in PT. Half Life RTX is ~90% faster as well. The difference is really large with the heaviest PT games, especially in games that have a lot of foliage and can use OMM on Nvidia hardware. These results are for a 5070Ti vs a 9070XT. The Nvidia card also uses 25% less power and is a a smaller and less dense chip on a worse node (5nm vs 4nm).Going by a single game isn't good enough.
Here is a proper test result.
The 9070XT (64 RT cores, 304W) vs the 5070TI (70 RT cores, 300W) is the closest comparison we can get, which isn't double the RT performance.
Radeon RX 9070 XT (ASUS TUF OC) review
![]()
![]()
Even if it's not 1tb, but closer I would digest it, but 640 is just diabolical levels of stupidity.This. PS6 should be at least 1TB/s. That's less than a 2X increase over the already 5 years old XSX, that by then will be 8 years old. It's not even an older generational increase. It's just bare minimum.
Here in Cyberpunk, which is a Nvidia favored game. At 1440p RT overdrive with no upscaling, the 5070TI is only 50% better and at the same watts.I'm not going by a single game. I listed two (although even one would be better than 3D mark of all things), but there are even more if you want, Doom The Dark Ages is ~80% faster in PT. Half Life RTX is ~90% faster as well. The difference is really large with the heaviest PT games, especially in games that have a lot of foliage and can use OMM on Nvidia hardware. These results are for a 5070Ti vs a 9070XT. The Nvidia card also uses 25% less power and is a a smaller and less dense chip on a worse node (5nm vs 4nm).
I think AMD can catch up with RDNA5 but exceeding Nvidia as to be twice as fast in RT? Not going to happen.
That's a very broad estimation lol. Its almost as if he doesn't know.There's no backtracking.
In the video he said it should land between the 9070 and 5080 in raster.
8:41 mark.
That's a very broad estimation lol. Its almost as if he doesn't know.
How can he know?That's a very broad estimation lol. Its almost as if he doesn't know.
Well Loxus wrote the 9070 which is significantly slower9070XT is sometimes on par with 5080 and sometimes below 4070ti Super, it's game dependent.
I did watch it and it was ridiculous, especially the 5090 RT reasoning. He might have some of the specifications but his predictions are, well dumb.How can he know?
If the documents states 2.5 - 3× rasterization vs PS5. Without a proper GPU benchmark comparison, we can only guess where raster will fall based on the 2.5 - 3x.
He even says this in the video.
It's as if you didn't watch the video.
One of the main reasons devs are swapping to Unreal Engine 5 is because it can do in real-time what other engines require baking for. Baking is slow and error prone and in development time is money - hence the trend. If next-gen consoles ship with enough RT grunt to standardise RT solutions for lighting and geometry culling, then dev will absolutely implement that as it'll save them money and time. This means we get better looking games faster with less financial risk to the devs.But would the next games actually use all of that
There's no backtracking.
In the video he said it should land between the 9070 and 5080 in raster.
8:41 mark.
Alright, so put it all together - All Sony is claiming in the documents I have is that they have a GPU as fast as the 6060 Ti or 6070 in 2027 that uses the same energy as a typical console lol. That isn't anymore insane than when we learned in 2019 that Sony would put a next-gen SSD in a $500 console...and they did do that despite what the armchair experts on Reddit said...
Alright, so put it all together - All Sony is claiming in the documents I have is that they have a GPU as fast as the 6060 Ti or 6070 in 2027 that uses the same energy as a typical console lol. That isn't anymore insane than when we learned in 2019 that Sony would put a next-gen SSD in a $500 console...and they did do that despite what the armchair experts on Reddit said...
My guess is that he's taking the 5060 ti and adding a 15% increase that is what generations are now about. But that 5060 ti is going to perform like a 4060 if bandwidth is not addressed. Can't believe is the same company that put 8GB of GDDR5 in PS4.More bullshit.
How the fuck did Sony get their hands on a 6060 Ti when Jensen himself doesn't have one?
Here in Cyberpunk, which is a Nvidia favored game. At 1440p RT overdrive with no upscaling, the 5070TI is only 50% better and at the same watts.
![]()
Or am I looking at it wrong?
I'm open for correction.
But cross-gen games and PS6 handheld will ensure that baked lighting will still be around at least until PS7 and PS7 handheld and PS6 cross-gen games.One of the main reasons devs are swapping to Unreal Engine 5 is because it can do in real-time what other engines require baking for. Baking is slow and error prone and in development time is money - hence the trend. If next-gen consoles ship with enough RT grunt to standardise RT solutions for lighting and geometry culling, then dev will absolutely implement that as it'll save them money and time. This means we get better looking games faster with less financial risk to the devs.
De-nial with another corkerDay 1.
Per CU/SM RDNA5 should beat Blackwell in RT, but 5090 has way more SMs, way more bandwidth and cache, and way higher TDP.So in a lot of games the difference isn't really that large, but those games have the most demanding PT feature sets while leveraging things like OMM, hence the massive lift in performance vs AMD. Microsoft has already stipulated support for OMM in the next version of DXR so I expect RDNA5 (and thus PlayStation) to have that capability as well. So catching up or even exceeding Nvidia in pure RT is possible, I just don't think it will be to the level that a ~9070 XT card in raster will perform like a 5090 in RT.
By the time this is released. Sony will be doing Day 1 PC releases making this console redundant.
Wccftech: Renowned leaker KeplerL2 confirms that RDNA 5 RT performance per CU beats Blackwell and as a result, PS6 RT performance is 6-12x that of PS5. But most importantly, he confirmed that MLiD is a dumbass that doesn't know how to interpret his own leaks.Per CU/SM RDNA5 should beat Blackwell in RT, but 5090 has way more SMs, way more bandwidth and cache, and way higher TDP.
The issue is MLID is interpreting the "6-12x faster RT" as "if a game uses RT on the PS5 and runs at 30 FPS it will run at 180-360 FPS on the PS6", when it actually just means the RT portion of the frametime is 6-12x faster (and the 12x is somewhat theoretical as it applies to things RDNA2 sucks at like multi-bounce GI or diffuse reflections which PS5 games pretty much never use).
Someone is triggered, I guess?De-nial with another corker![]()
Thank you. While I do believe that RDNA5 can best Blackwell pound for pound in RT…that's the whole issue with his 5090 statement. The 5090 is a super heavyweight and the PS6's GPU will be a middleweight. A top tier middleweight still isn't beating even a mid tier super heavyweight.Per CU/SM RDNA5 should beat Blackwell in RT, but 5090 has way more SMs, way more bandwidth and cache, and way higher TDP.
The issue is MLID is interpreting the "6-12x faster RT" as "if a game uses RT on the PS5 and runs at 30 FPS it will run at 180-360 FPS on the PS6", when it actually just means the RT portion of the frametime is 6-12x faster (and the 12x is somewhat theoretical as it applies to things RDNA2 sucks at like multi-bounce GI or diffuse reflections which PS5 games pretty much never use).
Launch titles
Horizon Zero Dawn : Director's Recut Remastered Edition
Horizon Forbidden West: Remastered
Last Of Us: Director's Cut Remastered Edition
Last Of Us 2: Remastered Remake
Are you sure Jensen doesn't have one?More bullshit.
How the fuck did Sony get their hands on a 6060 Ti when Jensen himself doesn't have one?
Check the results of Black Myth, Indiana Jones, and Doom. Doom doesn't look almost 100% but remember that with PT disabled Doom on the 9070 XT is quite a bit faster vs the 5070 ti.
So in a lot of games the difference isn't really that large, but those games have the most demanding PT feature sets while leveraging things like OMM, hence the massive lift in performance vs AMD. Microsoft has already stipulated support for OMM in the next version of DXR so I expect RDNA5 (and thus PlayStation) to have that capability as well. So catching up or even exceeding Nvidia in pure RT is possible, I just don't think it will be to the level that a ~9070 XT card in raster will perform like a 5090 in RT.
...he confirmed that MLiD is a dumbass that doesn't know how to interpret his own leaks.
I personally really like Id Tech engine, and especially the newest version in the Dark Ages, I wish devs would use more of thatOne of the main reasons devs are swapping to Unreal Engine 5 is because it can do in real-time what other engines require baking for. Baking is slow and error prone and in development time is money - hence the trend. If next-gen consoles ship with enough RT grunt to standardise RT solutions for lighting and geometry culling, then dev will absolutely implement that as it'll save them money and time. This means we get better looking games faster with less financial risk to the devs.
Both cards are using balanced upscaling. FSR3/DLSS3 perform very similar and are on the order of a few percent different.That is not a fair testing of RT performance due to DLSS.
With no upscaling it's about 40% faster.
Are you sure Jensen doesn't have one?
Here is what is said to be the 6090, slated to release at the end of 2026. Surely he has a 6060Ti.
Check the results of Black Myth, Indiana Jones, and Doom. Doom doesn't look almost 100% but remember that with PT disabled Doom on the 9070 XT is quite a bit faster vs the 5070 ti.
So in a lot of games the difference isn't really that large, but those games have the most demanding PT feature sets while leveraging things like OMM, hence the massive lift in performance vs AMD. Microsoft has already stipulated support for OMM in the next version of DXR so I expect RDNA5 (and thus PlayStation) to have that capability as well. So catching up or even exceeding Nvidia in pure RT is possible, I just don't think it will be to the level that a ~9070 XT card in raster will perform like a 5090 in RT.
Cross gen is based on install base, not tech capability. Everything will be cross gen with PS5.But cross-gen games and PS6 handheld will ensure that baked lighting will still be around at least until PS7 and PS7 handheld and PS6 cross-gen games.
I support the idea of cheaper PS6 in the guise of PS6 Canis (handheld and console), but I hope cross-gen games for next gen will be limited to PS6 Orion. PS7 "Canis" (handheld and console) will be comparable to PS6 Orion by then.
They have no choice but to do PC releases, there's a yuuuuge gaming population that will never buy Consoles. There's also a yuuuuge Cloud only population that can't afford consoles or refuse to buy hardware.That's why they won't do Day One releases.
If they are smart, they won't do any PC releases.
Well, there's the next article.Wccftech: Renowned leaker KeplerL2 confirms that RDNA 5 RT performance per CU beats Blackwell and as a result, PS6 RT performance is 6-12x that of PS5. But most importantly, he confirmed that MLiD is a dumbass that doesn't know how to interpret his own leaks.
I think only the AT1 die running in a Xbox PC with water cooling could make any claims about being faster than 5090 in either Raster or RT.Thank you. While I do believe that RDNA5 can best Blackwell pound for pound in RT…that's the whole issue with his 5090 statement. The 5090 is a super heavyweight and the PS6's GPU will be a middleweight. A top tier middleweight still isn't beating even a mid tier super heavyweight.
It makes sense, it is easier for him to get people excited I guess. It is a kind of "build it and they will come" as RT could be relied by devs more if the capability was there. Despite it all PS5 allowed RT more than people thought.Per CU/SM RDNA5 should beat Blackwell in RT, but 5090 has way more SMs, way more bandwidth and cache, and way higher TDP.
The issue is MLID is interpreting the "6-12x faster RT" as "if a game uses RT on the PS5 and runs at 30 FPS it will run at 180-360 FPS on the PS6", when it actually just means the RT portion of the frametime is 6-12x faster (and the 12x is somewhat theoretical as it applies to things RDNA2 sucks at like multi-bounce GI or diffuse reflections which PS5 games pretty much never use).
It's just a staggered release to boost high end sales.Absolutely not. Not only is this uncertain, but NVIDIA might not have even named their products yet. On top of that, the 60 Ti cards come out much later. We're possibly over 18 months away from it. No one has one because it doesn't exist yet
The last time I checked, MLiD is the one leaking the PS6 specs not Kepler.