Insider Gaming can confirm that the leaked PS5 Pro specs leaked earlier today are real and the PlayStation 5 is still tentatively targeting a 2024 holiday release.
Speaking with sources, who wished to remain anonymous because they were not authorized to talk about company plans, we can confirm that the leaked documentation from the YouTube channel Moore’s Law is Dead is real, despite the criticism of the leaker and the leaked specs. Insider Gaming can confirm that the documentation leaked is from a PlayStation developer portal, which was sent out this week to a wider band of third-party developers.
In early 2023, I reported via Key to Gaming that the PS5 Pro is under the codename ‘Trinity’ and will be targeting improved and consistent FPS at 4K resolution, a new ‘performance mode’ for 8K resolution, and accelerated ray tracing. In addition, it was reported that Trinity will have 30 WGP and 18000mts memory.
Today’s leaked documents also confirmed:
Insider Gaming, who was also shared documentation from the developer portal under the condition that it’s not shared publically or privately can also confirm that Devkits have been available to first-party studios since September 2023, third-party since January 2024, and from Spring 2024 Testkits will also be available which will be identical to the final product.
- Rendering 45% faster than PS5
- 2-3x Ray-tracing (x4 in some cases)
- 33.5 Teraflops
- PSSR (PlayStation Spectral Super Resolution Upscaling) upscaling/antialiasing solution
- Support for resolutions up to 8K is planned for future SDK version
- Custom machine learning architecture
- AI Accelerator, supporting 300 TOPS of 8 bit computation / 67 TFLOPS of 16-bit floating point
Insider Gaming understands that the PS5 Pro is currently aiming for a tentative holiday 2024 release, but the date could be changed due to the lack of first-party games released on the PlayStation 5 this year.
This shouldn't be delayed due to a lack of 1st party games, this isn't a new generation machine it's a refresh, what they should do is have improvements ready for existing 1st party games. If GTA6 is still supposed to come out before the end of the first quarter of 2025 getting this out early and having it be known that this would be the best place to play GTA6 at launch would be better than having a first party title to launch alongside it anyway.
More corroborations
It's 45% but that's only part of it, it'll be interesting to see what their upscaling solution is like and how much that allows for other improvements.35% performance increase? I’d have thought it would deliver way more from the leaked specs.
Suspect chart considering PS5 isn’t RDNA 2
Someone on Neogaf wrote this up, not saying it's 100% accurate because I don't know lol.
Naturally all the "Leak Bros" on X/YouTube are running with the high floppage.
77TF - FP16 Half-Precision
33.5TF - FP32 Single Precision Dual-Issue
16.75TF - FP32 Single Precision << The traditional unit of measurement and what we need to compare it to PS5's 10.28TF.
~1TF - FP64 Double Precision
So that's a 63% jump in TF.
The 45% faster "render" which was original referred to as "raster" seems to be a bit of a miscommunication. I'm guessing this is in regards to the GPU frontend jumping from 64ROPS to 96ROPS minus a small clock reduction.
Back to the flops... That's a 63% jump in raw compute, then you can add the isolated architectural gains of RDNA3 (and some RDNA4) for an idea of real-world perf (on console you might see ~20-30%). As for dual-issue, it's hard to say for sure, but if on average, ~10% of workloads can be ran this way then that's an extra 10% of additional perf. Probably looking at ~20TF equivalent or 2X.
Then if the RT acceleration capabilities are 2x as good, then that's 4x the RT performance in ideal circumstances.
Still surprised by the 2180MHz number, I'd have thought they'd just stuck with 2233 at least, the extra 53MHz is unlikely to break the bank. I can only assume this is in Pro only mode and that in base mode it runs at 2233 with 24CUs shut off.
Or, these leaks are only reporting the low end of the Pro's Continuous Boost scheme, just as the low end of the same was reported for PS5.
Suspect chart considering PS5 isn’t RDNA 2
Someone on Neogaf wrote this up, not saying it's 100% accurate because I don't know lol.
Naturally all the "Leak Bros" on X/YouTube are running with the high floppage.
77TF - FP16 Half-Precision
33.5TF - FP32 Single Precision Dual-Issue
16.75TF - FP32 Single Precision << The traditional unit of measurement and what we need to compare it to PS5's 10.28TF.
~1TF - FP64 Double Precision
So that's a 63% jump in TF.
The 45% faster "render" which was original referred to as "raster" seems to be a bit of a miscommunication. I'm guessing this is in regards to the GPU frontend jumping from 64ROPS to 96ROPS minus a small clock reduction.
Back to the flops... That's a 63% jump in raw compute, then you can add the isolated architectural gains of RDNA3 (and some RDNA4) for an idea of real-world perf (on console you might see ~20-30%). As for dual-issue, it's hard to say for sure, but if on average, ~10% of workloads can be ran this way then that's an extra 10% of additional perf. Probably looking at ~20TF equivalent or 2X.
Then if the RT acceleration capabilities are 2x as good, then that's 4x the RT performance in ideal circumstances.
Still surprised by the 2180MHz number, I'd have thought they'd just stuck with 2233 at least, the extra 53MHz is unlikely to break the bank. I can only assume this is in Pro only mode and that in base mode it runs at 2233 with 24CUs shut off.
Or, these leaks are only reporting the low end of the Pro's Continuous Boost scheme, just as the low end of the same was reported for PS5.
What does that have to do with this info? This is just for info and context lol but Teraflops don't tell the whole story so they do matter to a lesser degree than last gen because you need a larger gap to make a material difference.Teraflops matter again
Teraflops matter again
Some people seem to keep wanting this gen to be a rerun of the One X vs the PS4 Pro for some reason, the 1.8 TF advantage is basically the same but the percentage of overall performance that 1.8 TF number represents is much smaller.It always mattered to a certain extent but not for a standardized performance indicator especially for two consoles that are built different with some having advantages over the other in different areas.
Leak bros and floppage in the same sentence?Naturally all the "Leak Bros" on X/YouTube are running with the high floppage.
that’s just bad writing, it’s like the typical 2nd movie in a trilogy, always ends in a cliffhanger….pffftWhat does that have to do with this info? This is just for info and context lol but Teraflops don't tell the whole story
What am I expecting?
Could that then work with older games? I’m assuming that wouldn’t work with games that are streaming?Upscaled 4k output / 60 FPS in every game with AI Machine learning/ PSSR.
Could that then work with older games? I’m assuming that wouldn’t work with games that are streaming?