Sony PlayStation 5 Pro

s gibt aber auch Spiele, in denen RDNA 3 teils klar vorne liegt. Den größten Vorteil erkämpft sich die Architektur in Call of Duty: Modern Warfare 2, wo RDNA 3 bei gleicher Rechenleistung mal eben 21 Prozent performanter als RDNA 2 ist. Auch in A Plague Tale: Requiem gibt es mit 13 Prozent einen größeren Vorteil, in Cyberpunk 2077 sind es 10 Prozent und in Spider Man: Miles Morales 12 Prozent.
You don’t use outliers as the rule when determining a performance metric.
 
Upside down (horizontal obviously) without any additional device is the way. Advice from the Playstation boss and it works pretty well.
 
Btw whats te purpose of this stand ? ;d I have my ps5 vertical without stand
It's most useful for the Digital PS5, which has a very small base footprint when vertical, and a very awkward shape for placing it horizontally.

If you have a disc model, that helps a fair bit in both situations, adding a bigger footprint when vertical, while also provide a part base when horizontal.

And all this applies ever moreso to the original PS5 versus the newer 'Slim'.
 
  • Like
Reactions: snc
It's most useful for the Digital PS5, which has a very small base footprint when vertical, and a very awkward shape for placing it horizontally.

If you have a disc model, that helps a fair bit in both situations, adding a bigger footprint when vertical, while also provide a part base when horizontal.

And all this applies ever moreso to the original PS5 versus the newer 'Slim'.
Its a bit scummy that its recommended to buy a stand. I'll just use it with the plastic pins when I buy it next year and lay it flat. Or place it between two walled parts if its possible.
 
The RT units are RDNA4, what I said is that AMD came up and worked by themselves to create those new RT units without the need for Sony to motivate or advice them.
Cerny specifically said in the interview that they actually motivated them and they were really happy they did (because they were happy of the results?). Seems we wouldn't have that kind of accelerated RT if it wasn't for Sony.

Now what did he meant by "motivated" is another story and open to discussion. Like: can you give us something similar to Nvidia accelerated RT? Like this feature, that feature? I dunno.
 
Cerny specifically said in the interview that they actually motivated them and they were really happy they did (because they were happy of the results?). Seems we wouldn't have that kind of accelerated RT if it wasn't for Sony.

Now what did he meant by "motivated" is another story and open to discussion. Like: can you give us something similar to Nvidia accelerated RT? Like this feature, that feature? I dunno.

Sony didn't "motivated" AMD to create RDNA4 RT Unity/Capabilities, like Cerny himself says in that interview AMD itself wanted and it's engineers delivered. Sony only made an order from AMD Semi-Custom to put those RT unities on an older than RDNA4 GPU Arch. The PS5 Pro SOC wouldn't exist if Sony hadn't ordered that SOC for their PS5 Pro, but RDNA4 RT Unity would still exist with all it's capabilities regardless.
My gripe is that WCCF articles gives people the impression that Sony and Cerny deserve any credit for AMD's work, like if AMD needed Sony just to run their business.
 
So the PS5 Pro is using a precursor to AMD's Tensor core equivalent? Pretty exciting to see what AMD has in-store for us.

The last time we got a major graphics feature on consoles before PCs was Unified Shaders on the 360 no?
 
So the PS5 Pro is using a precursor to AMD's Tensor core equivalent? Pretty exciting to see what AMD has in-store for us.

The last time we got a major graphics feature on consoles before PCs was Unified Shaders on the 360 no?
If I remember correctly ps5 and xsx were before rdna2 desktop, also ps4pro has 16bit double rate before amd desktop gpu, thats common in history, not sure why surprise here
 
So the PS5 Pro is using a precursor to AMD's Tensor core equivalent?
It's not entirely clear what the solution for PS5 Pro will entail. A while back, some digging into RDNA4 found sparsity instructions and 8-bit precision improving on RDNA3's basic wave matrix instructions, which should give further AI/ML improvements. This doesn't sound like they're doing a dedicated AI/tensor core equivalent to me, and more just gradually expanding their existing strategy, providing a leaner(in terms of die space) solution.

Not an expert on this though, by any means.
 
Back
Top