Quote from: Sanjiv Sathiah on May 03, 2020, 14:25:07
@OzMomotaro You concede that "Yes, Sony may be partnering with AMD on specific tech..." That is the point of this article. And yes, it is equally true that Microsoft is doing the same. Yet, that is not what this rumor is about. Cerny isn't going to give away all the details this far ahead of the launch that is going to give Microsoft a chance to sniff out what Sony and AMD are doing on the PS5 -- particularly if it is instruction set-based and not hard wired into the silicon.
Instruction sets are "hard wired" into the silicon - either a piece of silicon supports a given instruction set (or subsection of an instruction set, like an extension of one), or it doesn't. There is no way to add this after the fact unless the required hardware is already in place (unless you emulate support through some sort of translation layer, in which case it will be slow and likely buggy). So if Sony had access to some sort of ISA-level RDNA 3 feature that MS didn't, they would be shouting it from the rooftops by now if it had any serious value - there would be no way for MS to implement the same at this point. Both of these SoCs have been taped out already and are likely in test production with mass production ramping up in the coming months.
Quote from: Sanjiv Sathiah on May 03, 2020, 03:27:02
@Valantar "Which is just outright false. VRS is an RDNA 2 feature" I didn't claim it was an RDNA 3 feature -- I was writing about it as an RDNA 2 feature but also as an example of the kind of technology/approach that might result in Sony getting access to further optimized instruction set support. But perhaps you are just trying to find fault?
I'm not "trying to find fault", I'm reading your article as it was written with the barest minimum of critical interest - which I would say is how one should read anything one cares about, no?
You present VRS as an example for both parts of a two-pronged argument. The first part of this argument is that Sony has previously been given early access to GPU design elements for their consoles - which they did with parts of early Vega designs for the PS4 - and you present VRS as an example of what such access might look like for the PS5. The issue here is that VRS is a bog-standard RDNA 2 feature, so how would it be representative of getting early access to anything, given that the XSX is confirmed to be fully RDNA 2? PCs with AMD GPUs will have VRS around the same time as the PS5 arrives, as will the XSX. All VRS exemplifies is how RDNA 2 unlike RDNA (1) focuses more on adding features. If you were trying to exemplify some sort of hitherto unknown future graphics feature, using VRS as an example doesn't fit the bill.
The second part of the argument is that you claim that "RNDA [sic] 2 and RDNA's architectural changes have been focused on reducing instruction sets and latency" and that VRS as a new feature contrasts to this approach. This part of the argument is flawed in two ways: not only does RDNA 2.0 come with quite a few significant new graphics features (VRS among them, as well as real-time ray tracing, alongside the primitive shader from Vega (which was finally enabled in RDNA (1)) being updated to a mesh shader, and likely more that haven't been presented yet) making the factual basis of this part of your argument fundamentally erroneous, but even saying RDNA (1) focused on "reducing instruction sets and latency" is ... a weird misrepresentation of its changes. While RDNA indeed does lower latency, it does so through a fundamental rearchitecting of its wavefront to increase SIMD utilization, which amounts to much more than "reducing instruction sets and latency". It also carries over 100% compatibility with the GCN ISA, so saying it "reduc[es] instruction sets" is also false; it might implement a new and simplified ISA, but it is not a reduced instruction set as such, just a condensed one added on top of the existing one.
Quote from: Sanjiv Sathiah on May 03, 2020, 14:30:43
As for me personally, I'm equally excited by both consoles and can't wait for them to be released. I have no skin in the game one way or the other as I will be buying both consoles, as I always do. The reason is two-fold -- as a tech writer I just love technology and I like to play the exclusive titles on each platform.
Just an FYI: taking unsubstantiated rumors from untrustworthy sources and running with them by writing uncritical articles speculating willy-nilly about how it might look if the rumor turns out to be true a) makes you look biased towards whoever stands to gain from the rumor being true (and calling it a rumor in the title doesn't help when there's nary a trace of critical thinking in the rest of the article), and b) is
terrible journalism. I don't mean this to come off as personal in any way, but writing and publishing articles of this type makes both you and NotebookCheck look bad - writing like this belongs in garbage-quality rumor-monger publications like WCCFTech, not serious publications wanting to maintain a reputation for journalistic integrity and rigor.
It's great that tech journalists are enthusiastic about the tech they cover (I'd not quite call it a requirement, but close), but the ability to maintain a critical distance and ask the right questions is part of what differentiates quality journalism from rumor-mongering.
Deutsch
English
Español
Français
Italiano
Nederlands
Polski
Português
Русский
Türkçe
Svenska
Chinese
Magyar