AMD: RDNA 3 Speculation, Rumours and Discussion

Status
Not open for further replies.
The SeriesX launched 3 months ago, and when DigitalFoundry couldn't observe its predicted 15-20% performance advantage over the PS5 in multiplatform titles, we started hearing about how there's a future devkit that will unlock the full power of the console.
Aside from Hitman 3 which for all we know it's the exception to the rule (and where we can't really compare both consoles because they're running different resolutions at over 60FPS), we're yet to see anything running better on the SeriesX in any measurably substantial way.
This comes across as purposefully ignorant to data interpretability and is entirely unnecessary to attack sites who are putting in significant effort to reduce variability in measurements for the sake of providing data to its viewers.

There is no magic to either console, these are mathematical devices that are fixed never to be changed. The performance of the titles will have a distribution of performance over time, and as such over time the hardware can be placed on a line of regression like all other GPUs that have come before it in existence. No one, including digital foundry or any other journalist, is making any sort of controversial claims if their claims are to just follow the trends of that line of regression anyway.

There has never been a controversial claim that within any family of architecture cards, that the more compute units and more bandwidth would lead to greater performance. If that is not controversial, then one should ask why you are making out such a big deal of this claim. It is _expected_ that XSX outperform PS5 by this margin because that is what is expected on the line of regression.

When we look to determine why XSX is not performing better than PS5, we do not look at PS5, we look to the line of regression. Is the XSX performing where it should be relative to the data points around it? Many could easily look at the surrounding data points around XSX and determine that the answer is no. We find it's performing worse than a 5700 in a great many of the initial launch titles, and this is without any sort of magic the PS5 could have. Until recently do we see it is performing better than the PC GPUS that we expect it to with Hitman 3.

Is PS5 performing better than where it lies on the line of regression? And it would appear, with proper and aligned settings to the PC, the answer is no. We see with Hitman 3 it's performing in line with the expectations of where it belongs with other data points around it. Does that mean PS5 can't be better? No. We expect data points to actually fall above and below, and hit directly on the line of regression.

As you can see, the reason why there is more faith around DF, and not the other sites is because they (DF) aim to illustrate of a variety of data points around where the consoles exist. They do not simply pit one console versus the other. They bring in additional data points from PC to cross check and validate results, also known as cross validated evidence, all of which you are largely ignoring the amount of painstaking process it requires to perform this. Digital Foundry is not just providing evidence via repetition, they have and are providing evidence via redundancy. The importance of this is that to pressure test any theory, it should be able to survive such theories from a great deal of a number of attack vectors. And even if the measurements are not nearly exact, they (DF) are still able to provide sufficient context on the subject to ballpark where the systems should land given if they could find exact settings.

There are 72 months in a generation of consoles, there are 70 months remaining approximately before the next generation begins. You're implications that Hitman 3 could be an exception to the rule without explicitly saying is as far as I'm concerned misinformation campaign. 2/72 months of data less than 1.3% of the data has arrived but you're putting a majority of the weight of your statements on the first 1/72th of data. Everything is going to be an exception to the rule until the first 30% of titles are released if you want to get technical, singling out Hitman 3 is quite frankly willful ignorance. To remove a data point while simultaneously attacking the people who do the work to defend an argument, that as far as I can see, contain no relevance to the discussion pertaining to the performance of the geometry engine.
 
Last edited:
This comes across as purposefully ignorant to data interpretability and is entirely unnecessary to attack sites who are putting in significant effort to reduce variability in measurements for the sake of providing data to its viewers.

Who did I attack and why the hell are you tying me to these conspiracies?

Please find some other target for fuck's sake. I'm done with this harassment.
 
Who did I attack and why the hell are you tying me to these conspiracies?

Please find some other target for fuck's sake. I'm done with this harassment.
Both RGT can be correct that PS5 has an advanced geometry engine, and simultaneously Dictator can also be correct that he hasn't heard anything from his sources that the geometry engine is improved or better.

I wouldn't have said shit if you just said, okay, RGT said it's there, but you haven't heard anything and that's okay.

I have no issue with PS5 having an advanced geometry engine. If I were set out to prove that it was, this tiny piece of this huge render pipeline, I would look for video games where the geometric details and complexity was high but the rendering resolution was low and in order to over emphasize that part of the pipeline. I would look at games like AC Valhalla and perhaps Demons Souls as anomalies where the geometry engine there could have an impact overall and try to keep as technical as possible about what could be going on behind the scenes. AC Valhalla with it's improved foliage above ultra is an excellent point to make. I wouldn't have said a thing.

But then you went and said this with respect to the context of secret sauce:
and when DigitalFoundry couldn't observe its predicted 15-20% performance advantage over the PS5 in multiplatform titles, we started hearing about how there's a future devkit that will unlock the full power of the console.
Which has nothing to do with your argument at hand and is completely a different topic. Nothing at all except to try to discredit DF, what else would the purpose of this statement be? DF was the ones who indicated that the devkit could be at fault as it was likely not ready early enough for launch titles. But once again, I'm still confused as to why you would bring in xbox series x into a discussion about how PS5's geometry engine works.

I can't make this shit up, because you wrote it. If you stick to the topic that is specifically PS5 and geometry engine, people might actually help you look for information, I actually don't have an issue to help looking for this and I would love to learn more about the Geometry Engine in PS5. But your posts aren't about discussing how the geometry engine works or what could be in it. You're not asking people to help, you're just telling us it is, and shutting down everyone else by discrediting their sources. You may not know you are doing this, but this is certainly what it comes across as.
 
I wouldn't have said shit if you just said, okay, RGT said it's there, but you haven't heard anything and that's okay.
Which is exactly what I wrote in response do @Dictator and you know it.



Which has nothing to do with your argument at hand and is completely a different topic. Nothing at all except to try to discredit DF, what else would the purpose of this statement be?
It was an answer to this post, which had been directed to me. Stop with this blatant dishonesty.

pKYKsvw.png



You're trying to make it like I'm addressing @Dictator in this harassment quest of yours when I was answering to a different user.
This was not the first nor the second time you pull this shit. You're starting a personal defamation campaign to accuse me of attacking other users and very specifically DF staff, which then results into me receiving all kinds of trolling/dogpiling and problems with the moderation team.
This is first-class underhanded trolling you're doing. Why?

I didn't do anything to you personally. I don't know you. I even tried to resort to direct messages to ask you politely to stop this harassment and then all you did was double down on the thread.
If you detest my presence so much then use the fucking ignore functionality.
 
I am beyond perplexed at users not wanting what they write in a discussion to be discussed. It is beyond baffling.

All that is requested is that you keep things civil, which the standing posts have been.

If this is too much for you, then perhaps take some time away from discussion forums.
 
I am beyond perplexed at users not wanting what they write in a discussion to be discussed. It is beyond baffling.

All that is requested is that you keep things civil, which the standing posts have been.

There is little discussion to be had when the first sentence of the "discussion" starts with an accusation of attacking websites and people.
It wasn't the first nor the second time I've been a target of this, and it's hardly a civil type of discussion.

It's also beyond baffling that posts like this one that consist of mockery and lies are deemed civil, IMO.


I have zero problems with discussing what I write. I have close to 12 000 posts in this forum and most of them are doing exactly that. To suggest this is my problem sends quite the message.




If this is too much for you, then perhaps take some time away from discussion forums.
If this is the goal then you can be more transparent about it.
 
That post should have been worded better as a direct response to your claim of not wanting to be tied to exact thing you had posted.

I did not write anything about RDNA3 being in the PS5.
It's a mockery and dogpiling post using a lie as basis for it.
It's pure trolling.
 
Okay, I see where that particular portion spiraled from, even though it could be interpreted as such by the casual user. Given your new context around what it doesn't mean, I can see your side of it there.

That is where writing short posts in discussions can falter, not providing enough context for your thoughts and leaving it up to others to interpret.
 
Can the chiplet/mcm design help with larger memory bus design ? Like easier to wire on the pcb or something like that ?

If I had to guess I'd go with no, it won't make much of a difference.
Being MCM means it's two chips in the same substrate, opposite to current solutions with one chip per substrate. All the memory bus paths need to go towards the same substrate area.

If these were two separate chips in the PCB, each with their own substrate then it would probably make it easier to get wider buses. We could see that in the dual-GPU videocards like the R9 295X2 with a whopping 1024bit combined width between the two chips, but in a MCM setup there's little reason to believe the external PCB can have a simpler design for wider memory buses.
 
Plenty of opportunities for AMD to release a more mainstream GPU to replace the 5700 XT and compete with the RTX 3060.

This thing better be good...
 
Status
Not open for further replies.
Back
Top