• Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
Status
Not open for further replies.

vivftp

Member
Oct 29, 2017
19,754
Let me ask this, what sort of database was rogames info mined from? Is this a source where actual hardware would have had to be tested for this info to appear? Or is this just info typed by someone in a database somewhere? What's the track record of data being mined from this source?
 

Deleted member 12635

User requested account closure
Banned
Oct 27, 2017
6,198
Germany
One Thing that I think is kinda reasonable that I see being posted here is the idea that earlier devkits were using GCN. Just makes sense to me giving how devkit Revisions work.
Yeah, I can get my Vega 56 up to 12TF with high clocks. A Vega 64 would be over 13 TF at same clocks. If this was a first iteration of dev kit setups I can get all the confusion ...
 

anexanhume

Member
Oct 25, 2017
12,913
Maryland
Going with this, Oberon first showed up in these drivers in August and then the updated 'spin' B0 version in October. It isn't old info and there is no evidence of any other chip so far. I do wonder why there is still nothing for Arden, though?
AFAIK benches for a MS part have never leaked. Tonky answered the rest of your question.

So if it's a public driver, could this be retail gpu not console gpu?
I don't think so. The codenames wouldn't make sense.
 

gundamkyoukai

Member
Oct 25, 2017
21,105
It came from Matt, who did not guarantee it but did state that it was the plan for some time.

Matt also said that was change ages ago.
Jason said since early this year consoles were 2020 for a good while.
Klee since 2017 plans chnage.
The reason people acting like it matter is to say PS5 tech is old or using there own.
 
Oct 25, 2017
13,246
No one has verified the information from Matt, Jason, and others, either. Klee has also talked to an actual dev.
There's literally no difference. We're basing the accuracy of their info on their reliability and credibility. It's your prerogative who you choose to believe.

I'm sorry. There is a world of difference between someone like Jason Schrier and a forum insider with unverified information.

A world of difference.
 

sangreal

Banned
Oct 25, 2017
10,890
here is the context they aren't giving you:
This info is not datamined from drivers
It is not related to devkits
It is not complete so I wouldn't read into "missing features." It's not a spec sheet
It is from internal AMD performance testing documents inadvertently posted publicly
They cover some specific set of test cases mostly comparing Ariel and Oberon results
It is dated June/July and is the same source for info "leaked" around that time

The poor sap who uploaded it will surely be fired regardless but I won't link to it because it contains personal data including their resume. It only took like 30 minutes to find, just search for it -- this guy has been quoting whole pieces verbatim

Since people keep quoting this I'll add my later disclaimer:
to be clear I have zero opinion on the usefulness of this data or applicability (if any) to the retail PS5. It's detailed to an extremely low level and at the same time a bit of a mess with gaps at the high level. I am glad there are people smarter than me who can read the tea leaves from it. I just think everyone should have the context of what it is and isn't instead of debating for 50 pages whether it is real (it is)

Also, rogame later explained where the data comes from so this post isn't really helpful anymore
 
Last edited:

Jeffram

Member
Oct 29, 2017
3,924
You know what I find really interesting..
-We know there were two SKUs from MS for a while. They denied Lockhart's existence and su subsequently showcased the Anaconda console at E3, which we now know to be Series X.

-Andrew Reiner was sitting at this event and said devs told him the PS5 is superior to it.

-Devs couldn't have been mistaken, you know, getting confused between Anaconda and Lockhart, because MS were showcasing Anaconda, and still denied Lockhart's existence.

-Then we have Cerny saying specifically that PS5 has HW RT.. after much speculation from some it used software RT.

-We know 2019 AMD GPUs dont support RT. We also know AMD is known to pull tech from future products and incorporate them into console semi custom products, case in point the PS4 Pro. At the same time though.. RT isn't one of those things as their specifically saying it's a RDNA2 feature and wont work with any existing 2019 cards.

-Now we also have rumors of a big GPU being used in PS5, and what we have here isn't that...

-We also have Jason saying the machines are within 10%, or very little difference, and Klee saying the PS5 actually edges out the Series X..

-We also have words like "PREMIUM, Niche" etc being thrown around by Sony execs and $399 is anything but.

-Lastly... zero mention of RT being data mined... and literal words by Cerny saying the RT is Hardware accelerated.

I dont know guys... I have alot of doubt about these rumors, or datamined info.
If Sony is making a 36cu part with no next gen features, it's a PS4 revision and maybe can boost to 2ghz in amd testing.
 

Kibbles

Member
Oct 25, 2017
3,418
dors this mean that original Lockhart rumor from like a year ago was right about everything? They mentioned 12TF Anaconda and a weaker PS5
 

CosmicBolt

Self-Requested Ban
Member
Oct 28, 2017
884
If PS5 is 9.2TF and all the credible insiders are saying both consoles are very close in power, probably less than 15%. Then the most likely scenario is that XBSX is not 12TF. It's probably less than 10TF, this might be the reason Phill gave the over 8x xbox one and 2x xbox one x performance figure.

EDIT: But it's likely that XBSX is 12 RDNA TF and PS5 is somewhere around the number and 9.2TF is incorrect
 

momottolo

Member
Aug 20, 2019
95
Why this morning someone was talking about a simultaneous launch of PS5 Pro alongside PS5 ? Is it possible ?

Anyway i don't really see the news in XSX being more powerful of PS5 if Microsoft really went with two SKUs (and it's almost certain at this point, isn't it ?)
 

Evodelu

Alt Account
Banned
Dec 19, 2019
558
You should be asking chris 1515 in my opinion. My novice take is that they have a complete pipeline for RT thanks to the Caustic legacy. They've been making hardware RT for almost a decade at this point, and John Carmack thought their tech half a decade ago was solid.



I think there are two possibilities. One is that they always conceived of pushing out a non-final APU for developers to play with. This could be an off the shelf Zen 2, IO, and Navi dies, with potentially only changes to the IO die needed. Additionally, it could be pin compatible with the final product. This would enable PS5-like silicon in their hands up to a year before they get the real development chips.

As a thought experiment - why does Navi 12 exist? A Navi 10 like design with hardware updates and HBM memory. Who is the desktop consumer for a product with high memory cost? Unless the memory cost isn't actually that high anymore, the product is simply a leveraged version of something used elsewhere, or both.

The second possibility is that this is plan B. At some point in early 2018 (At the latest), they called an audible and decided that Navi 10 wasn't good enough. Could this be a reaction to what they thought MS was cooking, how the narrative of X vs. Pro was evolving, developer feedback (this would have been the time Cerny was touring developers), or AMD pulling some roadmap items in as they went through some tumultuous times? Who knows. Small snippets like Chinese forum posters alleging a hardware revision would cut it close for 2020 launch are potential clues, though I am hesitant to bite on those things too hard.

What I do know is that virtually every insider is saying PS5 and XSX are close, and most of those are sticking to that story after the XSX reveal. We know hardware RT is confirmed. Sony said it, Matt said it, Kleegamefan said it. One of our grandmothers probably said it. Klee has also said VRS is included, and everyone is suggesting >10TF. All of these things Navi 10 is not. It would be surprising to see this many people with other valid info turn up wrong on this.

And to make my lack of allegiances clear - I am team parity. I own Pro and X and love them both.
You're the most wholesome guy in this thread.
 

Deleted member 1589

User requested account closure
Banned
Oct 25, 2017
8,576
So yeah, I would think this is a devkit and not final. Too many things doesnt make sense. 16 1GB GDDR6 chips, RDNA1 with no RT, and at 2Ghz and so forth.
 

Deleted member 40133

User requested account closure
Banned
Feb 19, 2018
6,095
Why is everyone ignoring that rogame said Ariel/Oberon does not have RT? That is clearly something that ps5 has but this data mining says it doesn't
 

M3rcy

Member
Oct 27, 2017
702
not sure why did you edit this out


so PS5 is really using 16 chips of 1GB with 16bit each, 14gbps, so they end up with 448GB/s, what a mess. why even go for this GDDR6 layout, so confusing.


I believe each GDDR6 chip has 2 - 16 bit busses (as compared to the 1 - 32 bit bus used by GDDR5). So, it would be 8 - 2GB chips.
 

Arkham

Member
Nov 12, 2017
322
Is there like any concrete information in this thread?

Just this:

common-used-concrete-admixtures-845036-color-GIF-V1-5243aa2acb8d4a90bb1193b7faa04b48.gif
 

Deleted member 18951

User requested account closure
Banned
Oct 27, 2017
5,531
One Thing that I think is kinda reasonable that I see being posted here is the idea that earlier devkits were using GCN. Just makes sense to me giving how devkit Revisions work.

As in both Sony and MS? It's that confusing that I cant remember when the XSX dev kits were supposed to be getting sent out, was this a recent thing or have they had dev kits in the wild for a while now?
 
Status
Not open for further replies.