Is this some kind of Spy movie?
Baseless Next Generation Rumors with no Technical Merits [post E3 2019, pre GDC 2020] [XBSX, PS5]
- Thread starter BRiT
- Start date
- Status
Target specs for devs and reality are 2 different things though. It's how we got the infamous WatchDogs downgrade. They were given a pretty serious alpha kit setup (intel + nvidia) with Xbox and it only came out to be a fraction of the power when the final silicon shipped when they switched to AMD + AMD
So are you expecting Sony targeting same or even higher price than Microsoft?
D
Deleted member 7537
Guest
I'm expecting them to have lower price and performance.
But I don't think it's related to the matter that I don't think we won't know the actual specs until the final reveals. When VGLeaks leaked PS4 devkits specs, PS4 was going to be a 4core CPU console with 2-4GB of GDDR5.
Well i'm pretty sure what we heard for PS5 is maximum performance if it's $100 lower than XSX unless you think Microsoft in 2019/2020 is same as 2013.
That said i always expect both target same price with 5% max difference(or no any difference at all), so recent rumors kinda surprise me, but if it's real i guess that's understandable(for price&performance balance).
D
Deleted member 11852
Guest
What's the source for this?
Durango kit starting out as intel and Nvidia
Ubisoft released the E3 watchdogs demo the E3 just prior to the dev kit being sold online on eBay.
So it’s time stamp is around the time in which developers would have been developing against this spec.
the features that were removed from the E3 demo were found on the PC build and enables with no stability issues. So the conspiracy is that they purposely hacked all variants of the game to ensure that the PC version didn’t get too far away from the consoles.
Last edited:
PS4 Pro has a boost mode, running with 18CUs (same number as PS4) and 911MHz (native clock of PS4 Pro).
OBR 36CUs (same number as PS4 Pro ) 2GHz (native clock) looks like the same way as boost mode of PS4 Pro.
IMO Gen 0, Gen 1 and Gen 2 are a series of BC testing.
Aquatiuszi (the Taiwanese leaker) said Ariel was arrived in the end of 2017 He also saw Sparkman layout and estimated it is half the size of Arden.
1. Basically sparkman in github is fake. How much fake info is there in github leak?
2. SONY had Ariel APU 2 years ago. This indicates Sony indeed has an early prototype.
1. Basically sparkman in github is fake. How much fake info is there in github leak?
2. SONY had Ariel APU 2 years ago. This indicates Sony indeed has an early prototype.
AbsoluteBeginner
Regular
https://www.extremetech.com/extreme...series-x-substantially-more-powerful-than-ps5
Interesting part on why Sony would go with 18/36CU design choice.
Interesting part on why Sony would go with 18/36CU design choice.
AbsoluteBeginner
Regular
Its not fake, its just that template for Arden is copied to Sparkman.
Might be the case, although that would not imply full chip is not 36CU, as doc actually reads in one of comment sections that "full chip is used". That, along with 256bit bus and fact that Arden Native mode is 56CU would tell me BC theory does not hold scrutiny.
I replace "fake" with "wrong". Basically sparkman in github is "wrong". How much "wrong" info is there in github leak?
Basiclly Aquatiuszi's info tells us SONY' has an APU of 36 ~40 CUs because Ariel in 2017 might use Navi 10 for GPU.
Therefore Ariel has no "hardware" RT just like the tests in the github.
And OBR A0/B0 (I dion't know if it is copied and pasted from other folders) is like the iteration of Ariel 2017, no "hardware" RT.
There should be a newer version of APU which adds "hardware" RT to APU. IMO it is the APU tested in October 2019.
So basically SONY has at least two different APUs: an older APU based on Ariel 2017, no hardware RT. The other is the newer version having hardware RT.
Last edited:
D
Deleted member 11852
Guest
I'm aware of the Durango hardware but you implied Microsoft pulled the performance rug from developers. What's the source for that - specifically, Microsoft telling developers that Xbox One would be a performance level of the devkit rather than what they shipped?
This sounds unbelievable.
AbsoluteBeginner
Regular
Its not "wrong" or "fake", its just a bit of a messy, incomplete template.
For Ariel/Oberon we can see progress of a chip. We can see BW improvements, we can see number of WGPs for full chip, we can see texture and pixel fillrate data which match with 2.0GHz 36CU chip and obviously full BC1 and BC0 results. Its not hard to make a relative safe conclusion of basic GPU data presented there.
For Arden/Sparkman we can see that it is likely later addition to repo, and there is only template of theoretical values that have been generated from which we can assume number of CUs (3584SPs) and bus width (320bit). In addition to this we can see that Arden and Sparkman have Xbox and Xbox One X BC template and that is it.
So, not wrong, just a bit messy and incomplete (but still, we can get alot of data from it).
Are we sure about that ? can we trust this leaker ? because this is really surprising IMO: 2017 ? It could mean:
- Navi 10 / Ariel could be the first RDNA GPU ever developed by AMD.
- The fact that RDNA is ISO compatible with GCN could be the direct consequence of BC requirements and tests for (and paid by) Sony.
- Ariel could be the PS5 planned for 2019 (1.8ghz, sweetspot-ish with 36CUs, 8.3 tf at 448GB/s). Then they decide a 2020 release date so they overclock the shit of it (following the advice of crazy Ken) and use faster GDDR6 chips in what is called Oberon B0 (2ghz, 9.2tf at 550GB/s).
Ariel = 2019 PS5
Oberon = 2020 PS5
With that bandwidth and GPU clocks of Oberon, there should not be big differences of performance between Oberon and the relatively bandwidth starved XSX.
Nah he didn't said that.
https://pttweb.tw/s/4airz
"Sparkman很小顆 只有Arden一半大一點而已 但從layout來看也應該是SoC 要期待它有NG console的效能不太實際 所以就有兩種可能 1. 串流主機,就之前傳的Lockhart 2. 掌機"
"Sparkman is very small, only little more than half of Arden, but from the layout should still be SoC, shouldn't expect it have Next-gen console's power, so there is 2 possible 1.streaming console, 2.handheld"
"然後Ariel從2017年底就出現在神教 我覺得他跟索索沒關係 跟OBR的時間差太多"
"Then Ariel already shows up in AMD at 2017, i think it's nothing to do with Sony, time difference was too large with Oberon."
"之前有講過Arden比OBR大（約15%） 這少掉的面積確實可能是RT ASIC"
"I said before Arden is bigger than Oberon(15%, see his posts at Oct and Nov), the smaller part could possible due to RT ASIC"
Anyway someone said he do have info but his predict about consoles are not good, so maybe just take the info part.
No source on that. I was implying what you wrote on your first sentence. Any sort of lack of communication with this hardware could have resulted in over ambition. Not that MS specifically told Ubisoft that this was the target to go against for final.
Unfortunately no knowledge on target specs back then or how communication was handled. I wasn’t into this type of thing or join the board until 4 month or so before launch
AbsoluteBeginner
Regular
10% difference is a difference of around ~50mm².
320bit bus will be ~16mm² extra over 256bit bus, how much will extra 20CUs be? If we go by Github leak, it will not feature additional ROPs, so 64 as Oberon.
320bit bus will be ~16mm² extra over 256bit bus, how much will extra 20CUs be? If we go by Github leak, it will not feature additional ROPs, so 64 as Oberon.
- Status
Similar threads
- Locked
