Baseless Next Generation Rumors with no Technical Merits [post E3 2019, pre GDC 2020] [XBSX, PS5]

Status
Not open for further replies.
I doubt it. Compression was already used on PS4 games (from Blueray) using a zlib decompression block (a real dedicated block then), they never ever stated the compressed numbers in their PR or elsewhere. To reach those conflated numbers Microsoft most probably use GPU decompression, as hinted by one MS insider on era, so that shouldn't be realistically useable for streaming, only loadings, just read closely their PR:



Anyways:
SSD -> using conflated compressed numbers from 2.4 to 7.
GPU -> using an incredible 25tf (RT equivalent) number, the new FP16 metric applied to RT

Seems like they needed to seriously conflate their numbers in those 2 metrics (not for CPU!), I wonder why ? because those numbers should already sound awesome, amirite ? ;)

They could have said 38 times faster then RTX2080Ti (one of current high end gpus), but they didnt, missed oppertunity there.
 
Dont get me wrong, I completely flabbergasted by anyone pushing the "25 TF" number without actually context.

I just dont see anything wrong with them giving out actual detailed specifications on the hardware, including the IO subsystem.
 
That 25 TF number is perhaps 'necessary' to 'quantify' the additional performance. How else can MS communicate what's effectively a qualitative difference? Well, pictures ideally, but I don't begrudge them trying to communicate relative performance by using a number that sort of does that. If comparing XBSX on screen with XB1X on screen graphics with the delta described by a number, it'd be 25 VTF (virtual teraflops) versus 6. Although then they ought to go the whole hog and use GCN > RDNA2 TFs and be maybe 35 VTFs? XBPU (XXbox Performance Units).
 
4 times faster then 2080Ti could have worked, or perhaps 38 times, depending on what's being done.
I wouldn't expect XBox gamers to know what a 2080Ti is. You'd want comparison with your existing platform that they already know. Comparing to XB1X, which is probably a desirable for a lot of XB1 owners but they aren't going to pay that for a mid-gen refresh, saying XB1X is 'four times faster than that other console you've got your eye on' gives a very communicable comparison.
 
I still go after data, not insiders. It doesn't matter if your wrong, or if i am. That's the point. Just as you are kicking on everything except an insider :)

No, I was kicking people who thought Github was fact and therefore anything that contradicted it was wrong (as if Github was proof any anything). And here we are, Github was wrong...which means all those that shat on insiders using the Github data were wrong to do so - regardless if the insider was/is right or wrong. This Tommy Fisher guessed/knew pretty accurately the XSX specs, he was closer than Github...regardless of if it was fact or guesswork.
 
16 Gbyte is not enough, no HbM2 or other Wide I/O Solutions is also very dissapointing, cleary not a generational Leap. Busarchitecture/Memorysystem is a big Bottleneck.
 
No, I was kicking people who thought Github was fact and therefore anything that contradicted it was wrong (as if Github was proof any anything). And here we are, Github was wrong...which means all those that shat on insiders using the Github data were wrong to do so - regardless if the insider was/is right or wrong. This Tommy Fisher guessed/knew pretty accurately the XSX specs, he was closer than Github...regardless of if it was fact or guesswork.

On the contrary, it wasn't wrong at all. Hate to dissapoint you, but both God and Gospell aren't of the cards yet ;)
 
16 Gbyte is not enough, no HbM2 or other Wide I/O Solutions is also very dissapointing, cleary not a generational Leap. Busarchitecture/Memorysystem is a big Bottleneck.
What's this based on, just wanting more?
What do you think would be min?
 
Github had 56CU @ 1.675 didn't it?

The github data was most likely a devkit, not final hardware, but resembling final specs in raw power. It was the full chip at 56CU's. Final hardware has 4 disabled for yields, final clocks adjusted to meet target specs (12TF).

Which is almost spot on?

The github data is from earlier then December. No idea about Sony, they could offcourse changed plans and upped to the 15TF that was leaked by a verified insider. Time will tell.
 
The github data was most likely a devkit, not final hardware, but resembling final specs in raw power. It was the full chip at 56CU's. Final hardware has 4 disabled for yields, final clocks adjusted to meet target specs (12TF).



The github data is from earlier then December. No idea about Sony, they could offcourse changed plans and upped to the 15TF that was leaked by a verified insider. Time will tell.

but devkits are nearly always more powerful with more ram and what not, not the same as the retail with lower clock and more CUs. The GitHub data was most likely the final console imo, I would expect a devkit to be the same clock but with no CUs disabled.
 
No, I was kicking people who thought Github was fact and therefore anything that contradicted it was wrong (as if Github was proof any anything). And here we are, Github was wrong...which means all those that shat on insiders using the Github data were wrong to do so - regardless if the insider was/is right or wrong. This Tommy Fisher guessed/knew pretty accurately the XSX specs, he was closer than Github...regardless of if it was fact or guesswork.
So guy who posted 10 wrong things for 1 right, after 12TF number was already confirmed, was more correct then document which showed :

56CU chip (which it technically is)
Actual confirmed chip codename - Arden
560GB/s
320bit bus
RT/VRS
Same 2.0GHz clock for PS5
Actual codename for PS5 chip - Oberon

Now, I already said this, with Xbox One X MS used full chip for dev kit and cut 4CUs for retail, so it could have easily happen again. But to throw away one leak because data from it was incorrectly assumed (because hey, there still are 3584 shader cores in XSX), and go with guy that posted 10 more wrong things for one correct one is...interesting.

Btw, if we are being pedantic - chip is 1.825MHz.
 
I see quality troubles of future Multiplatform games , many downports from the PC


Its a good Question we dont know how complex and Memory hungry upcoming Games will be in the Future. And will the Game developers take the trouble to optimize the games on the new Xbox to get around the restrictions?

I think you got it all backwards.
 
So guy who posted 10 wrong things and 1 right, after 12TF number was already confirmed, was more right then document which showed :

56CU chip (which it technically is)
Actual confirmed chip codename - Arden
560GB/s
320bit bus
RT/VRS

Now, I already said this, with Xbox One X MS used full chip for dev kit and cut 4CUs for retail, so it could have easily happen again. But to throw away one leak because data from it was incorrectly assumed (because hey, there still are 3584 shader cores in XSX), and go with guy that posted 10 more wrong things for one correct one is...interesting.

If the testlab leak was from summer 2019, was most likely long before final clocks and adjustments. Having spot-on to the number specs that long before is impossible, even for insiders, they couldn't have them either. It wouldn't surprise me if MS finalised the specs this month, and went to DF when it was all done. They are very pro-active in sharing hardware details and plans with us.

Edit:

Also, read his post, explains everything regarding the insider who guessed XSX specs.

https://www.resetera.com/threads/ps...se-read-03-16-staff-post.173318/post-29949494

From DrKeo

''I have to say, he had like 50% chance of guessing that. If you take 12TF for granted (because of the windows central leak), there aren't many ways of achieving that. You basically have two options:
1) 52CU @1.81Ghz.
2) 56CU @1.675Ghz.

Anything else doesn't make much sense. I mean, you can add two less likely options, 48CU @1.95Ghz, and 60CU @1.562Ghz but that's still well over 25% chance (considering the last two options aren't that likely). A lot of numbers have been thrown around, I mean I made a post about XSX memory having two speeds, 10GB @560GB/s and 6GB @336GB/s back in December which is spot on, but you don't see me claiming to be an insider. I just take the rumors and leaked numbers and play with them according to my technical knowledge and still I've provided better and more accurate "insider" information.''
 
Btw do you guys remember taiwan insider aquariuszi, he ended up pretty accurate about the SoC for XSX at least(he guessed ~350mm2), i was thinking he's wrong due to most people(including DF) guessed ~400mm2 for XSX, but nah.
Now let's wait for the PS5(he guessed ~300mm2).
https://pttweb.tw/s/3We5x
Scroll down and you will find this:
"Anubis現在已經沒有在做,可能是開發機代號 Arden沒意外應該會是要上市的版本 另外 Arden進度比較快,OBR還在初期驗證 而且Arden比OBR大了不只一號(350 vs 300 mm2)"
 
16 Gbyte is not enough, no HbM2 or other Wide I/O Solutions is also very dissapointing, cleary not a generational Leap. Busarchitecture/Memorysystem is a big Bottleneck.
Ha ha ha. Only you could look at this elegant machine and moan it's not enough. Your preferred $1000 machine is never going to happen. Go play on PC and be done with it, please!
 
Status
Not open for further replies.
Back
Top