Now going back to not posting in these next-gen threads anymore because you all keep taking what I say out of context and it is infuriating!
Yep. Colbert knew the whole time it seems. And now he's banned. Hopefully he's let back in sooner than later.Props to you and Colbert for sticking to your guns. I've always said that 36 CU's was possible, regardless of Github or not. I'm curious whether you think a 2.23 GHZ variable clock speed was Sony's original plan or did they get surprised by MS's plans and decided they had to go for much higher clocks.
he said they will use it only if the cpu isnt running at max speeds. you cant rely on that.we don't know this, radeon on rdna1 are almost always above basic frequency but indeed, 2.23ghz? waiting for some df benchmarks
According to Dark1x, XSX is quiet.
So any idea why they went narrow and fast, given this is contrary to traditional chip design? Did BC force 36 CUs? Or did the relatively larger cost of fabbing on 7nm lead them to build a smaller chip?
About 400 Mhz, 2.23 vs 1.825This is really weird...2.23 Ghz...that is insane.
That's literally over 500 Mhz higher than the XSX GPU. That will lead to big advantage in all those other aspects of GPU processing outside of the ALUs. This is why 10.3 vs. 12.1 is not really a good measure of performance. If I had to guess, they will perform probably essentially the same in the real world.
Uh, no, because there were no mods who said that.So, github is right. Are there a couple of mods who get banned for calling it garbage and wrong?
there is a bay to add going by what Cerny said. gimme a bit and I'll try and find the exact wording.Did anyone take away from the SSD talk the the SSD is replaceable? Or is there a bay to *add* and SSD?
Is a question of price. Unless the SSD drive tech is crazy expensive, the PS5 should be considerably cheaper than the XSX.36CUs is bizarre considering it didn't even guarantee 100% PS4 BC.
I didn't understand the logic behind this decision.
He said only in rare instances would clock speed have to be dropped 10%. Are you saying that's a lie?
games that load in 4 seconds on XSX will load in 2 seconds on PS5, assume there's no other overheads.Alittle confused with ps5's ssd performing around 8 gbs vs xbox 4 gbs how much of a difference in real world will you actually see
I am excited for the hardware as it is indeed very good (if it ends up being 399) but boy, this was a stupid way for them to reveal."Most exciting hardware in 20 years."
Not really just the SSD.
I don't doubt that Cerny is a genius, but I think they may have made some wrong moves here:
- The SSD sounds amazing, but I think something more standard may have been a better play
- Variable freq for CPU and GPU means that 10.3TF is best case scenario
- Overinvested in 3d audio
I think MS did it better. I'm saddened.
Austin Evans also said it ran all day and stayed quiet and cool
Not really, no. It's certainly an interesting choice.So any idea why they went narrow and fast, given this is contrary to traditional chip design? Did BC force 36 CUs? Or did the relatively larger cost of fabbing on 7nm lead them to build a smaller chip?
Let's be honest, these threads don't look good in retrospective, in many many ways...Yep. Colbert knew the whole time it seems. And now he's banned. Hopefully he's let back in sooner than later.
If you think it'll be loud you skipped a whole section of the video.you can expand ssds that won't exist at launch and will be prohibitively expensive when they do.
this thing is going to run expensive, hot, loud, and slow
DF said that the unit that was running the entire time they were at Redmond was really quite
What are your thoughts on how narrow and faster vs broad and slower will translate into performance? Is there situations where a higher clocked 36 CU GPU will outperform a 52 CU lower clocked one, especially when the disparity is in the order of 400+ Mhz? Or will that never be the case.
OK, thanks. I suppose another factor could be retaining similarity to whatever RDNA1 predecessor chip they might have been working with - i.e., Ariel.
Cerny showed 16 GB GDDR6, I didn't catch the speed.Lol GitHub was it, it just ended up being clocked higher than tested.
Cpu
8% faster xbox
Gpu
15% faster xbox
Ssd
2.3x faster ps5
Ram
Both have 16GB GDDR6
Xbox
10GB 576
6GB 336
PS5
Still unknown(?)
BC
Xbox
Every game works on xbox from day 1
PS5
Every game will work eventually, but expect the top 100 at launch(?)
he said they will use it only if the cpu isnt running at max speeds. you cant rely on that.
LMAO. 9.2 tflops confirmed.
I love how they say its actually 10.3 tflops when we know it will almost never hit the 2.23 ghz clock.
Colbert DukeBlueBall DrKeo I was wrong. you guys were right. github was real. Well played.
Klee, jschreier , BGs , osiris, thanks for giving us false hope for over a year. lmao.
@Team14tflops Time to buy Xbox Series X. 12 rdna 2.0 tflops will do.
But doesn't that also mean RT is going to be rather gimped, relative to XSX?Is a question of price. Unless the SSD drive tech is crazy expensive, the PS5 should be considerably cheaper than the XSX.
Yeah, I caught that in my next post. Thanks.
Jason said in Splitscreen a few months ago that the 10.7TF comment was a guess, not from knowledge, but I was too lazy to dig that up. And hey, he was right, PS5 wipes the floor with Stadia.jason said they were aiming to beat stadia number. 9.2 is not even close to that. 10.3 is basically never going to happen in game.
klee said ps5 specs were higher.
BGs and osiris were 12+ tflops.
This was my guess months ago when people said "2Ghz doesn't make any sense" and it's still what I think right now. Sony had already designed a 36 CU chip, probably because of BC limitations and a 399$ goal, but at some point Sony found out MS is going for 12TF. No one is redesigning their APU in early 2019 when the console is aiming to launch at the end of 2020, so all they could do was stepping in order to raise the clocks. And we even saw the process with Gonzalo (Ariel A0?) @1.8Ghz, Flute + Github (Oberon B0) with 2Ghz and now the final PS5 with 2.23Ghz. That's probably at least one stepping out of the extra 3 stepping Sony had from Oberon B0 to E0.Props to you and Colbert for sticking to your guns. I've always said that 36 CU's was possible, regardless of Github or not. I'm curious whether you think a 2.23 GHZ variable clock speed was Sony's original plan or did they get surprised by MS's plans and decided they had to go for much higher clocks.
If you think it'll be loud you skipped a whole section of the video.
Don't be foolish, RDNA 2 flop does not equal to GCN flop, Cerny said it himself.jason said they were aiming to beat stadia number. 9.2 is not even close to that. 10.3 is basically never going to happen in game.
My first prediction for PS5 Pro:
CPU: 8c/16t max freq 4.0 GHz
GPU: 21.7 TF, 72CU, max freq 2300 MHz
RAM: 20-24 GB, 640 - 816 GB/s
SSD: 2 TB at 6.0 GB/s (raw)
Launch 2023 or 2024
So, Unified pool vs separate. 360 vs PS3 again, with with roles reversed!? :O
This is one of the rare moments on Era, thank you for that. I expect you are one of the only ones that will stand corrected in hindsight.LMAO. 9.2 tflops confirmed.
I love how they say its actually 10.3 tflops when we know it will almost never hit the 2.23 ghz clock.
Colbert DukeBlueBall DrKeo I was wrong. you guys were right. github was real. Well played.
Klee, jschreier , BGs , osiris, thanks for giving us false hope for over a year. lmao.
@Team14tflops Time to buy Xbox Series X. 12 rdna 2.0 tflops will do.
256 bit bus at 448GB's