The reference card has a Vapor Chamber and it performs worse than the Asus Strix. In clock, in noise and in temps!
The reference card has a Vapor Chamber and it performs worse than the Asus Strix. In clock, in noise and in temps!
If RT is included in the cores how could 365-380mm² not be more powerful then 9.2tf?
It isn't wrong because Reiner was already talking about what was the case at the time of his Tweet.
what did he say?
I am using 1755Mhz cause that is AMDs advertised "game clock" for their 5700XT GPU. Which is said to be the clock people should be expecting most times with that card.Interesting points.
Why only 1700 - 1755mhz? Why not higher like 1800mhz?
Eurogamer suggested 48cus (4 disabled) 44cus usable would be possible in a 380mm2 SoC.
Things will be really interesting (and brutal) if PS5 has a small GPU advantage but scarlett has a CPU advantage.
I wonder what GPU + cooling system would be cheaper
36cus @ 2000mhz
Or
40cus @ 1755mhz
I mean, they could both end up really close with PS5 having a slight edge. PS5 ending up stronger (or vice versa) doesn't have to mean the other will be that much weaker.Well see.
Doubt it.
Scarlett being 8TF box not gonna happen.
Some users calculated the size based on GDDR6 Memory size from the Scarlett Reveal.
less than 4x the PS3's GPU clock (550Mhz) is not too much guys!
what did he say?
For those who are so curious about the TFLOPS of the PS5 -- 9TF (~1080 performance)
I want to add something more to my Debby Downer Persona:
I think 5700 chips are down binned chips that would be not capable to reach clock speeds of a 5700XT. For me that is the reason you see lower clocks on 5700 in general (reference design & AIB designs).
To put this into a console perspective: This would mean a lot of binning for consoles. The more binning the more cost and the less yield. IMO not a good idea for a console.
My Nostradamus Persona tells me that the #2000MhzBelieversClub is out for a disappointment ...
My Nostradamus Persona tells me that the #2000MhzBelieversClub is out for a disappointment
Yes.
My point is, the preview can be anything...
A 52CU @ 1.0GHz, a Vega 56 GPU. We don't know.
But the Scarlett SOC is bigger, much bigger. Like 40~60mm2 bigger.
Doubt a 380mm2+ SOC will be weaker than 9TF.
Even if Scarlett is 44CU GPU, they can clock the GPU higher than 1.6GHz.
The rumour from Reiner is wrong for sure.
Yup.
But the size alone is bigger, 60mm+
No L3 Cache is this size.
Don't ask me, but yields should improve. If quality of the chips improve too I do not know.I presume the binning improves as the manufacturing process matures? Or no?
Yield improves, yes.I presume the binning improves as the manufacturing process matures? Or no?
Reiner/Colin rumour?
No evidence at all Scarlett is weaker.
Probably the rumour is wrong based on early preview devkits.
If we are looking at Scarlett video, the SOC around 380mm2+... Scarlett will be more powerful for sure.
Because the early preview Devkits.
Like many said.
Reiner/Colin rumour is wrong.
Or half of the 380mm+ SOC is Raytracing/L3Cache.
Or MS don't give a fuck and will release a Lockhart 4TF box and not the Anaconda.
lol
GTX 1080 level of performance in late 2020 is disappointing asf!
my entire point is that a 380mm2 Scarlet is going to be MORE powerful than a 9.2 tflops. im not sure how you missed that.
Okay, I will identify myself as a Debby Downer because I still doubt we see 2000Mhz in a console.
The reason is simple: If I look at the cooling requirements of current board partner cards to actually reach those clocks, I cannot imagine it.
Just at this stage the only AIB card that is actually achieving 2000 Mhz air cooled on Navi is the ASUS Strix and if you have not seen that card, it is big and has 3 fans! The card also still gets loud unless you select the Silent BIOS, but then temps go up quite a bit.
So no, I don't see it in a small form factor a console is.
I am using 1755Mhz cause that is AMDs advertised "game clock" for their 5700XT GPU. Which is said to be the clock people should be expecting most times with that card.
Well, if Scar has 44CU usable @1755Mhz that would make it sit at 9.8TF to the PS5s 9.2TF. I can even at that point see MS pushing for ~1800Mhz to hit 10TF. Or its 40CU @ 175Mhz which yields 8.9TF in which case I can also see MS pushing for 1800Mhz to land at 9.2TF too.
Either way whatever the case there isn't going to be a lot between the two.
I mean, they could both end up really close with PS5 having a slight edge. PS5 ending up stronger (or vice versa) doesn't have to mean the other will be that much weaker.
The arguments based on the die size of a flimsy cg mock up seem flimsy at best...Can't believer you are putting so much emphasis on that. That is literally grasping at straws.
Let's put this one to rest. In a tweet, Reiner clarified that what he is hearing is that ps5 was stronger not based on devkits but the TARGETS each one set. So apparently the target for Sony was higher than ms
9.2tf navi is above rtx 2070 levelCan't be. 9TF GCN is crushed by the GTX 1080. Even 12TF Vega 64 loses to the GTX 1080 at times.
The reference card has a Vapor Chamber and it performs worse than the Asus Strix. In clock, in noise and in temps!
whats wrong with what i said and what does that have to do with you assuming that i said scarlett < 9.2 tflops.Oh?
You said Sony can also change clocks? Implying that the ps5 could grow in power just as much as scarlett could.
Can't be. 9TF GCN is crushed by the GTX 1080. Even 12TF Vega 64 loses to the GTX 1080 at times.
If Scarlet is weaker, there will be meltdowns. The power narrative is something Microsoft is actively courting.If PS5 is 2Ghz what the hell is Scarlett doing to be weaker?
That is an interesting development, which probably means we are getting something in the 36-44cu range.
This makes sense. Next gen won't be competing with high end 2020 cards. Best we'll get is around 5700 XT performance.
I would say the video that was made to create hype is the odd man outOr Reiner rumour is totally wrong, or the Komachi info's are totally wrong, or the Scarlett video is totally wrong.
And 9.2tf with 36CUs should be slighlty better (in most games) than 9.2tf with 40CUs. We need benchs of 5700 OC near 2ghz
It is for sure not the solution to everything when it comes to cooling.It's almost like vapor chambers aren't this magical cooling thing.
It will be interesting to see how the fuck Sony will cool this thing. I mean, it's not impossible, there are gaming laptops dissipating more than the 220W tops this thing will be putting out, and they do so in cases that are relatively more cramped than your average console.
It's almost like vapor chambers aren't this magical cooling thing.
Yes.
My point is, the preview can be anything...
A 52CU @ 1.0GHz, a Vega 56 GPU. We don't know.
But the Scarlett SOC is bigger, much bigger. Like 40~60mm2 bigger.
Doubt a 380mm2+ SOC will be weaker than 9TF.
Even if Scarlett is 44CU GPU, they can clock the GPU higher than 1.6GHz.
The rumour from Reiner is wrong for sure.
Yup.
But the size alone is bigger, 60mm+
No L3 Cache is this size.
If i remember right sony had a patent on cooling but i can't remember what it was for .
Some methods to dissipate heat through a package and a PCB - that is, cooling a chip from both sides. It's pretty crazy shit and the kind of wacky stuff I totally can see Sony pulling.
Yes.
That's not the problem.
The problem is... Why @60mm2+ bigger than PS5?
No L3 Cache is this size.
More Raytracing?
More CPU Cores?
More GPU Cores?
SRAM 2.0?
Can PS5 achieve 2.0GHz with more than 36CU?
All the info's we have don't match.
Or Reiner rumour is totally wrong, or the Komachi info's are totally wrong, or the Scarlett video is totally wrong.
The Scorpio CG was 100% right.
Yes, we know that.
Target and final specs are different things.
We need to know if Microsoft is going with a better Lockhart or Anaconda.
NiceI know I am. 🤣 I'll reluctantly get the PS5 if it turns out to be more powerful, though.
whats wrong with what i said and what does that have to do with you assuming that i said scarlett < 9.2 tflops.
we have literally seen the ps5 clocks jump from 1 to 1.8 to 2.0 ghz in a matter of six months. i am pretty sure Phil didnt reveal the tflops count because he is looking to do the same. but that doesnt mean cerny has been sitting on his ass this entire time. Reiner said devs told him PS5 is more powerful based on what they know. if you guys are holding out hope that MS will increase the clocks to beat the PS5, great by all means but look at the recent developments and know that Sony could do the same.
👍🙏Frankly I don't care which one is more powerful, I will get both eventually, both platforms have game devs that are to good to miss.
Scarlet's SOC is bigger than PS5's SOC. You saw somewhere PS5s SOC size? Like you said: Scarlett WILL BE more powerful for sure".
Any source on that or proof?
I don't get what you're saying. If one is targeting higher than the other....it stands to reason that the one with higher target is going to be looking at better tech
Oh for sure, no console will be "weak".If Scarlet is weaker, there will be meltdowns. The power narrative is something Microsoft is actively courting.
I think Scarlet will be more powerful but this PS5 sounds delicious.
No.
That't the point.
If the Komachi thing is right, PS5 is 36CU 100%. The only way you can achieve 2.0GHz. Even so, some users don't believe it, like Colbert
They can't clock a console SOC GPU @ 2.0GHz with 44, 48, 52CU, with 36CU the Die size can't be bigger or the same size as @380mm2. So the Scarlett video still valid.
If Reiner is right and Komachi is right, we need to explain why the Scarlett SOC from the video is so big and less powerfull. So the video is fake/wrong?
RX 5700 XT Die Size (40CU): 251mm2
PS5 with 40CU / 4 Disable / 36CU + Zen 2 ( @70mm2 ): @321mm2?
Scarlett SOC from the video: @380mm2?
60mm2 for what? 320-Bit Bus? More CUs? More L3 Cache? More Raytracing?
I'm in.Team 12.9 tflops (which means mostly me and Xeontech),
It's with great sorrow and regret I am announcing my departure from Team 12.9 tflops. In light of recent events, our crazy 12.9 tflops prediction must be reevaulated and reassessed. 12.9 Tflops at 2.0 Ghz is simply impossible. I can no longer sanction this buffonnery.
Which is why I am now reviving Team 14.2 tflops. 56 CUs at 2.0 Ghz = 14.2 tflops. Recruitement is underway. We are currently accepting all applications regardless of race, gender or platform affiliation*.
All Avatar bets will be honored retroactively.
*Nintendo fans need not apply
If Reiner is right and Komachi is right, we need to explain why the Scarlett SOC from the video is so big and less powerfull. So the video is fake/wrong?