• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Next-Gen PS5 & XSX |OT| Console tEch threaD

Status
Not open for further replies.
DemonCleaner

DemonCleaner

Member
CrustyBritches said:
I've had a running bet with SonGoku that the 5700 XT will beat PS5 performance.
Click to expand...

SonGoku said:
wft? I thought the bet was a stock 5700XT (1950MHz)/RTX2070
4
Click to expand...

R600 said:
We'll place our bets. Mine are well known by now, so are yours, so lets see 😊
Click to expand...

i wanna join with a side bet.

main bet is still that at least one of the consoles will have 56 active CUs.

side bet is that clocks will be no where near 1.8ghz [because of power ...d'oh] but like 1.5-1.6 though gonzalo leak was true and the "/18" means something completely different.
 
Last edited:
pawel86ck

pawel86ck

Banned
R600 said:
We dont know how RDNA acts in syntetic benchmarks. We do know Vega56 ~ 2070 in 3DMark, yet 2070 is much faster GPU on average.

One thing I saw few days ago was syntetic bench with 5700 being 3% of RTX 2070, yet its only big Navi XT, with 2TF more that is competing with 2070.

My point is, 20K is great score, but with Zen2 6 core @ 3.4GHZ beating same specced PC with 2700X, it might be that Zen2 scores a bit higher then old Zen+ and AMD continues with solid synthetic scores that get outperformes in actual games.
Click to expand...
Even if Navi RDNA is indeed stronger in synthetic benchmarks it's unrealistic to expect 8.3TF Navi beating 12.5TF Vega 64 (It's way above 25% IPC performance gain) and especially if we keep in mind PS5 drivers for windows and 3dmark should be unoptimized (while vega64 is very well optimized for 3dmark and windows). If Gonzalo score is legit (and you clearly believe that's the case), then we are looking at much more capable GPU than your 8.3TF estimations. SonGoku estimations (11-12TF) make much more sense if you really want to take Gonzalo 3Dmark score seriously.
 
CrustyBritches

CrustyBritches

Gold Member
ArabianPrynce said:
What kind of performance are expecting out 8tflop Navi. I’s it really that good? Curious to know why people are invested in double digits tflops
Click to expand...
Vega 56(OC)/Vega 64/GTX 1080/GTX 2060 performance. We've already seen the benchmarks for 5700 Pro. Take that as best case scenario with optimized titles.
index.php


I look at benchmarks to see where that lands depending on the game. Metro Exodus, for instance, where 6% faster is just nipping at the heels of the GTX 1080 and Vega 56.
 
xool

xool

Member
DJ12 said:
So you have got a Zen 2 Apu with Navi you and under clocked it, my mistake.

...
Click to expand...

Mudslinging aside - the idea if an underclocked large die size Navi (assuming it performs similarly to Vega) is interesting - the evidence is that AMD is moderately pushing clocks (and thermals) in their PC cards to be competitive on FLOPS not FLOPS per watt.

People aleady sort of expect console parts to be at lower clocks that PC cards .. but by how much ?

Assuming 7nm+, 6nm, etc are just around the corner, and as easy as TSMC (and maybe even counting on 5nm to give a break after ~1 year- just like during the PS3/360 gen getting its ass saved by v early 65 and early 45nm) .. I wonder if Sony/MS would go for an unusually large chip by console standards, but very conservatively clocked - getting both FLOPS and FLOPS/Watt at the cost of a large piece of silicon ..
 
Lort

Lort

Banned
Aceofspades said:
What made it a dumpster? Last I remember it was running all games better than Xbox one, by a huge margin. That what matter to me.

I don't care what a bunch of hackers are trying to achieve here, the hacking community is still butthurt by Sony's removal of "other OS" on PS3...It was funny seeing them cry over that feature 🤣
Click to expand...

So i guess your loving the xbox one x then....

So your logic is .. hackers were so pissed off at Sony that they convinced Mark Cerny to put the HDD through a USB layer to reduce performance? Or are you arguing they infiltrated the EU to set power standby standards so low that Sony had to include a whole seperate SOC?

You have no rational point .. your just reacting.
 
SlimySnake

SlimySnake

Flashless at the Golden Globes
CrustyBritches said:
DF says that a smaller chip is less costly and easier to cool. They mentioned the outside chance that it could also be built in a way to make BC easier.

Perhaps. I think Gonzalo is it and it is powerful, but that could a way bigger chip than DF expects. I think Navi just has better clocks than Polaris and Vega. Pretty much every GTX 1060 could go over 2GHz. Maybe Navi doesn't mind 1.8GHz?
Click to expand...
oh that makes much more sense now.

dammit. 36 cu and 8 tflops it is then.
 
SlimySnake

SlimySnake

Flashless at the Golden Globes
ArabianPrynce said:
What kind of performance are expecting out 8tflop Navi. I’s it really that good? Curious to know why people are invested in double digits tflops
Click to expand...
vega 56 more or less. 10 gcn tflops. not bad, but not great. RT will tax it even more. RTX 2060 basically.
 
Lort

Lort

Banned
ColdToffee said:
The Southbridge CPU is ARM, and at the time windows for ARM wasn't even a thing. How can you say it is windows compatible? Why are you judging a console board from a windows pc pov?
Click to expand...

Sorry i should have provided some context as it was a few pages back.. there was a discussion about Playstation 5 booting ( in dev test mode) windows.. then someone said that the PS3 could run DX11 ( which is cant) .. then the post i was replying to said that the ps4 couldnt boot windows because of the SOC south bridge ( which is correct)... after i watched the video it became obvious that Sony wouldnt do that design again. The PS5 im suggesting will contain a more normal southbridge design .. will not have HDD via usb and will have memory and IRQ mapping in the same way as most “IBM compatible pcs”. This means the PS5 dev kits will likely be able to boot windows. Obviously they wont support this officially in any normal capacity... but potentially the windows run benchmark might have been run as originally suggested.
 
SonGoku

SonGoku

Member
R600 said:
Its calculated for 64mm² (256 bit bus). For 384 its 32mm² more.
Click to expand...
look up your post i replied to 🤦‍♂️ffs:
R600 said:
PS4 die size = 348mm²
PS4 CPU + GPU = 70 + 212 (282mm²)
66mm² remaining is IO / 256 bit bus
The IO and 256 bit bus already accounted on the GPU 212mm2 die size
PS5 die size = ???
PS5 CPU + GPU(40CU) = 75 + 251 + ~10RT (336mm²)
IO + 256 bit bus = 38mm² + 64mm² (102mm²)
The IO and 256 bit bus already accounted on the GPU 251mm2 die size
Click to expand...
R600 said:
For system with 384bit bus we are looking at ~370mm² without adding any new CUs and front end
Click to expand...
Lets see:
75mm for CPU
97.92mm2 for 384 bit bus (24GB)
46.66 mm2 for 2SEs (ROPs, cache, etc.)
IO 20.72mm2
10.85mm2 GPU CC (CP, ACEs etc.)
97.8mm2 32DCUs with RT (40CUs)
Total
348.95mm2 on 7nm with a fat 384 bit bus
279.16mm2 on 7nm EUV
R600 said:
without adding any new CUs and front end, so ~400mm2+ is safe bet.
Click to expand...
lol how did you come to that conclusion, you are way off
R600 said:
So system with 40CU, Zen2, RT and 256bit bus would be ~340mm².
Click to expand...
75mm for CPU
65.28mm2 for 256 bit bus
46.66 mm2 for 2SEs (ROPs, cache, etc.)
IO 20.72mm2
10.85mm2 GPU CC (CP, ACEs etc.)
97.8mm2 32DCUs with RT (40CUs)
Total
316.31mm2 on plain 7nm
 
Last edited:
R

R600

Banned
SonGoku said:
75mm for CPU
65.28mm2 for 256 bit bus (24GB)
46.66 mm2 for 2SEs (ROPs, cache, etc.)
IO 20.72mm2
10.85mm2 GPU CC (CP, ACEs etc.)
97.8mm2 32DCUs with RT (40CUs)
Total
316.31mm2 on plain 7nm
Click to expand...
I know I said I wouldn't post anything anymore, and this REALLY is a last time but your prediction on Zen2 + 40CUs on 7nm...316mm²

Now, where have I seen this before...

monolithic die ~22.4mm by ~14.1mm
Click to expand...

Lol 😂

 
Last edited:
Geki-D

Geki-D

Banned
I don't understand anything in this thread anymore.
SonGoku said:
75mm for CPU
97.92mm2 for 384 bit bus (24GB)
46.66 mm2 for 2SEs (ROPs, cache, etc.)
IO 20.72mm2
10.85mm2 GPU CC (CP, ACEs etc.)
97.8mm2 32DCUs with RT (40CUs)
Total
348.95mm2 on 7nm with a fat 384 bit bus
279.16mm2 on 7nm EUV

75mm for CPU
65.28mm2 for 256 bit bus (24GB)
46.66 mm2 for 2SEs (ROPs, cache, etc.)
IO 20.72mm2
10.85mm2 GPU CC (CP, ACEs etc.)
97.8mm2 32DCUs with RT (40CUs)
Total
316.31mm2 on plain 7nm
Click to expand...
This looks like a cat jumped on your keyboard to me. :messenger_grinning_squinting:
 
SonGoku

SonGoku

Member
R600 said:
I know I said I wouldn't post anything anymore, and this REALLY is a last time but your prediction on Zen2 + 40CUs on 7nm...316mm²

Now, where have I seen this before...
Lol 😂
Click to expand...
lol that just shows anybody who knows how to estimate die sizes faked it right after amd computex. Everybody who believed that leak was predicting 40CUs
But if that's the hill you wanna make your stand on: 316mm on regular 7nm

See ya at the reveal :)
 
Last edited:
R

R600

Banned
SonGoku said:
lol that just shows anybody who knows how to estimate die sizes faked it right after amd computex. Everybody who believed that leak was predicting 40CUs
But if that's the hill you wanna make your stand on: 316mm on regular 7nm

See ya at the reveal :)
Click to expand...
Issue is, this was May 21st. Before Computex and before E3.
 
Last edited:
CrustyBritches

CrustyBritches

Gold Member
SlimySnake said:
oh that makes much more sense now.

dammit. 36 cu and 8 tflops it is then.
Click to expand...
Could be hardware RT on top, too. Aside from 2060 the other chips don't support it. I see SonGoku's rationale and it makes sense to me I'm just doubtful it will get us up to 56CUs/~11TF(Navi). I'm happy with Vega 56/64 performance. I know some people were wanting a long term future proof console at $499, hoping for ~2080 performance. All the better if that happens. It's all about reading different opinions and learning for me, I play PC primarily and plan on having both consoles. I think both consoles will be like 10-15% apart in power.
 
Last edited:
R

R600

Banned
SonGoku said:
How can i tell the date on reddit? all it says is one month ago
Regardless of it being before or after, it being in close proximity is suspect

lol this might come as a surprise to you but im not glued to gaf 24/7
Click to expand...
www.resetera.com

Next-gen PS5 and next Xbox speculation launch thread - MY ANACONDA DON'T WANT NONE

I think some do. And those who do probably dont want to be the source of info, but instead just confirm the ones that are true. I wont be surprised if some journalists have a gist on the specs of both consoles, but can't really risk posting about it because of the lack of credible sources that...
www.resetera.com www.resetera.com

Its 20th of May actually.

Close proximity? Perhaps, but before Computex and E3 no body knew design of Navi CUs and the fact that CUs are so much bigger. You have to say, this is interesting coincidence.

Posting this again:

The profile also states "May 21st 2019" as cake day but it's May 20th so the user is probably from a timezone where the 21st started a bit over an hour ago... or that's how I believe this stuff works. I'm pointing it out because what if this is someone who works where development kits are being made/manufactured?
Click to expand...
 
SonGoku

SonGoku

Member
R600 said:
Close proximity? Perhaps, but before Computex and E3 no body knew design of Navi CUs and the fact that CUs are so much bigger. You have to say, this is interesting coincidence.
Click to expand...
The date is suspect, someone could have gotten access to 5700 info and took it from there, it can also be a lucky shot in the dark. The possibility of Sony making a 7nm 316mm2 chip on 2020 is null.

The only way i see that size being real is on 7nm EUV
 
R

R600

Banned
SonGoku said:
The date is suspect, someone could have gotten access to 5700 info and took it from there, it can also be a lucky shot in the dark. The possibility of Sony making a 7nm 316mm2 chip on 2020 is null.

The only way i see that size being real is on 7nm EUV
Click to expand...
Doubt someone would get die size info 3 weeks before official 5700 gets presented, and then proceed to post completely technical rumor describing PS5 alleged dev kit and not even touching on 5700.

Maybe someone actually has worked at manufacturing and knows PCB/chip properties? Occams razor and all that, coincidence is crazy if you are fair, because back in May everyone thought Navi CU = GCN.

Funny thing is, rumor is not even talking about specs, only whats on dev kit PCB and die size of the chip.
 
Last edited:
Insane Metal

Insane Metal

Gold Member
CyberPanda said:
Good. Competition is healthy, and will push Intel to do better.
Click to expand...
Interesting to see the single core performance more than 20% ahead... of course it's overclocked but apparently in normal clocks it's very close, something AMD was always a lot behind.

Edit: also that is probably an engineering sample since it runs at 3.3Ghz base and the final chip will run at 3.5Ghz. So the final product might actually fare better than in this leak.
 
Last edited:
DJ12

DJ12

Member
R600 said:
Doubt someone would get die size info 3 weeks before official 5700 gets presented, and then proceed to post completely technical rumor describing PS5 alleged dev kit and not even touching on 5700.

Maybe someone actually has worked at manufacturing and knows PCB/chip properties? Occams razor and all that, coincidence is crazy if you are fair, because back in May everyone thought Navi CU = GCN.

Funny thing is, rumor is not even talking about specs, only whats on dev kit PCB and die size of the chip.
Click to expand...
Damn, did I nap that long, has it really been a few months, or is someone an attention seeking drama queen....
 
SonGoku

SonGoku

Member
R600 said:
Doubt someone would get die size info 3 weeks before official 5700 gets presented, and then proceed to post completely technical rumor describing PS5 alleged dev kit and not even touching on 5700.
Click to expand...
Not touch but access to pr release documentation would be enough
R600 said:
Maybe someone actually has worked at manufacturing and knows PCB/chip properties? Occams razor and all that, coincidence is crazy if you are fair, because back in May everyone thought Navi CU = GCN.
Click to expand...
Maybe or someone knowledgeable enough who lurks obscure tech info made a lucky educated guess, all that was posted could be found publicly except 5700 size
These are die size estimates and the RT part is based on nvidias cards, the hypothetical final product won't be an exact match so it being so close to the estimate should ring alarms.
Its SSD info also contradicts Sonys patent and it being confirmed to be custom

But to toy with the possibility of it being a 7nm EUV spec:
75mm for CPU
65.28mm2 for 256 bit bus
93.32mm2 for 4SEs (128ROPs, cache, etc.)
IO 20.72mm2
10.85mm2 GPU CC (CP, ACEs etc.)
156.48 mm2 32DCUs with RT (64CUs)
Total
421.65mm2 on 7nm
337.32mm2 on 7nm EUV

Mind you this is with 128ROPs
 
Last edited:
Ovech-King

Ovech-King

Gold Member
Benchmark on ryzen 2700 3.2 ghz base clock (even more powerful than the leaks) . Again , if both Microsoft and Sony are striving both a 4k 60fps baseline (yes I'm assuming but still) , here's why I believe we're in for more power than most thinks



IYjeDoE.jpg

Numbers on the right is average fps and those are the latest games ... imagine in 2 years and beyond.
 
Last edited:
SonGoku

SonGoku

Member
ethomaz said:
I don’t know but it feel that here are something off with these estimates... in what they are basing to make the estimate?

Like in the size of what?
Click to expand...
They are based on Navi 3700 die shot
4d900c.png
Full 4SE setup with 128ROPs, 384 bit bus on on 7nm EUV (20% reduction)
64CU chip 363mm2
72CU chip 379mm2
80CU chip 394mm2

Keep in mind these estimates are very conservative: The RT bit is based on NV cards and consoles are likely to use different design rules that prioritize density over high clock speeds found in 5700 (empty spaces), so expect further 10-20mm2 deducted from each tier (if not more)
CrustyBritches said:
They stated "opportunistic boost clock", so I'm guessing that's just a function of the silicon lottery and good cooling.

P.S.- If it ends up somewhere in between our predictions then we will call a truce.
Click to expand...
But let's make it clear anything above 10TF Navi and i win
Are we using stock RTX2070?
 
Last edited:
vpance

vpance

Member
I think for PS5, 10TF will be reached on 44 to 46 CU 7nm+.

The question is if MS also went 7nm+. If they did, Scarlett 385mm2 should be much more powerful (unless they low clocked it I guess). That kind of power difference would've definitely leaked out.
 
ArabianPrynce

ArabianPrynce

Member
The only monkey wrench in you guys speculation are the trump tarriffs. Current plan is 25 percent, trumps says it can go down to 10 percent if things do go well. The g20 summits where its china and merica are gonna have trade talks. The trade talk will happen in 6/28-29. Pray to god things go well. If tarriffs are 25 percent you can kiss ur 499 ps5 goodbye lol. Sony/ms might end up selling a weak console for higher price.
 
Status
Not open for further replies.

Similar threads

AMD RDNA 5 To Be A Completely New GPU Architecture From The Ground Up, RDNA 4 Mostly Fixes RDNA 3 Issues & Improves Ray Tracing
News Rumor Hardware 2 3
128
6K
SoloCamo replied
Was 8 CPU cores enough for 9th gen consoles? Strix Halo, the prefect console.
Opinion Analysis 2
62
3K
Tertullian replied
Tom Warren - Sony’s PS5 Pro is real and Sony is asking developers to get games ready over the summer, with a push for ray-tracing support
News Hardware Hype 2 3 4
186
15K
DanielG165 replied
Report: Upcoming 'Turin' Zen 5 CPUs will support 4TB of RAM at 6000 MT/s
Rumor Hardware
43
3K
OverHeat replied
Leak - Zen 5 is now in mass production, release in late 2024
Hardware 2
61
6K
CuteFaceJay replied
Top Bottom