• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

[MLiD] PS6 Early Specs Leak: AMD RDNA 5, Lower Price than PS5 Pro!

I've never heard of a console maker forcing profiles based on resolution or FPS. Developers will have to target all official Xbox SKUs, resolution or fps are not a thing that platform holders dictate to developers.

What is the point of a pro console that no game supports? Besides, it's not even possible; partners can't just throw in more CUs or widen the memory bus. Parteners get ready-made chips, all they can do is play with the clocks and hardware that surrounds the SOC, like bigger or faster SSD.

ASUS isn't going to contact TSMC and get a new line for fabbing custom Xbox chips. Xbox owns the SOC design; no one, including AMD, has the right to make Xbox chips.

Partners are usually very boring. What you will see at best is the vanilla Xbox GPU running at 2.8Ghz and a partner releasing a variant that runs at 2.95Ghz. Nothing to write home about, just a 5% performance boost to make people prefer their flavor.
Those would be the targets aka the goal to strive toward, with or without the use of AI and upscaling. If a game doesn't quite reach 4k/120 but something like 4k/90, MS would still probably certify the game to be published. The reason for 120 fps target is simple, Magnus will also be going to be used for Xbox Cloud. Every doubling of FPS reduces latency by 30% roughly, so going from 30 to 60 or 60 to 120. Offering up to 1080/120 or 4k/120 for xCloud will be MS goal, it's precisely what Nvidia GFN Ultimate tier offers with their 4080 equivalent GPUs. Up to 1080/240 or 4k/120 with the help of DLSS and Frame gen. So if Magnus is equivalent to 5080, I can see 4k/120 being a target goal.

Remember, there's no actual devkit hardware for Series S consoles, the Series X devkits handle it via Series S profiles. MS has gathered data from Series S usage and trained devs to build for two profiles. MS can set whatever profiles for devs to build for.

Also, the Xbox hardware won't be a 7 year generation, it is designed to evolve much quicker with AMD GPU and CPU advancements. So roughly 2-3 years. So MS wouldn't want the devs to build to fixed hardware only but rolling targets, so resolution and fps based profiles make sense, but we don't have enough info on how they plan to handle that yet.

The Elite or Pro variants of the devices would be for future proofing. Games are more heavy towards later in the gen. Path Tracing also will require more beefy hardware. And don't forget, the device would be playing PC games as well. PC games are designed to scale to hardware, while Console games are optimized to fixed spec.

So if someone is willing to pay $2k for an Xbox PC or Console with 152 CU GCD, and the OEMs see enough of a market to profit from, MS could order the chips for them. Again, it won't happen at launch but could be possible down the road.

AMD wouldn't be building that high CU count GCD die if they didn't have plans to use them.

Yes, but not from 1st party.

but leaving the high-end to OEMS, they get to differentiate better. OEM won't be able to differentiate easily from 1st party devices if they're in the same tier.
True, there are enough hard-core userbase salivating at the thought of a 152 CU Console, including me.
So, which Console Wars(tm) are we on right now? Time to crank up the "Fortunate Son".

It is streaming and subscription service Wars now. Content being exclusive to your streaming/subscription service, but still sold everywhere at full price. Just like how Disney+ or HBO Max compete.
 
Last edited:
Is there any point in running PS4 Pro versions, since the handheld itself will likely have a 1080p screen and PS4 Pro versions are just PS4 games @ 1440p/1800p/CB 4K?
I guess some people would care due to some of the 60fps modes? Especially with those that also got patched for PS5, though I have no idea how that would translate to Canis.
 
Isn't the new handheld rumored to be $300?

$500 + $300 is still less than a $1000.
No, you got that MLID part mixed up. He said the handheld will be favorably priced in comparison to Switch 2, so around $449 or $499. Then he speculated that Sony could use the internal components of the handheld to create a cheap $299 entry point console without a screen.
 
Yes, and N3/N5 wafers are rumoured to cost ~$17k/$18k respectively, vs. ~$10k for N7/N6. So you're talking about a 70/80% increase in silicon costs.
That's true, but unlike PS5 Pro that will only sell ~10 million over 4 years.

PS6 will sell that much within it's first year, so bulk discounts would mean the PS6 would be cheaper than the PS5 Pro.
 
That's true, but unlike PS5 Pro that will only sell ~10 million over 4 years.

PS6 will sell that much within it's first year, so bulk discounts would mean the PS6 would be cheaper than the PS5 Pro.
Read the bloody news. TSC ain't doing no discounts. They are rising prices if anything.
 
Last edited:
No, you got that MLID part mixed up. He said the handheld will be favorably priced in comparison to Switch 2, so around $449 or $499. Then he speculated that Sony could use the internal components of the handheld to create a cheap $299 entry point console without a screen.
So you're telling me the chip within the handheld will cost more than the PS6 chip if we look at the PS Portal $200 price as a base?

That makes no sense.
 
Read the bloody news. TSC ain't doing no discounts. They are rising prices if anything.
Clearly you are not understanding the orders Sony puts in (DRAM, NAND, cooling components, etc.) and the cost of the PS6 chip itself.

The way you are talking, you are making it sound like the PS6 chip by itself will cost $500.
 
Last edited:
Jerry Seinfeld Reaction GIF
 
You can do the math dude.
300mm² on 3nm is ~$150.
The Portal uses a low end Qualcomm chip used in <$100 pre paid phones. So $199 for the portal plus $150 for Canis, that's $350 right there.

Portal screen is 60 hz, adding a 120 hz VRR capable screen is more money. So $399 likely bill of materials. Plus Sony may want $50 profit, or to offset the tariffs.
 
The Portal uses a low end Qualcomm chip used in <$100 pre paid phones. So $199 for the portal plus $150 for Canis, that's $350 right there.

Portal screen is 60 hz, adding a 120 hz VRR capable screen is more money. So $399 likely bill of materials. Plus Sony may want $50 profit, or to offset the tariffs.
Canis die size is much smaller than the PS6's chip, so it can't be $150 as well.

It's still cheap than $1000 for both, which is how all this started.

But I like how you guys want this thing to be expensive. It's like you guys don't like gaming.
 
Last edited:
That's true, but unlike PS5 Pro that will only sell ~10 million over 4 years.

PS6 will sell that much within it's first year, so bulk discounts would mean the PS6 would be cheaper than the PS5 Pro.
True, but we're still talking about $50 extra in silicon costs, 50% more memory, and possibly 1 TB of extra storage. If the PS5 DE breaks even at $450 say, then Sony have only $50 to play with before they start taking losses, if they're targeting $500.
 
Canis die size is much smaller than the PS6's chip, so it can't be $150 as well.

It's still cheap than $1000 for both, which is how all this started.

But I like how you guys want this thing to be expensive. It's like you guys don't like gaming.
Well, didn't MLID also say it will be more powerful than the Z2X in the upcoming Xbox Ally? That device will be going on sale soon for $899. OEMs usually have 20-30% profit margin. So a $899 device would still be $599 with or without tariffs. Unless MLID is just wrong......
 
If this is a true than great news.

Maybe the graphics won't be much better but the much better processer and power efficiency would be which would still allow the newer PS6 games to look and play much better compared to the PS5. 2TB SSD please.

550 for the digtal and 650 for physical which also acts as a decent 4k hd bluray player.
 
Well, didn't MLID also say it will be more powerful than the Z2X in the upcoming Xbox Ally? That device will be going on sale soon for $899. OEMs usually have 20-30% profit margin. So a $899 device would still be $599 with or without tariffs. Unless MLID is just wrong......
Alright, you guys win. It will cost the same or more than the PS6.
 
Canis die size is much smaller than the PS6's chip, so it can't be $150 as well.

It's still cheap than $1000 for both, which is how all this started.

But I like how you guys want this thing to be expensive. It's like you guys don't like gaming.
Valve sold the $399 Steam Deck model at a "painful" loss back in 2022, which is a similar class of device as Canis is rumored to be.

There is just no way that five years later, on much pricier silicon the PS6 portable would be able to cost $299, as you were trying to argue some posts back.

Best case scenario, it would be just as expensive as the Switch 2, and that may be with Sony selling it at a slight loss. If not, then $499 is my bet, $599 for PS6 (which wouldn't be the low-power joke that AMD pitched in 2023 imo).
 
Last edited:
Valve sold the $399 Steam Deck model at a "painful" loss back in 2022, which is a similar class of device as Canis is rumored to be.

There is just no way that five years later, on much pricier silicon the PS6 portable would be able to cost $299, as you were trying to argue some posts back.

Best case scenario, it would be just as expensive as the Switch 2, and that may be with Sony selling it at a slight loss. If not, then $499 is my bet, $599 for PS6 (which wouldn't be the low-power joke that AMD pitched in 2023 imo).
Glad to see games have money to spend and shouldn't complain in game prices increase again.
 
Alright, you guys win. It will cost the same or more than the PS6.
I think the point is that if Sony sells both devices at a breakeven point, the only saving for bundling the two together is the controller cost + possibly economies of scale on the handheld. Against that, every $100 of BOM added by the handheld is $100 extra on the retail price, in a highly price sensitive market.

Those who would have purchased the handheld anyway get a slight saving, while everyone else pays considerably more.
 
I think the point is that if Sony sells both devices at a breakeven point, the only saving for bundling the two together is the controller cost + possibly economies of scale on the handheld. Against that, every $100 of BOM added by the handheld is $100 extra on the retail price, in a highly price sensitive market.

Those who would have purchased the handheld anyway get a slight saving, while everyone else pays considerably more.
That's usually what Sony does and make a profit on game sales, subscriptions and fees.

That's why it's pointless to compare with PC handhelds that add ~1.5× markup. Sometimes even a 3× markup.
 
Last edited:
That's usually what Sony does and make a profit on game sales, subscriptions and fees.

That's why it's pointless to compare with PC handhelds that add ~1.5× markup.
What you're suggesting makes it much worse for anyone just looking to buy one of the two between Orion and Canis.
 
Why not?
PS6 is aiming for $500 and that's with 8 Zen6 cores, 48 RDNA5 CUs, 24-32GB GDDR7 and probably a 2TB PCIe 5 SSD.

PS new handheld with 4 zen6 cores, 12-18 RDNA5 CUs and LPDDR5 bundle with the console and absorbing the cost of a controller probably would make the cost be $600-$700.

That's a far better deal than the PS5 Pro.
Have you been sleeping for the last 5 years?
 
What you're suggesting makes it much worse for anyone just looking to buy one of the two between Orion and Canis.
First I said a hybrid console/handheld approach would be better for next-gen instead of two separate devices.

Then i switched to suggesting two separate devices but a cheaper price for the handheld is what to expect.

Then I said you guys win and it'll be an expensive handheld.

Now an expensive handheld is worse for who wants to buy both.

I'm definitely certain you guys just like to complain.
 
Last edited:
That's usually what Sony does and make a profit on game sales, subscriptions and fees.

That's why it's pointless to compare with PC handhelds that add ~1.5× markup. Sometimes even a 3× markup.
What 1.5-3 times markup?


OEM profit margins range between 15-30%. Even Apple which has highest profit margins does 35-40%.

The point of both Sony and Xbox is they no longer want to subsidize hardware, so their game sales, subscription revenues are much higher profits.
 
What 1.5-3 times markup?


OEM profit margins range between 15-30%. Even Apple which has highest profit margins does 35-40%.

The point of both Sony and Xbox is they no longer want to subsidize hardware, so their game sales, subscription revenues are much higher profits.
I already agreed with you guys saying the handheld will be expensive, jeez.
 
MLiD said it'll be a cheap console as well to get PS4 owners to upgrade.
I don't understand the arguments behind it. PS5s at launch were super expensive pretty much everywhere (except in US obviously as those consoles are always subsidized by the rest of the world) and they were very hard to find for like 2 years. They even raised the price. For 4 years Sony had no problem selling a very expensive console. I has only become a problem after 4 years so now they are doing sales. Why should it be a problem for Sony now?

Besides PS4 Fortnite players spend as much as PS5 Fornite players. Do they really think people are going to buy PS5s to buy their lastest 70$/80€ AAAA boring games or their in-house GAAS games and to pay to play online? Because that would be a delusion.

People are going to buy PS6s to play Fornite and COD, the real stuff, and not pouring money into any of their one word named woke clones.
 
Last edited:
First I said a hybrid console/handheld approach would be better for next-gen instead of two separate devices.
How do you make PS5 Pro BC work, then?
Then i switched to suggesting two separate devices but a cheaper price for the handheld is what to expect.

Then I said you guys win and it'll be an expensive handheld.
The handheld hardware is basically done, it's gearing towards a dedicated device and not exactly a Portal 2.0, therefore you can't just suggest that Sony should bundle two proper consoles into one pack as the only option for next-gen. At the very most, you will only get a niche, optional bundle like this (which was real, but I have no idea if it actually released):
phJLTW3WPZETjZyo.png

Now an expensive handheld is worse for who wants to buy both.

I'm definitely certain you guys just like to complain.
Your only option to buy that expensive handheld being a bundle with a home console is objectively worse for most consumers.
 
I don't understand the arguments behind it. PS5s at launch were super expensive pretty much everywhere (except in US obviously as those consoles are always subsidized by the rest of the world) and they were very hard to find for like 2 years. They even raised the price. For 4 years Sony had no problem selling a very expensive console. I has only become a problem after 4 years so now they are doing sales. Why should it be a problem for Sony now?

Besides PS4 Fortnite players spend as much as PS5 Fornite players. Do they really think people are going to buy PS5s to buy their lastest 70$/80€ AAAA boring games or their in-house GAAS games and to pay to play online? Because that would be a delusion.

People are going to buy PS6s to play Fornite and COD, the real stuff, and not pouring money into any of their one word named woke clones.
I kind of agree with him.

I mean look at the prices where I'm from.
KrXGbNMivPijRgQ7.jpg


Look at that Switch 2 price and you'll see way I wanted a hybrid design or a cheaper handheld.

We are paid at the same rate but stuff cost more than 2× the price.
 
I kind of agree with him.

I mean look at the prices where I'm from.
KrXGbNMivPijRgQ7.jpg


Look at that Switch 2 price and you'll see way I wanted a hybrid design or a cheaper handheld.

We are paid at the same rate but stuff cost more than 2× the price.
Bummer That Sucks GIF
 
Im really hoping that they decided to bump the specs up to reasonable levels especially since its been years since that initial meeting and having a console come out significantly weaker than the xbox 2 years after just isn't a good look for the go to shiny console for the masses. I mean it would be even harder to sell the upgrade from the ps5 to the ps6 with the paltry specs touted. Ps4 to ps5 was a large and robust upgrade in specs and still it was hard to put a real gen on gen difference like before...I'm not sure a lot of consumers will feel the meagre upgrade going from ps5 to ps6 will be worth it especially with the pro out and most games being cross gen but I fear Sony has decided fuck it to pushing hardware and is happy to push safe and lazy 60 fps marginal upgrades like yotei and will design games around the handheld and if so these specs would line up possibly.
It's just 1 year after and I don't get what makes the difference this time around when you're saying that most consumers didn't exactly see a real gen on gen gap between PS4 and PS5 and that still sold just fine. PS5 Pro is obviously not a factor when only 15M AT MOST of the userbase is going to buy it.
 
True, there are enough hard-core userbase salivating at the thought of a 152 CU Console, including me.

One thing that's puzzling me is the the vast amounts of CUs that are disabled for the gaming GPU if the MLID's chart is legit. For desktop GPUs that's the norm, but for a console, it's definitely unprecedented to have more than a quarter of the GPU's disabled.
 
Yeah and these guys are trying to tell me a hybrid console with a cheaper handheld as the controller is worse for the consumer.

If Sony did that at the same price as the PS5 Pro, outside of the US that's insane value.
They ignore the fact that most steam users on pc use a $300 graphic card in the US.

I agree cheaper is a better value for the consumer. Plus it makes the install base grow faster.
 
Last edited:
Let's assume the the rumour that the PS6 is faster than a 5080 in ray-tracing, and thus a 4080, we can get a glimpse at the performance by looking at some of the benchmarks of AC Shadows from the developers, who compared the PS5 Pro to the RTX 4080, and we can see the performance difference in RT render time was significant, keep in mind this is the 4080 and not the 5080.

So, in the latest DF Direct, Alex mentions a Ubisoft presentation where they revealed some eye-opening numbers regarding ray tracing and global illumination.

In short, getting Assassin's Creed Unity quality and method of GI in Shadows wouldn't be possible due to the size of the game world and the dynamic time of day.

ntX1Fc6.png


As you can see, it would have taken 1.9 TB of data just for the GI file.

Another cool thing is the rendering time for the ray tracing in the pipeline.

SlYme6O.png


itedLgx.png


A few things surprised me: The Pro is slower than I expected. I thought it would be twice as fast or take half as long as the PS5. It takes a little under 2/3 of the rendering time. The 4080 is much faster than I thought being over 4x the speed of the PS5.
 
Those would be the targets aka the goal to strive toward, with or without the use of AI and upscaling. If a game doesn't quite reach 4k/120 but something like 4k/90, MS would still probably certify the game to be published. The reason for 120 fps target is simple, Magnus will also be going to be used for Xbox Cloud. Every doubling of FPS reduces latency by 30% roughly, so going from 30 to 60 or 60 to 120. Offering up to 1080/120 or 4k/120 for xCloud will be MS goal, it's precisely what Nvidia GFN Ultimate tier offers with their 4080 equivalent GPUs. Up to 1080/240 or 4k/120 with the help of DLSS and Frame gen. So if Magnus is equivalent to 5080, I can see 4k/120 being a target goal.

Remember, there's no actual devkit hardware for Series S consoles, the Series X devkits handle it via Series S profiles. MS has gathered data from Series S usage and trained devs to build for two profiles. MS can set whatever profiles for devs to build for.

Also, the Xbox hardware won't be a 7 year generation, it is designed to evolve much quicker with AMD GPU and CPU advancements. So roughly 2-3 years. So MS wouldn't want the devs to build to fixed hardware only but rolling targets, so resolution and fps based profiles make sense, but we don't have enough info on how they plan to handle that yet.

The Elite or Pro variants of the devices would be for future proofing. Games are more heavy towards later in the gen. Path Tracing also will require more beefy hardware. And don't forget, the device would be playing PC games as well. PC games are designed to scale to hardware, while Console games are optimized to fixed spec.

So if someone is willing to pay $2k for an Xbox PC or Console with 152 CU GCD, and the OEMs see enough of a market to profit from, MS could order the chips for them. Again, it won't happen at launch but could be possible down the road.

AMD wouldn't be building that high CU count GCD die if they didn't have plans to use them.
Again, no platform holder is going to dictate resolution and FPS for a developer, certainly not the one that is going to struggle to sell 20 million units. Rockstar isn't going to make a 1440p/30fps RDR3 on PS6 as the leading platform and then a 4K/120 version on Xbox just because Xbox said pretty please. Limiting developers doesn't serve anyone, and as the weakest platform holder, they have no way to force developers' hands.

You are overthinking it. Xbox will introduce x amount of SKUs to developers, and developers will have to optimize their games to these SKUs, just like Playstation and Xbox always did. There is no reason to think otherwise.

An yeah, if MS will make a 152CU Xbox SKU, then a 152CU Xbox SKU will exist. But partners can't do that on their own; only Xbox can.

You can do the math dude.
300mm² on 3nm is ~$150.
A 300mm^2 3nm chip will cost Sony well more than 150$. The PS5's 7nm SOC probably cost more than $150 at launch. A 3nm SOC the same size as PS5 will probably cost almost twice the PS5 SOC price at launch, likely somewhere between $300 and $350. Chiplets will make it cheaper, but it's hard to tell by how much.
 
Last edited:
A 300mm^2 3nm chip will cost Sony well more than 150$. The PS5's 7nm SOC probably cost more than $150 at launch. A 3nm SOC the same size as PS5 will probably cost almost twice the PS5 SOC price at launch, likely somewhere between $300 and $350. Chiplets will make it cheaper, but it's hard to tell by how much.
This is how I did.
Keep in mind this is just an estimate and I could be completely wrong and is open to corrections.
This is assuming the PS6 is 300mm².

Step 1: Calculate wafer area
Wafer diameter (D) = 300 mm (12-inch wafer)
Wafer area = π × (D / 2)² = π × 150² = 22500 × π ≈ 70,685.83 mm²

Step 2: Calculate number of whole dies per wafer
Using the formula:
N = (π × D²) / (4 × A_die) – (π × D) / (√2 × √A_die)

Where:
N = dies per wafer
D = 300 mm (wafer diameter)
A_die = 300 mm² (die area)

Calculate:
(π × 300²) / (4 × 300) = (22500 × π) / 1200 = 18.75 × π ≈ 58.9049 × 4 = 235.619

(π × 300) / (√2 × 17.3205) = 942.4778 / 24.495 ≈ 38.4765

Subtract:
N = 235.619 – 38.4765 = 197.1425 ≈ 197 dies per wafer

Step 3: Calculate die yield using the Poisson model
Convert die area to cm²:
A_die = 300 mm² = 3.0 cm²

Defect density (mature 3 nm assumption):
Y = e^(–D × A_die) = e^(–0.10 × 3) = e^(–0.30) = 0.7408

Step 4: Calculate number of good dies per wafer
N_good = N × Y = 197.1425 × 0.7408 = 146.05 good dies

Where:
N = dies per wafer

Y = yield
N_good = good dies
Step 5: Calculate fab cost per good die
Assume:
Wafer cost = $22,000
Wafers = 50,000
Mask cost = $20 million

Mask cost per wafer = 20,000,000 ÷ 50,000 = 400
Mask set cost amortized per wafer = $400


Calculate fab cost per good die:
Fab cost per good die = (Wafer cost + Mask cost) / N_good
= (22,000 + 400) / 146.05 = 22,400 / 146.05 = $153.38

Step 6: Add packaging cost to get total cost per good die
Assume packaging cost = $8.00

Total cost per good die = Fab cost per good die + Packaging cost
= 153.38 + 8.00 = $161.38 USD

So each PS6 chip = $162 USD
 
One thing that's puzzling me is the the vast amounts of CUs that are disabled for the gaming GPU if the MLID's chart is legit. For desktop GPUs that's the norm, but for a console, it's definitely unprecedented to have more than a quarter of the GPU's disabled.
I was wondering what they would use for the laptops and handhelds. The Z2X in Xbox Ally is 16 CU and Z3X is discontinued reportedly. Is it possible they could use a 48 CU chip, then disable 32 CUs for use in handheld, then disable less CUs for gaming laptops?
Again, no platform holder is going to dictate resolution and FPS for a developer, certainly not the one that is going to struggle to sell 20 million units. Rockstar isn't going to make a 1440p/30fps RDR3 on PS6 as the leading platform and then a 4K/120 version on Xbox just because Xbox said pretty please. Limiting developers doesn't serve anyone, and as the weakest platform holder, they have no way to force developers' hands.

You are overthinking it. Xbox will introduce x amount of SKUs to developers, and developers will have to optimize their games to these SKUs, just like Playstation and Xbox always did. There is no reason to think otherwise.

An yeah, if MS will make a 152CU Xbox SKU, then a 152CU Xbox SKU will exist. But partners can't do that on their own; only Xbox can.


A 300mm^2 3nm chip will cost Sony well more than 150$. The PS5's 7nm SOC probably cost more than $150 at launch. A 3nm SOC the same size as PS5 will probably cost almost twice the PS5 SOC price at launch, likely somewhere between $300 and $350. Chiplets will make it cheaper, but it's hard to tell by how much.
Again, not dictating to devs, or a mandate but a guidance or encouragement. A profile isn't a minimum threshold for the devs to build at, it's a maximum threshold of the hardware capabilities. Xbox will certify whatever the devs come up with, that's reasonable.

Besides, PS6 is also targeting 4k/120. Likely dynamic resolution and FSR4 but that's the goal.. And unlocked FPS. However, the thought of Canis holding back games for Magnus is very amusing.

MS is very accommodating to their OEMs, if they ask for something reasonable, MS would help.
This is how I did.
Keep in mind this is just an estimate and I could be completely wrong and is open to corrections.
This is assuming the PS6 is 300mm².

Step 1: Calculate wafer area
Wafer diameter (D) = 300 mm (12-inch wafer)
Wafer area = π × (D / 2)² = π × 150² = 22500 × π ≈ 70,685.83 mm²

Step 2: Calculate number of whole dies per wafer
Using the formula:
N = (π × D²) / (4 × A_die) – (π × D) / (√2 × √A_die)

Where:
N = dies per wafer
D = 300 mm (wafer diameter)
A_die = 300 mm² (die area)

Calculate:
(π × 300²) / (4 × 300) = (22500 × π) / 1200 = 18.75 × π ≈ 58.9049 × 4 = 235.619

(π × 300) / (√2 × 17.3205) = 942.4778 / 24.495 ≈ 38.4765

Subtract:
N = 235.619 – 38.4765 = 197.1425 ≈ 197 dies per wafer

Step 3: Calculate die yield using the Poisson model
Convert die area to cm²:
A_die = 300 mm² = 3.0 cm²

Defect density (mature 3 nm assumption):
Y = e^(–D × A_die) = e^(–0.10 × 3) = e^(–0.30) = 0.7408

Step 4: Calculate number of good dies per wafer
N_good = N × Y = 197.1425 × 0.7408 = 146.05 good dies

Where:
N = dies per wafer

Y = yield
N_good = good dies
Step 5: Calculate fab cost per good die
Assume:
Wafer cost = $22,000
Wafers = 50,000
Mask cost = $20 million

Mask cost per wafer = 20,000,000 ÷ 50,000 = 400
Mask set cost amortized per wafer = $400


Calculate fab cost per good die:
Fab cost per good die = (Wafer cost + Mask cost) / N_good
= (22,000 + 400) / 146.05 = 22,400 / 146.05 = $153.38

Step 6: Add packaging cost to get total cost per good die
Assume packaging cost = $8.00

Total cost per good die = Fab cost per good die + Packaging cost
= 153.38 + 8.00 = $161.38 USD

So each PS6 chip = $162 USD
You forgot the most important part of the calculation. TSMC's profit.
 
That's already included in the wafer cost.

I mean, you can just do your own research, instead of just saying random stuff.
Well, something doesn't add up. Don't we already have the list of BOM for PS5? What were the costs for the PS5 chip? A 5080 equivalent chip for $160 sounds quite cheap.
 
I suspect in a couple of/few months time it may make sense to have Next-Gen XBox threads and PS6 threads separate as the leaks have more concrete info and less speculation - right now there's not quite enough yet, but we're getting close and there will be all sorts of stuff (less hw based and more business side/games etc) which would make it more sensible to have two separate threads.
 
Well, something doesn't add up. Don't we already have the list of BOM for PS5? What were the costs for the PS5 chip? A 5080 equivalent chip for $160 sounds quite cheap.
5090 for example, which is a 761 mm² die on 5nm.

Spitballing Nvidia's RTX 5090 GPU die manufacturing costs — die could cost as little as $290 to make
Considering defect density and yields, Nvidia may have to spend around $290 to make a GeForce RTX 5090 graphics processor, though it could also increase to $340 if only the perfect dies were sellable.
 
Oh, then it sounds pretty traditional. PS4, X1, and PS5 APUs have been sold as PC parts in some markets. This makes much more sense.

So it seems that approach is dual?
1) A more or less traditional Xbox with multiple SKUs similar to the XSS/XSX, but the SOC will be available for third parties to white label as their own Xbox, similar to Nvidia's Founder's edition and third-party OEM cards.
2) A PC centric OS, basically Windows Lite for gamers, that will compete with SteamOS and will be used in PC machines like the ROG Ally.


If Xbox sells thier chips for third party to use, but it will still be a treditional box, then I don't think third parties will mess with clock speeds or memory sizes. For a console to be a console, developers need a fixed spec to optimize to. Even if there are 4 different Xbox SKUs, that's still better than a messy market with hundreds of configurations. Maybe Xbox will allow messing with the clocks, and the guidelines will be to develop to the baseline machine, and if some ROG Xbox has 10% higher clocks, it will hold the DRS and FPS better.

According to rumours there will not be XSS. There is only a single 68CU APU in development. Which is interesting and I guess they think XSS was a failure but I think XSX sold just as bad if not worse.

Anyways I think Microsoft will make the console only and since the rumours also suggest the same APU will be used for Xbox PC. I think other OEM like Asus will have ROG Xbox PC, same with Lenovo with Legion Xbox PC. If they decide to go with other OEM then MSI with their Suprim and Gigabyte with Aorus etc .... these will be on steroids with aggressive clocks, faster memory, SSD etc ... and come at a price. I think Xbox will be clocked conservatively to meet console power requirements and cheaper cooling solution to keep cost down. Memory and BW will also be limited unlike Xbox PC. PS6 is much smaller with 40-48CU so they will again go with higher clock speed both console I should think will use same amount of power. 160W for PS6 according to rumours though seems too low.
 
Which is interesting and I guess they think XSS was a failure but I think XSX sold just as bad if not worse.
Well, according to MS numbers it did. But I think they saw the damage the Baldur's Gate 3 and Wukong situation did to the brand and don't want to be there ever again.
 
They ignore the fact that most steam users on pc use a $300 graphic card in the US.

I agree cheaper is a better value for the consumer. Plus it makes the install base grow faster.
The quickest way of growing the install base for the PS6, is reducing the price, which you can do by not including any handhelds.

The "bundle everything together and compete on value" is the Kinect strategy.
 
I kind of agree with him.

I mean look at the prices where I'm from.


Look at that Switch 2 price and you'll see way I wanted a hybrid design or a cheaper handheld.

We are paid at the same rate but stuff cost more than 2× the price.
But People bought expensive PS5s for years and they are doing the same with Switch 2. Currently PS5s are still selling very well at a incredible high price compared to PS4 same timeframe.

Besides with a cheaper-ish handheld (where they actually can attract PS4 customers!) there are no reasons anymore for a cheap and weak console (when there was before, but they were still selling everything they produced for years).

For decades Sony have been saying a weaker / cheaper console is not a valid strategy and suddenly it is?

We know what they want. They want people to buy their 80$/90€ director's cut single player or paying GAAS PS6 games (because they didn't buy their 70$/80€ PS5 games!) instead of playing free games like Fortnite, GTA5 or fun and cheap AA games on their PS4s or even PS5s.

There is a reason TLOU2 or Spider-man 2 are in constant sales everywhere and are usually less expensive to buy than Astro-Bot or Demon's Souls. Because that is the kind of games they want to promote as they have not faith anymore in the kind of games their company have been doing for decades even if that strategy was actually working better than ever in 2020 just before PS5 was released.
 
Last edited:
a weaker/cheaper console has never been a valid strategy until it is.

A weaker/cheaper console is viable because we are even more deep in the area of diminishing returns.

Microsoft releasing an Xbox branded Rog Ally device that is roughly equal to a PS4 in raster means the Flagship Xbox/Acti games will have to run on that baseline hardware at the very least for the next generation.

Hell is Us running on a 5080 vs a PS5 Pro is nearly identical, you won't spot the difference unless they are side by side and even then it's very difficult to tell the difference.

Raster performance won't need to increase, it will be marginal, RT/TOPS will be increased at the top end (PS6) and stay in the same ballpark but at lower power for the baseline (PS Handheld).

It will be a much smaller jump next gen than it was going from PS4 to PS5, that's something people need to understand and rationalise.

We won't see a huge leap again until we get to a point where a console has a render pipeline that doesn't rely on a hybrid raster/ray traced render pipeline.
 
Top Bottom