-
Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
-
We have made minor adjustments to how the search bar works on ResetEra. You can read about the changes here.
Their production target seemed overly ambitious. The new figure seems more realistic for a console launch during a pandemic and it should also ensure a higher quality product, less prone to failures. If I buy a PS5 soon, I want it to last for the whole generation at least.
I wonder about this as well, is msoft is facing similar issues then everything would be harder to get
Honestly if there were any mic drop moment to be had in that scenario. It would be the Series S lowered to $249. There is no possible way the cost difference (BoM + Manufacturing + Logistics) between the XSS and PS5 DE is only $100.
While I don't agree with Snake's sky is falling take, it wouldn't surprise me if the story is true. Yield issues were rumored several months ago and people took turns dunking on the insiders who suggested it.
That being said, it still sounds like we should have slightly more PS5s available than PS4s in the same time frame, and I'm not convinced Sony would change long term pricing to address any increased costs from initial yield issues. So I don't think anything has changed.
I don't understand all the concerns, 50% yield is not great but is not that bad neither. We don't have the dimension of the PS5 SOC, but since we have the dimensions of the XSX SOC, about 22x16.4mm we can easily estimate the yield. TSMC 7nm node as a defect density of 0.09/cm2, which gives 108 good dies out of 148, about 73%. Out of those 108 dies, not all of them will be able to reach the intended frequency target (that's the binning process), so the actual yield would be closer to 60%.
So yeah Sony may have lower yield, but their SOC is smaller. If the PS5 SOC is about 320mm2, 20x16mm, they can make 173 of them on a 300mm wafer. At 50% yield in the worst case scenario, that makes 87 SOC that pass the QC. XSX SOC at 60% yield would be 88 certified dies. So price per SOC should be very similar for both of them, at least in the beginning.
So yeah Sony may have lower yield, but their SOC is smaller. If the PS5 SOC is about 320mm2, 20x16mm, they can make 173 of them on a 300mm wafer. At 50% yield in the worst case scenario, that makes 87 SOC that pass the QC. XSX SOC at 60% yield would be 88 certified dies. So price per SOC should be very similar for both of them, at least in the beginning.
Not trying to stir the rumour pot, but is it normal for yields to be identified as a problem at this stage? 2 months ago Sony was increasing production and surely they'd have access to yield data then?
And this is where the stirring comes in - increasing the CU count over the summer could be a plausible reason as to why they've had to re-evaluate their yields.
I'll take 28 hours and 41 mins of dreaming this to be true, until Sony's presentation tomorrow when the 36 CU reality is once again set in stone ;-)
And this is where the stirring comes in - increasing the CU count over the summer could be a plausible reason as to why they've had to re-evaluate their yields.
I'll take 28 hours and 41 mins of dreaming this to be true, until Sony's presentation tomorrow when the 36 CU reality is once again set in stone ;-)
Perhaps the increase in production was initially based off a sample that wasn't representative of what would become the norm. And later on had to revise due to yields not meeting prior expectation.
Or the increased rate that they tried to produce the PS5's at had a negative influence on the yields so they had to take it a little slower than they wanted. 15 mil by March is absolutely crazy and 11 mil is still a huge amount aswell. They are going to be fine.
Lol 36 cu is set in stone since 2019. Never gonna change haha .
10tf is an amazing baseline for PS. Anything more powerful will come in 2023 /2024 as pro with say 20 tf on 3nm node
Can we please ignore that user for our own sanity and actually talk about the issue at hand. Hopefully 11 million units will be enough for me to secure a console this year. I did not have too much of a problem securing a PS4 and since this is in line with that it should be okay.
Ps5 is clocked insanely high, which affects yields negatively. Xbox is clocked, I wouldn't say conservatively, but more in line with what you would expect.
I don't understand all the concerns, 50% yield is not great but is not that bad neither. We don't have the dimension of the PS5 SOC, but since we have the dimensions of the XSX SOC, about 22x16.4mm we can easily estimate the yield. TSMC 7nm node as a defect density of 0.09/cm2, which gives 108 good dies out of 148, about 73%. Out of those 108 dies, not all of them will be able to reach the intended frequency target (that's the binning process), so the actual yield would be closer to 60%.
So yeah Sony may have lower yield, but their SOC is smaller. If the PS5 SOC is about 320mm2, 20x16mm, they can make 173 of them on a 300mm wafer. At 50% yield in the worst case scenario, that makes 87 SOC that pass the QC. XSX SOC at 60% yield would be 88 certified dies. So price per SOC should be very similar for both of them, at least in the beginning.
Hope people read this since it should give them a idea of things.
Hopeful this also make AegonSnake more clam lol
Nope, I take it back. Apologies. I didn't know this was a part of your schtick. Carry on sir.
I know now.
I'm getting a PS5 if it's the last thing I do
Omg...
Remember, when y'all were dissecting the hell out of the github data, the assumption was it was on RDNA 1.
It took AMD to spill the beans before Sony, and even then some didn't and probably still don't think it's on RDNA 2.
There are rumors there will be some RDNA 2 cards clocked as high as the PS5.
We just need the hardware to finally come out to see, the PS5 and RDNA 2 desktop gpus.
Y'all put forth compelling cases for why 2Ghz was insane. I thought we all agreed that RDNA 2 made it make sense.
I have close to 10,000 post on this forum. Most of them in the next gen speculation threads fighting off accusations of being a PS fanboy from MS fanboys. I find it hilarious that people think i have been pretending to be a Sony fan for 3 years over 10,000 post waiting to be activated at the right time as if im some kind of sleeper agent. lmao.
I remember that. it's crazy how quickly people have forgotten that I pretty much led the 14 tflops team for closed to two years. I always maintained that 2 ghz on rdna 1.0 was not possible, and that it had to be 7nm+ or a smaller node than 7nm. We dont know much about RDNA 2.0, but its clear now that the 50% perf/watt rate made the higher clocks possible.
lol
i blame sony. i wouldnt be making this post in september if they had revealed the price back in June. All this waiting and one bad news after another is nerve racking. i mean Lockhart is finally a reality. I am still not over that and this after Matt told me that it wont hold back next gen that much.
Glad I'm going to chill for a bit on the PS5.
I'll pick it up after we get a good library of exclusives. Did the same with my PS4.
I'm trying to decide if I'm going to keep my PS4 for now, or if I should trade it in while it's worth more??? 🤔
Not sure what a Slim and a handful of games would be worth.
I'll pick it up after we get a good library of exclusives. Did the same with my PS4.
I'm trying to decide if I'm going to keep my PS4 for now, or if I should trade it in while it's worth more??? 🤔
Not sure what a Slim and a handful of games would be worth.
It's not about faith, it's about not jumping to the most extreme conclusion based on the currently available and potentially incomplete information.
The GitHub leak was absolutely real, but don't forget the conclusions drawn from them, everything from feature set/teraflops/clocks turned out to be wrong.
Voicing concerns is totally fine, but when AMD still hasn't shown final specs and especially frequency ranges for RDNA2, we're likely missing some pretty crucial data points still.
Yeah, this is the way it works. Now if yields are low because TSMC screwed up something in the manufacture stage then they eat that cost, but if it is low yields due to design (not enough chips hitting your minimum clocks) then that is on the owner/designer of the chip. We don't know exactly what AMD and Sony have on their contract, so unsure of how they are eating the costs here. We know that AMD is seeing worst case yields of 70% on 7nm currently, but is likely at 75-80% by now. We don't have good comparisons for the consoles though because they are unlike any other product in AMD's line.
Very emotional moment in that film. Can't believe I'm laughing at it. Well executed gif. 😛
Still hoping for a $499.
For the people so concerned just don't preorder I'll be one of millions of guinea pigs and we'll tell you when it's safe to buy 😁 but I'm definitely getting a warranty
People are laughing because he's trying really hard to link poor-but-improving SoC yields to poor overall console reliability – something he's pulling directly from his ass.
The fact is that the PS5 GPU has 40 physical CUs. They could change their mind. They almost certainly won't, but "set in stone" is too strong a phrase.
I have not read anyone explain that wouldn't be the case. Again I don't know much about this but if a shit ton of chips are failing doesn't that mean it isn't that reliable
Speaking of pulling stuff out of asses, I never linked poor yields to console reliability. this is the post everyone is quoting with laughing gifs but i just re-read it and i dont see anything here linking yields to console reliability.
I will post it here so you can find what im missing here. seems to be in your head.
he's wrong. I never linked it to console reliability. in fact, I asked about it in the ps5 technical thread after you didnt get an answer here and people over there are accusing me of making you up as if im asking for a pornstar's name or something.
I think quite the opposite. If they assumed that they would get 68% yield per wafer, but ended up getting 50% then SOCs would become a manufacturing bottleneck, and they wouldn't be able to ship as many as they expected on time, so they pony up for Air Freight so they can meet their holiday targets as their manufacturing is slightly behind their expectation...
The way Microsoft leaks information, I think we would have heard about any issues with their yields. Also, you could argue that their consoles are underclocked. I suppose anything is possible though.
Honestly the only thing I'm doubting now is whether the digital console can be 399. Guess we'll see tomorrow!
They don't even need to drop RRP, just the price they sell to retailers at and tell them to lower the price or they can just bundle games & accesories at current RRP.
I follow the PS5 tech thread too you know
Why even bother connecting this yield issue to "hot PS5s" and all the related & unrelated FUD that's popped up this year (which you now say you believe)?
$399 has to be the price for PS5 Digital. You know Sony don't want to take the loss, but they have to because of the XSS.
Nope, I asked that in the PS5 technical thread on Miamiwesker 's behalf after he didnt get an answer here last night. i never linked the reliability in ANY of my posts in this thread, i literally just asked a question in a different thread. i can provide you our whatsapp chat transcripts but i get the feeling you will just accuse me of photoshopping it.
Although your post is kinda hyperbolic and I don't agree with your price conclusions because I think Sony is willing to eat the extra cost. However, as someone who has been closely paying attention to the tech side of these consoles this post has a lot of truth to it. Sony going with a 2.23 clock was completely shocking. To support BC, Sony went fast and narrow and, from a business perspective, it's looking to be the wrong gamble. I think when we look back on these consoles in like 5 years, the PS5 technology design will be considered something of a disappointment. Not a disappointment for most gamers, but a disappointment for people who analyze this stuff from a tech and business perspective. It isn't all bad news - the one possible upside for the PS5 is long term they may be able to more aggressively reduce costs than the XSX due to the lower CU count, etc.
That said, I think Sony is willing to eat these costs now to sell a ton of consoles, and for most gamers none of this will really matter. However, I think Sony biz folks and shareholders aren't loving the PS5 design right now and I suspect Sony is regretting some of their decisions.
IMO, of course.
Last edited:
What about your post from this thread that you just quoted
"pretty much all the FUD we heard about the ps5 has come to be true. i remember hearing about hot ps5 earlier this year as well. i have stayed strong and dismissed everything as fud but the hits keep on coming.
at this point, im inclined to believe everything. "
Again I ask, How do you get here from poor SoC yields?
Hot PS5 devkits dont indicate poor SoC yields?
I am failing to see the link to reliability here. I made no such comments.
According with MS it's yields aren't that good either.
Lower than previous gen =/= not that good.