Changed my lower bound prediction to a different CU setup and clock.
Threadmark
Flutter
The secret sauce I was talking about 2 days ago was the analysis I posted today.
Eh? I saw 325 vs 360. Maybe it's the extra cache on the X, I'm not sure what that takes up.
Isnt Colin Moriarty a Playstation fanboy too? Or was that Greg?
About Lockhart:
I am sure based on what I know that Lockhart was part of the plan and the leaks about it were not planned to confuse the competition. This is just too much of a conspiracy theory in my opinion. Sony is the best example that plans can change, same is valid for MS.
I think my confusion comes from the fact that the result talks about TDP.
For me TDP was defined as the maximum amount of heat generated the cooling is designed to dissipate under any workload.
In the video he is actually talking about "Leistungsaufnahme" which is "power consumption".
I cannot remember if you can assume the power consumption of the GPU and the heat dissipation as equal. I actually thought not all of the power consumption would dissipate as heat.
Just thinking about Phil saying they only started to discuss Scarlett with third-parties in Fall 2018. Could that be when he started to get back negative feedback about Lockhart and that is why it wasn't talked about at E3?
That really does sound like a mistake, iirc according to cerny, they already visited developers on next gen at late 2014 or early 2015.Ledbetter did say he's heard lots of skepitcsm from devs about the lower end SKU. But like....how do you spend all that time and money and not consult with any third party devs? That sounds crazy, and to be frank a mistake
Eh? I saw 325 vs 360. Maybe it's the extra cache on the X, I'm not sure what that takes up.
Are we really doing this? The x cost $100 dollars more and released a year later. It's SUPPOSED to be more powerful, it's extremely well built machine but there's nothing genius to it. It's like being shocked that a 2020 e class is better and has more features than a 2019 c class
Ledbetter did say he's heard lots of skepitcsm from devs about the lower end SKU. But like....how do you spend all that time and money and not consult with any third party devs? That sounds crazy, and to be frank a mistake
PS4Pro is 325mm²-ish. PS4 Slim is around 225 to 230mm², IIRC.
X has two extra memory interfaces that take up quite a lot of space (15mm²-ish), besides the extra CUs + RBs (12mm²-ish). There's also probably some difference to some of the support silicon and layout optimization.
Die shot comparison.
Hard choice ps5 vs apple monitor stand
That really does sound like a mistake, iirc according to cerny, they already visited developers on next gen at late 2014 or early 2015.
Bullshit. There's genius in all of these machines if you aren't too blinded by platform loyalty to see it. How much additional performance would an additional year and an additional $100 on the price be expected to deliver? What amount of additional performance for the X vs the Pro short of this would be a disappointment? What amount exceeding the baseline expectation would be considered a "genius" level achievement? Think on whether you are at all qualified, based on the amount of technical knowledge you possess, to set those benchmarks and perform that evaluation.
I have a very hard time believing they didn't consult with devs earlier on. I think it's more likely they knew they were down on it but took a gamble that once they saw they final machine they'd be okay with it, and evidently they weren't. Still a gamble but it would make more sense than just flubbing that badly.
I agree. I am just talking based on what the general consensus had been. It was that Ana would be $500, PS5 $399 and lock $299. And even back then when that was what most seemed to have been thinking I was insisting that I believe Ana would be the same price with the PS5 and that MS can't put the PS5 in a position where they seem the most balanced and best buy by default. Which is why I also said that's not how I would do it.I think that the sandwich strategy was doomed to fail, all they are doing is making the PS5 a really nice balance of price and power. If they want the Anaconda/Lockheart to succeed, they need the Anaconda and PS5 to be priced the same with about the same performance, even if the PS5 out powers them a bit, while Lockheart is at least 150$ cheaper than both. The whole idea is having a cheap console for the more casual audience, an SKU to get these 3rd and 4th-year adopters early in order to build a big base fast.
They only started talking to devs less than a year ago? That seems crazy to me.
They only started talking to devs less than a year ago? That seems crazy to me.
Good. Smart engineering and better design as far as I'm concerned. Though it seems the Pro still has a smaller APU, I'm not sure why.
... Yes? That's exactly what I was saying, that the Pro's price and launch window are responsible for the vast majority of the difference between the two consoles.
They very well might have. What I was arguing against was the baseless assertion that such a console could never have hoped to match the X simply because the Xbox engineers are supposedly more talented, and the idea that the PS5 CAN'T be more powerful than Scarlet because of this.
In the end, one reason why I'm more appreciative of the Pro is that it pushed reconstructed resolutions into mainstream development in a big way. My hope is that that continues into the next generation, which I think is more likely now than it would've been if Sony had gone the route of just trying to push the most native pixels. Still think they should've done all this at $499, though.
They only started talking to devs less than a year ago? That seems crazy to me.
ArrogantThey only started talking to devs less than a year ago? That seems crazy to me.
My phone contract is £35 a month for 2 years. Gives me the phone and the service. Back of a napkin calculations mean Sony could sell me a brand new PS5, bundled with PS+ and some kind of PS Now/Games pass style service for £30 to £35 a month for two years. I'd have the console and online play and tons of games to play with no big up front free. £30-ish a month basically gives Sony £15 a month for those services and £15 towards the console (£400/$499) or any particular split for them that works. The services obviously make a lot of profit and they have me locked in for years spending more money. I hope they do this.They could do that and I guess another option wouldn't hurt but for simple me it is just making things overly complicated with so many options. Buy console, buy games and if a multiplayer gamer, buy Gold or PS+. Like you say they have tried similar before and I'm guessing the take-up didn't make it worth continuing?
It could change this time round with streaming and possibly increasing hardware costs, though.
I question better, at least as an unqualified statement. From a development standpoint you received a target that required more work to get optimal results from and was more restrictive in what you could do given that you had less memory to work with, to realize the benefit of the Pro design you had to re-write shaders to use fp16 (which only benefitted that specific SKU on the console side and a maybe very small percentage of PC users), and had to use the specific reconstruction technique enabled by the hardware, even if another technique mapped better to your engine. This doesn't make the design chosen a worse one either when you take everything into account, but like almost every other design decision made when designing a console there are pros and cons to the choices that were made.
As for why the One X SoC is bigger, that's obvious, more CUs and a wider memory bus.
That's not exactly what you said in the post I quoted. You dismissed the XBone X's power advantage as being down to, again, just having the benefit of time and money. As in my response to console lover, I think most of us are highly unqualified to make that type of judgment.
That the X proves that MS's engineers are better is not an argument I would make or support.
Fair. I guess making it necessary instead of optional did push development forward.
They only started talking to devs less than a year ago? That seems crazy to me.
Yeah, that just doesn't make sense. Is it really possible that MS would just set off on development of Scarlett, simply assuming that they knew exactly what the devs would want? I mean, I can see how the path for this generation might be a little more obvious when it's so similar to the current one, but Christ.
Yeah, that just doesn't make sense. Is it really possible that MS would just set off on development of Scarlett, simply assuming that they knew exactly what the devs would want? I mean, I can see how the path for this generation might be a little more obvious when it's so similar to the current one, but Christ.
My guess they started to talk to devs last year what their plan is. Like Cerny said when they unveiled PS4, they traveled to developers and asked what they wanted but never mentioned the console itself. What they were doing etc. i'm sure MS have done the sameYeah, that just doesn't make sense. Is it really possible that MS would just set off on development of Scarlett, simply assuming that they knew exactly what the devs would want? I mean, I can see how the path for this generation might be a little more obvious when it's so similar to the current one, but Christ.
It is an excellently designed system, it can put out at 4k, it's small, it's silent, it's stylish. Tell me what's so miraculous about its design without using the word vapour chamber which somehow people became convinced was some sort of exotic custom tech.
Hovis method is very innovative, but really it's relevance is in chip yields and power draw.
I question better, at least as an unqualified statement. From a development standpoint you received a target that required more work to get optimal results from and was more restrictive in what you could do given that you had less memory to work with, to realize the benefit of the Pro design you had to re-write shaders to use fp16 (which only benefitted that specific SKU on the console side and a maybe very small percentage of PC users), and had to use the specific reconstruction technique enabled by the hardware, even if another technique mapped better to your engine. This doesn't make the design chosen a worse one either when you take everything into account, but like almost every other design decision made when designing a console there are pros and cons to the choices that were made.
As for why the One X SoC is bigger, that's obvious, more CUs and a wider memory bus.
That's not exactly what you said in the post I quoted. You dismissed the XBone X's power advantage as being down to, again, just having the benefit of time and money. As in my response to console lover, I think most of us are highly unqualified to make that type of judgment.
That the X proves that MS's engineers are better is not an argument I would make or support.
Fair. I guess making it necessary instead of optional did push development forward.
I agree for a midgen target the RPM FP16 was probably too much work for dev to be useful, the ID buffer was a good idea...
Yeah, sounds like 9tf maximum.they are saying that they expect 40 CU, 4 disabled and around 8.3~8.4TF based off the die size.
One interesting aspect of the presentation that Chi delved into was yield. With the original batch of SDM855 processors, they had quite a lot of bad parts rejected by their partners due to high power consumption. A more in-depth look revealed that there was quite a large spread in their Vmindistribution. Naturally, there are two main approaches to handling this specific situation. You can lower the operating voltage or you can tighten the spread distribution. For Vmin higher than the Vdd, the operating voltage has to be raised to pass. Dynamic laser stimulation (DLS) was employed in order to determine the location of sensitive areas. The analysis located flip-flop devices located at a cell boundries. Further analysis revealed physical defects that cause systematic transistor Vt shift, impacting the operating voltage of the critical path. In collaboration with TSMC, design and process changes had to be made to improve the timing margins and reduce the physical defect. Multiple such problems showed up on the 7-nanometer process. A number of key modules that were particularly prone to generating low-voltage defects were isolated including the polycut and RMG clean. Through the DTCO collaboration, the yield loss due to low Vt operations was reduced by 9x. With the help of TSMC, the spread of variation in Vmin was tightened up using device tuning, optimizations across the fin, epi, and the metal gate. All in all, the result is much better uniformity across wafers with power consumption spread being reduced by around 60%. All of this effort is gone into ensuring that the share of parts that are rejected is significantly lowered.
TSMC also developed a 2nd generation of their 7nm process. This is an optimized process which uses the same design rules and DUV and is unrelated to 7nm+ which is EUV-based. This process is entirely design-compatible with the first generation but enjoys additional power and performance enhancements. For their second generation process, TSMC made some additional optimizations.
All in all, the 2nd-generation 7nm process is said to deliver over 5% improvement in performance. Additionally, at the same leakage, at high frequencies, the second-generation 7nm process has improved the Vmin by 50 mV.
- Fin profile Optimizations
- Epi Optimizations
- MOL resistance Optimizations
- FEOL capacitance
- Metal gate Optimizations
For both?they are saying that they expect 40 CU, 4 disabled and around 8.3~8.4TF based off the die size.
First party devs have different goals than 3rd party. Also at least one less platform to develop for.
they are not talking off the knowledge of next gen consoles. they are guessing their ballpark based off the navi information.