Silicon lottery doesn’t just apply to the consumer end, it applies when you’re the one actually making the product too.
Pushing stuff higher inherently means you’ll have more rejected wafers/components that can’t stay stable or function or have defective cores (re: chip wafers), which drives up the cost.
The reason is heat. I actually saw this in an Xbox video where they explained thermals and under clocking the GPU
While it’s able to handle the load right now, what happens in two years when the inside of the console is dusty and the fans don’t spin quite so easily?
There are longevity concerns that have to be accounted for - not just short-term gains.
We arnt talking about doubling the GPU speed here, or winding it up to PS5 levels. But if you don’t think the Series S could have its clock increased without it causing issues then I really don’t know what to tell you. It could easily be increased. Its the slowest RDNA2 GPU on the market. Some RDNA 2 cards are getting over 2.4ghz Do you think that MS suddenly hit the power envelope just when they got to 4tflops? No they didn’t. Do you think that MS hit the limits of the Series X when they just got to 12tflops? No, they didn’t.
Microsoft set out to get a 12 tflop Series X and a 4tflop Series S and that’s all they wanted. Microsoft advertised the Series S as 1440P/60 machine, however it is not even able to keep 1080P in alot of games. This really is disappointing, and not what MS initially thought it would be. Its CPU and SSD are better than the One X, but its GPU looks less capable at this stage. In my eyes MS should have aimed for a 6tflop Series S, to at least match the One X. It really is under performing, and a clock increase to the GPU would be a good thing.
Increasing the clock speed on a GPU in a silicon shortage year is asking for a lot. You can find a Series S at least, you can’t find these higher clocked GPUs anywhere 
They designed a system that’s smaller, quieter, cooler, and more powerful than their competitor with Series X, and they got their mic drop $299 moment with Series S. No need to lose more money than they already are by reducing viable silicon yields even a fraction of a percent.
My brother in law has a PS5, and its quiet. I haven’t noticed any difference between my Series X and his PS5. The Series S is only 1550mhz. I doubt very much there would be any real reduction in yields if they took the Series S GPU to the same speed as the Series X for instance. I mean, otherwise MS wouldn’t have clocked the Series X as they did. That would at least give the Series S a 4.65tflop GPU, and would help keep the resolution above 1080p. We are not talking about adding expense to the product. It would still come in around the same cost. As the real world results have shown, it really doesn’t have enough performance. We have titles where devs have removed Ray Tracing from a game that has it on all other platforms. That’s more than just a drop in resolution, it removing features. Just how do you think its going to cope as this gen moves forward? Its only going to get worse
I am an Xbox player. I support Microsoft in gaming, but I am prepared to call bullshit when they do something wrong.
I’m not sure what point you are trying to make. The foundries are still producing the same amount of wafers, increasing the Series S GPU from 1550mhz to 1825mhz won’t alter their output.
this alters the output because a lower clocked chip has a different yield than a higher clocked one.
Sure, a GPU clocked at 2.3ghz has a lower yield than one at 1.5ghz but there won’t be that much of a difference going from 1.5 to 1.8. AMD, Sony and MS arnt going to use a chip that has noticeable increases in yield drop offs going to 1.8ghz, and then ramp it up to 2+. In the scheme of things, a 1.8ghz RDNA2 GPU is quite low speed wise. Infact not one AMD card is as low as the Series X. Both Series X.and Series S are the ONLY RDNA 2 GPUs that do not exceed 2ghz, and you think there would be yield issues going to 1825mhz for the Series S chip?
I get how you feel. I do the same. 20GB, some higher clocks for the Series X 12GB, some higher clocks for the Series S
Both with the same housing. But the simple answer. It’s not worth it. The consoles hit the target and are good enough compared to competition and previous gen.
It’s all about the money. Series S should already break even and they expect that console to be the volume driver. The money is on software and services. The Series X is already super cheap for what you get.
Once you start bumping the GPU clock speed on the Series S you will need to start bumping memory speed and/or memory width, which makes it vastly more expensive.
yes
I don’t think Series X is leaving anything on the table RDNA 2 is clocked pretty high on PS5 but it is a 36 CU chip and can afford the heat. The bigger Navi gets the lower the clocks and Microsoft isn’t going to water cool to try and get sustained boost clocks on it. Series S on the other hand could be clocked way higher but would need cooling on par with PS5 which would drive costs up and erode the difference between S and X. So its a combination of costs and market segmentation.
I mean, maybe, but your post is in the long tradition of “guy on forum is certain he’s figured out something obvious that a team of expert hardware engineers have missed.”
It’s possible, but it seems more likely to me that raising clocks actually does affect yield, or cooling costs, or expected failure rates, or whatever to the point that it wasn’t worth modifying. Remember that taking on an additional five dollars of cost per unit is probably a $200 million dollar decision for the volume they expect to move with Series S. What’s the expected financial gain from locking to 1440 or 1080 somewhat more often? If it’s less than the total cost of that change over the lifetime of the console, then why bother?
I get what you’re saying about RDNA clocks, you’re not wrong that they could have clocked higher. But their goal is to make money. In console hardware that means minimizing losses. And you do that by being relatively conservative with your choices.
It costs way more to push a design up to a certain perf envelope than to have designed for it in the first place.
See PS5. For the same retail price Series X is smaller, quieter, cooler, and more powerful. Why? Because X was designed for its perf envelope. PS5 was modified to hit a new one ~20% higher.
The most they could have done is bump the TFs to something like 4.5TF. Anymore they would have had to use 16gbps GDDR6, revisit cooling and/or console size, resulting in something they would have had to sell at $399.
Also worth noting that with the tools maturing SX is consistently delivering higher than 20% more performance, so not only they ended up with a better overall designed console, they also are able to hit their target performance more often than a console that was overclocked to that level.
I think MS misjudged the performance they would get out of the Series S. Go back wand watch the video of Jason talking about the Series S when it was announced. He says that the Series S was designed to play games at native 1440P and 60 FPS. That is so far from the reality of it. Marvel is running as low as 720P on Series S. Microsoft has even removed all mention of 1440P and are now saying its a 1080P machine in its advertising.
No. We’ve been over this on here somewhere. That was for the title featured, Forza Horizon 4. Microsoft said they would be doing future ads featuring other titles which will have their respective resolutions/fps listed.
Here it is once again:
Ok, missed that. But they have stopped referring to it as a 1440P console. I’m not sure what went wrong, but it certainly hasn’t met expectations.