This is all nuts.
A bit torn between "how the f can they mess it up so much"
A four month delay for a AAA title is beyond normal in this industry at this point. Like, it's not even a fuck-up if you haven't announced the product yet. There is no "delay" in the sense of a consumer product having one announced release date and then getting another. There is an increase in the internal development timeline.
Release size has come down, but release cadence sure as fuck hasn't. The hardware is still selling incredibly well, they
Wonder, the fastest selling Super Mario game in franchise history just released, and their two competitors are going out of their way to stick their foots in their mouths, right before shooting them.
Totally speculating - Nintendo wouldn't have delayed the hardware had Switch 1 launch not gone so well. They know a March launch can work (as every Famiboarder has spent the last 2 years reminding us), and that meant they didn't have to crunch to make Holiday if something wasn't working out. I don't agree with folks who said this, but just last week you had people in this thread ask why Nintendo would release Switch 2 with Switch 1 going so well?
Switch doing well doesn't dictate Nintendo's release schedule, but it certainly enables them to make long term decisions. Having an absolutely rock solid launch year for this thing is much more important than having a slightly weaker 6 months at the end of the Switch lifetime.
I think this also implies there will be a lot less cross-gen games
It's pretty much a confirmation that there will be little to no cross-gen period and that all of Nintendo's output will be for the Switch 2 immediately.
I don't think you're wrong, but i don't think you're right either. I can see the argument either way. Nintendo burns every cross-gen title to keep the Switch afloat during the delay. Or Nintendo delays the cross-gen games to give Switch 2 the fattest possible launch lineup, and just lets Switch have a softer final year.
I think the "delay it" strategy is the better long-term move, and Nintendo is in a good position to make the long term move. If you're a Metroid fan, maybe this is the best possible world, where a summer title that was going to get paved over by the Switch 2 launch now because the Switch 1's holiday title swan song.
That or we get "break glass, release Zeldas" this year, and I never played Twilight Princess, so I'm alright with that.
Then how was the Xbox Series X able to be released in 2020 with an RDNA 2 based GPU (RDNA 2 GPUs released same time as Series X) and Zen 2 CPU cores (which were only a year old)?
This question doesn't make sense, respectfully. Why would that matter? I don't see what you're getting at. The fact that RDNA 2 GPUs came out at the same time as other RDNA 2 hardware doesn't say anything about the design timeline, except that they roughly matched, however long they were. Do you have evidence that RDNA 2 design wasn't completed till the last minute?
An APU isn't existing chips squished together, it's a whole new chip. RDNA 2 isn't just one chip, it's
four, each with a separate tape-out and electrical design. The overall design (how is RT going to work? What's the memory subsystem going to look like? How are we going to update the rasterization engine?) has to happen before electrical design can even properly begin. And the electrical design process takes it's own time.
The fact that Series X had a "full" RDNA 2 implementation just suggests that APU tapeout was happening in parallel with the first RDNA 2 GPUs. Which seems like it was a highly accelerated design, as PS5, releasing the same time apparently needed to start the electrical design process before the RDNA 2 design (not electrical design, but feature design) was finished.
RDNA 3's first dies were
probably taped out 13 months before release. RDNA 3's MCM/chiplet based design was intended to accelerate the tapeout-to-release window for hardware, but one which won't be used by APUs (because they're sticking to monolithic designs). While none of this is definitive, it certainly suggests that a 2-3 year tape-out to release window isn't insane for a console chip.
That graph just maps out the timeline of development - work can be done on SoC design and engineering that runs in paralell to the design of the future architectural components themselves at AMD that have yet to be released.
I mean, it says when the SOC is planned on being taped out. That's what we're talking about. Honestly, when laid out in a calendar like this, a 3 year to launch tapeout seems
tight!
On the silicon track, they have a first tape out, 3 months to fix any issues discovered in the first tape out, 6 months to build device verification and QA processes, and turn those processes into functional factories for mass production of a chip. I mean, I'm not seeing any fat on that timeline. That a first sample to mass production schedule of 9 months.
The consumer device track also seems pretty fat free. 3 months to build an entire console prototype. 3 months for a revised prototype. 3-6 months for device verification and mass production processes. 9-12 months
The developer track is 3 months for an internal devkit, 6 months for a complete final devkit.
The problem of course is that not all of these tracks can happen in parallel. You can't revise the chip design without testing the console prototype, which can't be built without a chip. A0 tapeout being 3 years in advance, at least on this calendar, seems like the minimum.
This was the case for the Series X and PS5 which launched with very new components that had just been made available for consumers to buy separately. The Switch 2 will release with components that were available to consumers 5 years prior to its release date.
I have no idea what you mean by this. If you mean RDNA 2 and Zen 2, versus Ampere and A78C, I think you're comparing apples and a reese's peanut butter cup. Show me all the handhelds that users can buy with ray tracing and DLSS, and then we'll talk.