> If a microprocessor architecture consensus finally exists
I don't think that's a good goal to have. Without competing architectures we'll enter a microprocessor dark age. Although the glory days of Alpha and IA64 are over, plenty of that experience was used to bolster today's instruction sets. Diversity is a good thing.
Microarchitectures are the new architectures. The game these days is all about finding a way to make your code go fast on the latest and greatest AVX thing, while still maintaining compatibility for old machines via CPUID. Getting those 10x speedups (sometimes 50x w/ things like crc32 pclmul) is much more fun than caring about things like s390x (i.e. doing IBM's work for them), SH (i.e. folks emulating Sega Saturn on FPGAs for fun), MIPS (hacked home routers), etc.
I'm not sure I buy that, you can have a lot of competition on the same ISA. After all on the desktop/server x86 has been king for a while, and on embedded architectures ARM has all but taken over the competition. Yet there's significant competition on both fronts.
Meanwhile attempts at creating new ISAs over the past couple of decades has been met with relatively little success, or only in very specific niches. Itanium being an obvious example. People talk about RISC V a lot but in practice it's not really making a dent into the ARM market share yet.
These days competition appears to come mostly from companies reimplementing and extending existing ISAs with better performance.
I would just like to add in that RISC-V is still a lot of talk because there are still some important standards that really need closer attention before we start mass-producing server CPUs. Specifically, the Platform-Level Interrupt Controller[1] and the Vector Instructions[2]. The former is super-important because we now know interrupts/traps can be used in Spectre/Meltdown style attacks, especially if your chip has something like a DSP/iGPU, which can be used in covert side channel attacks. The latter will be important for the same reason that SSE/AVX is important today.
Just my opinion, but SiFive has been a huge driving force in even giving RISC-V a shot at breaking into server space. Most other companies (Google, Nvidia, Western Digital) have invested in RISC-V with the apparent intent of focusing on building microcontrollers/embedded processors. AFAIK, there's only one player who is investing in building RISC-V server-class CPUs and that's SiFive.
Please correct me if I'm mistaken! I've been following RISC-V pretty closely for the last year or so and there's quite a lot to catch up on.
I don't think that's a good goal to have. Without competing architectures we'll enter a microprocessor dark age. Although the glory days of Alpha and IA64 are over, plenty of that experience was used to bolster today's instruction sets. Diversity is a good thing.