Discussion in 'Article Discussion' started by bit-tech, 16 Sep 2019.
Is it just me, or does ARM not have the slightest clue why anyone is interested in RISC-V?
Because they seem to be flailing about trying to address the problem with "no money down" promotions, while simultaneously placing terms in these free licenses that seem actively calculated to drive people away from ARM.
It's panicking. All it has is its IP, so it can't really give it away like RISC-V, while it's seeing a decent-size portion of its customer base flirting with or outright moving to FOSSi alternatives. Heck, just this month GigaDevice - fair-sized Chinese chipmaker - launched a pin-compatible drop-in replacement 32-bit microcontroller which is exactly like its previous model except RISC-V instead of Arm architecture. Swap the chips out, recompile for RV32, job done. Is there any wonder Arm's getting antsy?
And that's the rub. RISC-V can eat ARM's long-tail, but when it comes to cases where you have a legacy codebase and performance requirements beyond "it runs", ARM has a similar inertial to x86. It's not so much a choice of 'RISC-V is free, and ARM is $x00,000, so RISC-V!', but 'ARM is $x00,000 to continue with our current codebase & engineering experience and with fab-houses whose bread and butter is ARM, RISC-V is $y00,000 to completely rearchitect all our code, replace out devtools and retrain engineers (to a new architecture without a mature talent pool), and hope no snags are hit when it comes to etching wafers'. Even if the long-term cost of multiple 'x'es is more than 'y', if your company needs to survive then y being greater than x may prohibit a switch in the short term if all the budget you have is 'x'.
Whether ARM can survive as a 'squeezed middle' between commodity uC going to RISC-V and low-power-high-performance being edged ever closer on by x86 is another matter. ARM's efforts to break into high-performance have repeatedly flopped hard (the closest being Apple's ARM-derived cores, but the gap is further than benchmarks may lead you to expect), and their efforts to create the "yeah they're not so fast, but there's a lot of them!" server market is still chugging after a decade and multiple failed companies, and soon to be eaten by AMD's "there are a lot of them and they aren't crap!" chips. Smartphone, tablet, and smart-devices are the remaining middle ground, and while that market continues to expand there aren't all that many players in it to sell IP to.
Separate names with a comma.