3 ms·
That's probably the strongest counter-argument. New generations will learn programming on Arm and RISC-V (whether it's something like an iPad or a Raspberry Pi)
by mtgx 8y ago
That's probably the strongest counter-argument. New generations will learn programming on Arm and RISC-V (whether it's something like an iPad or a Raspberry Pi).
Linus very much sounds like a grumpy old man saying "get off of my lawn!" (as he usually does).
- deng 8y agoWell, it's difficult to not get grumpy when you hear the same promises for many years. It's not like "ARM for servers" is something new. The exact same things were already promised over a decade ago, and somehow, it never really came to pass. And I think his reasons are spot on: aside from very specific workloads, it just did not make sense to buy ARM-based servers. They were usually not much cheaper, and they usually did not save much power, compared to x86-based machines. Combine that with cross-development, and you have a loser.
- AstralStorm 8y agoExactly. Modern X86-64 chips got improved power saving, more cores to increase density at reduced power cost, much more powerful GPU for running compute. Plus, server space buyers are conservative, they want support of a big company behind their hardware. Someone the size of Cisco, Dell or HP. Or hardware that is used by Google or Microsoft. Combine that with anticompetitive long term agreements and you get the picture. Somehow network routers have a variety of architectures inside and nobody has a problem - because they have a bunch of reputable companies behind each of them. (Variants of MIPS, ARM, ARM64 and x86 are in the wild.)