Why is RISC-V shilled as the replacement for x86 by Schizo's? Power PC is just as open , has better performance and has mature software support so you can actually run a Linux desktop or Server on it in 2023. I feel like RISC-V is more a alternative to ARM rather then x86.
It's just better okay???
IntelJeet thread?
>Intel
Power PC is IBM retard
Are you retarded?
>Power PC
PowerPC is extremely expensive, and really only a product for commercial companies. Even then it's extremely expensive to buy an I/P series box over something from Intel/AMD or even Amphere at this point.
PowerPC is defiantly expensive .I looked into replacing my old server with a Talos 2 and it would cost 6k vs 2.5k for a Xeon machine. The same price to performance is true for RISC-V right now but I believe both RISC and power PC are expensive due to lack of volume and that would change if they got mass adoption
>expensive
Or you can go for the soft core approach:
https://github.com/antonblanchard/microwatt
>defiantly
not everything is about some lamp/node app on a linux server, and not every instruction/abstraction is the same on each architecture
>Why is RISC-V shilled
The RISC-V people have control over academia, and if you don't bleat "RISC good, CISC baaaahd" they will fail you on your exams. You can tell it is academia heavy since it is all piles of discarded VHDL/Verilog projects on Github plus a few Google/whatever projects that are marketed for what it is worth and then some. Most businesses have not joined this religion.
Reality is that x86 and x64 are radioactive sludge but are saved by super advance micro architectures, and will remain with humanity for years to come. Similarly ARM is established with a gigantic library of tools, libraries, compilers, debuggers, whitepapers, design rules, reference manuals and more, which is the sum total of 30 years head start.
When RISC-V has caught up with these, the competition has moved another 30 years forward.
Because they don't understand that open ISA != libre implementation.
CISC is shit, there is no reason that it is used other than legacy. Although even that use case is falling away as ARM chips into the client market. Even Intel has admitted they have failed multiple times and have tried to come up with a new architecture to replace x86 (but both iAPX and IA64 managed to be worse).
That's absurd considering the Talos 2 is what, 5 years old at this point? You could buy a car or pay for 6 months rent at an apartment with that money.
This brings up a good point. The poison pill of RISC-V is the ability of manufacturers to add in proprietary instructions. At some point we will have competing architectures based on RISC-V, not RISC-V itself.
Best blue ball machine v1.337 🙂
>CISC is shit, there is no reason that it is used other than legacy.
Yeah, this is kinda true. Modern processors operate kinda like RISC core + legacy support rather than CISC as it's originally intended.
>Modern processors operate kinda like RISC core + legacy support
No, they do not. Pre Pentium-M (which became Core architecture) at least tracked memory uops separately from arithmetic, but modern CPUs do not do this - internal uops are complex mem+arith. And AMD processors never tracked memory uops separately from arithmetic.
>CISC is shit, there is no reason that it is used other than legacy.
ARM64 is CISC though. Way too many instructions that need more than 1W/2R register file ports. Typical instruction streams is more complex than what Intel has (excluding decoding).
Execution of rarely used CISC instructions are significantly different than the optimized more common instructions which is what I was referring to.
x86 processors have been RISC internally for decades
Agreed. I can download Ubuntu/Suse/ Fedora/ Debian for ppc64le but not for RISC-V
> Power PC is just as open , has better performance and has mature software support so you can actually run a Linux desktop or Server on it in 2023.
It's way too expensive.
>It's way too expensive.
See
and also
https://github.com/orgs/OpenPOWERFoundation/repositories
The latest iterations of PowerPC seem to be getting more and more focused on HPC, if this continues it will be increasingly more difficult to develop average and low profile designs.
Meanwhile RISC-V is highly modular. Theoretically this high level of customization will be able to cover many distinct market needs. Also, because of this, things will need be open and portable all around or it will fragment and fail.
>https://en.wikipedia.org/wiki/RISC-V#ISA_base_and_extensions
>HBC focused
Power 10 sadly this way but since power 10 is a open platform raptor computing and solid silicon have been working on a instruction set compatible platform with desktop and servers in mind.
This isn't useful for the reasons us Schizo's like Power . It's about having trustable firmware at the lowest level. Emulating power PC doesn't help with that .
Was Raptor not skipping Power10 altogether? Even if they get a replacement for the off-chip OMI DRAM bridge, the PPE I/O processor is on chip.
Hopefully these design decisions were due to COVID-19 and Power11 is properly polished.
Raptor is working on a platform that is PowerISA 3.1 but doesn't use IBM processors or architecture. Sort of a fork of Power 10 made to be compatible but without all the problems hopefully. The new CPUs will be from a start up so hopefully they can compete with IBM , Intel and AMD for performance but who knows.
https://www.talospace.com/2023/10/the-next-raptor-openpower-systems-are.html?m=1
>Theoretically this high level of customization will be able to cover many distinct market needs.
Practically, it means that it's never a single hardware architecture but a family of them. In particular, no OS designed for working with an MMU is going to be happy on hardware without one, so if that hardware module is optional (and it usually isn't there on anything for embedded) then you'll never have a unified OS. Firmware is just software for a lower level of abstraction.
Schizos just don't understand how different things are. Being able to share an ISA is nice and helps get a compiler toolchain going, but it sure isn't the whole story.
>I feel like RISC-V is more a alternative to ARM rather then x86.
Usually semiconductor companies work on the next big thing rather than what the customers have currently. Go to any semiconductor company and talk to marketing people, they will all say "Yeah PC market is not growing as fast as we'd like" regardless whether it's CPU or any other peripheral. They always ask "So we're selling this already, what's next" and right now what's next is more power efficient processors or that's what it seems like. Hence you see a lot of RISC-V stuff.
At the company I'm working for, we use RISC-V stuff as a replacement for ARM stuff but we're not in CPU business. I just know it because I have a lot of friends working at Intel and AMD.
Your right, RISC-V could be a viable replace for lower power ARM SoCs. I can see a lot of appeal avoiding the licensing fees of ARM.
In the very long run though, x86 is a stagnant architecture/platform. ARM is slowly eating away the x86 market share from the bottom up. It's not inconceivable that RISC-V could one day do the same.
I wouldn't hold my breath though.