Alibaba Crafts A 16-Core RISC-V Chip @ 2.5GHz

[Michael]

For those interested, finally added CoreMark to PTS today - https://openbenchmarking.org/test/pts/coremark-1.0.0

[Michael]

phoronix maybe you could run coremark on some modern x86 processors as a comparison, in multi-threaded and single-threaded modes, maybe also at a locked clock frequency to more accurately measure coremark/mhz. I would assume coremark is small enough to run entirely in the L1 (or L2?) cache, due to not using much memory since it's a embedded benchmark.
That sounds like it would make an interesting comparison -- seeing how far risc-v or arm is behind ryzen 3000 or intel 9th gen in this admittedly limited fashion.

https://openbenchmarking.org/test/pts/coremark-1.0.0 and will have some more CPUs in there shortly.

[coder]

The fact that this design is proprietary is a perfect example of why permissive licenses are bad. If RISC-V was copylefted, people would be going through the publicly available design right now, and could consider grouping together to order these chips from a fab.

I don't think it works like that. RISC-V is just the ISA - the functional description of the architecture. An implementation thereof wouldn't necessarily inherit the license, in a similar way (although not exactly analogous) to how code compiled with a GPL compiler doesn't become GPL'd.

And if the ISA were licensed in some way that forced all implementations to be open source, I doubt it would have the kind of traction that it's been getting. I don't care to argue this point - it's just my opinion.

[coder]

For those interested, finally added CoreMark to PTS today - https://openbenchmarking.org/test/pts/coremark-1.0.0

Um, that doesn't seem comparable to what the article quoted:

This Alibaba design achieves a 7.1 Coremark/MHz rating, a great deal faster than any other publicly announced RISC-V processor.

By comparison, your Xeon E-2288G gets 77.2 Coremark/MHz, if we assume it ran at 5 GHz. So, maybe you're running with different parameters?

[coder]

Anyone willing to analyze this two sentences and write down on how many levels this is wrong (or clueless)?

It's a given that such statements will be challenged, on here. However, you don't need to bash the original poster - just stick to the facts. We were all n00b, once.

[Michael]

Um, that doesn't seem comparable to what the article quoted:
By comparison, your Xeon E-2288G gets 77.2 Coremark/MHz, if we assume it ran at 5 GHz. So, maybe you're running with different parameters?

I'm running a stock threaded CoreMark build. So any difference would be tuning by Alibaba. But seeing the Xeon 10x faster than a RISC-V chip wouldn't be surprising considering it's a new(er) architecture and that the software support and compiler around RISC-V are far from mature, etc.

[coder]

I'm running a stock threaded CoreMark build. So any difference would be tuning by Alibaba. But seeing the Xeon 10x faster than a RISC-V chip wouldn't be surprising considering it's a new(er) architecture and that the software support and compiler around RISC-V are far from mature, etc.

IMO, for any half-decent out-of-order CPU, 10x is way too high... unless CoreMark can seriously utilize stuff like AVX.

Should we be dividing by the core count? That would yield 9.65 Coremark/MHz, if we treat it as 8-core. If 16-core, then half that... but obviously those aren't 16 full cores.

[schmidtbag]

Anyone willing to analyze this two sentences and write down on how many levels this is wrong (or clueless)?

He's not totally wrong. RISC-V is basically the anarchist's CPU architecture. Without some sort of order and control, anyone can do anything to the architecture, whether that's good or bad for the user. Companies that do their own tweaks to it (like Alibaba) have no obligation to share or open-source what they're doing. When something is an unregulated free-for-all, selfishness kicks in and nobody wins. The core architecture may remain constant, but nobody gives a shit about the core architecture if the rest of the hardware (and its associated drivers) doesn't work for your needs.

Granted, I don't think RISC-V was really ever intended with the average user in mind. It's always been the most appealing to companies that want a cheap way to tweak hardware for their specific needs, like for embedded devices. ARM and MIPS were ok for this, but those have enough licensing issues and restrictions that they aren't perfect. So, I don't necessarily think RISC-V is bad, but for us end-users, I think we should look elsewhere. At least, until someone releases something based on RISC-V that is comprised entirely of open-source hardware.

[coder]

When something is an unregulated free-for-all, selfishness kicks in and nobody wins. The core architecture may remain constant, but nobody gives a shit about the core architecture if the rest of the hardware (and its associated drivers) doesn't work for your needs.

Could be, but I guess the idea is that it's in their self-interest to stick to the officially-blessed ISA + extensions, specifically so they don't have to do a bunch of tool-chain, driver, and kernel work. We'll see how well that works, in practice.

I can say I'm at least glad they're using RISC-V, rather than some entirely homegrown ISA. Whatever replaces x86/ARM, in China, isn't likely to stay in China. It'll start leaking out inside various devices and hardware they make. Eventually, they might start exporting computers built around it. Even if we don't adopt them in North America and Europe, they could make serious inroads in South America and Africa. At which point, it's only a matter of time before we end up having to at least deal with them, in some way.

[programmerjake]

...
At least, until someone releases something based on RISC-V that is comprised entirely of open-source hardware.

Working on that (libre-riscv.org)