with out-of-order and branch predicting CPUs, those metrics are highly questionable.

Better "benchmark" real stuff: CPU bound AAA games, etc.

Most AAA games have not been ported to RISC-V yet.

But I bet pytorch has been, and also openCV and other compute-heavy stuff. (And Doom, of course.)

I'm guessing pytorch probably doesn't branch as much as a game.

One of the best "real" Benchmarks is compiling the Linux kernel, the issue is that on an emulator/verilog simulator that could take a very long time.

I also use this benchmark for a real world estimate. Here's some numbers. For the pretty ginormous Ubuntu kernel config for RISC-V, it's 7.5 hours on the HiFive Unmatched (at 1.5 GHz) versus 25 minutes cross compiled on a E5-1650 V4 Dell box.

1.5GHz, oooof... I guess they really need to catch up on the GHz front, not only the micro-architecture. Once they are in the 5GHz range, we will see how far the micro-architectures and compilers have to go.

5-6 GHz is a party trick of low core count CPUs (e.g. 4 or 8 cores) that use 250+ W of power to get less than twice the performance they can get at 25 W.

There is a sixty four core OoO RISC-V chip (SG2042) being demoed on a dev board in the last weeks. It runs at 2.0 GHz and has quite a bit of aggregate performance, at lower energy usage than an 8 core i9 running at 6 GHz.

AAA games want raw thread performance more than a huge core count.

I am afraid, but the GHz of RISC-V core for high performance desktop will have to go significantly higher.

That said, for self-hosted servers, and even some servers, RISC-V may already be ready for prime time.

There's nothing (other than loss of efficiency, which matters in these server-focused chips) in the way of higher clocks.

oh, and when I said compilers, that's for legacy code, since with a standard world-wide ISA, it is the start of the assembly era.

Huge table here: https://en.wikipedia.org/wiki/Instructions_per_second

Best x86 and ARM are around there.