It's worth noting that this is based on ARM's Neoverse N1 IP, which is also used in the AWS Graviton2. The Graviton2 benchmarks damn close to the best AMD and Intel stuff, so this chip looks very promising [1]. It's really looking to be a breakthrough year for ARM outside of the mobile market.
Phoronix paints a very different picture, especially in non-synthetic workloads[1]. Gravitron2 looks like a nice speedup over the first generation but either the optimization isn't there yet or there are areas which need additional work to become more developer/HPC competitive. That said, I'm thrilled we have competition in the architecture space for general purpose compute again.
Disclosure: I work for AWS on cloud infrastructure
My personal opinion is that the Phoronix way places quantity over quality. Measuring performance is an important part of shining a light on where we can improve the product, but I get little practical information from those numbers, even when they are reported as non-synthetic. There are HPC workloads that are showing significant cost advantages when run on C6g, like computational fluid dynamics simulations. See [1].
I expect the scalability of HPC clustering to improve on C6g in the future, like C5n improved cluster scalability compared to C5 with the introduction of the Elastic Fabric Adapter. The Phoronix and Openbenchmarking.org approach doesn't give much insight into workloads like this.
My advice for an audience like folks on HN is is to test it for yourself. For me, being able to run my own experiments is how I come to understand infrastructure better. And the cloud lowers the barrier of running those experiments significantly by being available on-demand, just an API call away. I'd love to hear what you think, either in a thread here or you can contact me via addresses in my user profile.
Didn't go too deep into it, but the AMD cpus being compared are different. Anandtech has an AWS-only EPYC 7571 (2 socket, 32 cores each, 2.5ghz), Phoronix has EPYC 7742 (1 socket, 64 cores, 2.2ghz). On top of that, Anandtech is using another AWS ec2 instance and Phoronix is testing on a local machine on bare metal.
Still would be interesting to know what differences caused the gap in results, but their setups were pretty different.
Ugh, yes; one of the perks of an Intel monoculture was that at least you only had one target to worry about, and inter-generational quirks were mostly limited to minor things. Now we have to deal with "this was optimized for (Intel|AMD) and doesn't work on (AMD|Intel)" and "the devs tested this on their x86 laptops and then it got weird when we went to run it on ARM" and "ARM is less of a platform and more of a collection of kinda-similar-looking systems that are mostly compatible". Don't get me wrong, I'll take this over a monoculture, especially an Intel monoculture, but there are some bumps on the road to a more diverse ecosystem.
In my experience, the Arm ecosystem has an excellent track record regarding compatibility across conforming implementations of the architectures (e.g., Armv7-A, ARMv8-A). I can draw a practical comparison to MIPS, where I had to deal with a lot of variability based on various vendor extensions. This is reflected in the "-march=" documentation for GCC:
[1]: https://www.anandtech.com/show/15578/cloud-clash-amazon-grav...