Hacker Read top | best | new | newcomments | leaders | about | bookmarklet login

Also the delta between theoretical performance and benchmarked performance is much smaller for Frontier (AMD) than for Aurora (Intel).

That being said, note that the software is also different on the two computers.



sort by: page size:

The two compare differently depending on the benchmark that is used.

Do we have any specifics on how the perf compares to Aurora? Is the 2x mentioned on the blog based on any specific benchmark?

There's a bit of drama in that there are unofficial reports of two systems in China with higher performance [0], the arXiv paper listed below talks about a 40 million core system with around double theoretical performance than Frontier, and there's apparently a second system online with similar performance. I personally suspect that they didn't submit benchmarks to the top500 simply because those don't run well enough in the systems

[0] https://arxiv.org/pdf/2204.07816.pdf


Not sure about this benchmark, but the differences in the AMD presentations have been Intel unpatched comparisons.

You're comparing different software rather than different CPUs.

Hmmm... these are two different benchmarks.

Do you have any benchmarks illustrating the difference?

Looking at the data in your second link, the performances differ by less than 3%. Looks like paulmd’s assertion that both are the same cpu core is fairly well supported by the data.

It's not similar at all. AMD's cores are 10-20% slower while Niagara was 80-90% slower. And AMD isn't intentionally slower; they designed the Zen core for maximum single-thread performance but they just didn't do as good a job as Intel because their budget is vastly smaller.

Do you have appropriate benchmarks? I would appreciate seeing the comparison.

That is true, perhaps its a difference in drivers and operating systems?. it hard to tell if the benchmark is bad without having the same setup to test against because anything could impact it really.

Are you aware of any interesting benchmarks? I’m curious what the performance difference is in different kinds of applications.

I'm interested. Any links with benchmarks or example applications demonstrating the difference?

Interesting! It'd be nice to benchmark to see if there are any practical difference in terms of speed.

Could you elaborate on that? For some reason that eludes me, realistic performance comparisons between different CPUs is a hard to find information.

We haven't migrated yet but we expect to do some benchmarking this quarter for Aurora.

For EC2 we run on spot and spot c5.metal are cheaper per vcpu than c6g.metal, so we haven't prio'd benchmarking our compute loads.


Can be very different. I recently benchmarked it with a simple python based benchmark (as I use python most often) and in some cases the differences were even 2x.

Some results and benchmark here: https://github.com/vprelovac/python-speed

These were coming mostly from the fact that some providers updated to the latest gen AMD chips which was making big difference compared to 3-4 yr old Intel chips.


Really? Benchmarks I've seen put it in the same ballpark

Nice, thanks for clarifying, I stand corrected. This is great to hear/see, do you guys also have benchmarks yet?
next

Legal | privacy