Currently there are few microbenchmarks on http://www.peachpie.io/benchmarks Expecting to be updated soon

PHP 7 improved performance by around 3x compared to PHP5, so I wonder how true that still is.

The benchmarks http://www.peachpie.io/benchmarks show peachpie outperforming PHP7 easily by 100x.

For some of the operations highlighted, which are (naturally?) selected to impress. For others, the multiple is in single digits.

WordPress doesn't run 100X faster on Peachpie, but it'd be very interesting to see "real" benchmarks. Even a 2X improvement (vs the latest PHP 7, of course) would be compelling.

On Azure, without Opcache enabled. That's not a very good comparison.

I would strongly suspect that the fact it's without opcache by itself makes it pretty much meaningless for any kind of a real-world application.

Yes, and look at the avg response times. It looks likely that the webserver can't handle the load. The number of concurrent requests should be limited.

While I do know that Azure is owned by Microsoft I'm not so sure on why it's such a valid argument.

In the end it's not like they're actively trying to ruin languages other than C# on their servers? Especially since that would ruin it for a lot of customers?

I would guess that .NET is more optimised on Azure, whereas PHP is not. So you're comparing an optimised installation of .NET against an unoptimised installation of PHP. In general, a benchmark saying "compares various metrics" is pretty meaningless.

I know, but is there any proof of this? And how can we be sure that the .NET environment is better optimised than the PHP environment? I would find it weird for them to shoot themselves in the foot like that. Since if someone were to make a decent blog post that showed proof they'd only have C# programs on Azure...

This all being said they're pretty transparent with their benchmarks. They supply all the code that they've used and are even comparing it to another project which aims to do the same.

I guess I'm just a bit angry because people are just saying "It's unfair" while they can easily do the tests themselves and then call someone out for it.

It's less about being unfair, more about it lacking much value as a benchmark.

The context of the parent to which I originally replied was suggesting that .NET was 100x faster than PHP 7. So I'm pointing out that claim is rather like saying "I'm faster than Usain Bolt". You're right that I don't think the original article was making such a claim.

Unfortunately those benchmarks don't compare against HHVM :/.

We will produce these benchmarks soon, but HHVM is pretty difficult to configure and we have not focused on optimizing our compiler much yet, so the performance is likely to change drastically over the coming months.

a few comments above said php7 was running without opcode cache, hhvm is even more about caching setups than php7...

Tried the TechEmpower plaintext benchmark using Peachpie: Requests/sec: 305,612.35

https://github.com/benaadams/PeachpieBenchmarks