Thought I’d never see the day when Firefox would match Chrome on Speedometer.
There’s also a few other benchmarks got a sizable boost. https://arewefastyet.com/
Thought I’d never see the day when Firefox would match Chrome on Speedometer.
There’s also a few other benchmarks got a sizable boost. https://arewefastyet.com/
@Molecular0079
I disagree with that. While your statement is correct in the sense you layed it out, it is not what I was saying. I am saying that benchmarks are testing what the benchmark is testing, not what the actual web usage of an individual is. And I am just speaking from my experience. Benchmarks do not represent reality. In example people have different configurations and addons, which has a huge impact on performance. Synthetic Benchmarks are often done with default configurations. I am not saying the results are wrong, they just don’t hit the reality (sometimes they do).
But what is the actual web usage of an individual? How do you define that? The problem is you can’t. People browse and read at different speeds, click on different things, open different numbers of tabs, visit different websites and webapps, etc. It is entirely subjective. I’d argue that this idea of the average web usage of an individual is actually even less representative of reality because of its subjectivity. Every one has their own personal experience, so which one of them is real? How do you even measure and compare these results?
If you wanted to build a fast car, you would optimize every aspect of the car piecemeal. You isolate the engine by itself, improve its fuel efficiency, RPM, and horsepower. Similarly, you isolate the body and chassis of the car, optimize it for less drag and lighter weight. Eventually you do this for all the different components and they come together to make an objectively fast car. If you started with the idea of “an average driving experience”, you’d end up with the typical use-case of going to the grocery store and picking up some food. Then EVERY car is considered fast. That IMHO is relatively meaningless.
Likewise, benchmarks help breakdown the problem of making a fast browser into smaller problems that can be tackled in a meaningful way. A lot of these smaller problems being solved or made faster come together to craft a truly fast browsing experience.
Then the argument here is that benchmarks should INCLUDE different configuration and addon permutations in their testing, not that benchmarks are useless.
@Molecular0079 That’s my point. You described why synthetic benchmarks are just that. I also didn’t speak about average web usage. Which one of the personal experience is real? Each of them! How do I compare? We don’t need to compare my experience to yours. What’s important is, that we compare what we want to find out: Firefox vs Chromium. And that’s what I do. I use both and in my experience Firefox wasn’t slower than Chromium.
Because browsers have many components that work much different under specific circumstances. Your car example is a perfect example of what I mean. Cars are different fast depending on the circumstances. Rally cars are optimized for Rally, tires are differently optimized for weather and street. Browsers are even more complex.
May I remember you what I wrote “I never felt” and “At least on my PC”. Do you feel a difference that Firefox is faster in usage than before? I bet you don’t. And that’s what I’m talking about. Synthetic benchmarks are only that, testing a specific use case. Don’t forget, I did not deny that the results are wrong! I believe that the components they tested on this particular setup is as fast as Chromium. It just does not represent reality (in my experience).
Engineering doesn’t operate based on subjective feelings though.
I really don’t think what I said aligns with the point you’re trying to make. If I may be so bold in my assumptions, I think your web usage probably is closer to a grocery store run than to a rally race, which is perfectly fine, I am not casting any judgement. But a rally race demands a lot more out of a car and each component needs to be benchmarked and optimized in order to get that performance. In this case, improvements in those benchmarks directly correlate to how the car will operate in the field.
I guess the point that I am trying to make is that synthetic benchmarks matter more for some usecases compared to others, but to say they’re always just synthetic and not indicative of reality is being dismissive of other people’s usecases that are probably more performance oriented than yours.
I just said that I switched from Brave back to Firefox because I felt that performance has been improving recently. I don’t think it’s a coincidence that it just so happens to align with the recent benchmark improvements. And no, my add-ons have not changed.
@Molecular0079 I think we won’t get much farther than this.
I disagree. Engineering requires subjective feelings too, depending on what is being engineered off course. But that’s going too far now. My points are still valid and your replies are confirming, such as you feel a difference (you did not measure it, you feel see it), and I do not. That’s my point.
Engineering uses objective measurements and techniques to work towards a certain subjective feeling, but that’s the extent of where subjective feelings matter.
No, your point was that synthetic benchmarks are just synthetic and do not matter in reality. My point is that synthetic benchmarks matter a lot in some usecases, particularly in high performance oriented ones. These two statements are not the same, regardless of how many times you say they are.
@Molecular0079 You are wrong. Synthetic benchmarks do not represent the normal daily usage of users. I did not say the results are wrong and could not help in finding slow parts. But they are not indicative to what browser is faster than the other, because no single benchmark can answer than question. I think you don’t get what I’m saying here. As I said, I think we won’t get much farther than this. My statements are all correct.
The “synthetic benchmark” that OP mentions is simply a to-do list app written using different web frameworks, populated with some Todo items and then reloaded. This is done many times and then measured. I don’t know about you, but this seems pretty real world usecase to me.
Of course not, but a sum of many of what you call “synthetic” benchmarks will give an indication of which browser is faster.