Vivaldi Benchmarking
-
Chrome recently claimed a large speed boost based on compiler refactoring (see https://forum.vivaldi.net/topic/69199/chromium-going-clang)
I don't use Vivaldi based on benchmarks, but I given my once-upon-a-time interest in benchmarking when I ran the old Opera wiki, I was curious to compare Chrome beta, Vivaldi snapshot and Opera developer. They all use the same engine so one may naively expect broadly similar performance, but as alluded to by the Chrome team, compiler tooling can make a significant difference. I used Speedometer and MotionMark from https://browserbench.org on an Intel Macbook Pro running macOS 12.3 — plots are the results as returned, which include what I assume are the standard errors of the measurements (I wish they'd use the 95% CI):
At present both Chrome beta and Opera Developer use M100 whereas Vivaldi snapshot is using M98. It will be interesting to see once Vivaldi gets to M100 what the performance difference is, but currently these differences are huge and, eyeballing the error bars, probably highly significantly so (in scientific terminology). Maybe this is platform specific (different compiler toolchains and all that), it would be interesting what other platforms got.
Vivaldi is still magnitudes better in terms of workflow and feature set, and benchmarks mean little in day-to-day use, so long may her baroque tones be heard across the lands!
-
@non-troppo I get with 5.2 Nightly (has Cr/100) for Speedometer 99,8 with Standardprofile (Ad/Tracker activated on Welcome page).
-
@doctorg — Interesting, though I think it is better to compare relative performance on your machine, i.e. Chrome vs. Vivaldi as the absolute number will be affected by your hardware as well as the OS... i.e. for me Vivaldi snapshot is only 67% of the performance of Chrome on Speedometer; maybe with 5.2 nightly it would be 85% or something?
-
Possibly Vivaldi is somewhat slower than other browsers in these tests, but at least I find it irrelevant, since in practice the differences are tenths of a second and it is well supplied by the functionality of Vivaldi, much superior to any another browser.
-
With the new 5.2 snapshot that contains M100, there is a nice bump to performance, with motionmark now equivalent to Opera Developer and Speedometer creeping up too...
-
@non-troppo Mine scores 94,9 in speedometer on a laptop with an i7 from 2017 and GNU/Linux,
but problem is last time I installed Hipst(op)era Developer to show off its beautiful wallpapers a year ago or so, Ι was confronted with Erinyes , from their wrath I managed to only escape with pleads for leniency, so I can only compare with Chromium when it reaches 100 (now on 99)
-
You mean gut feeling? Absolutely, yes.
-
Speedometer seems to be unreliable, since network latencies (and the missing http/2 for example) can hit hard on the final score.
Edit: Oh well, shame on me. Network latencies obviously have no effect...
-
@gnappi said in Vivaldi Benchmarking:
I know for a fact that most often benchmarks are made for the likes of PC mags and speed fetishists who care more about numbers than actual performance
Yes, I think you are correct. On top of this is the fact that "performance" is incredibly subjective, we humans end-users can be easily deceived. However for developers themselves, benchmarking can be critical, and obviously the large amount of resources Google and others put in all this automated tooling serves a critical purpose during development, it isn't only about showing off...
-
Benchmarks conducted on configurations and systems which differ to mine and no doubt some others are quite useless in my opinion.
I am the master of my own benchmarks on my own individual system and make as assessment from what i myself observe and vivaldi is more than good enough for my personal usage.
-
-
@npro — yes it seems the differences may be smaller on Linux compared to macOS (differences in the build tools maybe?)...
From a geeky perspective it would be interesting to hear from a Vivaldi developer as to whether they may consider tweaking their build toolchain?
-
Ppafflick moved this topic from Let's talk about Vivaldi on -
I cannot believe that current Opera is faster than Vivaldi. The last time I used it (after it had been sold to the Chinese), trying to download anything was almost glacial. Page loading was also slower than with other browsers.
* 
