Earlier this month, Microsoft Edge users, including Neowin readers, found themselves annoyed when the browser would fail to load websites with out of memory (OOM) errors. Fortunately, though, Microsoft was quick to react to the issue and proactively fixed the bug within a day, explaining how a deprecated Defender feature was responsible for it.
And while that was merely a bug, it is not a stretch to say that the modern web is more intense than ever and you can no longer get away with internet surfing on an entry-level PC or mobile phone.
The recent release of the Speedometer 3.0 browser benchmark reflects this with major industry players Apple, Google, Microsoft, Mozilla, and more, coming together to create a benchmark based on current web workloads.
Rising security standards, like Microsoft making 2048-bit keys mandatory on Windows, will also contribute to this. Hence, the recent internet speed bump by the FCC is certainly welcome (test your speed in this article).
While a browser benchmark like Speedometer 3.0 can tell you how fast a web browser is, at the end of the day, a lot of the performance is dependent on the hardware itself. Twitter (now X) user Dan Luu measured this to see how well different classes of hardware can handle the modern-day internet.
Although Neowin was not on the list of tested sites, there were several popular ones like Quora, Twitter, Reddit, Medium, and more. The tested CPUs were Apple's M3 Max and M1 Pro Macbooks, as well as an M3 with the throttling set to 10x via Chrome Dev tools, indicated by "M3/10" in the chart below.
Aside from those, Tecno Spark 8C with an octa-core UniSOC T606 chipset and an Itel P32 with a quad-core MediaTek MT6580 chipset, were also included from the mid-range and low-end SoC categories.
If you are wondering, here is what the measured metrics are meant to test according to Dan Luu:
.. every row represents a website and every non-label column is a metric. After the website name column, we have the compressed size transferred over the wire (wire) and the raw, uncompressed, size (raw). Then we have, for each device, Largest Contentful Paint* (LCP*) and CPU usage on the main thread (CPU). Google's docs explain LCP as
Largest Contentful Paint (LCP) measures when a user perceives that the largest content of a page is visible. The metric value for LCP represents the time duration between the user initiating the page load and the page rendering its primary content
What's surprising is that even an 8-core SoC has struggled with the Tecno Spark 8C failing to load Quora. And the 4-core MediaTek managed to fail nearly as many times as it had passed, highlighting the level of bloatware the modern web has, and the difficulty of surfing the internet on an inexpensive device.
Source: Dan Luu (X / Twitter)
19 Comments - Add comment