I can’t remember when they changed it, maybe after Mojave, but with Catalina there were definitely opportunities to upgrade safari without the OS. They released the leanest safari before they released big sur
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Chrome for M1 Macs Runs Up to 80% Faster Than Rosetta 2 Version
- Thread starter MacRumors
- Start date
- Sort by reaction score
I can’t remember when they changed it, maybe after Mojave, but with Catalina there were definitely opportunities to upgrade safari without the OS. They released the leanest safari before they released big sur
Did you use butterfly keyboard?
To be tracked completely by Google.
I told you guys Apple was innovating with silicon, but people consistently dismissed it. Their chips are and have been, amazing. The A series chips are what iPhones such a joy to use for such a long time.
The sad truth is a year from now if Apple has nothing new to offer (outside of the M chip) then the usual haters here and everywhere else will say Apple can't innovate, while they quietly hope nobody calls them out for ignoring the NON-innovations from Dell, HP, Microsoft and Lenovo.
After years of almost zero innovation (just spec bumps and repackaging) I was skeptical of Apple's M1 claims.
BUT FOR ONCE THE HYPE WAS REAL!!!!
The M1 machines are revolutionary (power and thermal).
Ordered a 16/512 Mac Mini M1 which hopefully will be here before Xmas.
Apple finally DELIVERED - REAL INNOVATION!!!!
Makes sense that this is one of the apps where native vs rosetta would be so distinct.
A lot of a browser's job these days is to just in time compile JavaScript into a more CPU friendly instruction set.
If Rosetta Chrome is compiling JavaScript into instructions optimised for x86 then that's a lot of wasted processing.
A lot of a browser's job these days is to just in time compile JavaScript into a more CPU friendly instruction set.
If Rosetta Chrome is compiling JavaScript into instructions optimised for x86 then that's a lot of wasted processing.
A series chips have always been innovative, among many other things Apple has done. Their learning from those chips brought you M1.After years of almost zero innovation (just spec bumps and repackaging) I was skeptical of Apple's M1 claims.
BUT FOR ONCE THE HYPE WAS REAL!!!!
The M1 machines are revolutionary (power and thermal).
Ordered a 16/512 Mac Mini M1 which hopefully will be here before Xmas.
Apple finally DELIVERED - REAL INNOVATION!!!!
Is that screenshot for real??? Chrome is taking up nearly 60% of the M1's CPU usage!!!
Not quite. Activity monitor showing 100% means one core is used 100%. So Chrome doesn't even fully use one of the four performance cores. CPU usage on an M1 can go up to 800%.
And note that if you use 100% of one of the power saving cores, which run at maybe 10% the speed of a performance core, you also get a display of "100%" because this software doesn't expect cores running at different speeds. Geekbench has a related problem; that benchmark shows the number of cores, and the amount of RAM and caches, but it shows the cache size of the low power cores, not the much larger cache size of the high power cores.
So what I take from this is that Activity monitor whilst saying % CPU and showing it's usage, it's not actually reporting the usage correctly because only 1 core is being used? If that is the case then shouldn't activity monitor be showing how many cores are being used and what core is being used because to the untrained eye, when a computers activity monitor shows a heading of % CPU, it's implying that ALL cores of the CPU are being used.
The Intel Core Duo iMacs and Macbook Pro from early 2006 were superior to the G5 iMac and the G4 Powerbook. And soon core 2 duos (iMac, Macbook Pro) and 4-8 core Xeon Mac Pros followed, making G5 PowerMac obsolete. It took less than a year. My G4 Powerbook compared to a core duo Macbook Pro, ugh, it was ugly.
M1 is fast, but the M2, or D1, or whatever they are called, next year will remove Intel for good.
The M1 is more impressive than the intel Macs were in early 2006. But the early intel Macs obliterated the PPC Macs quickly, they were a lot better immediately. The M1 (and successors) have the chance of putting Apple ahead of the competition, in a way that Apple with Intel never had. There would always be a faster or cheaper PC. There won't be any PCs with M1 CPU. (There will be ARM64 Windows computers though, and Apple won't be the fastest forever)
I’m not sure how much it says about native vs Rosetta even. Chrome (or web browsing, in general) is a worst case scenario for something like Rosetta 2, as there’s very little code Rosetta 2 would be able to optimize at install time. JavaScript performance, for instance, would have to be done via V8 for x86 as Chrome usually does then JIT compiled for ARM. Likely, much of the rendering engine code and any code dealing with extensions or plug-ins is going to take a penalty hit, as well. I am interested in seeing comparisons for other categories of software, though, as we’d get a much better picture of how good Rosetta 2 really is from more static types of software. Expecting a web browser to run fast in something like Rosetta 2 is a lot like expecting Python or a JVM running in x86-64 to run fast in Rosetta 2.
Dang, and M1 doesn't seem to have hyperthreading. It gets even more confusing with that involved.
Say only one process is running on your machine at 800% on an 8 physical core, 16 logical core machine. If you run two such processes at once, you'll see each using 800% like before, but each is getting half the time vs before PLUS whatever benefit the hyperthreading brings, which depends heavily on the task and could even be slightly negative. For that reason, flexibly parallelizable tasks will usually set num threads = num physical cores, not num logical cores. Even more confusing, with one task running you'll see 50% overall CPU usage even though nearly the entire CPU is in use, and with two you'll see 100%.
You are confused. Both versions of Chrome in this case ran on the same computer. This report has nothing to do with M1 vs x86 comparisons. It just exposes a use case where Rosetta2 will be a poor solution for running x86 apps. This might explain why Google was in a hurry to build a native version of Chrome for new Macs.
Have you been following along? Most applications do not have anywhere near a 50% hit, more like 15-20%
I’ve heard the arguments about JavaScript, maybe they are correct, maybe not. Seeing chrome is such a dog to begin with, maybe google finally got it right on the Apple silicon version and hasn’t rolled out the more professional code to the Intel version yet. Just saying, I’m guessing like everyone else who does not have access to source code
Hyperthreading only serves to keep the pipeline full when commands on threads are waiting for completion. With 8 cores, there should be little benefit on ARM because of the shorter command processes, so you simply might not see it
That said, on a Mac, I don’t really bother with browsers other than Safari. Speed and battery performance, Safari outperforms any other browser (especially battery performance, it’s like Google doesn’t know laptops are a thing or something). On non-Macs, I’ll use Chrome more often, but I prefer a privacy oriented fork like SRWare Iron. On Linux, there’s a lot to be said for using Firefox, though. It feels more like a native application than Chrome (and it’s a sorry state when Firefox feels more at home on a given platform than another program).
