> It takes advantage of modern hardware, parallelizing the work across all of the cores in your machine. This means it can run up to 2 or 4 or even 18 times faster.
@bwDraco Eh... it should scale DOM nodes but not for operations on each node.
which is still great, mind
Also, depending on how they implement the work stealing, you'll hit a limit somewhere where the contention caused by adding more cores ends up starving some cores
Though that's probably in the region of 16+
wait, they were stealing fairly large chunks, right? that should mitigate it
It's a work in progress and an active area of research.
The best part? This lets web browsers take advantage of the new breed of manycore consumer processors that Intel and AMD have been coming out with recently.
Now everyone benefits from extra cores.
I doubt it'll actually max out an i9-7980XE or Threadripper 1950X but this makes 6C+ processors more relevant to the average consumer.
It should be obvious what processor this refers to. This sort of near-perfect core scaling in a web browser's rendering engine represents a huge technological advance.
Hit Reload All Tabs and peg 36 hardware threads with just a handful of (complex) web pages open?
at least without changing your entire paradigm and using a pure language like Haskell
The downside is this assumes Rust itself has no bugs
(hint: they've already found a few of those)
@bwDraco It's not entirely new, but doing it in an accessible way is interesting.
Unfortunately, it's also not 100% -- and it relies on the user following some fairly strict rules
IIRC completely detecting whether concurrency is safe would be equivalent to solving the halting problem
Here, they've managed to solve a subset of the issues with concurrency (mostly revolving around integrity; data corruption), but there's still availability (deadlocks, etc.) to consider
@Avery debatable, but I meant even if the article isn't about that subject. E.g. it was discussing the possibility of using a stable ABI/API for drivers. Somehow after 10 comments people are calling each other idiots and spamming anecdotes.
Driver issues in Windows: All the time. Hunt for standalone drivers. Huge, bloated installers. Issues ranging from simple to ibzarre. Lots of troubleshooting guides on Google.
Driver issues in Linux. Rare, but when they happen, oh dear.
Howdy folks. I'm on 8.1 x64 and the output of echo %path% does not match the contents of system>advanced system settings>advanced>system variables>Path. Any hints?
#Firefox #QuantumCSS "can run up to 2 or 4 or even 18 times faster."
Peg the Core i9-7980XE?
Oh. My. God. 🤯
https://hacks.mozilla.org/2017/08/inside-a-super-fast-css-engine-quantum-css-aka-stylo/
I thought you just need the browser to support it, i.e. install into cache or whatever.
similar to pdf I think you can use any font if you embed it right. And Chrome has a pdf reader. So I assumed it could render fonts without the OS helping.
How about using web fonts as a fallback if native fonts can't render a character. Maybe some AJAX api that lets you get just the characters you need? A mixed emoji style is still better than boxes.
I tried installing win10 on a laptop, it doesn't load up.. but I made the usb by putting an iso on the usb with rufus, I wonder if perhaps I should've instead chosen the option in the media creation tool to put on usb.. i'll give it a try and see if the same thing happens
Windows Drivers Vs Linux Drivers can mostly be summed up by: If your driver is missing/doesn't work, you can 1. google for it (windows) or 1. fix it yourself (linux).
or 2. bug the manufacturer (windows) or 2. bug the developer (linux)
Then you realise, for most linux developers, they actually are using the devices they wrote drivers for, and so fixing things makes sense
for windows driver creators, it's generally a device they no longer actively sell, so why should they care?
Obviously there is odd outliers in both worlds. I've had devices under linux which will just never work 'correctly' and then magically work under windows by using a 'similar' driver.
@barlop rufus lets you choose between gpt or mbr. guess you chose the wrong option? Iso should be fine unless you managed to find a version without bootloader.
@JourneymanGeek if so, my advice is try and stick it out for 12 months, that's long enough on a CV that people don't ask why you jumped ship so fast, and then start looking
have you ever tried gandalf win10pe? I find it runs fine but no network access, there may be a fix for that to add drivers with it, but i'm wodnering if there is another win10 pe that has wifi working fine straight away?
@Bob A protective MBR is part of the GPT. Like a backwards compatibility feature. So BIOS kinda needs MBR, even if it isn't called MBR but 'protective MBR'.
And with MCT a single ISO can't be both a superfloppy and be bios compatible, since a superfloppy is just a filesystem. So MCT should output two types right?
see for example this comment superuser.com/questions/630833/… "BIOS doesn't "support" either MBR or GPT; it's the boot loader that's important. The Windows boot loader is limiting in this respect, but other OSes have no problems booting from GPT on BIOS-based computers"
I can't seem to find a good definition of superfloppy. In theory it could have the filesystem start at the first sector, leaving no space for boot code other than whatever the filesystem offers.
> On non-partitioned storage devices, it is the first sector of the device. On partitioned devices, it is the first sector of an individual partition on the device, with the first sector of the entire device being a Master Boot Record (MBR) containing the partition table.
@jiggunjer UEFI requires FAT32, so any superfloppy intended for booting should use FAT32.
Well, it doesn't require FAT32, but FAT is the only one that's definitely supported.
You could, in theory, assume the existence of a NTFS driver ... or provide one yourself via a FAT32 partition.
> Maps contribution Points earned Review 5 points per review Rating 1 point per rating Photo 5 points per photo Answer 1 point per answer Edit 5 points per edit Place added 15 points per place added Fact checked 1 point per fact checked
@barlop looks like Rufus can't do hybrid BIOS + UEFI on a single disk. MCT can. But on your old machine you only need BIOS anyway.
@jiggunjer No. I mean actual MBR.
MCT creates a FAT32-formatted MBR disk with BIOS boot via MBR boot sector, plus UEFI boot via standard UEFI paths.
I'll say it again: both BIOS and UEFI don't care whether your disk is MBR or GPT. BIOS works on MBR and GPT. UEFI works on MBR and GPT.
GPT is preferred, especially when dealing with UEFI boot, but it is not required unless you're booting an actual Windows OS (as opposed to the Windows insaller). And even then that's largely an artificial requirement -- the firmware doesn't care.
@jiggunjer UEFI as a standard, and as far as firmware support, should work fine off MBR. Windows OS won't boot, but that's not the firmware's fault.
@jiggunjer Ehhhh. Standard bootable disk is usually hybrid because it's not taht hard and very compatible. That means superfloppy with MBR-style boot sector, plus UEFI folder structure. In this case, I appear to have a MBR-formatted disk, but same story with the boot sector and folder structure.
@jiggunjer Hmmmmmmmm. On closer look, it's possible that the UEFI standard actually requires MBR or GPT and is not guaranteed to work on a superfloppy o.O
probably because the UEFI standard says use GPT, so MBR support is kinda unofficial/not guaranteed, but still works most of the time. As I understand UEFI firmware can choose not to understand the old partition layout.
