Root Access

Journeyman Geek

12:41 AM

@Mendeleev we discuss everything

also I do the opposite from @CanadianLuke and run 64 bit everything unless I can't ;)

RegularGDPR

> Whatever model you choose, the compact convertible is powered by a 1.6GHz 2C/4T Intel Pentium Gold Processor 4415Y

Yuck at that processor naming

Journeyman Geek

something intel has been terrible at throughout most of their recent history.

RegularGDPR

Yuck at that processor spec too, last generation 1.6Ghz with no turbo

@JourneymanGeek Yes but all the sysadmins swooning at their new Xeon Platinum processors they managed to convince finance/management to buy!

Way better than those Xeon Gold processors ~~those peasants~~ our competitors across the road use!

Journeyman Geek

naw

most sysadmins just get the same brand/model and such.

they're really concervative

Nick

12:58 AM

@JourneymanGeek On a 64 bit system, does it make sense for a 32 bit application to run faster than its 64 bit counterpart?

Journeyman Geek

@Nick it depends.

bertieb

IIRC Luke is right about what he said about performance in the early days of 64 bit

Journeyman Geek

however in many cases -the extra memory available can be nice.

bertieb

I thought most compilers are optimised now to use wider registers and extra instructions etc

Journeyman Geek

@Nick the relative difference is likely small

and most stuff isn't heavily performance oriented

Nick

1:05 AM

Ok, I want a system that cleanly handle 300 open chrome+firefox tabs every day, what sort of a system configuration would you reccommend for me? (feel free to use pcpartpicker)

Bob

Huh. When did @OliverSalzburg become @DerHochstapler?

@CanadianLuke May I introduce you to MobaXterm? :D

@CanadianLuke 128 is pushing it. Granted, I've gotten a minimal debian container (no kernel) running on 32, but you start having trouble getting apt going.

@CanadianLuke Unless you're actually short on RAM (for real use, not just the numbers), always go 64-bit. The CPU register situation is much better with 64-bit than 32-bit, ignoring the RAM limit entirely.

@CanadianLuke Any speed differences aside from RAM should be in favour of 64-bit.

Though I suppose on an old enough CPU the bigger points could have a negative impact on cache.

If that's a concern, use x32.

64-bit code with 32-bit pointers. Best of both worlds if you don't need the higher memory limit (though some programs might not be compatible).

x86 is notorious for register spilling

@RegularGDPR It's a gimped Core-M3, effectively.

Journeyman Geek

@Bob I used to run my blog on 128 ;p

offloaded the db tho

Bob

@Nick RAM >= 4 GB, ditch Chrome.

That's all you really need.

Or, if you insist on using Chrome, RAM >= 8 GB (16 GB recommended).

Everything else is general advice (CPU >= Intel Core-M [any i-series works]/AMD Ryzen of any description)

RegularGDPR

@Bob 4GB? Pfft, I have that many tabs open on Chrome on my phone and it has half that RAM!

Bob

@RegularGDPR tbh I've never erally tested tab limits on Android Chrome

Android Firefox exceeds 600 pretty happily though

Desktop Chrome is ... rather less happy

RegularGDPR

1:19 AM

All I know is the tab count turns into a smiley above 100, so I have no reliable indication of how many tabs I gots open

Bob

Namely, you can't even click on the tabs anymore by like 50

Nick

@RegularGDPR makes a good point. I've had tabs open on my phone until the :D appears and beyond. Why doesn't sort of performance reflect on desktop?

RegularGDPR

@Bob Eh that's why I have 20+ Chrome windows open

>_>

@Nick Because the phone only ever has to show one tab at a time.

Nick

... and how many do you think the laptop shows? I don't have every tab open simultaneosly on my laptop scaled to 1px*1px here.

Bob

@Nick Android Chrome aggressively unloads tabs.

That tends to make desktop users unhappy when their background tab stops processing anything.

Nick

1:26 AM

@Bob i don't get why static pages can't be unloaded like that.

Bob

@Nick How would you judge a page to be static?

How many static pages do you even load in a day? Probably exactly 0.

Static pages use almost no resources anyway.

Journeyman Geek

@RegularGDPR You're channeling tomtom ;)

Nick

even right now, with 15 tabs open across two windows, I've faced like 5 to 6 black screen and (Not Responding) moments.

@Bob Anyhoo, I think features like that provided by 'The Great Tab Suspender' extension should be available on desktop browsers by default.

Bob

@Nick That says approximately nothing.

It's not even a question of "tabs". It's a question of which sites you have loaded (and which resource you're running out of).

Every OS has a task manager of some description. Learn it. Use it. Love it. It will tell you what to upgrade next.

4

Or what to reduce usage of, if you'd prefer to go in that direction.

Journeyman Geek

@Bob work system has a strange non standard one....

Bob

1:31 AM

The alternative is pure guesswork and throwing money at the wall to see what sticks.

Journeyman Geek

it drives me mad

Its called glance? Never seen it anywhere but HPUX

Bob

@JourneymanGeek That's because you use HP-UX :P

Journeyman Geek

Work does!

IDK why

whether the client requested unix specifically...

Bob

@Nick Browsers provide the APIs. Extensions can provide the UI. That's actually a much better model than browsers trying to include the kitchen sink.

Journeyman Geek

since both major unix platforms run on... unusual hardware. (OS X doth not count)

Bob

1:32 AM

In any case... Firefox handles huge tab counts much better than Chrome.

No manual suspending required.

And I say this as someone with probably >5000 tabs open across 2 computers.

@JourneymanGeek Pretty sure that FreeBSD and OpenBSD are effectively Unix :P

Journeyman Geek

OH

@Bob they need certification I assume

and various wierd shit I can't talk about here.

Bob

@JourneymanGeek Blech.

Journeyman Geek

and other stuff I can't talk about cause none of it makes any sense.

Bob

2 tabs open in Chrome for its devtools, 2000 tabs open (ok, maybe more like 200 truly "open") in Firefox... guess which one's using more RAM?

Though measuring memory's become harder since they went multiprocess

@JourneymanGeek I suppose you should be glad you're not using Solaris :P

(Solaris isn't really bad ... but Oracle)

Journeyman Geek

@Bob ... I'd rather be using solaris...

Lets take this PII ;p

RegularGDPR

1:42 AM

@Bob Sticky money! Is sadly a thing of the past now they've switched to non-porous plastic banknotes

@Bob Chrome is a lot... less bad at it these days.

Bob

@RegularGDPR You won't convince me to even try until I see the tab bar scroll instead of collapsing to unreadable sizes

They're also lacking location-bar tab search/switching

@RegularGDPR I'll also note that it's not like I don't use Chrome at all... it's currently my primary dev browser.

RegularGDPR

@Bob I'm sure I've seen it somewhere, if not in Chrome but in Firefox... But I've hardly used Firefox in six years so I have doubts that's what I'm remembering it from.

Bob

It just won't even have a chance at becoming my browsing browser until it at least fixes its horrific tab UI.

And even then it'll only be a chance.

RegularGDPR

According to Chrome's task manager a single tab is eating 3.2GB of RAM at the moment.

Bob

@RegularGDPR It's probably been in Firefox for that long.

RegularGDPR

1:47 AM

@Bob Or I'm hallucinating because I'm stoned... but anyhows...

Bob

It's certainly not in Chrome right now. I just tested.

RegularGDPR

No the way it's improved is in the "freezing" or static-pre-rendering of background tabs.

As in, yeah they're all sitting there hogging memory, but unlike in earlier versions of Chrome, they're all using no CPU (at least, until I open/activate them).

Bob

@RegularGDPR That's a fairly recent thing. I've been sitting here watching my Chrome update feeds and trying to figure out how they've broken our app again.

RegularGDPR

Previously you'd easily get a quad-core CPU sitting at 50%+ all the time because of the CPU cycles eaten up by backgrounded tabs. Now the total CPU usage being reported by Chrome is only 90% (of one core). 85% of that the main browser process, and the remainder being the GPU process.

Bob

They have a habit of breaking web standards in the mask of performance.

(Reducing background timers isn't one of those, fortunately.)

RegularGDPR

1:50 AM

@Bob What, like Youtube?

Bob

@RegularGDPR Strangely, I don't think I've ever had that happen on Firefox

Ok, correction.

It happened all the time on 3.x and 4.x

I haven't had that happen since ... 40.x or 50.x or so.

RegularGDPR

@Bob That's because the last time I used Firefox it was still single process, and hence limited to maxing out 95% of one core and that's about it

Bob

@RegularGDPR Oh, it's perfectly capable of maxing out all your cores now.

RegularGDPR

It was incapable of hogging three CPU cores with backgrounded crap when it was all single-threaded anyway ^_^

Bob

But, and this maybe says more about what exactly I have open in background tabs, it's not exactly using much now.

That said Firefox has also reduced background timers. Not quite as aggressively as Chrome, but they did follow there.

RegularGDPR

1:52 AM

@Bob Oddly, Chrome seems to be getting worse at it - or rather, being more reluctant to max out all my threads on launch, probably because of much more conservative rendering of background frames to begin with

Bob

(That was only slightly worrying. The "intervention" for passive touch event listeners was much worse.)

RegularGDPR

@Bob Huh, hadn't realised that's what they'd done

Bob

@RegularGDPR Eh. FF doesn't truly load background tabs on application start

They're loaded on first click

But I usually keep an instance open long enough to accumulate a couple hundred of open & loaded ones

RegularGDPR

Chrome tends to load the visible tab in the focused/visible window on launch, then background load the visible tab on every background window, but then very selectively load any non-visible background tabs. Not all wait till you click on them but not all are auto-loaded on launch either. I haven't quite figured out the pattern.

Bob

@RegularGDPR If it's anything like FF, it'll probably do a quick preload when it thinks you're about to click on one

i.e. your mouse is moving/hovering there

RegularGDPR

1:56 AM

tl;dr it loads all the visible background tabs on launch, and saves most of the visible ones for on-click.

But that does right now still equate to in my case 200 tabs in the background that have been rendered, over time, out of maybe 500+ open

@Bob Unless it's very sensitive to what exactly qualifies as my mouse moving towards tabs, there's something else in play too, as I can see (via the task manager) that windows i've not even interacted with since launch have some tabs rendered.

Often cases more than one

In fact some windows all I've done with since launch is to go into the windows windows manager and move them to a non-visible desktop that I've never even tabbed into since launching Windows

> YouTube page load is 5x slower in Firefox and Edge than in Chrome because YouTube's Polymer redesign relies on the deprecated Shadow DOM v0 API only implemented in Chrome. You can restore YouTube's faster pre-Polymer design with this Firefox extension

Mozilla Technical Program Manager, Chris Peterson, commented

Bob

@RegularGDPR That one wasn't Chrome. It was Google, yea, but not Chrome.

Here was the big one: github.com/WICG/interventions/issues/18#issuecomment-250246788

RegularGDPR

@Bob Oooor was it? What if it was actually the other way round? What if Google thought they were implementing a standardised API that was equally slow in all browsers then Chrome came along and cheated the performance of said API, i.e. breaking standards in order to render Youtube "faster"?

Bob

2:12 AM

@RegularGDPR That API was never standardised.

"v0" was effectively pre-standard

v1 was v0 + changes proposed by non-Google browsers

RegularGDPR

I'm so stoned I thought sound of my toilet refilling was actually my laptop crashing and then autonomously reflashing its own BIOS.

Bob

In any case it's on the web site/app to not use nonstandard APIs in production, so it'd be the YT team that dropped the ball there

RegularGDPR

Which makes it not at all surprising Bob is out-knowing me on everything again

@Bob Would you consider the use of Microsoft proprietary shite like ActiveX on certain Microsoft sites to be "nonstandard"? Or "proprietary but standardised APIs"

Bob

@RegularGDPR That's a good question. I would consider that to be bad for the web, but not too sure what I'd call it.

Doing that in 2018 would be a big no-no though.

Bob

4:01 AM

> Not a real mechanical keyboard, but the feel, key-travel and efficiency are all close to full mechanical keyboard.

@JourneymanGeek ^ :P

Journeyman Geek

so... just an oldschool keyboard?

Bob

@JourneymanGeek it looks wonky

amazon.com.au/VicTsing-Backlit-Keyboard-Mechanical-Similar/dp/…

bwDraco

4:20 AM

dragonlord@li650-40:~> php -r 'class A { public function __clone() { clone $this; } } clone new A();'
[1]    7333 segmentation fault (core dumped)  php -r 'class A { public function __clone() { clone $this; } } clone new A();

Wat.

(phpsadness.com/sad/50)

> PHP Parse error: syntax error, unexpected '::' (T_PAAMAYIM_NEKUDOTAYIM), expecting end of file in Command line code on line 1

lol

> It is very important to be able to quickly debug issues in your application. When every second of downtime costs your company money, bad error messages can mean thousands of dollars in unnecessary losses and hours of wasted developer time. Languages posing to be used in large applications need to ensure that developers can quickly discern the cause of an issue.

bwDraco

kickstarter.com/projects/hexgears/…

So tempting...

anandtech.com/show/12924/…

I want this in Brown!

Bob

4:38 AM

@bwDraco Those keys look quite fragile.

bwDraco

I think these keyswitches have been around for a while...

The keycaps, you mean?

Bob

@bwDraco The caps, yea.

Also very easy to get gunked up.

They're just flapping in the breeze...

How are they attached?

What stops a side impact on a key from ripping it off?

Especially on a "portable" keyboard

bwDraco

Yeah, physical durability is a concern for me.

bwDraco

4:50 AM

So... requirements: Wired or Bluetooth wireless operation, low-profile key switches from Cherry or Kaihua, tactile (Brown) preferred. Backlit, RGB preferred but not required. 60% layout is best.

I'm kinda spoiled by the Corsair K95 RGB Platinum I have here. I want something of a similar level of quality I can carry with me.

Journeyman Geek

5:19 AM

@Bob erm...

so they stuck a cage with a dummy switch over a membrane?

@Bob funny thing is, if I saw this at SLS for... IDK, half the cost

I'd just buy it to see what the hell it was

then tear it down

Bob

5:46 AM

@JourneymanGeek lol, I almost bought it for the same reasons

then realised I've already run out of places to hide keyboards