/i:'mɪər/

@HackRazor1012 @Hosses21 skill issue

@VPCOMPRESSB @Hosses21 Me attempting to play Silver as a tank begs to fucking differ dude

You play an m1 carry, your team plays the game like it's overwatch and constantly feeds the enemy You play a frontliner, they put you against every cc ability spam in the game You play a spirit burst roamer, your team leaves every lane to crash and constantly team fights

And then he uses the ult to run away Via/epipen-gaming

@DeadlockAnon top worst takes

I hate the fact that you can’t build spirit carry in current meta. Gun damage is just overall more superior with more consistent damage and faster killing mid boss and towers. Not to mention some mage characters underperform late game if you don’t invest some gun damage.

@Colonthreee fldi0 looks like it's using the x87 FP stack? not familiar with this, but i'm sure you could stick with integers?

Comparing a branchless and branched approach to clamping a value between -1.0f and 1.0f for the GTA3-DC physics engine. The findings were... interesting. TwT)

@Colonthreee why can't you use cmov

@trisdentarius i start harassing the enemies who make bad decisions, like them leaving their ally.

@VPCOMPRESSB bro lat matched i literally started telling my enemies what to buy to counter us but they flipped me off😭

nvmd everytime i try to play everybody goes 0 10 cause they overextend and always team fight

@Epic_Edster matched against with the most aids characterd

When I'm laning against Doorman and Mo&Krill

@Pirat_Nation "shader compilation _stutter_" 😟

Nvidia, AMD, and Intel back Microsoft's plan to fix shader compilation stutter Nvidia announced support for Microsoft's new Advanced Shader Delivery technology. The feature, designed to address shader compilation stutter in PC games, will arrive on RTX graphics cards later this year.

@Colonthreee don't use logic operators. they're fake

I broke C...? I did not expect this level of assembly for boolean operations. ;w;) Here's clang and GCC:

@Colonthreee that's an intended feature. it's not really a bug either. you wouldn't `expect hash(string)==string`.

So apparently hashing integers in Python is broke...?

@valigo what were you working on

>be me >have slow code >parallelize slow code >it now fast, but has race bug >add mutex, bug fixed >it now slower than single-threaded was why tf it's so hard

@valigo just use Forth

I just had a revelation that LISP, in its superiority, does not need operator precedence. You always write everything in functional form, and there is no ambiguity what applies to what! Obvious in hindsight, but I got a real "whoa" moment after thinking through it explicitly!

@LUCKYY10P the scene where the guardian exits the city and begins to roam the wilderness was beautiful.

Ikora didn't even take herself out with that Nova.

As beautiful as the day that I lost you. The Red War campaign.

@wookash_podcast @antovsky you can bit-pack tree structures into an array and perform cursor-like (like in a text editor) operations on them via SIMD or other means of parallelism. this useful for syntax trees. no need for "nodes". just transform the bytes directly. you can even eliminate ASTs entirely.

Here is @antovsky walking me through a Dreams-like State Management known by many names and using couple of concepts (intrusive lists, slot maps, fat struct) I had no idea, that by putting pointers into elements, you can make everything be running based off an array, preserving parent-child, and sibling relationships! It was illuminating! We coined yet another term, Large Arrays of Things™. Do NOT confuse this with Entity Component System! @antovsky thank you!

In this blog, Tobias Hector, explains how the new extension fundamentally changes how Vulkan applications interact with descriptors. The working group is looking for feedback before making it part of the core specification. Learn more: khr.io/1n2 #vulkan #gpu

@pixperk "cell overflow" WTF?

why are they worried about 64 bytes? that's trivial. if you receive a bunch of 1 byte request, wasting 63 bytes each, the problem isn't the segment - it's the many small unpacked requests. then they solve it by introducing mandatory wasted bytes, unpredictable data addresses, pointer and block rearrangement madness, and pointer translations and offset searching.

i actually love reading