Ryan Flaherty

1.3K posts

Ryan Flaherty banner
Ryan Flaherty

Ryan Flaherty

@rflaherty71

round 2 Previously Thekla/Google.

Katılım Mayıs 2014
578 Takip Edilen2K Takipçiler
Sabitlenmiş Tweet
Ryan Flaherty
Ryan Flaherty@rflaherty71·
Assassin's Creed Valhalla - 170 GB, hour+ long download on Steam at 120 Mbit/s With Via you are in game in <5 min with <5 gigs of data on disk. Ubisoft's Anvil engine tech seems pretty good to me, good loading code and no fps stutters due to streaming.
English
2
1
60
19K
Ryan Flaherty
Ryan Flaherty@rflaherty71·
what if instead of a slow path and a fast path we just made the slow path fast
English
6
3
42
2.4K
Ryan Flaherty
Ryan Flaherty@rflaherty71·
@przekursor bootstrapped my own kernel in qemu, so the code here is running against the qemu virtualized nvme device
English
1
0
7
627
przekursor.
przekursor.@przekursor·
@rflaherty71 Congratulations, how do you test it? I did a simple driver in HDL.
English
1
0
2
888
Ryan Flaherty
Ryan Flaherty@rflaherty71·
got a successful round trip through nvme (I may have started writing an operating system...)
Ryan Flaherty tweet media
English
11
9
265
16.6K
𝅙
𝅙@bmattlife·
@rflaherty71 what language is this? looks like a mix between c++ and rust
English
1
0
1
777
Ryan Flaherty
Ryan Flaherty@rflaherty71·
@cmuratori I keep expecting to hit some wall but every time I go to implement something I just find it to be shockingly smooth and easy to do (I'll stay away from USB and the GPU for now..)
English
0
0
25
1.3K
Ryan Flaherty
Ryan Flaherty@rflaherty71·
finally got around to that fil and casey podcast on fil-c, fil is kinda awesome
English
0
0
17
755
Sakureil™
Sakureil™@lettuce_isgood·
@rflaherty71 Itanium is in order and x86 CPUs are out of order and OOOE is on the backend. This is somehow worse than saying that “x86 is internally RISC”
English
1
0
2
301
Ryan Flaherty
Ryan Flaherty@rflaherty71·
This was before my time but I would guess the problem wasn't compliers supporting it, that is easy. The problem was no back compat for already compiled x86 apps. That is kinda what we have today, x86 front ends on itanium backends.
neural oscillator of uncertain significance@mycoliza

English
6
1
50
6.2K
Don
Don@DonielOediv·
@lettuce_isgood Advertising that you were fired from Thekla is an interesting decision, I'm guessing this is one of the guys Blow implied was useless
English
2
0
0
122
Sakureil™
Sakureil™@lettuce_isgood·
Formerly Google btw.
English
3
0
4
241
Kevin Gray
Kevin Gray@graykevinb·
@rflaherty71 Look if I'm wrong on any of the points I welcome a computer scientist to show me exactly which tech claims I made were incorrect.
English
2
0
2
1.5K
Ryan Flaherty
Ryan Flaherty@rflaherty71·
@SebAaltonen Reminds me of how I had to make my file system driver report "NTFS" as it's name for windows to load and run an executable off of it, there is some user space code that hard codes the string compares
English
0
0
3
431
Sebastian Aaltonen
Sebastian Aaltonen@SebAaltonen·
HypeHype can't fallback to GLES. We have modified the Flutter renderer and we compose directly our Vulkan surfaces with Flutter's, so we need both to use Vulkan. We had to change the Flutter Vulkan app/engine name to workaround the Pixel 10 startup crash.
English
2
0
42
7.2K
Sebastian Aaltonen
Sebastian Aaltonen@SebAaltonen·
Google Pixel 10 vkCreateInstance fails on VK_ERROR_INCOMPATIBLE_DRIVER if the appName/driverName is "Impeller". This was fun to debug :D Likely some Google emergency hack for the new PowerVR DXT GPU to ban their faulty Vulkan driver in Flutter. Fallbacks to GLES.
English
8
7
161
25K
MEMPhistopheles
MEMPhistopheles@vivapolymetron·
You’re wrong but glad you are confident about it. Electricity runs through your entire system simultaneously. There is no real concept of an “off” with a computer that is a Low power mode that engineers decided was an off state. asynchronous is BS made up by devs like you 😃👍🏾
Ryan Flaherty@rflaherty71

Completely untrue. It is the opposite, everything is actually async but your cpu and the kernel tries really hard to sell you a synchronous model.

English
1
0
0
244
quantitative milkman
quantitative milkman@quantmilkman·
@rflaherty71 he’s talking about single core and ur talking about multi core and u guys are talking past eachother
English
4
0
9
4.1K
Ryan Flaherty
Ryan Flaherty@rflaherty71·
Completely untrue. It is the opposite, everything is actually async but your cpu and the kernel tries really hard to sell you a synchronous model.
Abhishek🌱@Abhishekcur

Hot Take: Nothing in computing is truly asynchronous at the hardware and instruction level. Everything is synchronous, until a runtime fakes it. People talk about async like it's a magical property of the universe. But at the lowest level, every system is ultimately synchronous. -CPUs appear to execute instructions in order (even though they reorder internally) -Syscalls block unless the OS schedules around them -Hardware is event-driven but still bound by clocks -“Async” functions don’t run in parallel, the runtime schedules them So what’s actually happening? Async is an illusion built by runtimes: -JavaScript: event loop + callback/microtask queue -Rust: executor + waker + poll-based futures -Python: event loop + coroutines -Go: scheduler multiplexing goroutines over OS threads -OS kernels: interrupts + preemptive scheduling Under the hood, it’s all synchronous execution broken into chunks and interleaved by a scheduler. -Async ≠ “parallel” -Async ≠ “multithreaded” -Async = “scheduled synchronous steps that look concurrent” The runtime just decides when your synchronous fragments run. In summary: Every system is synchronous at its core. Async is a runtime abstraction that slices sync operations into interleaved steps that feel concurrent. It's the core concept of Asynchronous. I am always open to feedbacks, because through your feedback I get to know a lot.

English
30
22
854
68.7K
Ryan Flaherty
Ryan Flaherty@rflaherty71·
10 replies about how unaligned reads are slow, quite the self report
wini@caramelows

@SheriefFYI what? it's true. wasn't meant to be a snarky comment about rust superiority or w/e. i just think it's nice that you get this performance-oriented behavior automatically also, if you're manually arranging your struct like this, why not just add the packed attribute?

English
0
0
8
1.9K