Jaap van Muijden
2.3K posts

Jaap van Muijden
@jaap_null
GPU and Metal guy at Apple via Playstation, Guerrilla Games. Creating beautiful things and building them well. Miami Vice Super Fan.
California, USA Katılım Ağustos 2015
447 Takip Edilen557 Takipçiler

@olson_dan @bkaradzic It was probably not the best idea to use a form factor that looks identical -but is a scaled down version- to a well known competitors product. The size difference gets lost and specs will look worse in comparison.
English

@bkaradzic It's clearly not for general use but could be cool in specific cases for some companies. I'd like to be able to use virtual/cloud desktops for my own work but I doubt it's economical vs rdc into a physical box.
English

@tsoding Leibniz: “I invented Calculus!” Newton: “No I invented Calculus!” This guy: “It’s all sums innit”
English

@SebAaltonen @chessMan786 There are many other (re)scheduling solutions to hide latency (and/or cache problems) in GPUs though. It's hard to make any statements on these things unless you limit yourself to specific GPU designs. RT has pushed a lot of rethink in the space.
English

@chessMan786 GPU does tolerate some latency as it has 8x-10x hyperthreading per core. But there's no out-of-order execution and slow tiny caches. CPU can tolerate memory load latencies much better in general. GPU completely chokes if you feed it a workload that trashes those tiny caches.
English

@politeia10 @EchoOfPixels @cmuratori That's the whole point of the comic; entropy still accumulates against dictionary attacks given the relatively large amount of possible words. (there is a separate discussion on word and word-list quality etc but it still applies)
English

@Rainmaker1973 This would be seen as Total Victory by the USBIF. Slowly eating up all the other standards until one remains... 🥇
English

@never_released @handleym99 It seems to be going the way of the IA-64... which is too bad because I like using 8x8x8 bitfields for math stuff.
English

@handleym99 Sorry, was a joke about Intel and (not) shipping AVX-512. They shipped it on 11th gen and then it vanished
English

@TheYoungJI ... and of course with "strength training" I mean learning to control my super strength so I stop crushing everything with my unbreakable titanium bones. :P
English

I find this a very interesting challenge. The question you really want to ask is «what is the entropy of this password, and is it above the threshold».
For example, 16 hex characters is 64bits of entropy. For alphanumeric you need 13, mixed case alphanumeric 11, and with 10 special characters also 11.
The issue is of course that this requires the characters to be truly random. Figuring out the entropy when they are not is likely impossible.
However, the solution used by most systems is simply misguided.
English

@EmmaVigeland Definitely frame this! Print it at your local copy shop and get a nice frame for in the office. 🖼️
English

@bmcnett Buffering makes you wait 10-20 seconds, TV makes you wait a week, and that is not taking 10 minute ads into account.
I'll take the buffering any day yes and thank you.
English


@happyghuru @twostraws You already can, Steam is offering a growing list of AAA titles with Mac support. We’re cookin’
English

I just watched a comparison of M4 chips against dedicated NVIDIA GPUs, and it's just astonishing – M4 Max on MBP came in second to a 4080 Super on a desktop PC. If you're not sure why that's incredible, you should see the size and weight of a 4080 😂 youtube.com/watch?v=0bZO1g…

YouTube

English

@KostasAAA 6.5 ms? wow that's a _lot_. Cool that they got to go wild on it though! Must've been a pretty tough needle to thread in terms of visual direction and perf management.
English

Some details about the great strand hair tech used in Dragon Age: Veilguard. A 6.5ms budget for hair rendering is very impressive! ea.com/technology//ne…
English

@aras_p @SebAaltonen Yeah it's vendor limitations trickling through the APIs all the way down to the wrapper layers. Same deal with pixel formats.
English

@SebAaltonen I bet the reason is that some gfx APIs need (needed?) 4 byte stride for vertex size. So this leaves float1, but not byte1 etc.
English
















