Jess Birch is a user on mastodon.gamedev.place. You can follow them or interact with them if you have an account anywhere in the fediverse. If you don't, you can sign up here.

Jess Birch @shivoa@mastodon.gamedev.place

Playing "guess which features nVidia will decimate in bringing their pro GPU design to gaming / consumer prices".
Very happy after a bit over two years with a Pascal card but still ever so slightly salty they dropped FP16 packed maths.

@_discovery Unfortunately not. I'm currently enjoying the slides etc from half-way around the world.

Extremely excited to dive into some new presentations / papers.
HPG: highperformancegraphics.org/20
Advances: advances.realtimerendering.com (slides coming)

Jess Birch boosted

Launching the 2018 State of Rust Survey! Please remember, you do not have to be a Rust user to give good feedback! We’d love to hear from all Rust programmers, past, present, and future: blog.rust-lang.org/2018/08/08/ source: twitter.com/rustlang/status/10

In , I'm somewhat surprised by extensive use of as_ptr() [generating UB on mistakes as lifetime info lost] when into_raw() is available [generates small memory leak on mistake as it must be free'd manually]. Much rather leak a string than CWE-416.

I removed the fps limiter from my current Vulkan test program.
*Approaches 4990 fps*
"Hmm, I wonder if the API has a 5k cap."
*GPU realises it should clock up under load*
*Goes up to 5150 fps*
"Clearly the answer is no!" 😀

Jess Birch boosted

Dev machine (nVidia) validation layer says "vkCreateFramebuffer: parameter pCreateInfo->flags must be 0."
However, I'm sure this is a lie (ie it is already set to 0, an assert even checks before handing the pointer over to Vulkan). Wondering if this is my fault or a bug.

Jess Birch boosted

@GregorGullwi As to optimisations, the tool I immediately thought of was Coccinelle's semantic patches (here targeting C, probably pretty far from the state of the art today) which would indicate a direction to go for detecting likely optimisation points. coccinelle.lip6.fr/impact_linu

@jakob @GregorGullwi Was just Googleing "cppcheck linter" myself and quite a lot of stuff comes up (for whatever IDE you want). A few specifically note they're designed just to decorate potential UB.
MISRA/CERT C++ detection should be pretty mature (even if many of the more complete linters are only available as commercial software).

More talk (and researcher moves) around real time ray tracing. ingowald.blog/2018/07/27/joini

Ready at Dawn planning to move their core tech modules to (importing C legacy code to start then slowly replacing with pure Rust). twitter.com/AndreaPessino/stat
Starting to feel like maybe Rust has a bright future in games.

This VSCode plugin is really nice (powered by the central Crates.io repository of libraries). Sometimes a small UI thing makes everything feel clean (and promotes keeping up to date with dependencies). github.com/serayuzgur/crates

@angelmixu For portfolio, Modern C++ is probably the safe choice.
That said, I'm currently working in Rust because I really like the compiler guarantees & think it (or even just the way the borrow checker requires us to think about memory) could be important in the next decade. But that's a risky bet.

Reading Rich Geldreich's tweets over the last few days has definitely offered some insights into how a self-organising tech company might operate (/has operated in the past).

Epic moving to a (backdated) 12% cut for asset store transactions is a great use of their newfound economies of scale. Also good for them to make it clear that this is still profitable for them - not a subsidised move just to capture market share.
Reminds me that Itch.io have been doing a default 10% + card fees cut for ages. We don't have to give 30% to whoever builds a shop & distributes our files. Time for other stores to react?
venturebeat.com/2018/07/12/for

Jess Birch boosted

Best practices advice for anyone building an engine with the new rendering APIs. youtu.be/6NWfznwFnMs