Jess Birch @shivoa@mastodon.gamedev.place

There has never been a better time to create things with code, to build on top of all the different platforms and ecosystems that have grown up.
That doesn't mean it is never frustrating, or necessarily easy, but we shouldn't despair. Progress is all around us.

"Everything is terrible & we need to throw all existing code away" is a companion concept to "why isn't everyone a 10x coder" (no, your fellow coders aren't a death cult cheering inefficiency). It may sound attractive when you're feeling a bit down but it's corrosive to actually building forward (from where we are now).
Requirements are complex and ecosystems take time to build. You also may not be aware of as many of the available tools or offered features as you think you are.

AnandTech has been around for a long time (with early writers now moved on to new thing) but continues to provide excellent and accessible coverage. Detailed while still offering explanations around the technical stuff.
A Turing architecture overview: https://www.anandtech.com/show/13282/nvidia-turing-architecture-deep-dive

Did Nintendo actually announce that the only place your online rank is saved is on your Switch's internal memory? (For games like Splatoon 2, which is why cloud saves are "not possible" for those games.)

Something I had completely missed (so maybe others have too): nVidia described their current Vulkan RTX extension with examples (months later and there still isn't a public beta - are we jumping straight to multi-vendor standardisation?).
http://on-demand.gputechconf.com/gtc/2018/presentation/s8521-advanced-graphics-extensions-for-vulkan.pdf

Hello...
"To find out how easy it would be to avoid using the Ray Tracing APIs, we wanted to see how fast "native" implementations would be compared to the ones that are using Ray Tracing APIs."
https://diaryofagraphicsprogrammer.blogspot.com/2018/09/ray-tracing-without-ray-tracing-api.html

Only flicked through it so far but the (free, online) chapter of Real-Time Rendering 4th Ed on ray tracing is out now. Looks pretty good for anyone wanting a fast intro up to today. http://www.realtimerendering.com/Real-Time_Rendering_4th-Real-Time_Ray_Tracing.pdf

High perf coding in two questions: what is the actual work that _needs_ to get done and how is the fastest way to do it.
Data-oriented allows focus on optimising the forms the data takes (in regard to what things need to happen to it) which tackles both questions at the same time.

Chatting to some #RustLang game devs recently, I heard the proposal that library code should never decide to abort. I strongly suggest that the ability to panic (when fully documented) is important to any project's final code quality.
https://blog.shivoa.net/2018/08/rust-fail-fast-and-loudly.html

Your current renderer is too sharp at native res.
Comparing static shots looks like there's stronger details in some versions with less AA but in motion that's going to cause shimmering.
Yes, enthusiasts on forums say TAA/etc is "too soft" and call it "just blurring" but you are making a better renderer if you always err on the side of softness.
Our expectation from film is holding a shutter open for 1/48th of a second and collecting all light that falls on the area of a sensor pixel. It's soft.

So, it's indiepocalypse time again?
I really hated the one in the early 2000s (I was in the middle of joining a brand new team; they're still around today) & the late 2000s one was the worst (tied to the wider Great Recession). I gather the one in the early 2010s wasn't amazing, nor was the one a few years ago. I wasn't old enough to experience the indiepocalypse of the mid '90s but I certainly heard a lot about it (mainly on forums from the veterans) when starting out in the late dot-com era.

Huh, that seems new. Valve must be starting to push back on purchasable badges/achievements (and trading card churn) which aren't bundled with a "real" game.

I wonder if Samsung will focus on expanding significantly beyond mobile SoCs or if this means our PC processors are now going to be entirely dictated by everyone on TSMC or Intel on Intel.
Even low volume (10% of market) competition feels like valuable R&D diversity so shame to see a player divest.
https://www.anandtech.com/show/13277/globalfoundries-stops-all-7nm-development

Hey, never do this even if your boss really pushes you on it: Figure 23 here makes me think this entire article is aiming to deceive me rather than inform me. If this was an academic paper, I'd be immediately on edge seeing those bar charts. https://software.intel.com/en-us/articles/checkerboard-rendering-for-real-time-upscaling-on-intel-integrated-graphics-v-12

Idle thought: AMD are surely working on using extra CPU power & compute to do gaming ray tracing fx + denoise. Something to show the potential of Ryzen + Vega systems.
(Radeon-Rays 3.0 - now accelerated with spare CPU threads for intersection calc / pre-processing so GPU can dedicate more to denoise stage & classic rasterizer work?)

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.