Home Creators Posts Import Register Favorites Logout

Content

Broken Silicon's next guest will be Bryan Heemskerk, a developer at Massive Damage Studios!  He has been on before multiple times, and has become a fan favorite due to his excellent knowledge not just in game development - but also in GPU architectures!

What's going on with "Console Ports"?

Should Sony / MS should bother with Pro Consoles?

R9 7950X3D & 7900X3D (Bryan has Zen 4 X3D)

Expectations of Phoenix and other Upcoming APUs

If RDNA 3 will be "fixed" this Summer

He can talk about anything with decent authority, so don't hold back!  Anything regarding recent (or upcoming) GPU, CPU, and console releases is fair game!

You have ~16 hours to submit below!

Previous Bryan Episode: https://youtu.be/k6qDwx--R-0

Bryan's Twitter Account: https://twitter.com/bryanheemskerk?lang=en

Comments

Resident Mexican

Hello Tom and Bryan, so recently I saw a video presentation on youtube talking about a new 3D rendering technology called Fovorender. The gist of it was that most 3D rendering techniques uses the linear perspective, (or the camera's perspective), where only so much of your field of view can fit in the image. But what Fovorender tries to do is recreate the same field of view and depth that the human eyes can see, so the image isn't distorted as much from the high field of view. From the demo's they showed, it's very impressive how much of a difference it makes, particularly in the first person perspective. So my question is would you, as a developer, like to implement this type of technology in to your games.

Samantha Vimes

Byran you're my favorite guest, glad to have you back! Your discussion of rendering pipelines, draw calls and other minutiae is fascinating. Two topics I'm particularly interested in now: First, how do devs, and art teams tend to view features like FSR 2/DLSS 2 are these "crutches," helpful additions to reduce demanding scenes/areas or optional extras/nice-to-haves for certain art directions or something else? How does that perspective compare with frame generation technologies like DLSS 3? Are these technologies interesting from an art/dev perspective? How are they viewed?

Gach

Hi Tom and Bryan, with DLSS, FSR, and VRS all reducing rendering workload, what techniques or features would you like to see included in future architectures to make the most out of each pixel?

AdanFS

We have seen an increase in CPU core count from lowly quad cores to 16 cores and beyond recently. However, Top end CPUs are still focusing on 8 fast cores. Examples: 13900k with 8 big cores, 7950x3d with just the single 8 core CCD running the game and the rest parked. How much effort will game developers need to put in to maximize the use of these new Heterogeneous Compute resources available to them?

Samantha Vimes

Hey Tom and Bryan, in your opinion how might increased cache sizes in X3D change how games are developed? Does this answer change depending on whether 3D cache comes to console next generation? How might devs leverage this cache better?

10GHz Netburst

Do you have any interesting game dev perspectives on Intel Arc?

Nightshade Laced Foxgirl

Bad take #4369. How do we definitively know RDNA3 is broken and can be fixed? What does it mean for an architecture to be broken? I mean sure it is underperforming expectations. By the metrics of not performing to expectations, wouldn't the 4090 be even more broken as it can't live up to the 2-4x performance Nvidia claimed.

Samantha Vimes

Hey Tom and Bryan! In BS 109 Byran mentioned that some AMD optimizations on consoles are stripped out during console ports. If next gen consoles have 3D cache, do you think optimization for this architectural feature would similarly be stripped out? Do you think these implementation habits will change as AMD continues to claw back market share? If the 3DV Cache optimizations are kept, how might this impact Intel's/Nvidia's future gaming performance?

Samantha Vimes

Hey Bryan, in BS 193 NX Gamer talked extensively about how changing specs on consoles during a refresh. He indicated that even if e.g. fully enabling a chip, unifying RAM, using better RAM, etc was relatively easy on a hardware level that the added difficulty of developing console games for another skew was perhaps too expensive to be worthwhile. Can you offer more insight on what some pain points are for this issue? What areas become difficult/require additional attention? Does this perspective change if a PC Port is being planned and the dev team already knows they're planning for an array of different hardwares not just e.g. the PS?

Tick Dickler

Hey Bryan and Tom, I wanted to pick your brains on whether or not current gen console versions of multiplatform games are adequate enough for the hardware at this point in the generation. And to clarify, by the current gen, I mean the PS5. Thanks!

AI Generated Fish

I have the reverse question on console ports but somewhat related. I've brought up in the past we've not had an NHL game on PC for near a decade. Do you foresee a scenario in the future where it will be possible to develop for PC, playstation, & Xbox simultaneously with the majority of resources being shared so it was efficient to release every game on every platform? What is keeping this from being the case now?

FloridaMan

{Adapted from my post on Reader Mail: General} Hi Bryan, when you use your X3D chip, I imagine there are situations where you want to play a game on one CCD and have a productivity application on the other, but this doesn't occur by default due to core parking. You must manually pick threads or have a multi threaded application span both CCDs. Firstly, does this bother you? Secondly, what would you prefer AMDs X3D thread logic to be, since it currently relies on the Xbox game bar?

Richard Yao

Why did Massive Damage Studios use Direct3D 11 over OpenGL or Vulkan to make Star Renegades? Will Massive Damage Studios ship Star Renegades with native Linux support? If reimplementing the graphics API calls with a native Linux API is a problem, has you looked at DXVK-Native to translate Direct3D 11 to Vulkan? Also, what about MacOS support? Intel modified DXVK to implement the D3D9 DDI to use their vulkan driver for certain D3D9 games. This was a surprise to me because they reportedly have not yet implemented vulkan graphics pipeline support in their Windows driver, which DXVK needs to mitigate shader compilation stutter. When that is implemented, how likely is Intel to drop their native D3D9 DDI implementation in favor of DXVK? DXVK also supports D3D10 and D3D11. How likely is Intel to begin using DXVK to do D3D10+D3D11 to vulkan translation in their Windows driver? What are your thoughts on SteamOS and the Steam Deck? Non-overclocked DDR5-8400 that does 67.2GB/sec per DIMM will arrive on the market in a few years. How do you see this changing integrated graphics?

Richard Yao

John Carmack wrote an article praising the use of static analysis when developing games at ID Software. Would you tell us about the use of static analysis at Massive Damage Studios?

Richard Yao

How would developers do optimizations for 3D V-Cache? I thought console optimizations were things like precompiling shaders and using non-standard API extensions to get the hardware to render things more efficiently, beyond the obvious things like reduced render scales and lower image quality presets. Are there any optimizations that fall outside of that category? If there are, are there very many of them?

Richard Yao

Why do so few studios release the source code to their games like ID Software used to do?

Richard Yao

Nvidia develops a unified driver that runs on Windows, Linux, FreeBSD, etcetera. It used to even run on MacOS before Apple’s hostility caused Nvidia to stop shipping a MacOS driver. AMD and Intel on the other hand split their development teams into two, developing independent Windows and Linux drivers while leaving MacOS graphics driver development entirely to Apple. Nvidia gains obvious benefits from deduplication of effort, but neither AMD nor Intel seem to be interested in doing the same despite the shader compilers largely doing the same thing on both platforms. The only exception is AMD’s vulkan user mode driver, which AMD develops for both Windows and Linux despite the Linux community developing an excellent driver one with Valve’s help. AMD could effectively increase its man power for free if it would just join the community development effort and reuse it on Windows; they already joined the community effort in all other areas on Linux, but do not reuse the result on Windows. Intel on the other hand reused DXVK, but nothing else. Why do AMD and Intel insist on maintaining this duplication of effort?

Gob

Hi Bryan and Tom, We have seen only a few games implement ray tracing effectively and maintain 60 fps. Mainly, Metro Exodus Enhanced Edition, Fortnite, Ratchet and Clank Rift Apart and the Spider-Man games. The studios that made these games also created the game engine. These games prove that the consoles are more capable of ray tracing than initially thought and these games run better on PC as well. When will we start seeing more games and engines capable of delivering ray tracing and 60 fps on console?

Diplomaticus

My armchair perspective on this is that the performance of games largely fall into 2 fairly distinct categories on RDNA3 hardware: ~25% better than RDNA2 and ~50% better than RDNA2 possibly highlighting those games that can make use of the dual-issue shaders and those that can't. I think "broken" might be a misnomer as I suspect it's down to the individual games' implementations whether or not they can take advantage of the architecture, but it would be interesting to see if anything can be done in drivers.

Matheus Duque

Hello Tom and Bryan! Why do you think that asynchronous time warp is only used in VR? I tried Comrade Stinger's demo on my LG C2 and the effect felt great, 60fps feels like 120fps at a minimal loss to image quality. Also can't fovieted rendering be used to some extent in regular games as well?

KingHarkinian

For GPUs, is the AMD toolchain better, worse, or roughly equal to Nvidia's? AMD has consoles and is highly committed to open sourcing a lot of things, while Nvidia has CUDA and much more maturity and support. Is either more pleasant to optimise a game for? Or is this just too low-level and most game companies don't bother with this?

Richard Yao

Mobile games are able to get amazing graphics from using tiled rendering that reduces memory bandwidth requirements to a level that integrated graphics can sustain. Why do PC and console games seem to avoid this technique?

Richard Yao

Why are a number of game developers hell bent on bundling malware in the form of kernel anticheat in their games? It allows for a third party back door into any system using it and risks destabilizing machines, yet is well known to be ineffective since it violates the basic security principle of physical access being game over. Server side techniques are far more effective since cheat developers cannot reverse engineer those, do not violate the security principle of physical access being game over and do not put backdoors into end user systems or risk destabilizing them due to bad programming or kernel updates from Microsoft that are incompatible with whatever hacks the kernel anticheat module uses.

yoda king

Hey Tom and Bryan! This is a curiosity of mine but how impact is denuvo on game performance? And another question I had but what alternatives are there to kernal level anti cheats? I see many people play call of duty or Valorant even though they have kernal level anti cheats. Is there a better way to combat cheaters? Thanks again!

PenguinsloveAI

Hey tom and Bryan, I know it’s a bit off topic, butIt dawned on me the other day when you were talking about the latest console gen becoming outdated that clx could bring us back to the days of expansion packs and other accessories to consoles. Aside from its use in the server market, it seems a perfect fit for expanding the functionality of consoles. Do you see this happening in future consoles. Would it make porting games from pc easier, due to the customization. On the other hand I can see it making things more difficult because you don’t have a set target anymore much like developers encounter in the pc space. Keep up the good work. It is very refreshing to have someone without bias do deep dives in hardware. -Shane