Jump to content
  • Advertisement


  • Content Count

  • Joined

  • Last visited

Community Reputation

5042 Excellent

About Krohm

  • Rank

Personal Information

  • Role
  • Interests


  • Github
  • Steam

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. Krohm

    Article 11 and 13 passed

    Excess bureocracy definitely never killed anyone. 🤦‍♂️
  2. Shadow volumes are still a thing? Well, with them you're most certainly screwed as they are basically vertex-ops. Back when I played parallax mapping I derived a true world position from the displacement. Ugly and slow, but mixed correctly and easily, a combo I strongly raccommend.
  3. Krohm

    Havok Physic Engine sold for $75 M

    It might have something involving executives. Intel behaviour in the last few years has not been smart as they used to. Their idea was, as previously noted, to counter the advance of GPU solvers. Those basically never surfaced so there was nothing to do - it probably involves users insisting on buying graphics-less CPUs. GPU design from a piece of software? I don't think so.
  4. Wow so we'll be getting back some form of new UnrealScript? I didn't expect this to happen so soon!
  5. Let's make some important distinctions: Are your resources loaded statically or dynamically? Are your resources modifiable? Do they move? Example: embedding a statue in world geometry. In this case, try to load it at 'map compile time' and bake it in the geometry. Pre-transform everything if needed. In line of principle this is very hard to do (if at all possible) for arbitrary shaders. Just keep it easy. OFC if you do this, at runtime you just get a big scene graph. Or perhaps save yourself the work and import as a dynamic model. Those kind of meshes are often immutable and non-moving. They can be embedded in the world geometry with ease and leverage the whole content pipeline. Your example however is somewhat more complicated as apparently you want to attach to an imported node. Example: attach rocket launcher to custom player model right hand Real bad: you don't know what model to load in advance, maybe you won't even have it but you only know the right hand is marked with a specific tag e.g. `right_hand`. I'm pretty sure fbx can export skeletal infos or other kind of annotations you can use to extract the information you need. Then you will need a dynamic system to pull out those informations and propagate them correctly. SVG for example does not have this information. Months ago I used specific syntax to identify the required points and reconstruct the information. OFC this was non-portable, non-standard and in general not recommended but you can do it. How you do it is up to you to design. I had things I dubbed 'value synchers' which would connect specific logic (even scripts) to GPU buffers. One syncher in would walk an imported mesh and extract special points to drive particle emitters, another could extract accumulated transforms. Inheritance will most likely be your friend here. Corollaries World geometry very special; it usually takes loads of geometry, covers large parts of the screen, most of the time, it is almost never fully visible. In probably needs its own scene graph to scale and a way to be partially drawn. Dynamic meshes are often instantiated multiple times, be sure to recognize the difference between a mesh and an instance of the mesh itself. An instance is usually small enough you can either draw or not draw it. In my experience, mixing dynamic stuff with static structures is recipe for disaster... but perhaps you might be fully dynamic and still be good enough for you. I would suggest against using fbx or any other content-creation oriented format for that matter. They can get really complicated real fast and you are pretty much guaranteed to be unable to support all their features, which implies you will have to keep a list of what is supported and what isn't. You always have a moment in which you know what is loaded. Be sure to know where to get that information. Most will come from 'level geometry' file. Some might come from network, you likely have been mangling this dynamic data at some point, just don't throw away the information!
  6. Krohm

    Comfortably Programming.

    Regular breaks are key. OFC the average manager approach here is just to prethend you can sit down 4+4 hours straight at 100% efficiency. As long as you're on you are in control of your own behaviour I strongly encourage taking regular breaks. I used to do some squats and physical exercise too. And be sure to have some sport outside of your programming hours, preferibly in the morning. You will be considerably less springy, so to speak.
  7. Krohm

    Merry Christmas GD Lounge!

    All the best Gamedev.net community. You guys rock. I remember being active here as a unique chance. I hope you will get the same vibe and hopefully tap it! As for me, I'm still arranging some FPGA jokes SystemVerilog is such a great fun!
  8. Dangerous article. Of course everyone is free to state his/her opinions but I feel like the cost of code maintainance is not taken in consideration at all. As far as I am concerned encapsulation is awesome... and I'm using Verilog those days! Proper code engineering is difficult. I've seen more than a company dominate the competitors thanks to well engineered codebase and more than one company biting the dust under the weight of unmaintaneable code bases.
  9. Krohm

    wavefronts in-flight

    I haven't had the chance to play with graphics in a while (nor compute for that matter) but I had the impression constant cache is no more a thing since maybe 2013. I don't recall GCN having it. From https://www.amd.com/Documents/GCN_Architecture_whitepaper.pdf, page 6 (2 minute google, I cannot find the detailed paper, my search bubble has grown very different lately). EDIT: apparently there is still a 'constant cache' somewhere in http://developer.amd.com/wordpress/media/2013/07/AMD_Sea_Islands_Instruction_Set_Architecture1.pdf including a reference to it being some sort of L1? LDS can take up to one read and one write each clock (it's dual-port) and serve one 32-bit value each clock for each 32-bit column. In most circumstances, I have seen the value being available on the next instruction from the point of view of a single wavefront. This will get unpredictably long in the case of address collisions, which are the reason for which there are memory guards at ISA level. I personally suspect the LDS to actually take 2 ticks to produce the result as in some cases I got occupancy 1 and observed a much decreased performance in LDS metrics. But again, at occupancy 1 all bets are off so I didn't investigate much.
  10. Krohm

    An idea for a video game

    It is absolutely sound as concept; maybe I've heard about something similar in the past. It's seldom explored AFAIK as it was super hard to do right. I think an old Blade Runner game made something like it. But, to be on topic, I've seen that happen in a tabletop D&D game I mastered years ago. By basically drawing the wrong conclusion, the playing characters turned their best allies against them. As you can imagine, it didn't end well. Perhaps a set of triggers would suffice. Surely targeting a specific character helps a lot.
  11. Krohm

    What are you working on?

    At work... I'm basically the whole QA team (and programmer; and sysadmin; and database mantainer; and web designer and... ). Maybe you mean in the would-be-spare time! I'm starting to work on a... well that's the problem, nobody knows what that's supposed to be in the end. For the time being I'm just setting up the demostration with a couple Pis, Omegas and a few Digi Zigbee shields so I guess we can call it some sort of wannabe-smart IoT system. The real goal for the time being is to get to the funding. I was positively impressed I found (good?) use for some of the junk I've accumulated over the years.
  12. Krohm

    OOP is dead, long live OOP

    Woah! So, is ECS still trending? That's just sad.
  13. This incredible news and I am honestly lifted from reading it. I have been looking at your work for years and found it solid, sometimes inspiring. I had the impression the industry was moving in a closed direction but I see there is still space for people to be competitive and awarded.
  14. Maybe I'm taking a long tour as I have been playing with FPGAs recently. As far as I've understood, Tensor Cores are 8-bit ALUs (or perhaps even 6-bit). RT cores are dedicated resources. If this isn't hitting a manufacturing wall I don't know what it is. Texture samplers have always been dedicated and work great, reflections have been traditionally difficult - in the last few years we got this screen-space-reflection trend being a trick I expected to never see again. Don't even get me started on shadows. With the NV 2k series starting at 500 I might believe this sounds 'reasonable' but zero chance this has an installed user base for the next couple years. Still, a shift is near. Unless you have invested considerably in your lighting infrastructure, it is coherent, competitive and works, I would suggest to not invest in most traditional schemes at this point. Shall I be back into graphics I will focus my interest in driving the hardware efficiently. Ray-tracing cores will stay there in a form or the other.
  15. Krohm

    Streets of Rage 4...OMG!

    I almost wondered why did they start so lo-res in the beginning I reloaded, █ me that's genius.
  • Advertisement

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

GameDev.net is your game development community. Create an account for your GameDev Portfolio and participate in the largest developer community in the games industry.

Sign me up!