|
Joda posted:Wasn't the point of Vulkan to get an API that more closely resembles modern GPU architecture than OpenGL? Not necessarily something that was more low level how is matching the hardware implementation "not more low level"
|
# ? Apr 7, 2020 17:43 |
|
|
# ? Jun 5, 2024 20:48 |
|
Falcorum posted:As a graphics dev, Vulkan's a lot nicer to work with than OGL (and to a lesser extent, than D3D). It's massively more verbose (until you build your own abstraction layer, same as you'd do with other graphics APIs anyway), but there's significantly less guessing at what the gently caress the GPU drivers are doing, and then being surprised at their "this is what you meant right" best guesses. It's been more annoying to try new things with it. I don't feel like the profiling tools are there yet, really, and the shader efficiency wasn't as good as OGL/DX11's. This was a while ago, I imagine support could have improved significantly. Also depends on the kind of work you do.
|
# ? Apr 7, 2020 18:35 |
JawnV6 posted:how is matching the hardware implementation "not more low level" Because OpenGL is already low level, it's just made for an architecture that hasn't been relevant since the early 00s and became more abstract only as hardware changed. At least that is my understanding.
|
|
# ? Apr 7, 2020 19:41 |
|
I'm not a graphics programmer but I've been in AAA games for 14 years and my general sense of things is that graphics programmers would prefer their platform APIs to be as low level as possible and handle things in the abstraction layer like Unreal's RHI. The original DX11 driver on XB1, for instance, was stupid slow in terms of how often it would just flush state doing "normal" things. Moving to DX12 was obviously a lot of work but makes things much more versatile.
|
# ? Apr 7, 2020 21:01 |
|
more falafel please posted:I'm not a graphics programmer but I've been in AAA games for 14 years and my general sense of things is that graphics programmers would prefer their platform APIs to be as low level as possible and handle things in the abstraction layer like Unreal's RHI. The original DX11 driver on XB1, for instance, was stupid slow in terms of how often it would just flush state doing "normal" things. Moving to DX12 was obviously a lot of work but makes things much more versatile. I mean, if you have a whole team fully dedicated to graphics programming then that's terrific and I'm jealous, but that's not always going to be the case, and when it isn't, it can be easier to support an engine based on a more hand-holdy API.
|
# ? Apr 7, 2020 21:06 |
|
I've actually been toying around with wgpu in Rust. It's the reference implementation of WebGPU so still somewhat rough around the edges, but it's quite successful at being a low-level API that still works on Vulksn, dx12 and Metal in a true cross-platform way.
|
# ? Apr 7, 2020 21:23 |
|
Joda posted:Because OpenGL is already low level, it's just made for an architecture that hasn't been relevant since the early 00s and became more abstract only as hardware changed. At least that is my understanding. I'd argue that OpenGL *maybe* matches some of the SGI Indy stuff from the 80s but ever since the addition of multitexturing in the early 90s it hasn't matched any graphics hardware. It's low-level in the sense that it's not a scene graph, but it's high-level in that the kinds of operations you do have a lot of driver magic behind the scenes. Absurd Alhazred posted:It's been more annoying to try new things with it. I don't feel like the profiling tools are there yet, really, and the shader efficiency wasn't as good as OGL/DX11's. This was a while ago, I imagine support could have improved significantly. Also depends on the kind of work you do. Not sure what you mean by "the shader efficiency isn't as good". The typical reason that might happen is that the driver does special optimizations to your shader behind the scenes, and that's the kind of thing that's that really inconsistent and makes it difficult to ship performant games on systems that don't match your own configuration. As for GPU profiling, NSIGHT Graphics is really your only choice on PC NVIDIA.
|
# ? Apr 7, 2020 23:00 |
|
As a graphics hobbyist, working with Vulkan is a pleasure. With OpenGL, the moment you want to do anything nontrivial you have to hack around every vendor's quirks and "helpful" optimizations, which is impractical without vendor support and a team; with Vulkan, it does what it says on the tin.
|
# ? Apr 8, 2020 16:46 |
|
Ralith posted:As a graphics hobbyist, working with Vulkan is a pleasure. With OpenGL, the moment you want to do anything nontrivial you have to hack around every vendor's quirks and "helpful" optimizations, which is impractical without vendor support and a team; with Vulkan, it does what it says on the tin. Yeah, it has a niche and specific goals and does very well meeting the goals. Vulkan isn't presented as the layer 'normal people' should be using, if it's too verbose and low-level for whatever purpose then move up a layer in the stack.
|
# ? Apr 8, 2020 17:57 |
|
https://twitter.com/cmuratori/status/1246930755860586496
|
# ? Apr 8, 2020 21:04 |
|
The slayerizer is taking it well https://twitter.com/theslayerizer/status/1246875912789995521
|
# ? Apr 8, 2020 21:28 |
|
someone should ask Casey about the ridiculous build system he made while at RAD Game Tools that has it's own custom language, and can't specify different compiler flags per-file.
|
# ? Apr 8, 2020 22:08 |
|
Suspicious Dish posted:someone should ask Casey about the ridiculous build system he made while at RAD Game Tools that has it's own custom language, and can't specify different compiler flags per-file. Please tell me you have further information on this. This sounds amazingly bad.
|
# ? Apr 8, 2020 23:09 |
|
it's called CDep, and it looks like this https://gist.github.com/rygorous/51bf4ec4078f3b07aa71
|
# ? Apr 9, 2020 02:00 |
|
Looks like a million $!
|
# ? Apr 9, 2020 02:24 |
|
Suspicious Dish posted:it's called CDep, and it looks like this https://gist.github.com/rygorous/51bf4ec4078f3b07aa71 a build system where handwritten makefiles would feel like a step up amazing
|
# ? Apr 9, 2020 05:58 |
|
lol even the loving comments have sigils: $//
|
# ? Apr 9, 2020 06:15 |
|
Winter Stormer posted:a build system where handwritten makefiles would feel like a step up It's actually superficially similar to some of make's more advanced/less useful functionality. Probably faster, though. See e.g.: https://www.gnu.org/software/make/manual/html_node/Conditional-Functions.html (I've written some things worth this thread with those when I was like 15 or something).
|
# ? Apr 9, 2020 06:55 |
|
A 40 minute video about twitter arguments is definitely some kind of horror, but I'm not gonna watch long enough to find out whether there's coding in it.
|
# ? Apr 9, 2020 09:09 |
RPATDO_LAMD posted:A 40 minute video about twitter arguments is definitely some kind of horror, but I'm not gonna watch long enough to find out whether there's coding in it. I watched half of it and he's just ranting, repeating himself, and saying the same things over and over. Apparently the only thing that should matter for project load times in an IDE is the size of the source code tree and the disk speed. "My source code is 1,5 MB and I have an SSD capable of 3 GB/s therefore the project should load in half a millisecond" is one argument he keeps repeating over and over. He's not wrong that by default Visual Studio does a shitton of things many developers will never need, and is slower than it needs to be as a result, but on the other hand I've never experienced those installation/upgrade failures he apparently has all the time.
|
|
# ? Apr 9, 2020 10:40 |
|
I’m not going to watch that video, but why is he restarting his ide frequently enough to care if it takes tens of seconds?
|
# ? Apr 9, 2020 11:18 |
|
So, this came out of CI today:code:
RAM/disk failure? Cosmic rays? Aliens?
|
# ? Apr 9, 2020 11:48 |
|
If I've learned nothing else from this thread it's that indie game devs are a weird bunch. He showed the directory listing for his game and it was relatively few files with many of them being 100KB+ in size which seems to be the indie game standard. He's also the type of person who feels that sharing the rant video paints him in a good light.
|
# ? Apr 9, 2020 12:57 |
|
I've triaged feedback tickets in Visual Studio before. This is the video equivalent of the "How can you possibly not see this bug that I hit every day! What's wrong with you people!" and getting mad when people try to help.
|
# ? Apr 9, 2020 13:05 |
|
His main point seems reasonable and if you take a step back and think about the broad themes of his video they're all reasonable but every detail is eyeroll-inducing in the extreme
|
# ? Apr 9, 2020 13:05 |
|
nielsm posted:I watched half of it and he's just ranting, repeating himself, and saying the same things over and over. Apparently the only thing that should matter for project load times in an IDE is the size of the source code tree and the disk speed. "My source code is 1,5 MB and I have an SSD capable of 3 GB/s therefore the project should load in half a millisecond" is one argument he keeps repeating over and over. Yeah, that's a weirdly large number of install failures IME e: JFC this guy bloviates a lot about how other people don't know how computers work for someone who apparently doesn't understand what the tools he's using are actually intended to do Munkeymon fucked around with this message at 15:17 on Apr 9, 2020 |
# ? Apr 9, 2020 14:12 |
|
Munkeymon posted:e: JFC this guy bloviates a lot about how other people don't know how computers work for someone who apparently doesn't understand what the tools he's using are actually intended to do More to the point, the video starts spending ten minutes lambasting a stranger for assuming that he’s incompetent instead of giving him the benefit of the doubt and he then spends thirty minutes assuming the VS team is completely incompetent instead of giving them the benefit of the doubt that their product owners are optimizing for use cases that aren’t his (it all kinda blurred together but i thought he mentioned that the otherwise empty solution is only using VS for the debugger and possibly release builds?).
|
# ? Apr 9, 2020 15:54 |
|
Soricidus posted:I’m not going to watch that video, but why is he restarting his ide frequently enough to care if it takes tens of seconds? He didn't. That's where VS support railroaded his question because the only person to pick up the phone on an (overly-broad) "debug slow" ticket was the VS Solution Loading In Ten Seconds PM. Nobody bit on "WW doesn't update while clicking" so he didn't explain it to them in that depth. The sub-10s load time was VS's inane focus. I know, I know, the video has context? Weird? Storysmith posted:More to the point, the video starts spending ten minutes lambasting a stranger for assuming that hes incompetent instead of giving him the benefit of the doubt and he then spends thirty minutes assuming the VS team is completely incompetent instead of giving them the benefit of the doubt that their product owners are optimizing for use cases that arent his (it all kinda blurred together but i thought he mentioned that the otherwise empty solution is only using VS for the debugger and possibly release builds?). He thoroughly demonstrates the ignorance of the drive-by comment accusing him of being unaware of "new features" and not bothering to file incident reports. Did you not think he had sufficient evidence that VS white knight who condescendingly dropped by was coming from ignorance? I don't really get this massive difference between "optimizing for use cases that aren't mine" and being a bumbling behemoth of abstractions that can't show the GUI in 10 seconds, but sure, he's Doing A Bad as well.
|
# ? Apr 9, 2020 16:17 |
|
JawnV6 posted:He didn't. That's where VS support railroaded his question because the only person to pick up the phone on an (overly-broad) "debug slow" ticket was the VS Solution Loading In Ten Seconds PM. Nobody bit on "WW doesn't update while clicking" so he didn't explain it to them in that depth. The sub-10s load time was VS's inane focus. I know, I know, the video has context? Weird? You're stanning pretty hard for a guy who got so burnt up over a twitter argument that he just had to make a 40 minute response video where he repeats the same points over and over. I don't know Casey Muratori and I definitely don't follow his twitter feed enough to understand the complete context that he just couldn't get around to providing in the video. Maybe he's a genius, maybe he's a dunce - but he's not coming off very well there.
|
# ? Apr 9, 2020 17:34 |
|
He's apparently programming for Windows, which is a weird edge case
|
# ? Apr 9, 2020 17:43 |
|
Space Gopher posted:You're stanning pretty hard for a guy who got so burnt up over a twitter argument that he just had to make a 40 minute response video where he repeats the same points over and over. Space Gopher posted:I don't know Casey Muratori and I definitely don't follow his twitter feed enough to understand the complete context that he just couldn't get around to providing in the video. Maybe he's a genius, maybe he's a dunce - but he's not coming off very well there.
|
# ? Apr 9, 2020 17:45 |
|
It’s almost like this is a thread on a dumb comedy forum that’s dedicated to ridiculing people we don’t know,
|
# ? Apr 9, 2020 17:50 |
|
JawnV6 posted:It seems like a streamer? Have you ever watched anyone stream.. anything? It's... verbose. Unedited. He didn't "have" to make it about "a" twitter argument, it was time-filler on a topic that comes up repeatedly in this person's world. But yeah, when you characterize it as a video essay instead of a stream segment, minimize the source annoyance to ignore the pattern it represents, once you drag the goalposts all the way over there? It might be as ridiculous as you're acting. the video was posted in this thread as a discussion item without any other context, in a format that didn't suggest "hey this is nothing but a capture of a streamer doing a 'time-filler', so it's probably not worth your time" but, oh, better not dare to discuss it unless you're familiar with the oeuvre of software development streamer king Casey Muratori. if you don't already know his debugger opinions and twitter beefs by heart, why even speak?
|
# ? Apr 9, 2020 18:18 |
|
omeg posted:So, this came out of CI today: Unicode fuckery inserting an invisible character combined with different systems is my guess.
|
# ? Apr 9, 2020 23:07 |
|
That reminds me of a recent issue we had at work. We updated our SQL generator to perform a "has any text" filter by using a DATALENGTH(column) > 0 check instead of the previous column IS NOT NULL AND column != '' check, because the newer one can make use of table statistics rather than evaluating the actual values. It turns out that SQL server's default collation settings will decide that any number of characters loosely defined as a "space" are equal to an empty string, including my favorite evil unicode character, the Mongolian vowel separator. So of course, we put out release notes and made it a configuration setting for the end user.
|
# ? Apr 9, 2020 23:40 |
|
Did you mean 'remquo'?
|
# ? Apr 9, 2020 23:47 |
|
I'm using visual studio for the first time with a few cmake projects and wow, what a poo poo show
|
# ? Apr 10, 2020 06:42 |
|
If you’re using cmake anyway then give clion a try
|
# ? Apr 10, 2020 06:44 |
|
omeg posted:So, this came out of CI today: I have absolutelly no idea. Maybe a problem with "ungetch"? most probably a out of memory problem https://users.pja.edu.pl/~jms/qnx/help/watcom/clibref/src/ungetch.html or that n in return is not a n, but a math symbol or other thing that happens to look like a n Tei fucked around with this message at 09:46 on Apr 10, 2020 |
# ? Apr 10, 2020 09:41 |
|
|
# ? Jun 5, 2024 20:48 |
|
Tei posted:or that n in return is not a n, but a math symbol or other thing that happens to look like a n Petition to start calling such Unicode symbols "mimics".
|
# ? Apr 10, 2020 11:42 |