|
The game is now worse because AMD exists as a company. Seriously gently caress AMD
|
# ? Aug 18, 2023 16:19 |
|
|
# ? Jun 4, 2024 03:33 |
|
I'm mad about video games!
|
# ? Aug 18, 2023 16:20 |
|
spunkshui posted:I'm legit mad they even considered this is an idea let alone went through with it. lol, y'all are mad for no DLSS, but where is all of the outrage for all of the poo poo ndivida has done to AMD to lock them out of features and degrade their performance. Like, ndivida does this too, but you're not mad at them? Just at the underdog?
|
# ? Aug 18, 2023 16:21 |
|
Rawrbomb posted:lol, y'all are mad for no DLSS, but where is all of the outrage for all of the poo poo ndivida has done to AMD to lock them out of features and degrade their performance. Like, ndivida does this too, but you're not mad at them? Just at the underdog? You don't understand. This inconveniences me slightly, and makes something I might enjoy a little less potentially enjoyable. The greatest of sins!
|
# ? Aug 18, 2023 16:22 |
|
Nvidia just has enough market dominance that nobody notices when they do it.
|
# ? Aug 18, 2023 16:23 |
|
spunkshui posted:The game is now worse because AMD exists as a company. STFU. AMD isn't the one with closed-off VRR (gsync) or temporal upscaling (dlss), both of which were only opened up thanks to AMD's efforts. Nvidia has abused their market position far more than AMD has ever done. CUDA is the gigantic whale in the room. Keep bitching about things you have no clue about though. You are apparently not a kid stuck in the back of the car yet you act like one
|
# ? Aug 18, 2023 16:28 |
|
Yudo posted:You don't understand. This inconveniences me slightly, and makes something I might enjoy a little less potentially enjoyable. The greatest of sins! Doesn't inconvenience me none as a 4090 haver. It hits the mid and bottom rungs disproportionately, which should upset people. power crystals posted:Nvidia just has enough market dominance that nobody notices when they do it. Just enough? They are an effective monopoly. I'm not going to bat for Nvidia here, but this ain't it chief. This is not the way to beat Nvidia, it's just wildly petty at best Zedsdeadbaby posted:STFU. Dude chill, lol
|
# ? Aug 18, 2023 16:29 |
|
Kibner posted:You can also do that with the Steam Deck. My gf does it all the time with hers. Sure, but then like, you gotta buy a Steam Deck. I already own a phone. Also a Steam Deck can't stream over data like a phone can, unless I guess you tether your phone to it. But carriers nerf tethering more than direct-to-phone data.
|
# ? Aug 18, 2023 16:29 |
|
Taima posted:Dude chill, lol
|
# ? Aug 18, 2023 16:32 |
|
Taima posted:Doesn't inconvenience me none as a 4090 haver. It hits the mid and bottom rungs disproportionately, which should upset people. No it doesn't. Those people aren't on 40xx series cards, and so dlss3 is irrelevant to them. And the low end is consoles, where dlss in any form is also totally irrelevant.
|
# ? Aug 18, 2023 16:32 |
|
The game is worse because AMD spent money to make sure it didn't run as well as it could on NVidia hardware. The suits making these calls will never boot the game once. Defending this is insane. We should all want far less of this to be happening.
|
# ? Aug 18, 2023 16:33 |
|
Taima posted:Just enough? They are an effective monopoly. I'm not going to bat for Nvidia here, but this ain't it chief. This is not the way to beat Nvidia, it's just wildly petty at best No I'm agreeing with you, poo poo sucks, but it's like how nobody notices when games are Steam exclusives. I think you read that "just" in the wrong place in my post there. All this upscaling stuff should be part of directx/vulkan/whatever rather than being vendor specific, but that doesn't sell GPUs, so here we are.
|
# ? Aug 18, 2023 16:34 |
|
it's good news for puredark at leastpower crystals posted:All this upscaling stuff should be part of directx/vulkan/whatever rather than being vendor specific, but that doesn't sell GPUs, so here we are. having a common abstraction layer for all flavors of TAAU is what nvidia proposed with streamline, and AMD said "lol no" to that would they say yes if streamline were a microsoft project instead of an nvidia one? maybe but i doubt it
|
# ? Aug 18, 2023 16:34 |
|
power crystals posted:No I'm agreeing with you, poo poo sucks, but it's like how nobody notices when games are Steam exclusives. I think you read that "just" in the wrong place in my post there. FSR isn't vendor specific in that it will run on any gpu, not just Radeon. It is the people's upscaler.
|
# ? Aug 18, 2023 16:36 |
|
Isn't XeSS also vendor agnostic? The ideal is still that it's not controlled by one of the GPU hardware vendors regardless of being able to run on anybody's cards.
|
# ? Aug 18, 2023 16:38 |
|
all of the big three are vendor specific to some degree, even FSR2 despite being open source is developed behind closed doors without taking any outside contributions so it's exclusively going to focus on what's best for AMD hardware (i.e. no ML until they catch up)
|
# ? Aug 18, 2023 16:41 |
|
Rawrbomb posted:lol, y'all are mad for no DLSS, but where is all of the outrage for all of the poo poo ndivida has done to AMD to lock them out of features and degrade their performance. Like, ndivida does this too, but you're not mad at them? Just at the underdog? "I am developing my own stuff that I am going to hoard and not share" is bad. "I am going to pay companies to not use technology to make my stuff look better" is worse You can be upset at AMD doing this without having to delve into whataboutisms
|
# ? Aug 18, 2023 16:41 |
|
power crystals posted:Isn't XeSS also vendor agnostic? The ideal is still that it's not controlled by one of the GPU hardware vendors regardless of being able to run on anybody's cards. And it has a version that runs better on Intel cards. I'm amazed that people are almost spinning this Starfield kerfuffle as a positive.
|
# ? Aug 18, 2023 16:41 |
|
power crystals posted:Isn't XeSS also vendor agnostic? The ideal is still that it's not controlled by one of the GPU hardware vendors regardless of being able to run on anybody's cards. To some extent, yes, but the XeSS implementation for Arc is far more complete than Radeon and GeForce. It isn't an even playing ground. repiv posted:all of the big three are vendor specific to some degree, even FSR2 despite being open source is developed behind closed doors without taking any outside contributions so it's exclusively going to focus on what's best for AMD hardware (i.e. no ML until they catch up) The whole point of FSR is that you don't need to do ML to do upscaling. I guarantee you that RDNA3 can handle the matrix math to do "AI" upscaling should that be a thing they wanted. Rinkles posted:I'm amazed that people are almost spinning this Starfield kerfuffle as a positive. It is more the deep, personal umbrage people seem to take over it that is ridiculous. Maybe I am just very old, but in my day if you didn't have a 3DFx card you were basically hosed. We have come a long way. Yudo fucked around with this message at 16:48 on Aug 18, 2023 |
# ? Aug 18, 2023 16:42 |
|
Yudo posted:The whole point of FSR is that you don't need to do ML to do upscaling. I guarantee you that RDNA3 can handle the matrix math to do "AI" upscaling should that be a thing they wanted. you also don't need temporal feedback to do upscaling but it turns out to be really useful, AMD initially touted FSR1s non-temporal approach as an advantage and we saw how that turned out
|
# ? Aug 18, 2023 16:55 |
|
Given how insanely advanced and accurate modern TV upscaling is, I wonder if it would be less complex to put a proper hardware upscaler in low-mid range cards instead of relying on ML and per-game integration.
|
# ? Aug 18, 2023 16:57 |
|
repiv posted:you also don't need temporal feedback to do upscaling but it turns out to be really useful, AMD initially touted FSR1s non-temporal approach as an advantage and we saw how that turned out But FSR2 is fine--we have seen how it has turned out too. I have used it a lot, now even side by side with DLSS2. There isn't some stark, game breaking difference.
|
# ? Aug 18, 2023 16:59 |
|
spunkshui posted:
No one's defending it. Just people making fun of when consumers pick and choose which anti-competetive practices they want to get mad at and pretend either AMD, Nvidia or intel are the 'good guy'. I mostly think it's very funny that AMD think this is in anyway going to help them.
|
# ? Aug 18, 2023 16:59 |
|
SlowBloke posted:Given how insanely advanced and accurate modern TV upscaling is, I wonder if it would be less complex to put a proper hardware upscaler in low-mid range cards instead of relying on ML and per-game integration. GPUs now come with Sports Mode
|
# ? Aug 18, 2023 17:00 |
|
Yudo posted:But FSR2 is fine--we have seen how it has turned out too. I have used it a lot, now even side by side with DLSS2. There isn't some stark, game breaking difference. if the difference is so small then why is AMD spending big bucks to keep DLSS off the table, if they perform about the same then it shouldn't matter which one users choose
|
# ? Aug 18, 2023 17:11 |
|
Taima posted:The whole "paying companies to not use DLSS or RTX" thing is such a perversion on all sides. I am curious though; how much money is AMD paying for something like Starfield to adopt this model? Do we know anything about how these agreements work, or is that all still behind closed doors? People need to stop thinking of AMD as "the little guy". They control an overall majority of the gaming market, and that's the leverage they're using here. NVIDIA controls a majority of some minority segments of the market, but if AMD wants to throw around their weight they can do it. Even in the CPU market they are not "the little guy" anymore, they are moving close to 40% marketshare in the server market etc. Anyway the angle AMD is selling to studios is "validate once, run everywhere". That if you have four upscalers in your code, that it's going to be a significant increase in validation work (does water look broken in XeSS? How about DLSS?), and many studios are bottlenecked by validation and time-to-market right now (a world of rushed ports and unpolished launches). The problem is that studios apparently don't think this way because AMD is having to pay them to keep competitors' code out. And also that even if you had to pick one, it still wouldn't be FSR because even XeSS DP4a pathway looks way better than FSR2. The problem is AMD doesn't support DP4a on anything pre-RDNA2 (not even RDNA1!) so it's a nonstarter for AMD to back this route. And from a market perspective, the hope is that if they can just get FSR2 in a bunch of poo poo and squash DLSS for the next year or two, that hype will die out, people will agree that FSR2 is "close enough", and move on from "proprietary NVIDIA tools". The freesync playbook - and it doesn't have to be better than DLSS for the freesync playbook to work! AMD slapped the label on absolute poo poo for years, but that allowed them to "build a market consensus" regardless of how unusable the product actually was in 90% of the monitors with the label on it. NVIDIA eventually gave up and started working on adaptive sync instead because that was where the market was going. And maybe they'd do that with FSR2 if they had obviously lost traction on DLSS. Zedsdeadbaby posted:STFU. Paying to develop proprietary technology that improves the experience of your customers is different from paying to keep your competitors' technology out of products. The latter is basically what AMD accused Intel of doing. They had the better technology, AMD argued that Intel wrote a check to keep it out of products and that it constituted anticompetitive behavior. It is what it is, proprietary tech is a fact of all brands in the tech world. AMD isn't paying to develop Infinity Fabric or Infinity Link and then open it up. They'll support their proprietary Infinity Fabric Coherent PCIe Interconnect over the openCXL standard even though it's a redo of the gsync-vs-adaptive sync situation in the enterprise world. The cloak of "AMD loves open source so much!" works on gamers so incredibly well, and open-source has been so incredibly successful that people are suspicious and skeptical of anything that isn't open. But there is lots of stuff that isn't open in the tech world, AMD isn't open-sourcing RDNA's VHDL design files either! You don't pay to bring a product to market and then give it away. Nor do you do that for individual technologies either. If you want these experience-improving things to come to market, companies expect to make a return on them. The second-mover always goes open and has an easier time copying the technology and knowing exactly what market segments to target, because the market has already been proven and the problem space has been explored a bit. Again, AMD isn't giving away their proprietary openCXL predecessor either. They paid to develop that. They built the market for it. And now you need AMD CPU+GPU to get the combo bonus. There are plenty of corporate partners who I'm sure would love to get in on AMD's internal capabilities for Infinity Architecture rather than having to push uphill getting OpenCXL specified and implemented and adopted. And while there might not be much sympathy for that, that's ultimately the same situation as g-sync was with NVIDIA. This is about corporate partnerships and licensing, not a relationship with end-users, and the negative effects work exactly the same. The products are worse because AMD is standing on their proprietary tech and refusing to open it up. Paul MaudDib fucked around with this message at 17:52 on Aug 18, 2023 |
# ? Aug 18, 2023 17:12 |
|
repiv posted:if the difference is so small then why is AMD spending big bucks to keep DLSS off the table, if they perform about the same then it shouldn't matter which one users choose Because monopoly is the only business strategy in tech. And FSR's relative performance isn't some mystery that AMD is obfuscating.
|
# ? Aug 18, 2023 17:19 |
|
Has there been a game on the PC in the modern era that has been an actual, full blown "Nvidia" or "AMD" exclusive and simply not playable otherwise?
|
# ? Aug 18, 2023 17:19 |
|
from software have no idea how much money they're leaving on the table by not taking money from amd for not adding temporal upscaling to any of their games anyway
|
# ? Aug 18, 2023 17:20 |
|
The nice thing about gsync monitors when they came out, as someone who doesn't use vrr/gsync/vsync at all, is they were almost all high quality monitors for the time. The non-gsync versions with the same panel from the same company were $50 or $100 less but sucked horribly. CUDA was infuriating as someone who wrote gpgpu code professionally during that era and I'm still mad people actually used it. Of course, the reason people used it is nvidia wrote a ton of useful libraries for people to use and deliberately didn't write it for OpenCL (which they were part of the committee on). Nvidia also would help you with cuda and respond to emails very quickly. Sometimes they'd even patch drivers in no time at all, and I worked for a nobody at the time. Nvidia also didn't properly support the opencl spec for a while and left vendor specific defects unpatched in their drivers, a big anti-competitive jerk move. It's only now, an eternity later, that people are not vendor locking to cuda as much. Is AMD actually paying companies to not do xess/dlss or is amd saying "partner with us and we'll write FSR support practically for you"? There's a pretty big ethical distinction there despite an almost identical outcome. kliras posted:from software have no idea how much money they're leaving on the table by not taking money from amd for not adding temporal upscaling to any of their games anyway It's great how little they care about pointless stuff and instead focus on making good games. Khorne fucked around with this message at 17:43 on Aug 18, 2023 |
# ? Aug 18, 2023 17:26 |
|
repiv posted:having a common abstraction layer for all flavors of TAAU is what nvidia proposed with streamline, and AMD said "lol no" to that they have said they'd refuse any API (open or not) that allows plugging proprietary code into it. So essentially the library and game would have to be GPL. Open game engines only. Streamline is MIT-licensed ("do anything you want with it, including non-redistribution of derived code, but you can't sue us") but you can still use it to plug in proprietary code so that's a no. Microsoft would have to follow a similar license because nobody will open-source their games just to meet an unreasonable demand from NVIDIA. it's the "give up the WMDs or we invade" tactic from AMD, make an impossible demand that they know nobody can follow, such that they can look like the good-guys crusading to protect open source instead of just flatly saying no. I still think NVIDIA could call their bluff and have their lawyers spend a week coming up with "GPL-lite" (CDDL or similar perhaps?) that requires pluggable modules to follow CC-NC-SA licensing or similar... so others would be free to improve the DLSS model but couldn't commercialize the resulting product. That would be the same kind of embrace-and-extinguish that AMD is attempting with FSR2 - you can join us, but everything you do only reinforces the ecosystem even farther. If you want open-source, here's open-source! But at this point NVIDIA is winning anyway so there's no point in poking the bear. I do think that if NVIDIA open-sourced the model that it could be adapted to other platforms just like LLaMA or Stable Diffusion. You can take the base Stable Diffusion model and quantize it to 4b or whatever and run it on apple silicon. And you could probably do the same thing with DLSS and XeSS, you could quantize it to run on XMX and DP4a and Apple Silicon and even AMD's own ML accelerator instruction on RDNA3. Ultimately that's where things are going to lead if streamline really takes off, one model in different places and forms. And that may not be something NVIDIA wants either. They can't stop Intel from doing XeSS and similar things, short of AMD-style anticompetitive shenanigans where they write a check to kill it off but they don't have to hand the model over either. But there's not anything super novel there other than the model. The glue code isn't complex and people have come up with glue code that hooks FSR2 into DLSS titles. I'm sure someone could come up with glue code that hooks a DLSS model that's been quantized for AMD GPUs into the DLSS DLLs in exactly the same way. Paul MaudDib fucked around with this message at 17:39 on Aug 18, 2023 |
# ? Aug 18, 2023 17:27 |
|
spunkshui posted:I'm legit mad they even considered this is an idea let alone went through with it. lotta funny posts but I think this one takes it papa nvidia will hug us all
|
# ? Aug 18, 2023 17:27 |
|
Paul MaudDib posted:Paying to develop proprietary technology that improves the experience of your customers is different from paying to keep your competitors' technology out of products. This. One of them is a bunch of engineering and the other is a bunch of money just paying to make the other companies product not work well. Like hiring a loving hitman to go out there on a race track and shove a potato in your tail pipe. Literally all AMD did here is just used money to make computers work worse. thanks AMD
|
# ? Aug 18, 2023 17:31 |
|
Paul MaudDib posted:And from a market perspective, the hope is that if they can just get FSR2 in a bunch of poo poo and squash DLSS for the next year or two, that hype will die out, people will agree that FSR2 is "close enough", and move on from "proprietary NVIDIA tools". The freesync playbook - and it doesn't have to be better than DLSS for the freesync playbook to work! AMD slapped the label on absolute poo poo for years, but that allowed them to "build a market consensus" regardless of how unusable the product actually was in 90% of the monitors with the label on it. NVIDIA eventually gave up and started working on adaptive sync instead because that was where the market was going. And maybe they'd do that with FSR2 if they had obviously lost traction on DLSS. Yeah I think that's a good point; most people don't have a baseline with both technologies, so I guess it's ultimately he said/she said. You can ultimately pay to sway that kind of discourse. And the point about freesync is well taken, that's an interesting angle that I wasn't thinking about. You could even argue it's already working. Tons of people really like FSR; I would personally argue that its nowhere near the quality of DLSS, but at the end of the day? That might not matter. They could be spending this money on improving their product but they would rather make the game experience worse. Gee I wonder why people don't like that? Yudo posted:No it doesn't. Those people aren't on 40xx series cards, and so dlss3 is irrelevant to them. And the low end is consoles, where dlss in any form is also totally irrelevant. Sorry could you elaborate? My understanding is that there's no DLSS, period. Personally I don't even like the DLSS3 features, despite having access to them. Frame gen is not good enough yet for me to care, but I'm totally open to it if the tech is improved. The whole "there is no low end consumer GPUs" is a really... interesting take though, I'll give you that. You're making personal attacks dude I don't know what to say. I don't know what you mean by that emoji either, am I not allowed to have opinions? Do you imagine in your head that anyone who doesn't like something is frothing at the mouth as they type about it? idgi. Anyways that's the last I'll discuss it, cheers. Taima fucked around with this message at 17:35 on Aug 18, 2023 |
# ? Aug 18, 2023 17:32 |
|
Zedsdeadbaby posted:STFU. VRR exists because of Nvidia, they adopted Freesync and unlike AMD set actual standards for displays. Temporal upsampling would have eventually become mainstream, but it's where it is now because of Nvidia, and they've built a framework for implementing each vendor's algorithm through one API. Nvidia has done some bullshit but your gaming-related examples are awful. And GPU compute... I mean everyone else has themselves to blame for letting Nvidia build a position of total dominance because they were too lazy and stupid to see that GPU compute would inevitably keep growing bigger and bigger. Yudo posted:But FSR2 is fine--we have seen how it has turned out too. I have used it a lot, now even side by side with DLSS2. There isn't some stark, game breaking difference. The difference is still quite big. DLSS Quality is usually at least as good as native TAA, I generally prefer it. FSR2 can sometimes compete with or beat native in static shots, but often it doesn't and it looks terrible in motion. If they were anywhere near equivalent AMD wouldn't be spending money to lock Nvidia out, because no one would care.
|
# ? Aug 18, 2023 17:33 |
|
Paul MaudDib posted:they have said they'd refuse any API (open or not) that allows plugging proprietary code into it. So essentially the library and game would have to be GPL. Open game engines only. someone should tell AMD that directx and vulkan allow for proprietary backends, including AMDs own windows driver the non-free software is coming from inside the house
|
# ? Aug 18, 2023 17:35 |
|
People getting way too worked up over fancy upscalers.
|
# ? Aug 18, 2023 17:37 |
|
Taima posted:
Its an “ironicat” (irony) My posts are absolutely just my opinion, so are yours. We both have strong opinions. Everyone here can post and disagree
|
# ? Aug 18, 2023 17:38 |
|
Taima posted:You could even argue it's already working. Tons of people really like FSR; I would personally argue that its nowhere near the quality of DLSS, but at the end of the day? That might not matter.
|
# ? Aug 18, 2023 17:40 |
|
|
# ? Jun 4, 2024 03:33 |
|
spunkshui posted:The game is worse because AMD spent money to make sure it didn't run as well as it could on NVidia hardware. So me on the charts where nvidia performs worse than amd.
|
# ? Aug 18, 2023 17:40 |