![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
It's a VERY small sum that spend that much. I just, for instance, kept my 6900xt and Instead saved my money. That was not hard to do. Now I have money and a still wicked quick gpu.
I hope you are right. Or at the very least, sure have your halo product but put everything under it back to sane prices. $1200 4080's and $800 4070 Ti's isn't good for anyone.
None of that is really related to gaming.
I also expect an uptick in 1080p screen sales, which then increases a need for 3060 class performance at a reasonable price point.
If you are into game engine development much of it is very relevant.
Sorta... most of this stuff related to data science and ML. Some of that can be helpful to game development... but quickly going through it also seems to suggest all these things would also wall you into having to use nVidia hardware? I'll keep reading, kinda cool anyway. Thanks for the share.
Their work there relating to the C++ libraries and the matrix transformations does a lot for greatly improving performance when you change camera angles or rotate objects on the screen, core stuff like that but yes it does use NVidia exclusive features, which are different than proprietary as there is nothing stopping others from inventing similar features, they just haven't.
Which is why they're going all-in on the machine learning angle since it's still a hotly competitive market with no established standards.
There is also me sort of standard all the calculations need low precision 8 and 4 bit. Something the current AMD lineup doesn’t do at all, they approximates those calculations using 16bit. But it makes the Nvidia hardware between 30 and 100x faster at those calculations than on the relevant AMD hardware.
Though I misspoke about what Gameworks "is", the ultimate conclusion on the affect is the same. Though nVidia themselves also refer to it as an SDK.
There are tons of articles out there about this subject. It was probably one of the most debated things in the graphics space in 2014-2015. In fairness to your point there is some level of he-said she-said. All things in the tech sector are opaque. Said another way: Intel suppressed AMD for 10 years and even that took a long protracted time in a court of law to reveal conclusively.
Sure I can. Especially considering all of the open source options that AMD is using are things nVidia can actually optimize for, whereas vice versa AMD cannot. If devs use "AMD technology" (that's all open source) they are free to share any and all optimization information with nVidia. If devs use nVidia Gameworks, they can share no optimization information with AMD. That's explicitly stated in that quote.
Because they have to be. nVidia has a vested interest in governing standards for graphics, only the Promoter level allows them to have a member on the board of Khronos group and gives voting rights. Considering that the dues to do this are "only" $75k per year, which nVidia can likely find as change in their couch cushions, this isn't surprising.
Fair. How many of those SDK's/libraries/API's are specifically for gaming? I'm not great at counting, but I'm going with 4? Maybe? (The ones listed under the Design, Visualization, and Simulation tab) In comparison to AMD which includes all of their SDK's/libraries/API's and all of the middleware?
If you are building a gaming engine, so Unreal, Unity, etc ... Much of what is listed is very relevant, if you are developing a game based on one of those engines then it isn't something you worry about because it's not something you are actively working with that is where you move over to Nvidia Game Works, which is Opensource but using Nvidia repositories which are of various source types.
It's "partially" open, but also requires a proprietary/commercial license. And they didn't even get to partial open source until well after AMD more or less forced their hand by creating GPUopen.
Updated 09/29/2021 09:58 AM
->
Article said:
Read what you just wrote. You just "proved" something by .. what. You said it?, that proves nothing. Read the quote from the Article you linked.. it disproves your point. (lol)
So you are still going on about Gameworks?? See the quote above as it applies to this too.
What a huge pile of bullshit you have here.
lol.
Well, your first link doesn't have margin information from EVGA, just a JRPeddle research graph with it's final point "2022 Est" which means estimated. Good try tho.
Probably.
Now your'e reaching.
G-Sync Ultimate monitors are pretty amazing... I played Cyberpunk on a G-Sync monitor with the proprietary chip, and it was smooth at 35 fps. I know that without that it wouldn't have played as well.
Agree to disagree, fine with me. You hate Nvidia, we get it.
The only thing closed source was G-Sync. RTX is open. Cuda is open. AMD could probably design their GPU's to work with DLSS, but they don't need to, they have FSR. What other technologies are left? Nothing. Simple shit like shaders and tessellation they can both do without any help from anyone.
AMD does this because it is in their best interest. Not everything AMD is "open", but those that come to market after something else has been established by a competitor, better be free or they have little chance of gaining any significant uptake. See previous post where I said they would die on the vine.
Yes.
Again, this isn't true. Gameworks sucked on AMD for 1 generation. AMD's raytracing sucking, that's not Nvidia's fault, and it is incredibly naïve to think a company should give it away. Raytracing works with DirectX API's that anyone can program for, game devs and GPU drivers from anywhere. Vulkan too. It's OPEN. Still sucks on AMD, which is AMD's own fault. they are always 2 years late and a dollar short. And they really have no excuse anymore, their CPU's are doing great, so where is the R&D inverstment? Maybe they just don't give enough of a shit... "We've had to give all of this open source shit away... why are we even doing this again? oh yeah, we can build GPU's into our CPU's... fuck!"
No they don't. Maybe 25 years ago when all of the hardware was different and you have to program directly to the hardware. This is why we have DirectX and Vulkan and OpenGL. The game dev programs to the standard API's, that (some of which) have been standard for more than a couple of decades. DirectX was already at v9 in 2002.
CUDA is propreitary and requires vendor lock in. It's not platform agnostic. AMD aside, I can't run this on Intel hardware either.
https://www.extremetech.com/gaming/...surps-power-from-developers-end-users-and-amd
What Middleware, API, or SDK does AMD use that is closed source?
FrgMstr this guy says you're a liar.
Indeed. If you don't understand nVidia's pricing structure then you can make that ignorant statement. nVidia placed a cap on how high any vendor can price a GPU . Meaning, if you can't make a GPU cost what nVidia "wants it to cost" then you lose money.
They don't until they do. Or are you still not aware that Gameworks was a huge controversy in the gaming community. It was talked about on every board including this board extensively.
I don't think anyone cares when nVidia makes a graphics advancement - in the sense of "complaining about it". That is obviously not the issue. The issue is what that does in the gaming space and to the other half of the market.
Not at all. It's been expressed that GSync modules add $100 to the cost of monitors, considerably more if you want HDR. If the free open standard didn't exist and we wanted to have some form of adaptive sync what do you think would happen?
Which with nVidia walls gives an inferior experience. If nVidia has their way anyway.
I used to love nVidia when they were the underdog in the late 90's and early 00's. It's not their being on top I object to. It's their anti-consumer behavior. But you're their lap dog, I get it.
Cuda isn't open. If it was then it would work on Intel and AMD cards and doesn't. It has a proprietary license and only works on "supported GPUs" (nVidia). AMD has to use OpenCL and the rest of their OpenGPU stack to compete.
If you wanted adaptive sync you were. If Freesync didn't exist, you would be forced to buy it in order to have that feature. You can't have it both ways.
Second time: what SDI/APK/Middleware does AMD use that are closed source? Still waiting.
So you have a reading comprehension issue. Because my first post on this directly addressed RT as an exception. And the major reason it's the exception is because it was adopted into the DX12 spec. That is in stark contrast to literally every other nVidia technology we're talking about.
You've read none of the Gameworks articles. The devs explicitly state that in the design of an engine they have to choose their dev path. If you use nVidia's middleware stack and optimize for them there isn't really a way to insert GPUopen Middleware without, guess what?, doing it twice.
This is OT.
Sadly the only way to make Nvidia stop being dicks is to offer something competitive against them, there is very little out there that is, there is no open alternative for CUDA, and the 3 or 4 projects that do exist that combined encompass what CUDA can do is 6-10 years behind in terms of performance and development libraries.
Well said.
Yes they are on features, as has been discussed repeatedly. As far as Nvidia and rt+dlss it is still a huge net image quality gain and Dlss is in 2.5x+ the number of titles fsr is, which can also be used by Nvidia. Framerates vary by game but unless you think 60+ is unplayable you'll be thrilled.
Yep, they deliver.
Exactly. I don't care about them being "dicks" or whatever Linux users seem to call them. I care about the awesome cards they deliver. I don't need their software to all be FOSS to be good tech. I speak as a gamer and indie game dev.
Did you speak with any of the card manufacturerers or read the GPP documents? What is your source? Would you care to provide us with links?
? Enough. It was overblown like so many other tech "scandals" in my opinion. Nice bait, do you have those yourself? You're the one who claims it was horribly unfair.
Nvidia is a corporation that will flex its power to expand, improve, and dominate the environment it is in. Every corporation does the same. I admire Jensen and the empire he has built. NV is a well-run company and innovates like few others. The strong survive and the weak expire. Hopefully, Intel and AMD won't back down to the challenge. Consumers need competition. Unfortunately, there isn't much at the moment.
I am playing most of my games at 120-144 fps at 4K using the features including RT. It’s not only RT, it is also DLSS, frame gen etc. 4090 is the first proper 4K card. Not sure I can say that about 7900XTX.
Like I give a shit what he thinks...
Hasn't 4k@60fps been solved for awhile now? I wouldn't call the 4090 "the first" proper 4k card.
