Nvidia and AMD, why is Nvidia the favored for development for most thing (based on what I have seen).

[glitchgodprime]

So im getting in to the creation world through the college realm and have found something interesting. Art and programming really favor Nvidia over AMD for everything and as an AMD user I feel left out. 

Adobe it would seem has code and addons meant for Nvidia for taking work from the CPU and pushing it on the GPU.
Unreal and Unity both have addons for Nvidia to speed up development of rendering.
3Ds Max has built in functions only supported by Nvidia.
But isn't Nvidia like super picky on their dealings of other companies and isn't AMD all for opensource drivers and support?

So why is Nvidia the go to since I would guess the requirements to work with their drivers would be much more time consuming to set up?

[jaslion]

Marketshare and money. Nvidia isnt stupid and by keeping themselves relevant with better software support for these creative and work tools they keep a bigger marketshare.

 

So basically if you go amd here you are shooting yourself in the foot if time is money so nobody does and thus support is lacking unless amd themselves will develop the tools or pay these companies money.

 

 

[mariushm]

Probably because nVidia has more market share (higher percentage of gamers have nvidia video cards)  and because nVidia has bigger pockets and pays programmers to implement those features or has teams of programmers go to studios and companies to "give advice" on how to implement nvidia specific features in their game engines or programs. 

 

Some hardware features are better on nvidia cards (like the video encoders) and some others are benefits of being first (cuda vs open cl) ... 

[Spotty]

3 minutes ago, glitchgodprime said:

So why is Nvidia the go to since I would guess the requirements to work with their drivers would be much more time consuming to set up?

NVIDIA is the one investing the time to develop those things in partnership with the software studios. Also marketshare. If 80% of your customers are using NVIDIA (probably actually much higher than that in the 3D design space) then it makes sense to develop your programs for that. No point investing time in to AMD if nobody in the industry is using it.

[jaslion]

22 minutes ago, Spotty said:

probably actually much higher than that in the 3D design space

From working as it support for a large visual developtment studio group (10k users) for half a year in 2019 its was basically 96% nvidia (about a 70/30 split quadro geforce, 3,...% intel hd stuff and the leftovers amd systems mainly for compatibility for deployments and development.

 

They are irrelavent in this sector except for end deployment then its just to make sure the things work.

 

Small edit: consoles not included

[igormp]

32 minutes ago, glitchgodprime said:

But isn't Nvidia like super picky on their dealings of other companies

Yes, they are.

32 minutes ago, glitchgodprime said:

isn't AMD all for opensource drivers and support?

Yes, they are, except that they're shitty at most stuff. Good luck trying to run ROCm and getting any ML framework to work on top of that, meanwhile you can get it working without any hassle with Nvidia's CUDA.

 

33 minutes ago, glitchgodprime said:

So why is Nvidia the go to since I would guess the requirements to work with their drivers would be much more time consuming to set up?

They provide way better support and they ecosystem is actually easier to setup compared to AMD. Their APIs are awesome to work with (as long as you use them as nvidia intended), and heavily invested into adding CUDA support to most professional tools out there. Meanwhile AMD isn't doing much and barely help any devs that are interested in their technologies, sadly.

 

Intel also has a pretty nice support, way better than AMD's, and most of their stuff is also open source. The sad part is that their hardware is trash.

[BuckGup]

1 hour ago, igormp said:

Yes, they are.

Yes, they are, except that they're shitty at most stuff. Good luck trying to run ROCm and getting any ML framework to work on top of that, meanwhile you can get it working without any hassle with Nvidia's CUDA.

 

They provide way better support and they ecosystem is actually easier to setup compared to AMD. Their APIs are awesome to work with (as long as you use them as nvidia intended), and heavily invested into adding CUDA support to most professional tools out there. Meanwhile AMD isn't doing much and barely help any devs that are interested in their technologies, sadly.

 

Intel also has a pretty nice support, way better than AMD's, and most of their stuff is also open source. The sad part is that their hardware is trash.

I agree with you on the Nvidia part but disagree on the Intel/AMD part. When AMD is testing new CPUs or there is an issue they send multiple techs on site to stay until the issue is resolved. They have great document sharing and don't stonewall you for corporate reasons. Intel is not as bad as Nvidia but still shit. A group of devs I know were working on a tool for flashing firmware on Intels hardware and they wouldn't give them the firmware. So they revere engineered the pieces they needed and Intel got pissed but at the end of the day Intel was the one that needed the work done.

[igormp]

1 hour ago, BuckGup said:

I agree with you on the Nvidia part but disagree on the Intel/AMD part. When AMD is testing new CPUs or there is an issue they send multiple techs on site to stay until the issue is resolved. They have great document sharing and don't stonewall you for corporate reasons. Intel is not as bad as Nvidia but still shit. A group of devs I know were working on a tool for flashing firmware on Intels hardware and they wouldn't give them the firmware. So they revere engineered the pieces they needed and Intel got pissed but at the end of the day Intel was the one that needed the work done.

Let me rephrase that: AMD is shit when it comes to software support for their 3rd party developers. Just take a look at their github, in special when it comes to ROCm stuff, it's a shit show all over the place with issues being ignored for years, which makes their software stack pretty much useless.

 

Meanwhile, Intel provides top notch support for their open source stuff, but they can be a PITA when it comes to anything other than that, pretty much like nvidia.

[BuckGup]

3 hours ago, igormp said:

Let me rephrase that: AMD is shit when it comes to software support for their 3rd party developers. Just take a look at their github, in special when it comes to ROCm stuff, it's a shit show all over the place with issues being ignored for years, which makes their software stack pretty much useless.

 

Meanwhile, Intel provides top notch support for their open source stuff, but they can be a PITA when it comes to anything other than that, pretty much like nvidia.

Hahaha I guess when I was thinking support I meant physical and accommodations not necessarily dev support. As yes AMD is not the leader of the pack for software support

[tkrsh_rage]

CUDA.

[mr moose]

Because Nvidia invested resources into developers, not just game development but all realms of GPU usage.  When you provide developers with cheap (if not free) tools that make their job easier and faster they are going to continue using your products thus ensuring your market share grows.

[Vishera]

This is the what the Dolphin devs think about each companies OpenGL drivers (last update is from 2019):

 

Quote

Excellent - NVIDIA

We had no occasion to test NVIDIA's response time to driver issues because we actually never had any issue with NVIDIA's OpenGL implementation, either on Windows or on Linux.

 

Quote

Good - Mesa

Surprisingly, open source drivers (Mesa) were far from the worst offenders: while one could assume they receive less QA and less professional testing, it turns out that we only found a few bugs in Mesa, all related to a single feature that was supported recently in Mesa 9.0/9.1. More than that, all of the bugs we found in Mesa were promptly fixed after we reported them through the correct channel (freedesktop.org bug tracker). Communication with drivers developers was always good, which is standard in the open source world: when asking on IRC, Mesa developers were very excited about having an application using their new, shiny feature.

 

Quote

Good - Intel HD Graphics on Windows

Intel integrated graphics chipsets are usually not a good match for Dolphin: while the GPU usage of the emulator can be decreased significantly, even our minimal profile is usually too slow to run on Intel IGPs. However, we are not aware of any bugs in the OpenGL implementation present in Intel’s IGP drivers on Windows impacting Dolphin. Unfortunately, the supported feature set is also very reduced

 

Quote

Mediocre - AMD

We had the most issues with AMD when using their proprietary graphics driver on Linux, fglrx/Catalyst. A lot of issues that do not happen on Windows are present on Linux, sometimes with a very visible effect in our emulator.

 

Quote

Bad - ARM/Mali

Let’s start talking about mobile GPU drivers. For us open-source developers, this is where things start becoming a nightmare:

• Only binary blobs, no good documentation about the architecture
• Drivers limiting performance of the devices when the GPU could perform faster
• Phone manufacturers and operating systems developers bragging about support of features their hardware could have had for the last 2 years (OpenGL ES 3.0) if drivers were not so bad
• Very few applications using advanced OpenGL features, thus making Dolphin reveal how broken some of the GLES3 features are.
• Driver developers that seem to be doing less QA than open-source teams like Mesa

The Mali GPUs, made by ARM, are not the worst offenders. In fact, Dolphin on Android now runs properly on Mali devices, after a lot of time spent working around driver bugs and general slowness.

 

Quote

Horrible - Qualcomm/Adreno

Where do we even start? First, a good thing: Qualcomm and Mali definitely seem to be sharing some code in their graphics drivers. It’s only unfortunate that this code is handling glBufferSubData and glMapBufferRange. That’s right: the Mali bugs causing these two functions to be slow and unusable (out of memory condition after a few seconds of runtime) are also present in the Qualcomm drivers for Adreno devices! At least our vertex streaming changes to support Mali devices were useful for more than one driver.

 

[Kisai]

On 9/19/2022 at 6:05 AM, glitchgodprime said:

So im getting in to the creation world through the college realm and have found something interesting. Art and programming really favor Nvidia over AMD for everything and as an AMD user I feel left out. 

Adobe it would seem has code and addons meant for Nvidia for taking work from the CPU and pushing it on the GPU.
Unreal and Unity both have addons for Nvidia to speed up development of rendering.
3Ds Max has built in functions only supported by Nvidia.
But isn't Nvidia like super picky on their dealings of other companies and isn't AMD all for opensource drivers and support?

So why is Nvidia the go to since I would guess the requirements to work with their drivers would be much more time consuming to set up?

 

This is basically a question of "which company provides the best support"

NVIDIA - essentially the best, first-to-release/last-to-open-source

Intel - Open sources most, if not all things

AMD - Open sources things and then generally ignores issues

Everyone else (ARM vendors)... this is why Android is awful.

 

Pretty much the only reason you see any AMD support for anything in the first place is because of Apple. If you are on the Apple platform, there is no NVIDIA. So the only option is AMD.  Apple Silicon came way way later.

 

So software like Adobe, Autodesk and various other companies like Blackmagic design (Davinci Resolve), and even open source software like FFMPEG and OBS build all their products around the assumption that the user has NVIDIA hardware, or is doing everything on the CPU. Nvidia has been extremely consistent with their driver support for features and not removing features until relatively recently. Where as Intel's support tends to be flakey at best. AMD? *cough 6500XT*, it's a question if a card even has it.

 

If you read between the lines however, you'll note that both Adobe and Autodesk software don't actually say they require an "NVIDIA GEFORCE".

https://knowledge.autodesk.com/support/3ds-max/downloads/caas/downloads/content/3ds-max-certified-hardware.html

 

Quote

 

There's two pages of NVIDIA Quadro/A-series cards, not a single Geforce card. Meanwhile, The Radeon Pro cards aren't even half a page, and Intel Arc... two lines. That's what Autodesk verified. Now, the same document also does list the consumer cards a page later, with every single one listed a "pending" GTX 960 to RTX 3090. Radeon HD7700 to R9 290x. Basically Autodesk doesn't test them. Even if you look at previous versions of the software. The last time they tested any consumer hardware was 3DS Max 2021.

 

Likewise AutoCAD, Revit, and so forth care more about the video memory bandwidth, which is why if a GPU doesn't have 8GB on it, it's essentially unusable for most tasks. These products all use OpenGL.

 

For Adobe products, They use OpenGL on Windows, but they also use the hardware video encoders/decoders on the video tools. They may use CUDA in the video effects stage. Davinci Resolve Studio uses CUDA for it's effects pipeline.

 

For it's part, AMD has tried to get people off the CUDA ecosystem.

https://www.phoronix.com/news/AMD-Radeon-GPUFORT

Quote

AMD Publishes Open-Source "GPUFORT" As Newest Effort To Help Transition Away From CUDA

But this is what OpenCL and "DirectCompute" were supposed to do anyway. Do you see any software using either of these? Not really. 

 

Part of this lock-in is PyTorch/Tensorflow, as developers pretty much build for "CUDA" within these libraries, and not actually build for a machine agnostic library.

https://www.khronos.org/sycl/

Quote

As you can see, even that is a supreme mess.

Obligitory:

[Neroon]

I remember playing a bunch of early access games, and many of them ran pretty shit on AMD because the support for indy early access, was pretty much 0.

 

While I wouldn't be recommending against getting an AMD gpu, I also wouldn't recommend it.

[wasab]

Nvidia simply has more market share and much better driver and support for other developers. Their SDKs are robust and their drivers work from day one. Same can't be said for AMD. E.g. 6xxx series open source drivers refused to work until months later. This is speaking from personal experiences. I went out of my way to return 6800 xt for a rtx 3070 for this reason. 

[Just that Mario]

Because Nvidia cares for that market and supports it. AMD is gamer oriented and is a gamer's wet dream for benchmark scores.

[BlueChinchillaEatingDorito]

On 9/29/2022 at 9:52 AM, Just that Mario said:

Because Nvidia cares for that market and supports it. AMD is gamer oriented and is a gamer's wet dream for benchmark scores.

AMD just wants big number. Don't mind the dozens of crashes and full system lockups it took to get there. BUT LOOK BIG NUMBER.