Initial Benchmarks Of The "NVK" Open-Source NVIDIA Vulkan Driver

[partcyborg]

Because scientists and business people are generally very stupid people?

What do I win for getting the correct answer?

You win nothing because that is the wrong answer. The kinds of scientists in question care a lot more that software work and be performant than whether it is open source or not. As has been covered here ad nauseaum, nvidia is leaps and bounds ahead of AMD in the gpu compute space.

[partcyborg]

The fuck you talkin about, bro? Go to bed lol

The only person who posted in this topic that's an Nvidia fanboy (that I'm aware of) is avis/birdie and he really didn't say much in this topic at all.

Just ignore this idiot. He regurgitates the same cringe horror show nonsense in every single thread with even a passing mention of AMD or nvidia

[airlied]

Is there any sort of ballpark estimate for these improvements to be completed? (a month? 6 months? a year? 2 years?)

The compiler is started written in rust. I'd say it would hope to have some useful functionality in 6 month timeframe

[wertigon]

Le sigh...

With Nvidia now having 90%+ of GPU market, and AMD down to like 8%, market has spoken, Open Source sucks and can we please have a $500+ premium on that GPU performance, pretty please?

WIth AMD still not done with the chiplets perf problems, with strong leaks hinting at AMD only releasing the monolithic NAVI43 and NAVI44 dies next year to consumers, those of us who like FOSS are going to have a 3+ year drought of FOSS high end performant hardware. 😢

I just hope this does not mean AMD GPUs are finished, because AMD GPU division is already sent reeling as it is.

[mrg666]

How can they trust if Nvidia will not throw a wrench in all this development effort by adding a small check in their firmware? In my opinion, they should just leave Nvidia alone. All Linux-based (MS + BSD as well) gaming devices have AMD for a reason. Apparently, Nvidia is only interested in open source platforms for AI and computing where they make profit.

[binarybanana]

Would make more sense to bench with the clock rate locked on the Nvidia driver to get an idea how well the drivers compare on even ground.

[HEL88]

What do I win for getting the correct answer?

Perhaps a reflection on your own judgments.

And that's a lot and hard.

[andyprough]

You win nothing because that is the wrong answer. The kinds of scientists in question care a lot more that software work and be performant than whether it is open source or not. As has been covered here ad nauseaum, nvidia is leaps and bounds ahead of AMD in the gpu compute space.

Ahh - so, these scientist friends of yours are generally very stupid people, as I said.

In fact, they are pushing the exact form of stupidity that has caused society to descend into chaotic patent wars, stifling innovation at every turn. You could say that they are frankly "anti-innovation", and therefore " anti-science". Thanks for clarifying.

[avis]

Would make more sense to bench with the clock rate locked on the Nvidia driver to get an idea how well the drivers compare on even ground.

That's not technically possible, as even when the lowest performance mode is enforced , NVIDIA drivers still reach clocks way above (three or fourfold) the boot clocks (which AFAIK are around 100MHz).

[CochainComplex]

Well I do work in science. Proper science. And well a lot of people here do favor opensource. Because its always availavle and you can adjust it. You don't know how many people hate windows but its running everywhere because the administration and the beancounters have issues dealing with a new UI. You will see the most presentations done with latex in science. Defacto writing tool is latex. Most favored scripting language is python....etc. A lot of my colleagues prefer Linux over windows because freedom is worth the burden of sepfmaintaining your system.