How do we fix this problem?
build more magic silicon brain factories
AMD or Intel make a graphics card that you can put in your ass, then show me the chart
>>108762287Improve ROCm/oneAPI, improve availability of FSR and XeSS, while maintaining lower GPU prices. That's all it would take, but it won't happen because Intel and AMD are not really trying to compete with Nvidia on GPUs.
>>108762287When is the last time nvidia made a low profile gpu? how is it that high?
>>108762410NVIDIA as in the company themselves or someone who makes makes cards using the gpus? Because PNY just made the 5080 slim.
>>108762287When the AI bubble pops, the problem will fix itself.
stop buying their stuff, develop cool things on AMD that only work with their hardware
Unironically demand splitting up Nvindia into 2 or more companies. When a company has a monopoly, everyone loses.Too bad this will never happen because cuckservatives needs to drink litres of ceo cum.
>>108762287make cuda software run transparently on amd with zero additional software setup process and provide at least 5-10% better performance per dollar
by making a product that actually works works. AMDs drivers are so fucking bad that the community has to write them from the ground up to have any semblance of stability. also like >>108762382 and >>108762501 said, without a functional compute stack their GPUs are worthless. not to mention feature roulette on things like HW encoders / compute modules is really dogshit.
>>108762491The problem with this is that every piece of hardware that Nvidia makes is dependent on their TSMC silicon and chip architecture. If they split up the company, any piece of it that didn't have the silicon would be entirely dependent and controlled by Nvidia anyways. Take the consumer grade graphics cards. If they split off that side of the business it would be nothing without the rest of Nvidia.
>>108762287Well from the manufactures side, one thing they could do is stop selling the same but worse. If I was running a gpu company I would come out with a low-level hardware native API rather than forcing everyone to use Vulkan garbage. The native api would expose the exact features of the specific gpu with no abstractions.
>>108762287intel stock is up 113% in the past 1 month, 181% in 6 months. that type of gain usually takes over a decade or two to attain. the investors know something we don't and intel is going to overtake amd soon.
>>108762287Maybe AMD and Intel should make more competitive hardware?
>>108763456/thread
>>108762287Make good software that doesn't need a fucking GPU to work
>>108762410RTX A2000, RTX 4000 SFF Ada, RTX Pro 4000 Blackwell
>>108762287We don't know. Maybe we should ask the Chinese.
>>108763456Unfortunately this, it's like AMD gimp themselves on purpose.
>>108762287I don't understand why it's like this, its not like AMD GPUs are hidden away in the back shelves, a lot of stores promote them and try to sell prebuilds with them because of the better performance/value ratio
>>108765788>only sometimes equal performance for games, and most games don't support latest Radeon AI features>worse for AI>worse for Blender>worse for basically anything not gaming really>red (inferior to green)
at 7% they're hardly a threat let alone a "problem".
>>108762287China
