3 ms·
Nothing is going to displace CUDA in the short term. ROCm is a piece of garbage. Intel is miles behind.
by greenavocado 2y ago
Nothing is going to displace CUDA in the short term. ROCm is a piece of garbage. Intel is miles behind.
- Jensson 2y agoCan't someone else just implement CUDA for their chips? I don't see that being worth trillions, it is hard but isn't that hard to replicate. Billions makes sense, not trillions.
- fragmede 2y ago"just"
- Jensson 2y agoCompared to a trillion dollars it is "just", it is just a software implementation moat, its a compiler it isn't on the scale of reimplementing windows or so.
- Detrytus 2y agoCUDA is propertiary, NVIDIA recently changed it's license terms to forbid implementing CUDA translation layers for non-NVIDIA devices: https://www.techpowerup.com/319984/nvidia-cracks-down-on-cuda-translation-layers-changes-licensing-terms https://www.techpowerup.com/319984/nvidia-cracks-down-on-cud...
- Jensson 2y agoI thought such contracts doesn't hold in court, you are allowed to implement interfaces for interoperability right?
- mnau 2y agoHaving a different API from CUDA is not the problem. Having working and performant product is the problem. AMD made a nearly clone of CUDA API with HIP (the implementation is garbage btw). The API mostly changes prefix from "cuda" to "hip" and even has semi-automatic source code modifier tool to make the switch. It's like any other software migration (e.g. from on-prem to cloud): If ("money to switch library" < "savings per quarter" * quarters) { "Do the switch" } else { "Status quo" }