Tweet
Hi all,
I am Frank Bruno and the GPLGPU is my core. Please feel free to ask questions if you'd like. I wanted to clear some things up:
1) This core is not intended for people to use to replace current graphics. It's not that powerful. It's faster than the original ASIC 15 years or so ago, but it's still the same price if you were to buy an FPGA large enough to compile to. That said, by all means if someone wants to use it to replace their graphics, feel free. I'd love to hear about it. Once I replace the PCI core with an AXI/ AMBA interface, you could put it in an FPGA w/ a PCIe interface.
2) There is a market for cores like this. Some companies need 2D only or basic 3D and have very long life cycles. FPGAs solve these problems. In many cases, to re-cerify software is much more expensive than replacing the hardware. The trade off I am making is to give people access to the code to play with, contribute to and to learn. I'm hoping that people on the software side of the house can help with drivers and hardware types might play with it and improve performance.
3) I took an OpenCL course over the summer to see how far graphics has come over the years. It's come a long way. I do want to work on an OpenCL core at some point, but real life often gets in the way. This core will not ever run OpenCL simply because it's a fixed graphics pipeline base on OpenGL 1.x
Anyways, I hope some of you see value in the project. I do. Andreas of Parallela fame tweeted about this and I hope to be collaborating with him to get the graphics onto his board shortly.
I know some see the value and some don't. Likely it's the HW people vs the SW. I think there is value at looking how projects are done. I know someone said it's a mess, but I think you'll find it's not. It's very well organized and not too hard to follow if you dig into it.
I am Frank Bruno and the GPLGPU is my core. Please feel free to ask questions if you'd like. I wanted to clear some things up:
1) This core is not intended for people to use to replace current graphics. It's not that powerful. It's faster than the original ASIC 15 years or so ago, but it's still the same price if you were to buy an FPGA large enough to compile to. That said, by all means if someone wants to use it to replace their graphics, feel free. I'd love to hear about it. Once I replace the PCI core with an AXI/ AMBA interface, you could put it in an FPGA w/ a PCIe interface.
2) There is a market for cores like this. Some companies need 2D only or basic 3D and have very long life cycles. FPGAs solve these problems. In many cases, to re-cerify software is much more expensive than replacing the hardware. The trade off I am making is to give people access to the code to play with, contribute to and to learn. I'm hoping that people on the software side of the house can help with drivers and hardware types might play with it and improve performance.
3) I took an OpenCL course over the summer to see how far graphics has come over the years. It's come a long way. I do want to work on an OpenCL core at some point, but real life often gets in the way. This core will not ever run OpenCL simply because it's a fixed graphics pipeline base on OpenGL 1.x
Anyways, I hope some of you see value in the project. I do. Andreas of Parallela fame tweeted about this and I hope to be collaborating with him to get the graphics onto his board shortly.
I know some see the value and some don't. Likely it's the HW people vs the SW. I think there is value at looking how projects are done. I know someone said it's a mess, but I think you'll find it's not. It's very well organized and not too hard to follow if you dig into it.
Comment