After the GPU, the PPU and.. the BPU, here comes RPU !!1!
category: general [glöplog]
http://graphics.cs.uni-sb.de/~woop/rpu/rpu.html
[quote]The RPU is a fully programmable ray tracing hardware architecture, with support for programmable material, geometry and lighting. The RPU combines the efficiency of GPUs with the advantages of ray tracing. The instruction set of the RPU is GPU like, which is optimal for shading purposes. In addition the RPU supports fast ray traversal through an k-D tree using a dedicated hardware unit and recursive function calls, usefull for recursive ray tracing. To increase efficiency always 4 rays are handled in a packet and multi-threading allows for high utilization of the hardware units.
A working prototype of this hardware architecture has been developed based on FPGA technology. The ray tracing performance of the FPGA prototype running at 66 MHz is comparable to the OpenRT ray tracing performance of a Pentium 4 clocked at 2.6 GHz, despite the available memory bandwith to our RPU prototype is only about 350 MB/s. These numbers show the efficiency of the design, and one might estimate the performance degrees reachable with todays high end ASIC technology. High end graphics cards from NVIDIA provide 23 times more programmable floating point performance and 100 times more memory bandwidth as our prototype. The prototype can be parallelized to several FPGAs, each holding a copy of the scene. A setup with two FPGAs delivering twice the performance of a single FPGA is running in our lab. Scalability to up to 4 FPGA has been tested.[/qoute]
*faints*
[quote]The RPU is a fully programmable ray tracing hardware architecture, with support for programmable material, geometry and lighting. The RPU combines the efficiency of GPUs with the advantages of ray tracing. The instruction set of the RPU is GPU like, which is optimal for shading purposes. In addition the RPU supports fast ray traversal through an k-D tree using a dedicated hardware unit and recursive function calls, usefull for recursive ray tracing. To increase efficiency always 4 rays are handled in a packet and multi-threading allows for high utilization of the hardware units.
A working prototype of this hardware architecture has been developed based on FPGA technology. The ray tracing performance of the FPGA prototype running at 66 MHz is comparable to the OpenRT ray tracing performance of a Pentium 4 clocked at 2.6 GHz, despite the available memory bandwith to our RPU prototype is only about 350 MB/s. These numbers show the efficiency of the design, and one might estimate the performance degrees reachable with todays high end ASIC technology. High end graphics cards from NVIDIA provide 23 times more programmable floating point performance and 100 times more memory bandwidth as our prototype. The prototype can be parallelized to several FPGAs, each holding a copy of the scene. A setup with two FPGAs delivering twice the performance of a single FPGA is running in our lab. Scalability to up to 4 FPGA has been tested.[/qoute]
*faints*
not that this is news, but it still looks like poor post 2000 3d-rendering.
"Wow".
Also note how they carefully avoid mentioning frame rates.
well, there's a video and they have an overlay with framerate and rendertarget size information. but it's fucking 66mhz (!) and it can render an UT2003 map with 18-25 fps and an edge filter for adaptive antialiasing !
Amazing!
I wonder if they solved the usual raytracing problems yet, like no support for dynamic meshes, AA/texture filtering only possible through bruteforce, post-processing effects only possible on CPU... etc.
I wonder if they solved the usual raytracing problems yet, like no support for dynamic meshes, AA/texture filtering only possible through bruteforce, post-processing effects only possible on CPU... etc.
een hardware TRIANGLE raytracert?
Scali: This card can only handle per-object-linear-animations (i.e. matrix-vector-multiplication) for a limited number of objects..
And only VERY limited scene-sizes..
And all BSP-construction has to be done on the CPU in a precalc-process (which takes up to 24h depending on object-size!)..
For a cooler demonstration of FULLY animated scenes check: Breakpoint Seminar: To Trace or not to Trace (shameless self-promotion)
And only VERY limited scene-sizes..
And all BSP-construction has to be done on the CPU in a precalc-process (which takes up to 24h depending on object-size!)..
For a cooler demonstration of FULLY animated scenes check: Breakpoint Seminar: To Trace or not to Trace (shameless self-promotion)
Well, judging from this RPU-movie, the answer would be: Not To Trace.
Someone just got their SIGGRAPH proceedings like I did :)
WANHA!
...and what about the LPU? =D
