| kristianpaul | If i'm sending data from a 80Mhz clock to a 16Mhz one, 3 flip flops should be okay for the clock domain sync, isnt?? | 03:40 |
|---|---|---|
| wolfspraul | how many frames per second does Milkymist render with right now? | 14:30 |
| lekernel | something like 24 | 14:37 |
| kristianpaul | how many bits for color? | 14:39 |
| lekernel | 16 | 14:39 |
| kristianpaul | and jsut to confirm resolution for render is 800x600 right? | 14:39 |
| lekernel | it's 640 atm | 14:39 |
| kristianpaul | (atm), this is limited by ADC chip?7 | 14:40 |
| lekernel | no by fpga design | 14:41 |
| kristianpaul | DAC** | 14:41 |
| kristianpaul | ok | 14:41 |
| lekernel | dac can do 1280x1024 60 fps 24bpp | 14:41 |
| lekernel | if you want render at this performance level, jfdi | 14:41 |
| lekernel | I think it's technically feasible, with a lot of work though | 14:42 |
| kristianpaul | we can go to then moon!, jfdi :-) | 14:46 |
| Action: kristianpaul kidding | 14:46 | |
| kristianpaul | lekernel: minimac2_psync.v line 28, can you give me a hint doing that level <= ~level ? | 14:48 |
| wolfspraul | 16 bits per pixel, right? | 14:49 |
| kristianpaul | maybe two signal changes are closer together?? | 14:49 |
| lekernel | see http://www.fpga4fun.com/CrossClockDomain.html | 14:49 |
| lekernel | wolfspraul, yes | 14:49 |
| wolfspraul | ok so 640x480, 16bpp, 24fps | 14:49 |
| wolfspraul | great, thanks | 14:49 |
| kristianpaul | sure i know is cross clock domain, but i wast aware if this kind of implementation.. | 14:49 |
| Action: kristianpaul reads | 14:49 | |
| lekernel | wolfspraul, this is hopefully subject to change with FPGA design upgrades, so don't write it all over the marketing material | 14:50 |
| lekernel | kristianpaul, the full minimac2_psync is well explained on the fpga4fun website | 14:50 |
| kristianpaul | ok | 14:50 |
| wolfspraul | no worries. first I need to understand where it is today, what is realistic with updates, what is the theoretical maximum of the current hardware | 14:50 |
| wolfspraul | that's important in many ways, for customers, also for potential contributors, etc. | 14:51 |
| wolfspraul | all fine, I think it's all great | 14:51 |
| lekernel | if we get to do the "pro" version i'll probably make a new soc architecture to support 30bpp 1280x1024 everywhere (or something like that) | 14:51 |
| lekernel | with improvements such as switching to myhdl, new bus architectures, faster memory controller, faster texture mapping unit, multiple LM32 cores, etc. | 14:53 |
| lekernel | as I said, lots of work | 14:53 |
| kristianpaul | (myhdl), :o | 14:54 |
| kristianpaul | "get to do the "pro"" you mean, basically is a funding problem? | 14:55 |
| lekernel | partly, it's also a problem of finding customers for it | 14:56 |
| kristianpaul | just to be sure, that 'pro' word scare me a bit from before projects i know ;) | 14:56 |
| kristianpaul | sure | 14:56 |
| kristianpaul | yeah, sure you are the experts about recomended specs for VJing stuff | 14:57 |
| kristianpaul | ha, flags i missed that part :-) | 16:31 |
| mumptai | hi | 17:16 |
| lekernel | hi | 17:20 |
| GitHub143 | [milkymist] sbourdeauducq pushed 1 new commit to master: http://bit.ly/pKALxU | 17:20 |
| GitHub143 | [milkymist/master] Update README - Sebastien Bourdeauducq | 17:20 |
| GitHub184 | [milkymist] sbourdeauducq pushed 1 new commit to master: http://bit.ly/rhXNY7 | 18:35 |
| GitHub184 | [milkymist/master] Update headers - Sebastien Bourdeauducq | 18:35 |
| GitHub23 | [milkymist] sbourdeauducq pushed 1 new commit to master: http://bit.ly/ojj6rG | 18:37 |
| GitHub23 | [milkymist/master] Bump version number - Sebastien Bourdeauducq | 18:37 |
| GitHub150 | [milkymist] sbourdeauducq pushed 1 new commit to master: http://bit.ly/nZH6tV | 21:37 |
| GitHub150 | [milkymist/master] fmlmeter: add memory address capture support - Sebastien Bourdeauducq | 21:37 |
| --- Mon Jul 11 2011 | 00:00 | |
Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!