_florent_ changed the topic of #litex to: LiteX FPGA SoC builder and Cores / Github : https://github.com/enjoy-digital, https://github.com/litex-hub / Logs: https://libera.irclog.whitequark.org/litex
tpb has quit [Remote host closed the connection]
tpb has joined #litex
matoro has joined #litex
ElfenKaiser has joined #litex
Degi has quit [Ping timeout: 248 seconds]
Degi has joined #litex
CarlFK1 has joined #litex
ElfenKaiser has quit [Quit: 🫀🫀🫀🫀]
CarlFK1 has quit [Ping timeout: 244 seconds]
TMM has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
TMM has joined #litex
CarlFK has joined #litex
oter has quit [Remote host closed the connection]
oter has joined #litex
<sajattack> I'm wondering what would be the best way to create a userspace/cpu-accessible framebuffer with litepcie, to make sort of a rasterizer card or something like that. I have an acorn which probably doesn't have enough io to actually drive a display (and no baseboard), has anyone tried something like this? I'm not too familiar with pcie or wishbone for that matter. But curious how someone would approach this
<sajattack> is there a way to just map a block of memory that both can share under linux?
<tnt> sajattack: you can create a memory mapped without ( BAR ) that just exposes your VRAM.
<tnt> s/without/window/
<sajattack> when you say vram, do you mean on the fpga or on my pc's gpu?
<tnt> on the fpga
cr1901_ has joined #litex
<sajattack> yeah so that's kind of what I was thinking, how does the memory mapping work?
cr1901 has quit [Ping timeout: 276 seconds]
<sajattack> using this example, is it just resource0 == the vram/dram?
<tnt> yes you could map bar0 directly to your vram or some part of it.
<tnt> you could also define another BAR
<tnt> (BAR0 is usually used for mmio registers)
<sajattack> cool
<tnt> Although not sure if/how litepci supports multiple BAR.
<tnt> Ok, it doesn't so you have to use bar0 (or a part of it).
<tnt> Note that this won't exactly be fast ...
<sajattack> I found this issue about a similar thing too https://github.com/enjoy-digital/litepcie/issues/54
<sajattack> this looks like it's for cuda though
<sajattack> yeah so I guess just do the 54 one and access the memory with the cpu one pixel at a time?
<sajattack> or try to find a way to map it to the gpu 🤔
<tnt> I don't understand what you mean. If you expose some RAM on the FPGA through a BAR, yeah, the CPU will access it through PCIe one pixel at a time.
<sajattack> nevermind
whitequark[cis] has joined #litex
<whitequark[cis]> i think you could maybe use MTRRs to make it cacheable?
Degi has quit [Ping timeout: 260 seconds]
<sajattack> vivado keeps failing to download lol
Degi has joined #litex
cr1901_ is now known as cr1901
ElfenKaiser has joined #litex
ElfenKaiser has quit [Ping timeout: 265 seconds]
FabM has joined #litex
so-offish has joined #litex
so-offishul has joined #litex
so-offish has quit [Ping timeout: 248 seconds]
so-offishul has quit [Read error: Connection reset by peer]
ElfenKaiser has joined #litex
FabM has quit [Ping timeout: 244 seconds]
TMM has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
TMM has joined #litex
<cr1901> If you are a Tera Term user (Windows), and use LiteX, would you be willing to try this plugin? https://github.com/cr1901/teraterm-litex
<cr1901> It _does_ work for me, including the installers. But I don't have a second machine to try a "no dev env" install on.