Wednesday, 2018-09-12

*** tpb has joined #timvideos00:00
fitzsimmithro: were you able to find the bit file, and load it on the iCE40 HX8K eval board?00:34
CarlFKmithro: did you move the prebuilt repo dir the hdmi2usb firmware lands in?01:20
CarlFKmithro: like is this it for this rev?
tpbTitle: HDMI2USB-firmware-prebuilt/firmware.bin at master · timvideos/HDMI2USB-firmware-prebuilt · GitHub (at
shorneCarlFK: FYI, I tried qemu in litex with the latest patch01:32
shorneit boots01:32
shornethe bios (just need this patch)01:32
CarlFKshorne: woot!01:32
shorne-       EXTRA_ARGS+=("-drive if=mtd,format=qcow2,file=$TARGET_BUILD_DIR/qemu.qcow2,serial=$SPIFLASH_MODEL")01:32
shorne+       EXTRA_ARGS+=("-drive if=mtd,format=qcow2,file=$TARGET_BUILD_DIR/qemu.qcow2")01:32
shorneThat will work because I hard coded a flash model01:33
shornewell, serial controller01:33
shorneI was just working on Some GCC stuff01:33
shorneI think I am done with it, so I will look at litex/qemu now01:34
CarlFKI've done a bit with qemu, so I kinda get -drive if=.. stuff.  but what did serial do?01:35
shornesee how device tree will work, I havent been able to connect with anyone as #qemu yet, I think its vacation/conference time for them01:35
shorneserial, sets the drive controller01:35
shornei.e. [cpu]-->[spi]-->[drive]01:35
CarlFKok I think I got it01:36
shorneit allows you to change the controller used, we have different ones supported by the bios01:36
CarlFKwhy did serial go away?01:36
shorneIts still there, but I just hard coded it in the hardware to always be a "m25p80" chip01:37
shornerather than configurable01:37
shorneWe are now supposed to configure it saying "-device m25p80,xxx"01:38
shorneI dont know how to configure the "xxx" part01:38
shorneI am going to look a bit more into that now01:39
CarlFKahh.. neat.01:46
cr1901_modernmithro: cr1901_modern: I mean there should just be switches which modify the template in some way rather than having to override whole template03:02
cr1901_modernRe this, you should talk to rjo about this. It was a concious decision to move away from what you suggest to move to templates03:02
tpbTitle: Icestorm Backend Improvements by cr1901 · Pull Request #84 · m-labs/migen · GitHub (at
_florent_cr1901_modern: do you remember if you tested the minimal lm32 variant on hardware?03:02
_florent_cr1901_modern: because it does not seems to work on my side when running code from rom (hangs)03:03
cr1901_modern_florent_: It was hit or miss whether it worked w/ arachne, even from ROM03:03
cr1901_modernI have a tag in my private git repo for a design that fails03:03
cr1901_modernminimal lm32 from RONM03:03
_florent_cr1901_modern: but fails to compile or fails to work on hardware?03:04
cr1901_modernthe latter03:04
_florent_ok. Here i'm testing on a Xilinx FPGA, i'm going to see if it's related to the multiplier or divider (or both)03:05
cr1901_modernOh I haven't tested recently on a Xilinx FPGA. Make sure your integrated_ram_size is a nice power of two, like 0x400003:05
cr1901_modernI remember having trouble on mimasv2 w/ this03:06
_florent_cr1901_modern:  is it really interesting to remove the multicycles multiplier/divider? do you have resources numbers?03:23
cr1901_modernit's about 500 LUTs or so to remove cache, mul/div03:24
_florent_ok so we don't have number for only the mul/div?03:26
_florent_ok, so on my side i need a minimal config with no caches03:30
_florent_but i'm not able to use the current one since it does not seems to work03:31
cr1901_modernDo things _work_ without the multipliers?03:31
_florent_so i think i'll revert it to the initial "minimal" variant03:31
_florent_i can try03:31
_florent_i tried with multiplier but no divider, it was not workng03:31
cr1901_modern_florent_: The minimal variant right now should have no cache03:31
*** rohitksingh_work has joined #timvideos03:31
cr1901_modernIt's the "lite" variant that has the cache03:32
_florent_cr1901_modern: yes but it's not working03:32
cr1901_modern>(11:25:23 PM) _florent_: ok, so on my side i need a minimal config with no caches03:32
cr1901_modernDo you want to reenable the cache?03:32
_florent_no i want a config with no caches, but that works :) (it seems current minimal is not working)03:33
cr1901_modernWell Idk how to help you there :(03:33
_florent_i'm testing now with divider enabled, multiplier disabled03:34
_florent_if that's working, i'll change the minimal variant config to that03:34
tpbTitle: cpu/lm32: re-enable multiplier/divider in minimal variant (does not s… · enjoy-digital/[email protected] · GitHub (at
_florent_cr1901_modern: it does not seem to work correctly without multiplier/divider04:09
_florent_cr1901_modern: i have a case where i need this config, so i reverted the changes on the minimal variant04:10
cr1901_modern_florent_: fascinating... well, maybe Xilinx has issues w/ it. I don't use the minimal anyway. I really need the cache04:14
*** rohitksingh_wor1 has joined #timvideos04:14
cr1901_modern_florent_: If I have time I'll take a look into why it fails. Def seems odd04:15
*** rohitksingh_work has quit IRC04:16
cr1901_modernwhat FPGA/board?04:16
_florent_cr1901_modern: i'm testing on arty04:16
_florent_cr1901_modern: but this is a specific design04:16
cr1901_modernHrm... weird04:16
_florent_cr1901_modern: if you don't need it, let's keep things like that04:17
cr1901_modern_florent_: I don't, just would prefer an annotation (maybe in the wiki) that lm32 minimal has mul/div enabled b/c it has trouble synthesizing otherwise.04:18
cr1901_modernI'm very meticulous about docs being consistent *cough*04:18
cr1901_modernA minimal core should otherwise be... well, minimal :)04:18
cr1901_modern_florent_: Also on a XIlinx FPGA, wouldn't it make more sense to use a non-pipelined mul?04:19
_florent_yes, but let's keep the it pipelined for small fpgas04:27
cr1901_moderntinyfpga: Btw, you may wish to /join #m-labs04:27
*** rohitksingh_wor1 has quit IRC04:34
*** rohitksingh_work has joined #timvideos04:37
*** cr1901_modern has quit IRC08:21
*** Shari2 has joined #timvideos09:41
*** cr1901_modern has joined #timvideos10:01
*** rohitksingh_work has quit IRC12:52
*** mauz555 has joined #timvideos13:50
Shari2mithro: is this used in opsis stable release?13:50
tpbTitle: HDMI2USB-litex-firmware/fx2_jpeg_streamer.vhd at master · timvideos/HDMI2USB-litex-firmware · GitHub (at
*** olasd is now known as olasd`14:04
CarlFKShari2: there isn't really a 'stable' release14:04
*** olasd` is now known as olasd14:05
CarlFKso if it looks like that code is used at all, then yes.14:06
CarlFKI guess if it was just added and I haven't had time to test it, then ... um.. still yes ;)14:06
Shari2CarlFK: that code should fragment a MJPEG frame from the encoder into multiple MJPEG payload transfers, but i can not see the fragmentation in wireshark14:08
Shari2CarlFK: im not sure if this code is not active in the v0.0.4-194-ge14edee version or its bugged14:09
*** mauz555 has quit IRC14:14
CarlFKShari2: v0.0.4-194-ge14edee is what I have used in production, so that answers that14:26
CarlFKcan't speak about implementations14:26
CarlFKand I have to run - gona try to sell video service to a convention center - bye!14:26
*** CarlFK has quit IRC14:27
*** rohitksingh has joined #timvideos15:00
*** cr1901_modern has quit IRC16:02
*** theshaun has quit IRC16:11
*** theshaun has joined #timvideos16:11
*** cr1901_modern has joined #timvideos16:28
*** Shari2 has quit IRC16:38
*** Shari2 has joined #timvideos16:38
*** CarlFK has joined #timvideos16:52
*** ChanServ sets mode: +v CarlFK16:52
CarlFKmithro: the opsis hdmi2usb build is timing out - I just re-ran the job, timed out again: "The job exceeded the maximum time limit for jobs, and has been terminated."
mithroCarlFK: It does that sometimes...16:55
CarlFKmithro: can you up the timout?16:55
mithroCarlFK: nope!16:55
mithroCarlFK: It's a timeout that Travis sets -- and it if hits the timeout it'll probably never complete anyway16:56
mithroHave to run!16:56
CarlFKmithro: I bet it is now failing because too many things are being done16:57
mithroCarlFK: Doubtful16:57
CarlFKmithro: I'm responding to Drews email, I think you will approve - bye!16:57
CarlFKmore on timeouts when you have time :p16:57
mithroCarlFK: It tends to go away if I merge another pull request16:58
Shari2CarlFK: USB stream capture works on your opsis, correct? Can you record some USB isochronous transfers with wireshark and upload the trace please?17:05
CarlFKShari2: works yes.  all that other stuff..  can you do it via an ssh connection to my box?17:06
CarlFKShari2: i'm a little strapped for time right now, and I've used wireshark, but I'm not at all an expert, so there would be lots of hand holding17:06
CarlFKi realize it is a GUI app, hoping you can do command line stuff or something17:07
Shari2CarlFK: not sure how to do this via command line, i do some commandline try runs on my setup17:08
Shari2dry runs*17:08
CarlFKShari2: what would be super is if you could come up with something that can be run from Jenkins as a test17:09
CarlFKso far I have 2 or 3 little things, but it basically tests that it is on and not completely broken: linux sees a video device, a serial device, and the version string retrieved over serial is the expected value17:11
CarlFKso anything would be a help17:11
*** rohitksingh has quit IRC17:33
*** swalladge has quit IRC17:39
*** swalladge has joined #timvideos17:41
*** rohitksingh has joined #timvideos19:07
*** Shari2 has quit IRC20:19
*** tdonahue has quit IRC22:12
*** tdonahue has joined #timvideos22:12
*** tdonahue has quit IRC22:21
*** tdonahue has joined #timvideos22:21
*** tdonahue has quit IRC22:25
*** tdonahue has joined #timvideos22:26
*** tdonahue has quit IRC22:34
*** tdonahue has joined #timvideos22:36
*** tdonahue has quit IRC22:40
*** tdonahue has joined #timvideos22:40
*** tdonahue has quit IRC22:43
*** tdonahue has joined #timvideos22:43
*** tdonahue has quit IRC22:51
*** tdonahue has joined #timvideos22:52
*** tac-tics has joined #timvideos23:02
tac-ticsCarlFK, you're always like "tac get in this channel, tacccc y u no here??"23:03
CarlFKhave you got the 25 gig tars?23:03
tac-ticsNo. I have the "I've been applying for jobs and playing video games to cope with the stress of that" :X23:04
tac-ticsBut like... soon.23:04
tac-ticsI got two promising companies I'm looking at right now. So hopefully I'll have a new job by the end of next month.23:06
CarlFKugh- sorry to hear of the stress23:06
tac-ticsOne of them is a hardware company.23:07
tac-ticsSo like23:07
tac-ticsIf that goes well, I'll 100% be doing FPGA stuff23:07
CarlFKI just went to the Crown Convention center next to ORD  to talk about doing video work23:07
CarlFKI didn't get a job, yet. I did get "I can't hire you, but I ca set up a meeting with my boss, I'll do that."23:07
CarlFKare you home?  I'm gonna turn up the presure on getting those 2 files downloaded23:08
CarlFKcuz that step is easy, but it takes hours23:08
CarlFKso get that out of the way so when you wan to do more stuff you can23:08
tac-ticstbh, I forgot the link23:08
CarlFKno problem - I but mithro for links all the time23:08
CarlFKso feel free to bug me - it may make him chuckle23:09
tac-ticswhat is mithro23:09
CarlFKmithro is Tim23:10
tac-ticshi mithro / Tim23:10
tpbTitle: HowTo LCA2018 FPGA Miniconf · timvideos/litex-buildenv Wiki · GitHub (at
CarlFKGo to - click download.23:11
tpbTitle: Downloads (at
tac-ticsWhat's Vivado?23:12
CarlFKconverts source code to blob for FPGA23:12
CarlFKdon't ask me why it isn't called a compiler :p23:12
tac-ticsDo I need a desktop environment to run it?23:12
CarlFKyou need the desktop to run the installer.23:13
tac-tics@[email protected]23:13
CarlFKafter that, no.23:13

Generated by 2.13.1 by Marius Gedminas - find it at!