Friday, 2016-08-05

*** tpb has joined #timvideos00:00
*** paradisaeidae has joined #timvideos03:53
*** rohitksingh has joined #timvideos04:06
*** rohitksingh_work has joined #timvideos04:07
*** rohitksingh_work has quit IRC04:10
*** rohitksingh_work has joined #timvideos04:12
*** rohitksingh has quit IRC04:14
*** rohitksingh has joined #timvideos04:16
*** ssk1328 has joined #timvideos04:43
mithrohey ssk132804:45
ssk1328mithro: Hi04:45
ssk1328mithro: The earlier problem was solved04:46
ssk1328mithro: Now I can't seem to figure out why, but the another DMA block is not working correctly04:46
mithrossk1328: Describe "not working correctly"04:47
ssk1328mithro: Okay, basically whatever I set for hdmi_out0_fi_base1_write(), it doesn't reflect in output04:48
ssk1328Only hdmi_out0_fi_base0_write() thing reflects on the output04:48
mithrossk1328: That is still not very descriptive04:50
ssk1328hdmi_out0_fi_base1_write() being the function used to set base address for second DMA port04:50
mithrossk1328: What do you expect to happen, and what is actually happening?04:50
ssk1328mithro: So I am mixing the outputs from DMA0 and DMA1, and the corresponding functions I stated before set the base address for these DMA ports04:51
mithrossk1328: Yes, that is the theory04:51
ssk1328mithro: As per the what output I see, I sense that, both the mixer block are getting same input from DMA004:52
ssk1328mithro: And changing base address for DMA1 doesn't affect anything04:53
mithrossk1328: Have you looked at what the hdmi_out0_fi_base1_write() function is doing?04:53
ssk1328mithro: So we are using hdmi_out0_fi_base0_write(), to set the base address from where DMA0 will read the contents04:55
mithrossk1328: So what you are saying is "I call the hdmi_out0_fi_base1_write function with a value and I don't see any changes in the output"04:55
ssk1328mithro: So for the other port address is set using hdmi_out0_fi_base1_write()04:55
ssk1328mithro: Yes04:55
mithrossk1328: That isn't a lot of information so far. Does the hdmi_out0_fi_base0_write() function do anything?04:56
ssk1328mithro: Its a function in csr.h, it stores the data in some CSR registers04:57
mithrossk1328: I know what it does - but doing a quick compare of that the working hdmi_out0_fi_base0_write and the nonworking hdmi_out0_fi_base1_write is a good idea04:58
ssk1328mithro: In the csr.h, only thing to compare, is they store the corresponding value in different memory locations in memory04:59
ssk1328mithro: What exactly do you mean by compare04:59
mithrossk1328: Look at the generated C code05:00
ssk1328mithro: Yeah, they just store the value passed at different locations, thats it05:00
*** Bertl is now known as Bertl_zZ05:11
mithrossk1328: confirming that they are writing to different locations is a good thing to do05:13
mithrossk1328: Next step would be confirming that the value is getting to your hardware05:14
mithrossk1328: I also find the code organisation in a bit confusing05:15
tpbTitle: HDMI2USB-misoc-firmware/ at float-arithmetic · ssk1328/HDMI2USB-misoc-firmware · GitHub (at
ssk1328mithro: I think this is because of  self.sdram.crossbar.get_master() function?05:16
ssk1328mithro: Do you know exact working of this05:16
ssk1328mithro: I think i need to call it two times, which I am not doign right now05:17
mithrossk1328: That could be the case - get_master does seem to create something ->
tpbTitle: misoc/ at legacy · timvideos/misoc · GitHub (at
mithrossk1328: You should try and draw the dataflow graph of G too05:34
ssk1328mithro: I was at the exact same line right, now, I guess I need to call it two times05:35
ssk1328mithro: It has an append function as well05:35
mithrossk1328: possibly05:35
mithrossk1328: btw I'll be around on the weekend and the start of next week for testing05:36
ssk1328mithro: So I am gonna try doing that right now05:36
mithrossk1328: great05:36
mithroI'll be in and out for the next couple of hours - have to get ready for a flight to Melbourne05:36
ssk1328mithro: Okay05:36
ssk1328mithro: As always I will ping here if I run into problems05:37
mithrossk1328: It can be helpful to just talk into the channel even if nobody responds05:38
ssk1328mithro: Yes05:39
mithrossk1328: Do you know about the idea of "rubber duck debugging"?05:39
tpbTitle: Rubber duck debugging - Wikipedia, the free encyclopedia (at
ssk1328mithro: Sounds interesting05:39
mithrossk1328: It is in some ways another way of saying "You don't understand something until you can explain it in simple language to someone else"05:44
ssk1328mithro: Yeah, most of the problems, I figured in something similar to this way!05:46
*** cr1901 has quit IRC07:27
mithrossk1328: any luck?07:33
*** cr1901 has joined #timvideos07:40
*** paradisaeidae has quit IRC07:40
*** ssk1328 has quit IRC07:55
*** ssk1328 has joined #timvideos08:26
ssk1328mithro: Not working right now08:27
ssk1328mithro: In status at HDMI2USB propmt I see ddr : read 0Mbps08:28
mithroDoes that normally work when using the video target?08:29
ssk1328Yeah, they don't work the moment I add another instatntiation of                 self.sdram.crossbar.get_master()08:30
ssk1328Otherwise it works08:31
mithroWhat do you mean? The bandwidth measurement doesn't work when you update your code for the extra master?08:31
mithroI'm about 5-10 minutes away from getting on a plane08:32
ssk1328mithro: When I connect the extra master in my block, the output itself doesn't work, and also ddr read is 0Mbps08:32
ssk1328mithro: I have been diving into dma codes as well08:33
ssk1328mithro: Can't seem to pinpoint the problem08:33
mithroI can't look at the code now, but how are you combining the outputs of the two dma engines?08:33
mithroIE that needs some type of synchronisation right?08:34
ssk1328mithro: Combinig I do later in my usual mixer , block, spcifcally using the floatadd08:34
mithroYou combined the pixels into a two pixel record, right?08:35
mithroWhich you then pass to the next block!08:35
mithroTyping on the phone is hard :-P08:36
ssk1328mithro: I changed layout at some places, yes08:36
mithroLook at how that bit works08:37
ssk1328mithro: The VTG has two Sinks instead of one now, each of which comes from a Cast think08:37
mithroThe output of those two sinks are combined into a single record which is then sent out a source, right?08:39
ssk1328mithro: No08:40
ssk1328mithro: The two sinks are separate08:40
mithroThat isn't going to work then08:41
ssk1328mithro: Okay, understood, I will focus on this part then08:41
mithroYou need to take the output from the two sinks, join it together and pass it onto the next stage08:41
ssk1328I guess the cast thing should combine both the streams in single layout08:41
mithroYeah, it might have some magic which does that type of thing already08:42
mithroBut I don't know08:42
ssk1328cast is a simple combination thing which just rearranges the connection08:43
ssk1328mithro: I am gonna resolve this08:43
mithroBoarding the plane now08:43
ssk1328mithro: This also explains this was working with single lasmi block08:43
ssk1328mithro: Okay08:43
ssk1328I will get back to you later08:43
mithroYeah, a single lamsi block is automatically synced with itself :-)08:44
ssk1328mithro: Yeah, seems logical08:44
ssk1328mithro: Anyways happy journey to you! :)08:44
mithroSee ya!08:45
*** danielki has joined #timvideos11:06
*** rohitksingh_work has quit IRC11:22
*** rohitksingh_work has joined #timvideos11:23
*** panther has joined #timvideos11:33
*** panther is now known as sab_12311:34
rohitksinghsab_123: ready?11:37
rohitksinghyou have the link right?11:38
*** rohitksingh has quit IRC12:02
*** sab_123 has quit IRC12:02
*** Bertl_zZ is now known as Bertl12:25
*** danielki has quit IRC12:38
*** rohitksingh_work has quit IRC12:56
*** sb0 has quit IRC13:20
*** danielki has joined #timvideos13:35
*** ssk1328 has quit IRC15:05
*** CarlFK has quit IRC17:22
*** CarlFK has joined #timvideos17:29
*** ChanServ sets mode: +v CarlFK17:29
*** sb0 has joined #timvideos17:32
*** Bertl is now known as Bertl_zZ18:17
*** danielki has quit IRC18:17
*** cr1901 has quit IRC19:28
*** cr1901 has joined #timvideos19:41
*** Bertl_zZ is now known as Bertl19:54
*** CarlFK has quit IRC22:22
*** CarlFK has joined #timvideos22:38
*** ChanServ sets mode: +v CarlFK22:38
*** miselin has quit IRC23:28
*** cr1901_modern1 has joined #timvideos23:32
*** cr1901_modern has quit IRC23:35

Generated by 2.13.1 by Marius Gedminas - find it at!