| --- Log opened Thu Sep 05 00:00:42 2013 | ||
| stekern | _franck_: I think that should be possible | 03:49 |
|---|---|---|
| stekern | there a slight risk that we have to modify the FIFO for it to work, but I think it might be ok as is | 03:50 |
| stekern | I'll take a little deeper look | 03:50 |
| stekern | ah, no sorry, it's the bufram that uses that I assume | 04:52 |
| stekern | but anyway, it *should* work, the generic ram has a different behaviour now when I look at it | 04:55 |
| stekern | it always return old data | 04:55 |
| stekern | is that supported by cyclone II? | 04:55 |
| stekern | perhaps it can't... | 04:57 |
| stekern | cyclone v can't at least | 04:57 |
| stekern | I had to do a change like this to mor1kx for that: https://github.com/openrisc/mor1kx/commit/164a8707d7b64354a584add5245b52c59e2f30c5 | 04:57 |
| stekern | but "read new" should be "better" for the bufram, so I think you can change that without worries | 04:58 |
| stekern | we need to change the generic RAM to have the same behaviour too | 04:59 |
| stekern | hmm, a question about version numbers... now we have v1.0 of mor1kx | 05:42 |
| stekern | and I have created a seperate mor1kx_v1 branch, where the idea would be that we roll in critical bug fixes and stable stuff for upcoming v1.1 | 05:42 |
| stekern | master is moving towards 2.0, i.e. new features and other crazy stuff go there | 05:43 |
| stekern | the question is: what should the version number of master be? | 05:43 |
| stekern | right now it's "1.0", but that feels wrong | 05:44 |
| stekern | perhaps our major releases should always start at 1.1, and x.0 indicating that it's the crazy stuff | 05:47 |
| stekern | a bit like gcc | 05:47 |
| stekern | I mean, always start at x.1 | 05:48 |
| stekern | so, a major release would start with a git tag just pointing to a "normal commit" and a new branch, no update in the mor1kx-defines | 06:02 |
| stekern | a major x.0 release, and then when enough "stuff" have built up on that we do the x.1 release with update in mor1kx-defines | 06:04 |
| stekern | and right after a x.0 tag would follow a mor1kx-define updating the x to x+1 | 06:04 |
| stekern | juliusb: are you reading this, what do you think? | 06:05 |
| stekern | http://pastie.org/8299621 <= perhaps a bit more coherent description of what I tried to say above | 06:25 |
| olofk | stekern: I *think* that bufhit & (&buf_clean_wb) has problems in startup. Could be that it never enters the first write because the condition isn't fulfilled until after a write | 06:32 |
| olofk | Or something like that | 06:32 |
| olofk | _franck_: I think it would be great to have test benches for all cores/systems in orpsocv3, but that's up to the creator | 06:32 |
| stekern | olofk: hmm, but that *should* be handled by the first_req | 06:33 |
| olofk | stekern again: I'm having the same problems with orpsocv3. My configure.ac says 3.0 now, but I'm also thinking about using odd minor numbers to indicate that it's a version between releases | 06:33 |
| olofk | Didn't try bufhit & refill_busy. Will do that right away | 06:34 |
| stekern | olofk: what I have in mind is more of x.0 == unstable, x.1..x.n == stable | 06:35 |
| olofk | Like KDE? :) | 06:35 |
| stekern | I'm not familiar with the KDE release process | 06:36 |
| stekern | but that would fit in to what our v1.0 release is | 06:36 |
| stekern | and to how we work | 06:36 |
| olofk | Just joking. There was so much complaints about KDE 4.0. | 06:37 |
| olofk | That could be a good thing for ORPSoCv3 as well | 06:38 |
| olofk | It's more of a beta release now really | 06:38 |
| stekern | it breaks the "I want to do a release, but I'd like to get what I'm working on done first" pattern | 06:38 |
| olofk | I agree | 06:39 |
| stekern | I mean, we're laid back people, our releases need to be laid back ;) | 06:40 |
| olofk | I rerun the test bench with (bufhit & !refill_busy) and I get the exact same results | 06:45 |
| stekern | "exact same results" as in 'bufhit & (&bufclean_wb)' or '(.. & (bufhit | ....)) & !refill_busy'? | 06:58 |
| _franck_ | stekern: so we have cyclone II fifo that can return old data but can't return new data AND cyclone V which is exactly the oposite....great ! | 07:08 |
| stekern | sans fifo | 07:09 |
| stekern | so cyclone II rams can only do old data? | 07:09 |
| stekern | note, that the NEW_DATA_WITH_NBE_READ is actually an read-before-write | 07:11 |
| stekern | => 'old-data' | 07:11 |
| stekern | but I think any behaviour should be ok for the bufram | 07:13 |
| stekern | we write into bufram at WRITE, then go back to IDLE and then into READ in write-read sequence | 07:14 |
| stekern | in a | 07:14 |
| _franck_ | ok I'll take a closer look. I thought NEW_DATA_WITH_NBE_READ was new data. Anyway, with NEW_DATA_WITH_NBE_READ quartus is not happy at all | 07:14 |
| stekern | so there should be a 2-cycle latency from the data been written to it's read, so the new data should in either case have time to propagate | 07:15 |
| _franck_ | stekern: ok, we just need to find something generic enough to avoid hacking your files everytime we change the targetted FPGA | 07:16 |
| stekern | maybe the TARGET parameter could be more precise, like ALTERA_CYCLONE2 | 07:17 |
| olofk | That sounds like a plan | 07:18 |
| _franck_ | I think so | 07:18 |
| stekern | and you just add a ram that fits your device | 07:18 |
| olofk | Xilinx have the same problem. V5/V6 FIFOs differ quite a bit | 07:18 |
| stekern | let's do that, and let the old ALTERA fall back to ALTERA_CYCLONE4 | 07:19 |
| olofk | One of my early ideas with orpsocv3 was to build a FIFO library where all FIFOs had a common interface and let orpsoc choose the correct fifo with the depend parameter in the .core file | 07:19 |
| stekern | actually, the parameter name was 'TECHNOLOGY' not 'TARGET', but same logic applies ;) | 07:21 |
| juliusb | stekern: (mor1kx version stuff) I think that's fine | 07:21 |
| juliusb | so the version number is only updated in the tagged version? | 07:21 |
| stekern | juliusb: yes and no. did you read the pastie? | 07:22 |
| juliusb | yes, still not 100% clear I guess | 07:22 |
| juliusb | but master is always .0? | 07:22 |
| juliusb | and tag | 07:22 |
| juliusb | tags are .1 .2 etc? | 07:22 |
| stekern | let's break it down to what we have now instead of x and y's | 07:23 |
| stekern | 1.0 is now in mor1kx_v1, the first commit where mor1kx_v1 branched off from master is tagged as v1.0 | 07:24 |
| stekern | the commit after the v1.0 tag in master updates MOR1KX_VERSION_MAJOR to 2 | 07:25 |
| _franck_ | why don't you make it like linux (old numbering) : 1.0 is tagged, you fix something here and you have 1.0.1. You tag a version with new feature and you have 1.2 then fix it it becomes 1.2.1 | 07:25 |
| stekern | _franck_: isn't that exactly what I proposed? | 07:26 |
| stekern | or close to at least | 07:27 |
| stekern | we only have two numbers to play with | 07:27 |
| _franck_ | let me read your file more carefuly ;) | 07:30 |
| stekern | juliusb: so basically what I want to do right now at this moment is to push a commit that changes MOR1KX_VERSION_MAJOR to 2 | 07:31 |
| stekern | but that does *not* mean we have released mor1kx v2.0 | 07:32 |
| stekern | only when we tag a commit as v2.0 and create a mor1kx_v2 branch | 07:32 |
| stekern | is that clearer? | 07:32 |
| _franck_ | it is for me | 07:33 |
| stekern | it seems like the most logical thing to do, since we are working on mor1kx v2.0 in master | 07:35 |
| juliusb | sorry, not ignoring you, be back a bit later! | 07:52 |
| juliusb | a bit hectic packing and have to power off this machine... | 07:52 |
| juliusb | (well, not this one, but the one im using to attach to this!) | 07:52 |
| juliusb | bbl | 07:52 |
| knz | hallo | 08:00 |
| knz | got my de0 nano today \o/ | 08:00 |
| stekern | another way of looking at it is of course "we are working ontop of v1.0 towards v2.0 in master" and just let the MOR1KX_VERSION_MAJOR stay 1 until we release v2.0 | 08:01 |
| stekern | I'm not sure which is best, that's why I'm asking | 08:01 |
| mor1kx | [mor1kx] skristiansson pushed 1 new commit to master: https://github.com/openrisc/mor1kx/commit/44940128a59f0b006ccbc4b2cfa25190318df3dc | 08:04 |
| mor1kx | mor1kx/master 4494012 Stefan Kristiansson: cappuccino/ctrl: fix stepping through mispredicted branches... | 08:04 |
| stekern | knz: yay! | 08:05 |
| knz | stekern: now I get to try your stuff :) | 08:05 |
| knz | but at the moment I have a bit of ire against all these fpga vendors that only release binaries for their tools | 08:05 |
| knz | it seems I can't work unless I ahve some intel processors around | 08:05 |
| knz | *and* forces me to a specific combination of cpu/OS | 08:08 |
| knz | the only intel box I ahve around runs freebsd | 08:08 |
| knz | what an annoyance | 08:08 |
| stekern | yes, that *is* an annoyance | 08:34 |
| stekern | olofk: orpsocv3 handholding bitte, I'm not sure I understand the orpsoc.conf thing | 09:16 |
| stekern | if I'm in orpsoc-cores, where should that be located? | 09:17 |
| stekern | do I even need it? | 09:18 |
| stekern | I've installed orpsoc to my system | 09:19 |
| stekern | because, as I understand how orpsocv3 works: orpsoc is the powerful build system, it should "just-work", users should not need to be concerned about how it works | 09:21 |
| stekern | orpsoc-cores is more of the old traditional orpsoc, where the actual cores and SoC live, right | 09:22 |
| stekern | ? | 09:22 |
| stekern | so, all my concentration after I've installed orpsoc should be in orpsoc-cores | 09:24 |
| stekern | right? | 09:24 |
| _franck_ | orpsoc.conf tells orpsoc where orpsoc-cores is located | 09:29 |
| _franck_ | stekern: right | 09:30 |
| stekern | ok, but by default it looks in ./, right? | 09:30 |
| _franck_ | the default we have in openrisc github looks in ../ I think | 09:32 |
| stekern | olofk: not so nice feature - a typo in one of the systems in orpsoc-cores will break all other systems as well | 09:32 |
| stekern | or one of the systems .core file | 09:33 |
| stekern | you might notice bugs quicker that way though ;) | 09:34 |
| knz | hum | 09:37 |
| olofk | stekern: Really? | 09:37 |
| olofk | Where? | 09:37 |
| knz | the de0 comes with this use 3.9GB software package from altera -- is this all necessary? | 09:37 |
| knz | s/use/huge | 09:37 |
| olofk | knz: Unfortunately, yes | 09:38 |
| olofk | I think the Xilinx stuff is close to 12GB nowadays | 09:38 |
| stekern | olofk: not an existing typo, a typo in my sockit.core file | 09:38 |
| knz | what part of this do you need to simply compile some verilog/vhdl modules an upload them to the board? | 09:38 |
| olofk | stekern: Oh, really? I thought it should just silently ignores that file then (which is also quite bad) | 09:38 |
| olofk | knz: All of it basically | 09:39 |
| knz | don't tell me I have to use a GUI in my workflow | 09:39 |
| stekern | knz: fortunately no | 09:39 |
| knz | ok | 09:39 |
| olofk | No, you don't. I guess you can remove the GUI parts if you like, but it's probably quite integrated with the CLI tools | 09:39 |
| olofk | You don't get any options to disable the GUI at install time | 09:40 |
| knz | what's the typical place-and-route compute time for a typical orpsoc? | 09:40 |
| stekern | altera tools are actually pleasently workable from the command line (in comparison to xilinx tools) | 09:40 |
| olofk | Less than an hour I would say | 09:40 |
| knz | ok | 09:40 |
| stekern | knz: on my machine, to build the de0_nano takes ~20 min | 09:40 |
| stekern | tops | 09:41 |
| knz | is there a tutorial somewhere already? | 09:41 |
| olofk | stekern: Amen that. I'm especially pissed of by their forceful line breaking in the log files | 09:41 |
| stekern | how to build it? | 09:41 |
| knz | yeah | 09:41 |
| stekern | knz: cd orpsocv2/boards/altera/de0_nano/syn/quartus/run/ && make all | 09:42 |
| stekern | and to program the board: make pgm | 09:42 |
| olofk | _franck_: Have you tried to build your de1 with orpsocv3 yet? I've been worried that it has bitrotted since I last tried it | 09:42 |
| knz | I'll check it in a moment | 09:42 |
| knz | many thanks | 09:43 |
| olofk | stekern: What happens when you have a typo? I'm a bit curious about that | 09:46 |
| stekern | olofk: http://pastie.org/8299942 | 09:49 |
| _franck_ | olofk: yes I built it and it runs | 09:55 |
| _franck_ | only problem I had was the wb_data_resize | 09:55 |
| olofk | _franck_: That's great to hear | 10:00 |
| olofk | stekern: Weird. Have to check it out. I guess that the sockit isn't in the dependency path of or1200-generic either | 10:01 |
| _franck_ | https://github.com/fjullien/orpsocv3/tree/master/systems/de1 | 10:01 |
| olofk | _franck_: I got your pull request. It scared the hell out of me, because I don't know what to do with it :) | 10:02 |
| _franck_ | :) I just canceled it since I have access to the orpsoccores repo | 10:03 |
| olofk | You do? | 10:04 |
| olofk | Can everyone in the openrisc group write to all repos? | 10:04 |
| stekern | hmm, yeah, but I'd say it's not advisibl to just push into others repos like that | 10:04 |
| olofk | hmm.. one annoying thing about the python docs is that I can't find info on which exceptions that a function can raise | 10:05 |
| stekern | olofk: no, but when you create a new repo, the default is that everybody can | 10:05 |
| _franck_ | stekern: that's what I thought when I did it | 10:05 |
| _franck_ | should have wait | 10:05 |
| stekern | so you have to manually give finer grained access rights | 10:07 |
| olofk | _franck_: Ahh.. I see that I messed up on the ordering in wb_data_resize. | 10:09 |
| olofk | That thing should have a test bench too :( | 10:09 |
| stekern | olofk: I'll create an 'orpsoc' group for you and add orpsoc and orpsoc-cores into that | 10:10 |
| olofk | _franck_: All patches for orpsoc-cores look good btw | 10:11 |
| olofk | stekern: Is that a completely new group, or like an ACL in the openrisc group? | 10:11 |
| stekern | ACL? | 10:13 |
| stekern | actually, the correct term is 'team' | 10:13 |
| stekern | http://en.wikipedia.org/wiki/Anterior_cruciate_ligament | 10:14 |
| stekern | that's not it... right? | 10:14 |
| olofk | Access Control List | 10:15 |
| stekern | yeah, I got it from the second entry google gave me | 10:15 |
| olofk | I told a guy at work that I had written a BFM. He came back after a google search and asked if I had written a Big Furry Monster | 10:15 |
| hansfbaier | knz: A great tutorial is here: http://kevinmehall.net/openrisc/guide/ | 10:16 |
| stekern | TLAs are a bitch | 10:16 |
| olofk | No, I don't like the Theatre of Living Arts either | 10:16 |
| hansfbaier | knz: building orpsoc for my ep4ce10 takes about 7 minutes on my machine | 10:17 |
| stekern | olofk: no, I meant Textile Labour Association | 10:17 |
| olofk | :) | 10:17 |
| stekern | olofk: "exact same results" as in 'bufhit & (&bufclean_wb)' or '(.. & (bufhit | ....)) & !refill_busy'? | 10:24 |
| knz | hansfbaier: many thanks again | 10:24 |
| hansfbaier | knz: y/w | 10:25 |
| olofk | stekern: Test 1: Added &!refill_busy to the original if. Test 2: Added (bufhit & !refill_busy) instead. Same results | 10:27 |
| stekern | i.e. 'bufhit & !refill_busy' works? | 10:28 |
| olofk | Yep | 10:28 |
| stekern | but 'bufhit & (&buf_clean_wb)' doesn't... | 10:29 |
| stekern | wtf | 10:29 |
| olofk | But bufhit is already included elswhere in that if statement, so it might actually be the same logic | 10:29 |
| olofk | Yes, that's also correct | 10:29 |
| stekern | umm, where in the if statement is bufhit already included? | 10:30 |
| olofk | Should the buffers be dirty on startup? Because with your wtf solution, it only runs a few cycles | 10:30 |
| stekern | buffers are dirty on startup | 10:30 |
| olofk | end else if (wb_cyc_i & wb_stb_i & wb_we_i & | 10:31 |
| olofk | (bufhit | first_req | | 10:31 |
| olofk | buf_adr != wb_adr_i[31:BUF_WIDTH+2]) & !refill_busy) begin | 10:31 |
| stekern | but first_req is asserted until first read | 10:31 |
| olofk | I could analyze it a bit more, but I need to clean some things up first | 10:32 |
| stekern | olofk: to be clear, this is the change I was proposing: http://pastie.org/8300030 | 10:32 |
| olofk | That I haven't tried :) | 10:33 |
| stekern | haha | 10:33 |
| olofk | 6000 transactions so far | 10:36 |
| olofk | 12000 | 10:36 |
| olofk | 16000 | 10:36 |
| olofk | ...and 20000! Seems to work | 10:37 |
| stekern | \o/ | 10:37 |
| olofk | Test case runs for 67 simulated seconds with ps accuracy. Thank god we have fast servers here :) | 10:40 |
| stekern | heh | 10:41 |
| olofk | My plan now is to clean up the test bench so you can put it in wb_sdram_ctrl | 10:43 |
| stekern | sounds like the best plan I've heard today | 10:43 |
| olofk | One problem however is that I don't want to duplicate my wishbone transactor (as I'm using the one in orpsoc's wb_bfm core). Is it ok to just put a note in wb_sdram_ctrl that extra files are needed to simulate it, and give pointers to the orpsoc-cores repo | 10:44 |
| olofk | For those who want to use it stand-alone | 10:44 |
| olofk | I'll probably split out wb_bfm from orpsoc-cores at some point too. | 10:45 |
| olofk | Another solution is to add a copy of the needed files from wb_bfm to wb_sdram_ctrl, but not list the in the .core file | 10:46 |
| olofk | But I'm so fucking tired of all this duplication | 10:46 |
| olofk | bbl | 10:47 |
| stekern | I'm ok with both solutions | 10:47 |
| _franck_ | I used the last solution for jtag_vpi | 10:54 |
| olofk | I prefer the first solution, but I'm totally ok with the second too. It's important not to make orpsoc try to dictate what people's cores should look like | 10:57 |
| knz | 15GB filesystem required for the tutorial... gee | 11:02 |
| olofk | knz: Welcome to the world of EDA | 11:04 |
| knz | tbh I find this a bit ridiculous | 11:09 |
| knz | apparently all the possible features are installed, but only get activated with license files | 11:09 |
| knz | this is 1980's computing | 11:09 |
| knz | unrelated: is there an i2c controller in the orpsoc? | 11:14 |
| olofk | Yes. The EDA world is terrible in this regard. We have to liberate the EDA world one tool at a time. For now, most focus is on the simulators | 11:14 |
| olofk | Yes, therei is | 11:14 |
| knz | thx | 11:15 |
| knz | next step, I need to acquire an usb to rs232 adapter | 11:15 |
| knz | this feels like a dungeon quest really | 11:15 |
| stekern | olofk: fwiw, I don't know what BFM stands for neither... | 11:16 |
| olofk | usb to rs232 convereters you can get from anywhere for less than $10 | 11:16 |
| olofk | stekern: Bus Functional Model | 11:16 |
| olofk | knz: But most of the are of questionable quality | 11:16 |
| knz | "now, roll a d2^34 for the quartus install. If you roll <10GB, go back to step 2, 'configure your VB instance'" | 11:17 |
| stekern | last I asked you, you accused me for not reading your blog post, but you just use the acronym there too | 11:17 |
| knz | he | 11:17 |
| olofk | stekern: I actuallly got a post in the works where I explain more about BFMs | 11:17 |
| knz | olofk: so what converter would you advise if most are crap? | 11:18 |
| stekern | olofk: looking forward to it ;) | 11:18 |
| olofk | knz: I haven't got one myself actually, so I would say buy two of different brands. Just in case | 11:18 |
| knz | ok | 11:19 |
| olofk | At least if you are planning on ordering them. | 11:19 |
| knz | k | 11:23 |
| knz | does the orpsoc uart controller drive the pins using ttl? | 11:24 |
| knz | or can I use a regular adapter? | 11:24 |
| stekern | it's ttl | 11:24 |
| olofk | knz: That depends on the FPGA. You set the I/O standard for each pin | 11:24 |
| knz | de0 nano | 11:25 |
| knz | default conf | 11:25 |
| olofk | Then stekern's answer is most certainly correct :) | 11:25 |
| knz | great | 11:26 |
| stekern | olofk: I think he means ttl vs rs232 | 11:26 |
| knz | http://dx.com/p/jy-r2t-v1-2-rs232-serial-port-converter-104325 <- I'm considering getting this one | 11:27 |
| olofk | Ah yes. rs232 that has the fun concept where -3 to -15 is a logical one, and +3 to +15 is a logical zero | 11:28 |
| _franck_ | with this you have TTL to USB: http://www.ftdichip.com/Products/Cables/USBTTLSerial.htm | 11:31 |
| stekern | or this: http://dx.com/p/ft232-usb-to-serial-breakout-board-black-silver-161170 | 11:34 |
| knz | looks good | 11:36 |
| _franck_ | olofk: how does modelsim simulation works in orpsocv3 regarding altera libraries ? | 11:37 |
| olofk | _franck_: I haven't started working on that yet | 11:40 |
| olofk | I think that some extra tags in the .core file is needed, becuase you need to specify the location with vlib, and then load the libs on the command line with -L | 11:41 |
| _franck_ | ok | 11:42 |
| stekern | still confused about where to be when building a system... | 11:43 |
| stekern | and the orpsoc.conf | 11:44 |
| stekern | I have a skeleton sockit system now | 11:44 |
| stekern | if I'm in orpsoc-cores/ and run 'orpsoc build sockit' | 11:45 |
| stekern | it populates a build/ directory | 11:45 |
| stekern | and then quartus start to scream: Can't analyze file -- file build/sockit/src/wb_intercon/wb_arbiter.v is missing | 11:46 |
| olofk | That's odd | 11:47 |
| olofk | What's in your orpsoc.conf? | 11:48 |
| stekern | I don't have one | 11:48 |
| stekern | that's the question, where should it be? | 11:48 |
| stekern | and why do I need one? | 11:48 |
| olofk | Ok... do like this: | 11:49 |
| olofk | Create an empty sibling directory to orpsoc-cores | 11:50 |
| olofk | Create an orpsoc.conf there and put in: | 11:50 |
| olofk | [main] | 11:50 |
| olofk | cores_root =../orpsoc-cores/cores | 11:50 |
| olofk | systems_root =../orpsoc-cores/systems | 11:50 |
| _franck_ | stekern: can't help you, I didn't installed orpsoc. Just run it in my orpsoc install directory | 11:50 |
| olofk | You shouldn't normally run stuff from your orpsoc-cores dir | 11:51 |
| olofk | The whole working directory can be considered a throw-away directory | 11:51 |
| stekern | why not? | 11:51 |
| olofk | You are of course free to run it from there if you want | 11:51 |
| olofk | Hmm.. come to think of it. If orpsoc finds the directories systems and cores in your pwd, it uses them | 11:52 |
| olofk | Because that was how it used to be before I split up the repos | 11:52 |
| stekern | yes it does, and that's what seems like the most inituitive | 11:52 |
| olofk | You should view orpsoc-cores as your portage tree basically | 11:53 |
| stekern | also, running in orpsoc-cores also seems the most intuitive | 11:53 |
| olofk | And nothing of this explain the missing wb_arbiter file | 11:54 |
| stekern | it complains about all the files | 11:54 |
| olofk | aha | 11:54 |
| stekern | because it's not a absolute path | 11:54 |
| stekern | (I guess) | 11:54 |
| olofk | Weird. It should be absolute paths everywhere | 11:55 |
| stekern | I'll try building outside orpsoc-cores | 11:55 |
| knz | ok I will get first an extra hard drive before anything else | 11:56 |
| olofk | hmm.. it looks like it's actually not using absolute files | 11:56 |
| olofk | s/files/paths | 11:56 |
| stekern | it founds them in the out of orpsoc-cores tree build | 11:58 |
| olofk | ok. You just found a bug. Thanks | 11:58 |
| olofk | A patch will be available in the paid support for ORPSoCv3 Enterprise | 11:58 |
| _franck_ | :) | 12:00 |
| stekern | I assume to get that I have to install the ORPSoCv3 update suite? | 12:00 |
| olofk | ORPSoCv3 Enterprise is basically a Java version of ORPSoCv3 where you write all your configurations in Excel files that gets passed to the backend with CORBA | 12:00 |
| olofk | And yes. The update suite also adds the ORPSoCv3 browser toolbar | 12:01 |
| stekern | that sounds excellent, and how big is the update suite? | 12:01 |
| stekern | if it's less than 10GB, we can't consider it being serious | 12:01 |
| mschulze | Hi! | 12:01 |
| olofk | We usually send it on 373 Zip Discs | 12:02 |
| stekern | but we only have ez drive readers... | 12:02 |
| stekern | mschulze: hi! | 12:03 |
| mschulze | tuesday I had my final exam and now I'm adding copyright and licensing to all my files :-( I seriously want to automate that. | 12:03 |
| olofk | That's a problem. ORPSoCv3 EE is only supported on Windows CE, and they don't have any ez drive drivers | 12:03 |
| mschulze | are you playing an RPG here? | 12:04 |
| stekern | haha, kind of ;) | 12:04 |
| knz | seriously, I'm considering proposing a meta-package with the board + development system already installed and ready to use | 12:05 |
| knz | on usb stick or something | 12:06 |
| mschulze | it's funny to read it' but 10 GB sounds small for me... | 12:06 |
| knz | mschulze: I generate terabytes of data in my architecture research with a toolchain that weighs less than 100MB | 12:06 |
| knz | the biggest two programs I ever used were civ4 and diablo 3, and they are full of graphics / cinematics | 12:07 |
| knz | I can't fathom what all these altera gigabytes are for | 12:07 |
| mschulze | knz: that's what toolchains are meant to be used for... producing data :-) | 12:07 |
| mschulze | you are complaining about the altera "thing" that ships on a neverending download disk? | 12:08 |
| knz | I got a CD with the de0 | 12:08 |
| knz | then I was greeted with a windows-y installer that asked me for 9GB of disk space right away | 12:09 |
| mschulze | but your CD is out of date!!! | 12:09 |
| knz | how should I know? | 12:09 |
| mschulze | yeah, how should you... | 12:10 |
| knz | hence my idea to propose a dev package with batteries included | 12:10 |
| knz | I might actually volunteer to do just that | 12:10 |
| olofk | knz: Problem is that you probably can't redistribute the Altera files | 12:10 |
| knz | doesn't matter, if I can automate the install | 12:11 |
| olofk | At ORSoC we provided a VM image (it's on opencores) that contained all the open source tools with instructions on how to get the altera stuff | 12:11 |
| mschulze | well I've been asked to provision a VM with qartus and the full nios 2 toolchain :-( | 12:11 |
| knz | anyway, fun has stopped for today | 12:11 |
| mschulze | and found this http://www.altera.com/support/kdb/solutions/rd03042004_8455.html earlier... | 12:12 |
| knz | I have to wait for my order to arrive | 12:12 |
| knz | rs232 adapter + extra hard drive | 12:12 |
| knz | so you say I should just throw my install away already, and just re-download everything from the altera web site? | 12:14 |
| mschulze | I got a short question: how does this github "organization" thing work? I created an account but cannot find a button to join. | 12:14 |
| olofk | mschulze: Now that's a hacky solution for automation :) | 12:14 |
| olofk | And windows only I guess | 12:14 |
| olofk | mschulze: First rule of the OpenRISC organization: You don't talk about the OpenRISC organization ;) | 12:14 |
| mschulze | knz: no! you can allways use the old versions that shipped with the boards. | 12:15 |
| mschulze | I was just kidding, sorry. | 12:15 |
| knz | ok | 12:15 |
| knz | oh well | 12:16 |
| mschulze | olofk: as the linux version actually is a windows version that comes with wine... and the installer is an windows installer running under the same wine... | 12:16 |
| knz | I knew from my colleagues that the situation with xilinx was not all rosy, but it seems it's barely better with altera | 12:16 |
| olofk | mschulze: Haha. I hadn't noticed that | 12:16 |
| stekern | knz: sadly, it's a magnitude better with altera | 12:16 |
| knz | I think this is appalling | 12:16 |
| olofk | Thought it was native QT | 12:16 |
| mschulze | olofk: okay, I learn the first rule and will come back later :-D | 12:17 |
| knz | so all the synthesis tools are windows binaries wrapped around by wine? | 12:17 |
| mschulze | knz: yes. | 12:17 |
| * knz goes to throw up | 12:17 | |
| knz | my. my | 12:18 |
| * olofk goes through knz's vomits to see if there is anything worth stealing there | 12:18 | |
| knz | :) | 12:18 |
| mschulze | don't give up, you will adopt to quartus soon... | 12:18 |
| knz | over my dead body | 12:18 |
| knz | nb: the entire process is motivated by a longer term project to design a configurable chip which does not require EDA tools | 12:19 |
| stekern | mschulze: I assume this is you: https://github.com/mschulze | 12:19 |
| knz | my work will succeed when I can throw all this away | 12:19 |
| mschulze | stekern: i have to admit that I would have registered this account if another mschulze wouldn't have been faster... https://github.com/martinschulze that is me | 12:21 |
| mschulze | knz: you can omit the installation of the Cyclone V family if you don't use it, that will save you ~500 MB of download and even a bit more disk sp | 12:24 |
| mschulze | ace | 12:24 |
| knz | I already did that | 12:25 |
| knz | hmm | 12:25 |
| mschulze | okay. | 12:25 |
| knz | I will soon get a parallela board | 12:25 |
| knz | with a Zinq 7020 on | 12:25 |
| knz | that's also from altera, right? | 12:25 |
| stekern | mschulze: congratulations, you are now a member of the sacred "organisation" | 12:26 |
| stekern | knz: no, that's xilinx | 12:26 |
| knz | argh | 12:26 |
| mschulze | one more toolchain to install... | 12:27 |
| olofk | I'm grateful that I don't have to deal with the Actel toolchain any more at least. _That_ was a serious pain | 12:27 |
| stekern | haha, no, the quartus binaries are not windows binaries wrapped around by wine | 12:27 |
| stekern | olofk: the kidder has been kidded ;) | 12:28 |
| olofk | ahh... :) | 12:28 |
| mschulze | quartus is native linux? | 12:29 |
| stekern | yes | 12:29 |
| mschulze | :-O it looks so unfamilliar... | 12:29 |
| olofk | Oh yes. Come to think of it. I talked to some of the Altera guys when Quartus 10 was in prerelease, and they showed me their fancy new QT gui | 12:30 |
| olofk | The old one might have been a windows one running through Wind/U | 12:31 |
| mschulze | wait, I confused it with modelsim :-D | 12:34 |
| olofk | But modelsim is written in TK, isn't it? | 12:34 |
| mschulze | okay, it might be an error in my brain, I will run an integrity check | 12:39 |
| stekern | olofk: are wildcards supported in the 'src_files =' directive? | 12:46 |
| olofk | stekern: It's on my TODO list | 12:47 |
| olofk | https://github.com/openrisc/orpsoc/blob/master/TODO#L15 | 12:49 |
| stekern | ok, that wasn't obvious to me that that implies "wildcards in src_files" | 12:50 |
| olofk | Oh.. you're so stupid! | 12:51 |
| olofk | That is totally obvious | 12:51 |
| stekern | that's my biggest shortcoming, being stupid ;) | 12:52 |
| olofk | sorry. That was mostly a note-to-self file that happened to get added to the git repo at one point. Should probably clean that up ab it | 12:52 |
| stekern | on the upside, I've heard that depression and insanity is far more common for intelligent people | 12:53 |
| stekern | so I just take that as you calling me a "happy and sane guy" | 12:53 |
| olofk | :) | 12:55 |
| mschulze | I have to bother one of my Professors, brb | 13:01 |
| stekern | but, great it's on the todo list | 13:04 |
| stekern | the mor1kx core description is broken right now and I bet there will be a bit of a pita to keep those src_files lists up-to-date | 13:05 |
| olofk | Yes, it's especially handy to have wildcard support when you're developing a core | 13:26 |
| olofk | And just to make the core files quick to write | 13:26 |
| olofk | But first on the list now is to finish the support for getting a certain tag/commit from github | 13:27 |
| olofk | Then I'll probably add library support | 13:27 |
| olofk | for modelsim | 13:27 |
| olofk | Did anyone see the forum post on opencores for the typo in or1200_alu? | 14:07 |
| stekern | yes, and the patch looked correct | 15:22 |
| hno | olofk, why spending time on makeing something that gets a specific commit? git has it built-in as submodules | 18:31 |
| stekern | _franck_: why do you have the intgen in your de1 port? | 19:50 |
| _franck_ | nothing I need to remove it | 19:54 |
| _franck_ | about my dpram problem, I did a patch and a pull request for you. Tell me what do you think of this | 20:10 |
| --- Log closed Fri Sep 06 00:00:43 2013 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!