| - Ar|stote|is (QUIT: Read error: Connection reset by peer) (~linx@149.210.32.218) | 01:26 | |
| + Ar|stote|is (~linx@149.210.32.75) | 01:34 | |
| + jacobk (~quassel@47-186-65-73.dlls.tx.frontiernet.net) | 02:29 | |
| - paperManu (QUIT: Ping timeout: 265 seconds) (~paperManu@107.159.71.33) | 02:38 | |
| - chrcav (QUIT: Quit: leaving) (~chrcav@user/chrcav) | 02:45 | |
| + chrcav (~chrcav@user/chrcav) | 02:48 | |
| - nsc (QUIT: Ping timeout: 252 seconds) (~nicolas@212-96-142-46.pool.kielnet.net) | 03:14 | |
| + nsc (~nicolas@152-99-142-46.pool.kielnet.net) | 03:16 | |
| - mjw (QUIT: Ping timeout: 252 seconds) (~mjw@gnu.wildebeest.org) | 03:44 | |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 04:42 | |
| - Gooberpatrol66 (QUIT: Quit: Konversation terminated!) (~Gooberpat@user/gooberpatrol66) | 04:43 | |
| + Gooberpatrol66 (~Gooberpat@user/gooberpatrol66) | 04:44 | |
| - mtm (QUIT: Ping timeout: 265 seconds) (~textual@47.202.75.129) | 06:04 | |
| + mtm (~textual@47.202.75.129) | 06:07 | |
| + mlarkin (~mlarkin@12.222.71.242) | 06:20 | |
| + chomwitt (~chomwitt@2a02:587:7a13:5400:1ac0:4dff:fedb:a3f1) | 06:37 | |
| - chomwitt (QUIT: Ping timeout: 265 seconds) (~chomwitt@2a02:587:7a13:5400:1ac0:4dff:fedb:a3f1) | 07:11 | |
| - amk (QUIT: Ping timeout: 252 seconds) (~amk@user/amk) | 08:27 | |
| + ohwow (~ohwow@82-132-218-71.dab.02.net) | 09:04 | |
| - ohwow (QUIT: Quit: Client closed) (~ohwow@82-132-218-71.dab.02.net) | 09:14 | |
| + ohwow (~ohwow@82-132-218-71.dab.02.net) | 09:14 | |
| - ohwow (QUIT: Quit: Client closed) (~ohwow@82-132-218-71.dab.02.net) | 09:20 | |
| + ohwow (~ohwow@82-132-218-71.dab.02.net) | 09:33 | |
| - ohwow (QUIT: Quit: Client closed) (~ohwow@82-132-218-71.dab.02.net) | 09:38 | |
| + ohwow (~ohwow@82-132-218-71.dab.02.net) | 09:39 | |
| - ohwow (QUIT: Quit: Client closed) (~ohwow@82-132-218-71.dab.02.net) | 09:50 | |
| + ohwow (~ohwow@82-132-218-71.dab.02.net) | 09:52 | |
| - jacobk (QUIT: Ping timeout: 260 seconds) (~quassel@47-186-65-73.dlls.tx.frontiernet.net) | 10:04 | |
| - ohwow (QUIT: Quit: Client closed) (~ohwow@82-132-218-71.dab.02.net) | 10:04 | |
| + jacobk (~quassel@47-186-65-73.dlls.tx.frontiernet.net) | 10:05 | |
| + ohwow (~ohwow@82-132-218-71.dab.02.net) | 10:05 | |
| - wickedshell (QUIT: Ping timeout: 248 seconds) (~wickedshe@2601:8c0:800:4baa:d69e:c297:3868:cba3) | 10:10 | |
| - ohwow (QUIT: Ping timeout: 240 seconds) (~ohwow@82-132-218-71.dab.02.net) | 10:15 | |
| + amk (~amk@user/amk) | 10:53 | |
| - jacqueline (QUIT: Ping timeout: 252 seconds) (~jacquelin@user/jacqueline) | 10:54 | |
| - MyNetAz (QUIT: Remote host closed the connection) (~MyNetAz@user/MyNetAz) | 11:09 | |
| + MyNetAz (~MyNetAz@user/MyNetAz) | 11:20 | |
| + jacqueline (~jacquelin@user/jacqueline) | 11:36 | |
| + paperManu (~paperManu@107.159.71.33) | 11:52 | |
| + wickedshell (~wickedshe@2601:8c0:800:4baa:7ad7:cac4:dfcc:fac3) | 12:18 | |
| + mjw (~mjw@gnu.wildebeest.org) | 12:43 | |
| grimmware | https://chimera-linux.org/news/2025/03/dropping-riscv.html interesting, chimera are dropping riscv because they can't get their hands on powerful enough build hardware | 12:50 |
|---|---|---|
| L29Ah | https://get.gentoo.org | 12:53 |
| hramrach | Not surprising, the situation with riscv hardware is pretty bad. You cannot buy the equivalent of Amprere One or even Ampere Altra even if you wanted to spend the money, powerful hardware simply does not exist. | 12:54 |
| josch | grimmware: riscv slowness is also a constant problem with debian | 12:54 |
| josch | the buildds we have are a continous source of annoyance for people who get blocked waiting for their stuff to build for days as there is another libreoffice, gcc or chromium build going on | 12:54 |
| josch | but it's not like mipsel for example is in a much better spot | 12:56 |
| hramrach | I think some distrbutions use cross-compilation as workaround but that has its own problems | 12:56 |
| + gustav28 (~gustav@c-78-82-53-248.bbcust.telenor.se) | 13:02 | |
| * mjw -> Guest3039 | 13:06 | |
| * Guest8856 -> mjw | 13:06 | |
| Manis | Does RISC-V have actual users? | 13:13 |
| - polymetis (QUIT: Read error: Connection reset by peer) (sid320610@id-320610.hampstead.irccloud.com) | 13:14 | |
| Manis | With hardware being so slow and buggy I could imagine only very few use cases for it. | 13:14 |
| + polymetis (sid320610@id-320610.hampstead.irccloud.com) | 13:14 | |
| - svp (QUIT: Read error: Connection reset by peer) (sid537750@id-537750.uxbridge.irccloud.com) | 13:14 | |
| - jbmorley_ (QUIT: Read error: Connection reset by peer) (sid21953@id-21953.tinside.irccloud.com) | 13:14 | |
| - [tj] (QUIT: Read error: Connection reset by peer) (sid609767@id-609767.lymington.irccloud.com) | 13:15 | |
| + svp (sid537750@id-537750.uxbridge.irccloud.com) | 13:15 | |
| - yuu (QUIT: Read error: Connection reset by peer) (sid267332@id-267332.ilkley.irccloud.com) | 13:15 | |
| + jbmorley_ (sid21953@id-21953.tinside.irccloud.com) | 13:15 | |
| + yuu (sid267332@id-267332.ilkley.irccloud.com) | 13:15 | |
| + [tj] (sid609767@id-609767.lymington.irccloud.com) | 13:15 | |
| + ohwow (~ohwow@82.132.218.71) | 13:22 | |
| - ohwow (QUIT: Client Quit) (~ohwow@82.132.218.71) | 13:23 | |
| - Ar|stote|is (QUIT: Ping timeout: 244 seconds) (~linx@149.210.32.75) | 13:38 | |
| + Ar|stote|is (~linx@149.210.32.40) | 13:42 | |
| + bkeys (~Thunderbi@38-146-94-247.echocast.zone) | 13:50 | |
| - bkeys (QUIT: Ping timeout: 260 seconds) (~Thunderbi@38-146-94-247.echocast.zone) | 13:58 | |
| [tj] | grimmware: I've never seen this complaint in freebsd, I'm not sure if its just a lack of package building or the quality of bsd user | 14:06 |
| [tj] | and a lot of the riscv work is happening through the cheri folks | 14:07 |
| [tj] | and they were running mips on an fpga before so they have a different persepctive on slow and are well capitalised | 14:07 |
| josch | just compiling gcc-14 on Debian riscv buildds takes 2 days *at least* | 14:09 |
| + bkeys (~Thunderbi@38-146-94-247.echocast.zone) | 14:12 | |
| [tj] | ah we aren't building packages so no one complains | 14:13 |
| [tj] | we can trivially cross build release | 14:13 |
| - Ar|stote|is (QUIT: Ping timeout: 268 seconds) (~linx@149.210.32.40) | 14:27 | |
| + Ar|stote|is (~linx@149.210.9.238) | 14:31 | |
| - bkeys (QUIT: Ping timeout: 248 seconds) (~Thunderbi@38-146-94-247.echocast.zone) | 14:36 | |
| + bkeys (~Thunderbi@38-146-94-247.echocast.zone) | 14:40 | |
| - bkeys (QUIT: Ping timeout: 260 seconds) (~Thunderbi@38-146-94-247.echocast.zone) | 15:10 | |
| + bkeys (~Thunderbi@38-146-94-247.echocast.zone) | 15:25 | |
| - bkeys (QUIT: Ping timeout: 276 seconds) (~Thunderbi@38-146-94-247.echocast.zone) | 15:34 | |
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50) | 16:57 | |
| + chomwitt (~chomwitt@2a02:587:7a13:5400:1ac0:4dff:fedb:a3f1) | 16:58 | |
| - chomwitt (QUIT: Ping timeout: 248 seconds) (~chomwitt@2a02:587:7a13:5400:1ac0:4dff:fedb:a3f1) | 17:02 | |
| - mlarkin (QUIT: Ping timeout: 260 seconds) (~mlarkin@12.222.71.242) | 17:03 | |
| hramrach | Manis: basically the use case for RISC-V besides enthusiasts that want as many platforms as possible (the more obscure the better) is digital paranoia. With the core IP being free you can theoretically throw money amd/or people at auditing it does what it says (unrelistic, all real chips are fabbed, tested, and refabbed until tests pass) (also you need peripherals, and those are not included) | 17:53 |
| hramrach | (also see xz fiasco), you can change the ROM to do what you want (that's somewhat realistic), there is no chip IP licensor to rugpull on you once you make substantial use of the chips (not for the cores, but non-free peripherals are still needed for most applications). Generally some step in the direction of opensource and open standards with little practical benefits so far. | 17:53 |
| josch | a step in that direction though nonetheless. With the core IP being free, a future becomes possible where different manufactures create riscv processors in their fabs and then the consumer has the choice which one you trust and which one you do not. Without riscv, projects like the european DARE also would not be possible or very practical. I agree that so far, the utility and gains in trust are small | 17:58 |
| josch | but progress in a nice direction it is nonetheless in my opinion. | 17:58 |
| [tj] | minute: how long do you think I will be waiting for a replacement charger board for my pocket? | 18:00 |
| Zaba | the risc-v instruction set is royalty free, this is not the same thing as silicon design being open source | 18:00 |
| Zaba | there happen to be some open source risc-v cores too, of course, but you're unlikely to find those particular ones in SoCs that have even remotely competitive performance | 18:01 |
| josch | yes, that's why i do not disagree that the step that was made by having the instruction set open is only a small step | 18:03 |
| - mtm (QUIT: Ping timeout: 260 seconds) (~textual@47.202.75.129) | 18:04 | |
| Zaba | realistically there is nothing free or open about most risc-v SoCs other than the fact the *vendor* did not have to pay royalties to use the instruction set | 18:06 |
| + mtm (~textual@47.202.75.129) | 18:06 | |
| + bkeys (~Thunderbi@38-146-94-247.echocast.zone) | 18:07 | |
| Zaba | and they're not any easier to audit than any other CPUs | 18:07 |
| - bkeys (QUIT: Ping timeout: 248 seconds) (~Thunderbi@38-146-94-247.echocast.zone) | 18:18 | |
| Manis | hramrach: but then just take OpenPOWER. It's even more open from what I heard and has performant offerings. | 19:08 |
| Manis | Also IIRC RISC-V is pretty locked down when it comes to making ISA changes. With OpenPOWER you can send change requests. | 19:10 |
| Manis | Firmware is all open (at least for POWER9) and works well. | 19:10 |
| - xktr (QUIT: Ping timeout: 265 seconds) (~xktr@user/xktr) | 19:20 | |
| + xktr (~xktr@user/xktr) | 19:20 | |
| * mjw -> Guest2904 | 19:52 | |
| * Guest3039 -> mjw | 19:52 | |
| - jacobk (QUIT: Ping timeout: 276 seconds) (~quassel@47-186-65-73.dlls.tx.frontiernet.net) | 19:59 | |
| hramrach | Manis: it uses proprietary IBM CPUs but AFAICT they are not boot CPUs and are brought up by a dedicated service processor which is something you can fully control. You can run your own firmware. The firmware for the machines is opensource, probably in the sense of AOSP that workable smaple codebase exists but needs some customization for specific hardware implementation. | 20:00 |
| Manis | hramrach: no, you can compile firmware and run it without problems and without limitations. | 20:02 |
| Manis | The CPUs are manufactured by IBM, but so ARE ARM, RISC-V. | 20:03 |
| Manis | There are also other implementations like Microwatt on the very low end. | 20:03 |
| Manis | And now there is SolidSilicon which announced a chip. | 20:03 |
| hramrach | you need a license to make your own afaik. Some Chinese company did get teh license to make POWER8 if you look through the CPUID tables in Linux but that's an outlier | 20:03 |
| hramrach | you can compile AOSP and run it without problems and limimtations too. Unless your platform has secure boot, or has custom bits not supported by the opensource upstream implementation | 20:06 |
| hramrach | and I expect it to be similar situation with the POWER firmware. | 20:07 |
| hramrach | or u-boot for that matter. it's completely opensource, and you can compile and run it without problems and limitations :) | 20:09 |
| + aloo_shu_ (~aloo_shu@85.51.18.10) | 20:22 | |
| * aloo_shu -> Guest8044 | 20:22 | |
| - Guest8044 (QUIT: Killed (osmium.libera.chat (Nickname regained by services))) (~aloo_shu@90.166.99.125) | 20:22 | |
| * aloo_shu_ -> aloo_shu | 20:22 | |
| Manis | hramrach: That's not correct, you can program your own secure boot keys into the open source firmware and it will use them. No limitations. | 20:37 |
| Manis | I don't know what you mean by license. If you want to fab IBM's designs, then probably yes. | 20:37 |
| Manis | But for other cores, no. You can clone Microwatt from GitHub and do with it what you want. | 20:38 |
| Manis | There is (at least as of now) also not such a downstream culture like there is with Android, EDKII etc. | 20:39 |
| Manis | Board config is usually in the upstream repos on https://github.com/open-power/ | 20:40 |
| hramrach | Manis: just like with AOSP or u-boot the OpenPower encompasses various commarcial vendors which have different ideas about opoenness of their platform. Some are enthusiast-oriented and very open, some are open because tehy thing it's a good idea. Nonetheless, it's clearly possible to take the opensource firmware, massage it to add some vendor-specific (mis) features, and require any updates to | 20:46 |
| hramrach | be signed with the vendor key. And clearly such firmware builds are out there. Sure, there is not nearly as much variability as with Arm SBCs or Android, manly because the ecosystem is much smaller. | 20:46 |
| Manis | hramrach: of course, but this applies to everything? | 20:48 |
| - midfavila (QUIT: Remote host closed the connection) (midfavila@sdf.org) | 20:48 | |
| Manis | Surely any vendor can do with the code whatever the Apache license allows. | 20:48 |
| hramrach | which means that you can have a device that is proudly part of OpenPower ecosystem which is in fact in no way open, depending on what you get. | 20:49 |
| Manis | I don't really know why you bring up AOSP all the time, because it's the perfect example of an anti-OSS project. | 20:49 |
| hramrach | because that's how OSS turns out when many commercial vendors get on it | 20:50 |
| Manis | hramrach: or you buy an appliance which runs Linux, technically... | 20:51 |
| Manis | I don't know what you're trying to tell me. | 20:51 |
| hramrach | u-boot and downstrema forks is the same story | 20:51 |
| hramrach | that 'you can compile firmware and run it without problems and without limitations.' is an illusion. Or delusion. | 20:52 |
| Manis | Why? | 20:52 |
| Manis | I compiled my firmware and run it. | 20:52 |
| hramrach | I did as well, when I carefully selected a piece of hardware that is not locked down and fully supported by the opensource version of the firmware. | 20:53 |
| hramrach | it comes with many qualifiers | 20:53 |
| Manis | Oookay, I think I still can't follow. How's this related to RISC-V again? | 20:54 |
| Manis | Because I said OPAL works better than the average RISC-V firmware? | 20:55 |
| hramrach | this is not related to riscv, this is related to your claim that OpenPower project has opensource firmware. Sure, it does. Qualifiers apply. | 20:55 |
| Manis | Like for all other architectures. | 20:58 |
| Manis | So I don't see the point of this discussion. | 20:59 |
| hramrach | 'Firmware is all open (at least for POWER9)' is not a correct statement. Sure, you can pick a POWER9 board that provides sources for the firmware that works on that board, and works with upstream to have that support merged. It's not 'all open', though. It's 'select boards open' as with most other platforms. | 21:03 |
| Manis | It's like me saying Reform has no open firmware because there are Android phones which are so locked down I cannot install a custom ROM. | 21:03 |
| Manis | Or Windows on ARM laptops with closed source UEFI. | 21:04 |
| hramrach | your approach to logic ... | 21:04 |
| Manis | It's what you're saying. | 21:04 |
| hramrach | you did not name a specific board, you only named the CPU. | 21:04 |
| Manis | And you felt like nitpicking? | 21:05 |
| hramrach | that's grossly incorrect | 21:06 |
| Manis | I don't think som | 21:06 |
| Manis | *so. | 21:06 |
| Manis | Can you name a POWER9 board which cannot run open firmware? | 21:06 |
| hramrach | it's like saying "All Arm firmware is open" - it just isn't | 21:07 |
| Manis | You're mixing things up. | 21:07 |
| hramrach | no, it's the same statemet except s/POWER9/Arm/ | 21:08 |
| Manis | POWER9 is the specific CPU design by IBM. And IBM released all the firmware for it. | 21:08 |
| Manis | I didn't say all Power ISA 3.0 cores are fully open. | 21:08 |
| Manis | What I said would be the equivalent of firmware for RK3399 is open. | 21:09 |
| Manis | Is this not true anymore because some manufacturers added a Wi-Fi card to their board which requires blobs? | 21:10 |
| hramrach | which it kind of is but depending one the specific board it may or may not be useful | 21:10 |
| hramrach | they may use a DRAM configuration not supported by open firmware, for example | 21:10 |
| hramrach | or may use some weird storage with no open driver | 21:11 |
| hramrach | or use secure boot to prevent loading custom firmware if it's actually working on that platform | 21:11 |
| Manis | I give in. Take from this discussion what you will. I don't understand what you're trying to tell me based on some minute detail I got wrong in my sentence. | 21:12 |
| hramrach | You are annoyed that I bring up 'anti-OSS' AOSP in connection with POWER9 firmware. It's exactly the situation we are in. IBM provided a canonical firmware implementation that they do NOT ship on their hardware, they use some modified derivative, and use secure boot to prevent you from loading your own firmware, just as many phone vendors do. Then you have Raptor computing that work with | 21:24 |
| hramrach | upstream to have their hardware fully supported, just like you have a few enthusiast-oriented phone vendors. And then you have some more vendors in the ecosystem on the spectrum between the two. | 21:24 |
| frickler | hramrach: Manis: maybe you could both take a deep breath? this is getting a bit offtopic IMO | 21:25 |
| minute | yeah. | 21:31 |
| + Guest95 (~Guest47@2600:387:f:7e14::9) | 21:32 | |
| - Guest95 (QUIT: Write error: Broken pipe) (~Guest47@2600:387:f:7e14::9) | 21:33 | |
| minute | btw, a small distraction: the reform next now has rp2350a which can be started in risc-v mode and the gateware source for those cpu cores is apparently on github :3 | 21:33 |
| + Guest47 (~Guest47@2600:387:f:7e14::9) | 21:34 | |
| frickler | minute: any plan yet regarding the emi shielding pocket backplane? ;) | 21:36 |
| - Guest47 (QUIT: Write error: Connection reset by peer) (~Guest47@2600:387:f:7e14::9) | 21:37 | |
| BoostisBetter | minute: great news, although I am not really sure about why that is. | 21:39 |
| BoostisBetter | minute: the next is very impresive. | 21:39 |
| minute | frickler: i ordered 100 fixed backplanes last week and they are currently in german customs | 21:40 |
| minute | [tj]: we'll take care of it tomorrow! | 21:40 |
| frickler | minute: cool. guess I should finally contact support also about my broken nvme module then and some other stuff, so you can get it all bundled up | 21:41 |
| minute | frickler: yeah | 21:43 |
| minute | custom trackpad is alive https://mastodon.social/@mntmn/114157061096017158 | 21:45 |
| + reform14955 (~porteur@2a02-842a-d87d-d101-bb16-1e91-64dc-7cd1.rev.sfr.net) | 21:49 | |
| reform14955 | test | 21:50 |
| reform14955 | exit | 21:50 |
| - reform14955 (QUIT: Client Quit) (~porteur@2a02-842a-d87d-d101-bb16-1e91-64dc-7cd1.rev.sfr.net) | 21:50 | |
| + chomwitt (~chomwitt@2a02:587:7a13:5400:1ac0:4dff:fedb:a3f1) | 22:02 | |
| - chomwitt (QUIT: Ping timeout: 252 seconds) (~chomwitt@2a02:587:7a13:5400:1ac0:4dff:fedb:a3f1) | 22:07 | |
| - gustav28 (QUIT: Quit: Quit) (~gustav@c-78-82-53-248.bbcust.telenor.se) | 22:15 | |
| + bkeys (~Thunderbi@38-146-94-247.echocast.zone) | 22:25 | |
| - bkeys (QUIT: Ping timeout: 248 seconds) (~Thunderbi@38-146-94-247.echocast.zone) | 22:33 | |
| + bkeys (~Thunderbi@173.186.16.211) | 23:17 | |
| - bkeys (QUIT: Ping timeout: 244 seconds) (~Thunderbi@173.186.16.211) | 23:32 | |
| + bkeys (~Thunderbi@h211.16.186.173.dynamic.ip.windstream.net) | 23:40 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!