| - jacobk (QUIT: Ping timeout: 248 seconds) (~quassel@129.110.242.224) | 00:06 | |
| + jacobk (~quassel@utdpat241106.utdallas.edu) | 00:40 | |
| - S0rin (QUIT: Ping timeout: 240 seconds) (~S0rin@user/s0rin) | 00:45 | |
| + S0rin (~S0rin@user/s0rin) | 00:46 | |
| abortretryfail | 8gb minimum sounds like a horrendous waste. | 01:28 |
|---|---|---|
| - mtm (QUIT: Ping timeout: 258 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 02:03 | |
| violet | josch: no its not contradictory. the throughput is the same- i get 200MiB/s read speed either way | 02:10 |
| violet | but it takes less CPU to do that | 02:10 |
| violet | so its more efficient | 02:10 |
| violet | in other words, the workqueue was not the bottleneck. but it was creating extra work for the processor to do for no reason | 02:11 |
| violet | so removing it did not speed things up because the bottleneck, whatever it is, is still there. but it did make the processor have to do less stuff | 02:12 |
| violet | per the topic of ram for building things - most of the time you dont strictly need that much, but you do if you want to take advantage of parallelism | 02:16 |
| violet | like- you cannot build boost on 4 cores with 4 gigs of ram because at some stages the compiler takes over 1 gig of ram per process | 02:17 |
| violet | but you can build boost on 1 or 2 cores with 4 gigs of ram | 02:17 |
| violet | however now your build is much slower because you have less parallelism, so having the extra ram means you can actually use all your cores | 02:17 |
| violet | plus i often like to do builds in tmpfs (keeping the files in ram) when possible to reduce wear on storage devices. however lately ive instead been using some intel optane for that purpose | 02:18 |
| violet | since intel optane has lower latency and much higher write endurance than flash-based SSDs | 02:18 |
| violet | which is great for builds | 02:18 |
| violet | great for swap too | 02:18 |
| violet | sad they killed the project but you can still buy perfectly good gen1 optane online at the moment | 02:19 |
| violet | unixpoet: NVMe SSDs' higher write endurance primarily comes from the drives just being bigger. and maaaybe the drive controller having better wear-leveling algorithms | 02:21 |
| violet | the bigger the drive, the more writes it can take before dying, because there's more flash cells to wear-level across. for the most part its simple as that | 02:21 |
| violet | bigger as in capacity, not physical size | 02:22 |
| - mjw (QUIT: Ping timeout: 246 seconds) (~mjw@gnu.wildebeest.org) | 02:37 | |
| unixpoet | thank you for the additional info :) | 03:06 |
| unixpoet | oh, and yes, builds on tmpfs are lovely! I tested that for the first time on this machine and built a linux kernel + modules in 12 minutes | 03:06 |
| - nsc (QUIT: Ping timeout: 244 seconds) (~nicolas@83-97-142-46.pool.kielnet.net) | 03:51 | |
| + nsc (~nicolas@103-99-142-46.pool.kielnet.net) | 03:53 | |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 04:10 | |
| - jacobk (QUIT: Ping timeout: 240 seconds) (~quassel@utdpat241106.utdallas.edu) | 04:47 | |
| + jacobk (~quassel@64.189.201.150) | 05:07 | |
| + klardotsh (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 05:30 | |
| - sterni (QUIT: Server closed connection) (~lukas@user/sterni) | 05:45 | |
| + sterni (~lukas@user/sterni) | 05:45 | |
| josch | violet: ah understood, thank you! | 07:11 |
| - chomwitt (QUIT: Ping timeout: 240 seconds) (~chomwitt@ppp-94-68-188-69.home.otenet.gr) | 07:27 | |
| + chomwitt (~chomwitt@ppp-2-85-137-223.home.otenet.gr) | 08:39 | |
| - klardotsh (QUIT: Remote host closed the connection) (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 08:47 | |
| + klardotsh (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 09:01 | |
| - XYZ (QUIT: Ping timeout: 240 seconds) (~XYZ@78-80-121-60.customers.tmcz.cz) | 09:38 | |
| + klardotsh_ (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 09:45 | |
| - klardotsh (QUIT: Ping timeout: 240 seconds) (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 09:46 | |
| - klardotsh_ (QUIT: Ping timeout: 252 seconds) (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 09:49 | |
| + klardotsh (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 09:51 | |
| + XYZ (~XYZ@78-80-121-60.customers.tmcz.cz) | 09:59 | |
| - XYZ (QUIT: Ping timeout: 244 seconds) (~XYZ@78-80-121-60.customers.tmcz.cz) | 10:57 | |
| - andreas-e (QUIT: Quit: Leaving) (~Andreas@2001:861:c4:f2f0:b0aa:56dd:ae73:8afa) | 11:12 | |
| + XYZ (~XYZ@37-48-59-110.nat.epc.tmcz.cz) | 11:15 | |
| violet | np! | 11:34 |
| - klardotsh (QUIT: Ping timeout: 252 seconds) (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 11:50 | |
| - chomwitt (QUIT: Ping timeout: 240 seconds) (~chomwitt@ppp-2-85-137-223.home.otenet.gr) | 12:09 | |
| + chomwitt (~chomwitt@ppp-94-67-236-45.home.otenet.gr) | 12:26 | |
| + eery (~eery@77.137.73.95) | 12:43 | |
| + mjw (~mjw@gnu.wildebeest.org) | 13:08 | |
| - S0rin (QUIT: Ping timeout: 240 seconds) (~S0rin@user/s0rin) | 13:43 | |
| - mtm (QUIT: Ping timeout: 244 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 14:03 | |
| abortretryfail | violet: In my experience building things on low-RAM systems, adding a lot of swap is usually just fine. I was bootstrapping Zig on a RPi 400 last year and it took something like 20GB total to successfully compile... terrible. | 14:24 |
| - GNUmoon (QUIT: Ping timeout: 246 seconds) (~GNUmoon@gateway/tor-sasl/gnumoon) | 14:28 | |
| + GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon) | 14:29 | |
| abortretryfail | the drawback to that on things like the RPi is swap on SD cards is often worse than on spinning disk, and it wears out your card. | 14:29 |
| - XYZ (QUIT: Ping timeout: 240 seconds) (~XYZ@37-48-59-110.nat.epc.tmcz.cz) | 16:01 | |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 16:10 | |
| + XYZ (~XYZ@37-48-33-138.nat.epc.tmcz.cz) | 16:14 | |
| josch | Debian linux maintainers uploaded 6.5.3 to unstable and some patches need to be rebased on top of the new versions as they fail to apply right now. Will do that later today. | 16:34 |
| sevan | I'm holding on to 6.4 until the new version of ZFS lands with 6.5 support. | 17:22 |
| josch | and then there is also the caam problem in 6.5 | 17:30 |
| josch | nice, 0002-drm-etnaviv-fix-dumping-of-active-MMU-context.patch got included | 17:49 |
| josch | minute: about adding cryptomgr.notests to the reform-system-image. Since this is a modification in /etc, we have no good way to remove this option later on via an apt upgrade. So even once this problem is fixed, systems will continue to boot with that option set. I think we should be very carful with which options we put into /etc via reform-system-image and make sure that they are unlikely to have | 17:53 |
| josch | unattended side-effects going forward. | 17:53 |
| - chomwitt (QUIT: Ping timeout: 240 seconds) (~chomwitt@ppp-94-67-236-45.home.otenet.gr) | 20:09 | |
| + chomwitt (~chomwitt@athedsl-25022.home.otenet.gr) | 20:48 | |
| + klardotsh (~klardotsh@c-67-170-115-80.hsd1.wa.comcast.net) | 20:50 | |
| - yankcrime (QUIT: Ping timeout: 240 seconds) (~nick@193.16.42.83) | 22:07 | |
| + S0rin (~S0rin@user/s0rin) | 23:25 | |
| - anzu (QUIT: Server closed connection) (~anzu@melkki.cs.helsinki.fi) | 23:33 | |
| + anzu (~anzu@melkki.cs.helsinki.fi) | 23:33 | |
| violet | josch: oh caam works if you disable tests? | 23:49 |
| violet | or what | 23:49 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!