| 09:08 | vpeter> | jnettlet[m]: I will try one samsung extreme just to compare. But EVO+ should be fine too. |
| 09:08 | vpeter> | Sadly my device has broken emmc - lots of errors when using. |
| 09:16 | Ke> | jnettlet[m]: you would think solid-run would be awesome enough to ship user replaceable socketed emmcs |
| 09:16 | Ke> | unless you use the combined DRAM/eMMC packaging |
| 09:17 | Ke> | my major complaint about eMMC is that it wears and is not replaceable |
| 09:18 | jnettlet[m]> | Ke: not on our SOMs no. the socket takes up too much space. |
| 09:19 | Ke> | sure, but on full boards |
| 09:19 | jnettlet[m]> | most the modern eMMC is rated for years of write cycles. However for replaceable storage we recommend m.2 ssds |
| 09:20 | Ke> | jnettlet[m]: you know me, always writing single bytes on btrfs and full flushing between each write |
| 09:21 | Ke> | I made my custom text edior for PeRfOrm4nC3 |
| 09:22 | Ke> | I actually do use btrfs and extent tree is known for it's avalanche like write behaviour |
| 09:23 | suihkulokki> | from my experience the boards around the eMMC tend to become obsolete before the eMMC wears out... |
| 09:23 | Ke> | I was actually amazed that my C201 Samsung SD rootfs has lasted this long |
| 09:23 | vpeter> | Ke: You can replace emmc. Just not easy :) |
| 09:23 | Ke> | I know, swiftgeek always tells me, when I complain |
| 09:24 | Ke> | it comes with the "of course you have full electronics workshop at home" |
| 09:24 | vpeter> | Or you know one. |
| 09:25 | suihkulokki> | I wonder if helsinki hacklab has the needed gear |
| 09:25 | Ke> | suihkulokki: most probably does, also my workplace does |
| 09:25 | Ke> | I still consider that there is a real risk of losing the board, when making hw mods |
| 09:26 | Ke> | that's why I most often discount the possibility of ~ |
| 09:27 | Ke> | and that's why I just ignore eMMC, if I don't have to use it |
| 09:27 | vpeter> | There must be real reason to do such replacement. Like when board costs ~1000 € |
| 09:27 | Ke> | suihkulokki: also my plan is not to buy obsolete stuff anymore, that's why I want upstream kernel support |
| 09:28 | vpeter> | Mhm, maybe I should just buy one small m.2 disk. |
| 09:29 | Ke> | suihkulokki: my 5 year old computer is still good enough for the foreseeable future, but it will become a liability, as its parts wear down |
| 09:48 | suihkulokki> | Ke: indeed, mine as well. for Arm based machine, if I would expect to use it for long term I'd treat the eMMC as a firmware storage |
| 09:48 | suihkulokki> | u-boot/efi and perhaps some known working kernel/initrd combo |
| 09:49 | jnettlet[m]> | really eMMC as the /boot folder should last more than a decade I would think. |
| 09:49 | Ke> | yup, right now my C201 eMMC holds the backup kernel and some partitions I did not dare to kill |
| 09:49 | jnettlet[m]> | u-boot / efi can also live on SPI-NOR |
| 09:50 | Ke> | that is 32MiB of linear write/week, almost whole device has been discarded |
| 09:50 | Ke> | so there is practically no wear at all |
| 09:51 | jnettlet[m]> | not to mention most modern eMMCs have nearly double the NAND they advertise. All for performance and wear leveling etc. |
| 09:52 | Ke> | hmm, I have heard 20% extra |
| 09:53 | Ke> | anyway, for me it's 500x extra at least, as I only have the kernel anyway |
| 17:04 | vpeter> | jnettlet[m]: fyi - I just boot sandisk extreme pro sd card with solidpc and same issue (bad speed). Card is not even recognized as uhs. |
| 17:05 | jnettlet[m]> | vpeter: not seeing 1.8V at all in sysfs? I wonder if there is a setting in the BIOS that got tweaked |
| 17:05 | jnettlet[m]> | you may want to go in and load the defaults |
| 17:08 | vpeter> | no sign of 1.8v. |
| 17:11 | vpeter> | no change after bios reset |
| 17:15 | vpeter> | Mhm, it works as uhs on cold boot. But after reboot it is again high speed only. |
| 17:18 | vpeter> | Seems another issue with my board. |
| 17:22 | jnettlet[m]> | oh interesting. I will test that. could be a bios bug. |
| 17:23 | vpeter> | Speed is almost 60MB/s. This is what I expect :-) |
| 17:27 | jnettlet[m]> | that sounds reasonable |
| 17:27 | jnettlet[m]> | that is on cold boot? |
| 18:20 | vpeter> | yes, cold boot where works as it should. |
| 20:34 | topi`_> | it seems *both* my ATX power supplies are dead |
| 20:35 | topi`_> | I tried to connect the PWR_ON to GND and measure the voltage between red/yellow pins and gnd |
| 20:35 | topi`_> | and no voltage :( |
| 20:35 | topi`_> | unless there's something I'm missing? |