Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.debian.user > #230005 > unrolled thread
| Started by | The Wanderer <wanderer@fastmail.fm> |
|---|---|
| First post | 2020-12-26 23:30 +0100 |
| Last post | 2020-12-30 04:10 +0100 |
| Articles | 20 on this page of 72 — 14 participants |
Back to article view | Back to linux.debian.user
No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-26 23:30 +0100
Re: No GRUB with brand-new GPU Georgi Naplatanov <gosho@oles.biz> - 2020-12-26 23:40 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-26 23:50 +0100
Re: No GRUB with brand-new GPU Linux-Fan <Ma_Sys.ma@web.de> - 2020-12-27 01:00 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-27 01:20 +0100
Re: No GRUB with brand-new GPU Linux-Fan <Ma_Sys.ma@web.de> - 2020-12-27 01:50 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-27 02:00 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-27 00:30 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-27 00:50 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-27 01:40 +0100
Re: No GRUB with brand-new GPU Sven Joachim <svenjoac@gmx.de> - 2020-12-27 09:00 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-27 13:20 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-27 16:30 +0100
Re: No GRUB with brand-new GPU Anssi Saari <as@sci.fi> - 2020-12-27 22:40 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-27 23:10 +0100
Re: No GRUB with brand-new GPU <tomas@tuxteam.de> - 2020-12-27 23:30 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-27 23:40 +0100
Re: No GRUB with brand-new GPU Andrei POPESCU <andreimpopescu@gmail.com> - 2020-12-28 09:20 +0100
Re: No GRUB with brand-new GPU Anssi Saari <as@sci.fi> - 2020-12-28 13:50 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-28 14:00 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-28 22:20 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-28 23:10 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-28 23:30 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-29 00:30 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-29 00:40 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-29 01:30 +0100
Re: No GRUB with brand-new GPU Linux-Fan <Ma_Sys.ma@web.de> - 2020-12-29 02:40 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-29 13:30 +0100
Re: No GRUB with brand-new GPU Anssi Saari <as@sci.fi> - 2020-12-29 11:20 +0100
Re: No GRUB with brand-new GPU "Thomas Schmitt" <scdbackup@gmx.net> - 2020-12-29 12:10 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-29 13:00 +0100
Re: No GRUB with brand-new GPU Andrei POPESCU <andreimpopescu@gmail.com> - 2020-12-29 19:10 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-29 19:20 +0100
Re: No GRUB with brand-new GPU Romualdas Taluntis <romualdas@taluntis.lt> - 2020-12-29 20:30 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-29 20:30 +0100
Re: No GRUB with brand-new GPU Romualdas Taluntis <romualdas@taluntis.lt> - 2020-12-29 21:00 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-29 21:20 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-29 21:40 +0100
Re: No GRUB with brand-new GPU Georgi Naplatanov <gosho@oles.biz> - 2020-12-29 21:40 +0100
Re: No GRUB with brand-new GPU Michael Stone <mstone@debian.org> - 2020-12-29 21:20 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-29 22:50 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-29 23:00 +0100
Re: No GRUB with brand-new GPU Michael Stone <mstone@debian.org> - 2020-12-29 23:40 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-30 03:50 +0100
Re: No GRUB with brand-new GPU Michael Stone <mstone@debian.org> - 2020-12-30 04:40 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-30 05:00 +0100
Re: No GRUB with brand-new GPU Doug McGarrett <dmcgarrett@optonline.net> - 2020-12-30 06:00 +0100
Re: No GRUB with brand-new GPU <tomas@tuxteam.de> - 2020-12-30 10:20 +0100
Re: No GRUB with brand-new GPU Michael Stone <mstone@debian.org> - 2020-12-30 15:00 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-30 19:40 +0100
Re: No GRUB with brand-new GPU Andrei POPESCU <andreimpopescu@gmail.com> - 2020-12-30 20:10 +0100
Re: No GRUB with brand-new GPU Michael Stone <mstone@debian.org> - 2020-12-30 20:20 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-30 22:10 +0100
Re: No GRUB with brand-new GPU Michael Stone <mstone@debian.org> - 2020-12-30 22:40 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-29 13:20 +0100
Re: No GRUB with brand-new GPU Anssi Saari <as@sci.fi> - 2020-12-29 22:50 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-29 23:00 +0100
Re: No GRUB with brand-new GPU Tony Rowe <torowe@gmail.com> - 2020-12-27 02:10 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-27 02:20 +0100
Re: No GRUB with brand-new GPU David Wright <deblis@lionunicorn.co.uk> - 2020-12-27 05:10 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-27 13:10 +0100
Re: No GRUB with brand-new GPU David Wright <deblis@lionunicorn.co.uk> - 2020-12-28 02:50 +0100
Re: No GRUB with brand-new GPU Andrei POPESCU <andreimpopescu@gmail.com> - 2020-12-27 08:40 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-27 13:10 +0100
Re: No GRUB with brand-new GPU Andrei POPESCU <andreimpopescu@gmail.com> - 2020-12-28 09:30 +0100
Re: No GRUB with brand-new GPU Romualdas Taluntis <romualdas@taluntis.lt> - 2020-12-29 14:00 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-29 14:00 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-29 18:10 +0100
Re: No GRUB with brand-new GPU The Wanderer <wanderer@fastmail.fm> - 2020-12-29 18:20 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-29 18:40 +0100
Re: No GRUB with brand-new GPU Michael Stone <mstone@debian.org> - 2020-12-29 19:20 +0100
Re: No GRUB with brand-new GPU Felix Miata <mrmazda@earthlink.net> - 2020-12-30 04:10 +0100
Page 2 of 4 — ← Prev page 1 [2] 3 4 Next page →
| From | The Wanderer <wanderer@fastmail.fm> |
|---|---|
| Date | 2020-12-28 22:20 +0100 |
| Message-ID | <Br85P-3Kn-1@gated-at.bofh.it> |
| In reply to | #230051 |
[Multipart message — attachments visible in raw view] — view raw
On 2020-12-28 at 07:55, The Wanderer wrote: > On 2020-12-28 at 07:47, Anssi Saari wrote: > >> One thing came to my mind, do you have the appropriate aux power >> connectors available and connected? Looks like the RX 5700 XT is >> quite power hungry at 225 W max vs. 151 W with your old card. >> Different connectors too (6 pin + 8 pin instead of 2x 6 pin). And >> these cards can be quite picky about having appropriate power >> connected. > > Yes - I noted the different power-connector requirements when I > first installed the card, and I've had them in place on every swap-in > since. > > If the card weren't getting its proper power, I'd expect anything > from no video output at all to a warning light to a motherboard beep, > none of which is happening. > > My PSU is overpowered for the system it's in, so I'm reasonably sure > it's capable of handling the added load for the upgraded GPU without > issues. I can now basically confirm that this isn't the issue. In my latest swap-out, to test the live-environment boot, I initially forgot to connect the power cables to the graphics card. I got no vide output at all, not even the POST screen, and from what reactions I was getting I suspect it may not have been booting even invisibly. The live-environment boot went nowhere; I got exactly the same hang booting that way. I tested on the old GPU too for comparison, and it worked there, although it brought up the GPU only in a low-resolution basic mode because the necessary firmware for hardware acceleration with that model isn't included in the Debian live images. I could probably dig up a non-Linux-based bootable medium (if nothing else, I have boot DVDs for Windows 10 install at my workplace, though I certainly wouldn't go through with the installation!), but I'm not at all sure I'd get any different result. At this point, I think my remaining options are as follows: [1] Get in touch with the people who made the motherboard, and ask them for a BIOS update to potentially get this working. Long odds of any success there, given that the last update for this motherboard model was eight-plus years ago, and if I'm not mistaken CPUs for the CPU socket on this motherboard are no longer even being manufactured. [2] Get in touch with the people who make the GPU (either AMD or MSI, the latter being the ones who did this particular card with the underlying processor) and ask them for advice. [3] Give up on getting this working with my current system, and start looking for parts to build a newer one that would be more likely to be compatible with this. Option 3 would be the most certain to work, but also the most expensive and unwieldy, particularly given that the hardware site I relied on for parts reviews when building past systems has died under financial pressure and been sold to faceless investors who run clickbait articles and don't engage with the community. It's made worse by the fact that I'd rather not invest in more storage devices so soon after building the current high-capacity RAID-6 array which holds such things as /home, and I don't want to reinstall onto the current pair of arrays and have to set back up my various installed programs et cetera, but I also don't trust things to work right if I move the existing hard drives en masse to a new hardware environment - and even if that *would* work, my potential motherboard selection would be limited by the need for a minimum of seven SATA ports. (And of course I'd rather not discard my $1000-plus CPU just because the motherboard doesn't want to work with newer GPUs, even if it's less performant and more power-hungry than CPUs which can be bought today for less.) -- The Wanderer The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. -- George Bernard Shaw
[toc] | [prev] | [next] | [standalone]
| From | Felix Miata <mrmazda@earthlink.net> |
|---|---|
| Date | 2020-12-28 23:10 +0100 |
| Message-ID | <Br8Se-4m8-3@gated-at.bofh.it> |
| In reply to | #230055 |
The Wanderer composed on 2020-12-28 16:13 (UTC-0500): ... 4-subscribe to one or both Grub mailing lists to ask for help. 5-Turn on UEFI if it's off, turn off if it's on, if you have a spare disk for making the switch, or your live media supports both. Grub's job is considerably different between the two. 6-Use Knoppix as your live media. It boots with Syslinux instead of Grub. 7-RMA the 5700. Have you tried slowing down the CPU, disabling hyper-threading, or slowing down RAM? Have you run memtest86 lately, or swapped RAM stick positions? Does your PS or motherboard have any swollen or leaky electrolytics? -- Evolution as taught in public schools, like religion, is based on faith, not on science. Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/
[toc] | [prev] | [next] | [standalone]
| From | The Wanderer <wanderer@fastmail.fm> |
|---|---|
| Date | 2020-12-28 23:30 +0100 |
| Message-ID | <Br9bz-4sD-5@gated-at.bofh.it> |
| In reply to | #230056 |
[Multipart message — attachments visible in raw view] — view raw
On 2020-12-28 at 16:59, Felix Miata wrote: > 4-subscribe to one or both Grub mailing lists to ask for help. The problem occurs with the live-boot environment, which AFAICT is using something other than GRUB to boot. Unless I'm mistaken about that boot method, I'm now reasonably certain that this isn't GRUB-specific after all. > 5-Turn on UEFI if it's off, turn off if it's on, if you have a spare > disk for making the switch, or your live media supports both. Grub's > job is considerably different between the two. As I indicated in the original post, this motherboard predates UEFI. (Or at least predates the universality thereof.) It has an actual BIOS. > 6-Use Knoppix as your live media. It boots with Syslinux instead of > Grub. I'm fairly sure the live environment I used (the amd64 image from https://www.debian.org/CD/live/) does likewise. I do have some other bootable media (of extremely primitive nature, graphics-wise) which I know for a fact uses ISOLINUX or similar, and I could try one of those if it might be fruitful... > 7-RMA the 5700. I'd be extremely hesitant to do that unless I know it's the problem. Also, although I just unboxed this within the past week, I ordered it more than a month ago; it's entirely possible they might not accept it back now. > Have you tried slowing down the CPU, disabling hyper-threading, or > slowing down RAM? No. All are possible options, but I'm nearly as hesitant to underclock the system as I'd be to overclock it (I did run this overclocked at one point, but that was within the first month or two I had the machine and I decided the added performance wasn't worth the increased cooler noise). I'm not sure I grasp why doing those things might help get things working with a newer GPU, when the existing one works just fine with the current settings. > Have you run memtest86 lately, No. > or swapped RAM stick positions? All 6 DIMM slots on this motherboard are occupied.The DIMMs come from a pair of three-DIMM kits, such that the DIMMs in each kit were tested pre-purchase with one another and confirmed to work; to the best of my recollection, the DIMMs are in the slots such that the three from a given kit are in the same memory channel. I'd be hesitant to go messing with that without solid reason. > Does your PS or motherboard have any swollen or leaky electrolytics? Not that I've been able to detect. I obviously can't inspect the inside of the PSU without in-depth surgery, but I did give the capacitors etc. on the motherboard a once-over during the last swap-out, and didn't notice anything apparently out of order. -- The Wanderer The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. -- George Bernard Shaw
[toc] | [prev] | [next] | [standalone]
| From | Felix Miata <mrmazda@earthlink.net> |
|---|---|
| Date | 2020-12-29 00:30 +0100 |
| Message-ID | <Bra7D-5gs-1@gated-at.bofh.it> |
| In reply to | #230057 |
The Wanderer composed on 2020-12-28 17:26 (UTC-0500): > Not that I've been able to detect. I obviously can't inspect the inside > of the PSU without in-depth surgery, but I did give the capacitors etc. > on the motherboard a once-over during the last swap-out, and didn't > notice anything apparently out of order. IME, the problem with electrolytics was solved a decade or more ago by switching motherboards to polys. The PS makers didn't and still haven't done that AFAICT. GPU needing both 6 wire and 8 wire dedicated power connectors smells like power hunger that an old PS with tired caps couldn't handle, notwithstanding its ostensible overprovisioning-rated output. Most PSes only require 4 screws removed to get the cover off far enough to inspect the usual cluster of candidate caps adjacent to where most wires feeding out connect to the board. -- Evolution as taught in public schools, like religion, is based on faith, not on science. Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/
[toc] | [prev] | [next] | [standalone]
| From | The Wanderer <wanderer@fastmail.fm> |
|---|---|
| Date | 2020-12-29 00:40 +0100 |
| Message-ID | <Brahk-5jN-17@gated-at.bofh.it> |
| In reply to | #230058 |
[Multipart message — attachments visible in raw view] — view raw
On 2020-12-28 at 18:26, Felix Miata wrote: > The Wanderer composed on 2020-12-28 17:26 (UTC-0500): > >> Not that I've been able to detect. I obviously can't inspect the >> inside of the PSU without in-depth surgery, but I did give the >> capacitors etc. on the motherboard a once-over during the last >> swap-out, and didn't notice anything apparently out of order. > > IME, the problem with electrolytics was solved a decade or more ago > by switching motherboards to polys. The PS makers didn't and still > haven't done that AFAICT. GPU needing both 6 wire and 8 wire > dedicated power connectors smells like power hunger that an old PS > with tired caps couldn't handle, notwithstanding its ostensible > overprovisioning-rated output. Given that I've already observed different behavior from the system when the GPU isn't plugged in to power at all, would it still be worth pursuing this as an angle? > Most PSes only require 4 screws removed to get the cover off far > enough to inspect the usual cluster of candidate caps adjacent to > where most wires feeding out connect to the board. I'd probably have to disconnect and unmount the PSU to get it far enough out to remove any such thing, which would be doable but still a pain. It's also a modular unit, so rather than "wires feeding out" of the PSU itself there are connector ports, but I imagine that wouldn't make much difference once it was open. -- The Wanderer The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. -- George Bernard Shaw
[toc] | [prev] | [next] | [standalone]
| From | Felix Miata <mrmazda@earthlink.net> |
|---|---|
| Date | 2020-12-29 01:30 +0100 |
| Message-ID | <Brb3H-5Pk-7@gated-at.bofh.it> |
| In reply to | #230059 |
The Wanderer composed on 2020-12-28 18:31 (UTC-0500): > Felix Miata wrote: >> IME, the problem with electrolytics was solved a decade or more ago >> by switching motherboards to polys. The PS makers didn't and still >> haven't done that AFAICT. GPU needing both 6 wire and 8 wire >> dedicated power connectors smells like power hunger that an old PS >> with tired caps couldn't handle, notwithstanding its ostensible >> overprovisioning-rated output. > Given that I've already observed different behavior from the system when > the GPU isn't plugged in to power at all, would it still be worth > pursuing this as an angle? I consider the process preventive maintenance once the warranty has expired. After 8 years it might be overdue for cleaning anyway. If all the bases aren't covered, the problem may never be discovered. >> Most PSes only require 4 screws removed to get the cover off far >> enough to inspect the usual cluster of candidate caps adjacent to >> where most wires feeding out connect to the board. > I'd probably have to disconnect and unmount the PSU to get it far enough > out to remove any such thing, which would be doable but still a pain. Absolutely removal from case would be necessary, but.... > It's also a modular unit, so rather than "wires feeding out" of the PSU > itself there are connector ports, but I imagine that wouldn't make much > difference once it was open. Modular should make removal from case easier than with conventional PSes. However, it could make it harder to visibly inspect the PS interior, maybe even harder just to get the cover off. I never buy modulars. -- Evolution as taught in public schools, like religion, is based on faith, not on science. Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/
[toc] | [prev] | [next] | [standalone]
| From | Linux-Fan <Ma_Sys.ma@web.de> |
|---|---|
| Date | 2020-12-29 02:40 +0100 |
| Message-ID | <Brc9t-6sk-21@gated-at.bofh.it> |
| In reply to | #230057 |
[Multipart message — attachments visible in raw view] — view raw
The Wanderer writes: > On 2020-12-28 at 16:59, Felix Miata wrote: [...] > > 7-RMA the 5700. > > I'd be extremely hesitant to do that unless I know it's the problem. > Also, although I just unboxed this within the past week, I ordered it > more than a month ago; it's entirely possible they might not accept it > back now. [...] This option 7 would have been my suggestion, too :) I think it is quite likely that the new GPU ist just being incompatible to the existent system. It may be possible to explain this situation to the vendor to get a refund. Here in Germany, many hardware parts (especially CPUs and GPUs) are available in very limited quantities only at the moment. I am not sure if that makes a difference or applies to your location at all... Apart from that, option 3 "upgrade more hardware" seems to be the next thing to consider from my point of view. I'd advise against getting anything compatible with the old CPU and rather go with a more recent platform. (My personal approach to keep the old CPU from going to waste would be to add a HDD, PSU and chassis to keep the old motherboard and CPU alive) > > Does your PS or motherboard have any swollen or leaky electrolytics? > > Not that I've been able to detect. I obviously can't inspect the inside > of the PSU without in-depth surgery, but I did give the capacitors etc. > on the motherboard a once-over during the last swap-out, and didn't > notice anything apparently out of order. [...] I have seen multiple modes of failure in my comparatively short experience with IT hardware: - PSU capacitor blows up, white smoke. This PSU was at least 20 years old! - PSU fails, system will not even turn on. This PSU was more than 10 years old. All inside electronics look OK (none of the capacitors blown, some dust as expected) - Motherboard caps fail (visibly bulging). System runs incredibly unstable (freezes at random moments). This motherboard was _cheap_ and 8 years old. - Server PSU (or the power switching unit?) on redundant-PSU server fails, server reboots at random time intervals due to this. Causes loud fan whine and clicking sounds in the switching unit, hard to miss. The PSU is 14 years old and the issue itself occurs only occasionally but then in short sequence. What I take from this: - In my cases, components either fail prematurely or issues occur at seemingly random intervals. - If it was a power issue at hand, I'd expect it to fail when starting something that needs 3D accelleration, at random points or upon power on (possibly high initial current draw). That it would always fail at the switch from BIOS to OS makes me believe it is not a power-related issue. My idea would be that at that point, initial power is already there (otherwise the screen would be black from the beginning) while at the same time, the GRUB screen should not increase power draw on the GPU by such large a value, that it would exceed the power draw from the previous GPU under load. One further question: When the live environments fail, where do they fail exactly? Whenever I run a live disk, it starts by printing something like "ISOLINUX Copyright... H. Peter Aanvin...", afterwards it will print/draw a menu. Does it do these things successfully and only fail to boot afterwards or is it black screen before anything from the live medium appears? (I do not have any answer in case of an immediate black screen.) HTH and YMMV Linux-Fan öö
[toc] | [prev] | [next] | [standalone]
| From | The Wanderer <wanderer@fastmail.fm> |
|---|---|
| Date | 2020-12-29 13:30 +0100 |
| Message-ID | <Brmiu-4Wy-5@gated-at.bofh.it> |
| In reply to | #230061 |
[Multipart message — attachments visible in raw view] — view raw
On 2020-12-28 at 20:38, Linux-Fan wrote: > The Wanderer writes: > >> On 2020-12-28 at 16:59, Felix Miata wrote: >> > 7-RMA the 5700. >> >> I'd be extremely hesitant to do that unless I know it's the problem. >> Also, although I just unboxed this within the past week, I ordered it >> more than a month ago; it's entirely possible they might not accept it >> back now. > > This option 7 would have been my suggestion, too :) > > I think it is quite likely that the new GPU ist just being incompatible to > the existent system. It may be possible to explain this situation to the > vendor to get a refund. > > Here in Germany, many hardware parts (especially CPUs and GPUs) are > available in very limited quantities only at the moment. I am not sure if > that makes a difference or applies to your location at all... I haven't been following availability very much, but there doesn't seem to be a shortage of what can be ordered online for delivery, at least. I did see items going out of stock as I viewed them when I was doing smartphone-related shopping this past weekend, but that's not necessarily going to translate to PC-hardware components. > Apart from that, option 3 "upgrade more hardware" seems to be the next thing > to consider from my point of view. I'd advise against getting anything > compatible with the old CPU and rather go with a more recent platform. > (My personal approach to keep the old CPU from going to waste would be to > add a HDD, PSU and chassis to keep the old motherboard and CPU alive) I've already started pricing out components to possibly build a replacement, but for financial reasons it'll probably be March at the earliest before I can potentially afford to do that, unless I'm lucky enough to win the lottery jackpot. >> > Does your PS or motherboard have any swollen or leaky electrolytics? >> >> Not that I've been able to detect. I obviously can't inspect the inside >> of the PSU without in-depth surgery, but I did give the capacitors etc. >> on the motherboard a once-over during the last swap-out, and didn't >> notice anything apparently out of order. > > I have seen multiple modes of failure in my comparatively short experience > with IT hardware: <snip> I've seen a few of those myself. None of them match very well with what I'm seeing here. > What I take from this: > > - In my cases, components either fail prematurely or issues occur > at seemingly random intervals. > > - If it was a power issue at hand, I'd expect it to fail when starting > something that needs 3D accelleration, at random points or upon power on > (possibly high initial current draw). That it would always fail at the > switch from BIOS to OS makes me believe it is not a power-related issue. > My idea would be that at that point, initial power is already there > (otherwise the screen would be black from the beginning) while at the same > time, the GRUB screen should not increase power draw on the GPU by such > large a value, that it would exceed the power draw from the previous GPU > under load. Seems reasonable to me. > One further question: When the live environments fail, where do they fail > exactly? Whenever I run a live disk, it starts by printing something like > "ISOLINUX Copyright... H. Peter Aanvin...", afterwards it will print/draw a > menu. Does it do these things successfully and only fail to boot afterwards > or is it black screen before anything from the live medium appears? (I do > not have any answer in case of an immediate black screen.) With the single live environment I've tried to date, it's the same immediate hard system hang as I get with booting to the internal hard drives; it goes straight from the last display that's part of what the BIOS handles to a black screen, and at the same time the keyboard stops responding (including keyboard-light toggle keys). I did get a similar black screen part of the time during a successful boot to that same live environment on the old, working GPU, but not right away, and IIRC the keyboard lights kept successfully toggling during that period; it was also only a few seconds, 5-to-15 at most, vs. nearly half an hour I've waited with no change on a (hard-drive) boot attempt with the new GPU. -- The Wanderer The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. -- George Bernard Shaw
[toc] | [prev] | [next] | [standalone]
| From | Anssi Saari <as@sci.fi> |
|---|---|
| Date | 2020-12-29 11:20 +0100 |
| Message-ID | <BrkgF-3IF-1@gated-at.bofh.it> |
| In reply to | #230057 |
The Wanderer <wanderer@fastmail.fm> writes: > I'm fairly sure the live environment I used (the amd64 image from > https://www.debian.org/CD/live/) does likewise. I do have some other > bootable media (of extremely primitive nature, graphics-wise) which I > know for a fact uses ISOLINUX or similar, and I could try one of those > if it might be fruitful... I can't tell without booting that, there are both grub and isolinux directories on that image (debian-live-10.7.0-amd64-standard.iso that is.) >> 7-RMA the 5700. > > I'd be extremely hesitant to do that unless I know it's the problem. > Also, although I just unboxed this within the past week, I ordered it > more than a month ago; it's entirely possible they might not accept it > back now. Well, it does have warranty, no? I'd say "I can get into BIOS all right but if I try to install Windows 10 I just get a black screen. Everything works fine with my old video card." Of course you'd want to test with Windows first but to me that's a sufficient diagnosis from a consumer. After all, at https://www.amd.com/en/support/graphics/amd-radeon-5700-series/amd-radeon-rx-5700-series/amd-radeon-rx-5700-xt#amd_support_product_spec AMD only lists sufficient power as a requirement. PCI express versions are compatible with adjacent versions and I think you have PCI express 3.0 on your motherboard & CPU. Still, I wouldn't mention Grub, Linux or the age of your hardware in a warranty claim. Of course, if you get a replacement and the new card doesn't work in your system either then it's more likely to be an incompatibility problem.
[toc] | [prev] | [next] | [standalone]
| From | "Thomas Schmitt" <scdbackup@gmx.net> |
|---|---|
| Date | 2020-12-29 12:10 +0100 |
| Message-ID | <Brl34-4h0-7@gated-at.bofh.it> |
| In reply to | #230067 |
Hi, The Wanderer wrote: > > https://www.debian.org/CD/live/) [ISOLINUX or GRUB ?] Anssi Saari wrote: > I can't tell without booting that, there are both grub and isolinux > directories on that image (debian-live-10.7.0-amd64-standard.iso that > is.) The ISOLINUX equipment is for legacy BIOS or EFI CSM, the GRUB stuff is for EFI native. Up to recently it was rare to see ISOs which boot from legacy BIOS via GRUB. Guix was about the most prominent. But Ubuntu 20.10 gave up on ISOLINUX and switched to GRUB for legacy BIOS. (This was not a cheap move, but given the maintainance state of SYSLINUX it seems inavoidable. The EFI image is not a file in the ISO but rather in the appended GPT partition number 2.) So if up to now it was always ISOLINUX which failed with that GPU, it might be worth to try https://releases.ubuntu.com/20.10/ubuntu-20.10-desktop-amd64.iso Have a nice day :) Thomas
[toc] | [prev] | [next] | [standalone]
| From | The Wanderer <wanderer@fastmail.fm> |
|---|---|
| Date | 2020-12-29 13:00 +0100 |
| Message-ID | <BrlPt-4xi-23@gated-at.bofh.it> |
| In reply to | #230068 |
[Multipart message — attachments visible in raw view] — view raw
On 2020-12-29 at 06:04, Thomas Schmitt wrote: > Hi, > > The Wanderer wrote: >> > https://www.debian.org/CD/live/) [ISOLINUX or GRUB ?] > > Anssi Saari wrote: >> I can't tell without booting that, there are both grub and isolinux >> directories on that image (debian-live-10.7.0-amd64-standard.iso that >> is.) > > The ISOLINUX equipment is for legacy BIOS or EFI CSM, the GRUB stuff is > for EFI native. That would indicate that this will have been using the ISOLINUX boot path, because as previously noted, this motherboard has a BIOS; if my memory is correct it predates UEFI entirely, and at the very least I've seen zero indication that it has one. > Up to recently it was rare to see ISOs which boot from legacy BIOS via > GRUB. Guix was about the most prominent. > But Ubuntu 20.10 gave up on ISOLINUX and switched to GRUB for legacy > BIOS. (This was not a cheap move, but given the maintainance state of > SYSLINUX it seems inavoidable. The EFI image is not a file in the ISO > but rather in the appended GPT partition number 2.) > > So if up to now it was always ISOLINUX which failed with that GPU, > it might be worth to try > https://releases.ubuntu.com/20.10/ubuntu-20.10-desktop-amd64.iso Unless the live-boot environment was actually using GRUB (which I'll admit I can't entirely confirm), to date it's always been GRUB that's failed with this GPU. I'll probably find time today for another shut-down and swap-out to test with an ISOLINUX-only boot medium; contrary to what I said in a previous mail, I'm not positive that I have any of them on hand at home, though I have plenty at my workplace. If I do test it, I'll report back here. -- The Wanderer The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. -- George Bernard Shaw
[toc] | [prev] | [next] | [standalone]
| From | Andrei POPESCU <andreimpopescu@gmail.com> |
|---|---|
| Date | 2020-12-29 19:10 +0100 |
| Message-ID | <BrrBv-8op-1@gated-at.bofh.it> |
| In reply to | #230069 |
[Multipart message — attachments visible in raw view] — view raw
On Ma, 29 dec 20, 06:52:12, The Wanderer wrote: > > That would indicate that this will have been using the ISOLINUX boot > path, because as previously noted, this motherboard has a BIOS; if my > memory is correct it predates UEFI entirely, According to Wikipedia in 2011 major vendors launched several consumer-oriented motherboards with UEFI. As far as I recall your board is from 2012. > and at the very least I've > seen zero indication that it has one. BTW, who about specifying exactly what motherboard you have and card you bought, unless it's confidential ;) Maybe some hardware expert (not me) reading the list spots something useful in the specs. Kind regards, Andrei -- http://wiki.debian.org/FAQsFromDebianUser
[toc] | [prev] | [next] | [standalone]
| From | The Wanderer <wanderer@fastmail.fm> |
|---|---|
| Date | 2020-12-29 19:20 +0100 |
| Message-ID | <BrrLc-8rr-15@gated-at.bofh.it> |
| In reply to | #230079 |
[Multipart message — attachments visible in raw view] — view raw
On 2020-12-29 at 13:05, Andrei POPESCU wrote: > On Ma, 29 dec 20, 06:52:12, The Wanderer wrote: > >> That would indicate that this will have been using the ISOLINUX >> boot path, because as previously noted, this motherboard has a >> BIOS; if my memory is correct it predates UEFI entirely, > > According to Wikipedia in 2011 major vendors launched several > consumer-oriented motherboards with UEFI. As far as I recall your > board is from 2012. No - the most recent BIOS update for this board was released in 2012. I think I bought it in 2010 or 2011; the purchase would have been on the basis of a review on The Tech Report (which was in full swing at the time), and the motherboard-review article for that model on that site is dated October of 2010. I'd have to dig deeper into my archives to identify the exact purchase date, but I don't believe it was as late as 2012. >> and at the very least I've seen zero indication that it has one. > > BTW, who about specifying exactly what motherboard you have and card > you bought, unless it's confidential ;) I thought I already did indicate the motherboard. It's this one (except purchased new, not used): https://www.newegg.com/p/N82E16813131665 Asus Sabertooth X58. The GPU I'm trying to get working is: https://www.amazon.com/gp/product/B07WNSP41M > Maybe some hardware expert (not me) reading the list spots something > useful in the specs. I doubt there's anything there to be found, but one never does know. -- The Wanderer The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. -- George Bernard Shaw
[toc] | [prev] | [next] | [standalone]
| From | Romualdas Taluntis <romualdas@taluntis.lt> |
|---|---|
| Date | 2020-12-29 20:30 +0100 |
| Message-ID | <BrsQV-CM-1@gated-at.bofh.it> |
| In reply to | #230081 |
You have an old system with first gen i7 which predates widespread UEFI adoption by at least 1 generation, your motherboard specs at https://origin-www.asus.com/Motherboards/SABERTOOTH_X58/specifications/ do not mention UEFI, only regular BIOS, so your system has no UEFI support. Newer AMD GPUs do not work at all or work randomly on some systems with legacy BIOS since RX 500 series, and you have a newer RX 5000 series card. Your system POSTs but after that you have a black screen, which is standard in such incompatibilities (you'd see the same on Windows). You can see a video from Level1Techs at https://youtu.be/26lGq_SOLkg?t=917 (Timestamp: 15:17-16:07) where they talk about this. So it's probably not GRUB and not your PSU, but hardware incompatibility, so as per my earlier post your course of action is: -upgrade/new pc -trying with two GPUs -mucking about with BIOS (I tried searching for your GPU and did not found anything provided by MSI, so if you want to go this (dangerous) route you can try by reading this thread: https://www.win-raid.com/t4262f16-How-to-disable-UEFI-for-legacy-only-boot-on-recent-AMD-Radeon-GPUs.html but keep in mind that they are talking about even older GPUs and you can destroy your GPU if something goes wrong). On 2020-12-29 20:13, The Wanderer wrote: > I thought I already did indicate the motherboard. It's this one (except > purchased new, not used): > > https://www.newegg.com/p/N82E16813131665 > > Asus Sabertooth X58. > > The GPU I'm trying to get working is: > > https://www.amazon.com/gp/product/B07WNSP41M -- Pagarbiai Romualdas Taluntis
[toc] | [prev] | [next] | [standalone]
| From | The Wanderer <wanderer@fastmail.fm> |
|---|---|
| Date | 2020-12-29 20:30 +0100 |
| Message-ID | <BrsQW-CM-5@gated-at.bofh.it> |
| In reply to | #230082 |
[Multipart message — attachments visible in raw view] — view raw
On 2020-12-29 at 14:20, Romualdas Taluntis wrote: > You have an old system with first gen i7 which predates widespread > UEFI adoption by at least 1 generation, your motherboard specs at > https://origin-www.asus.com/Motherboards/SABERTOOTH_X58/specifications/ > do not mention UEFI, only regular BIOS, so your system has no UEFI > support. Yes, I know - I pointed out the lack of UEFI on this system in the original post. > Newer AMD GPUs do not work at all or work randomly on some systems > with legacy BIOS since RX 500 series, That's news to me, and I'm surprised not to have seen it better documented; generational hardware compatibility breaks are usually made fairly prominent, in my limited experience. Aside from a video link, any sources / references on documenting this? Not that I doubt that it's true, I'm just trying to get more background, including on why and how it managed to slip past my attention. -- The Wanderer The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. -- George Bernard Shaw
[toc] | [prev] | [next] | [standalone]
| From | Romualdas Taluntis <romualdas@taluntis.lt> |
|---|---|
| Date | 2020-12-29 21:00 +0100 |
| Message-ID | <BrtjY-16J-5@gated-at.bofh.it> |
| In reply to | #230083 |
I have no official sources on this, but if you'd do an internet search for "new gpu legacy bios" you'd see a lot of people having the same issue, asking for guidance on buying a new GPU for an old PC, or wondering why they weren't better warned about such an incompatibility. I personally also had the same problem and found a rule of thumb when upgrading non UEFI PC with a new video card: use Nvidia (up to and including 10 series, newer also have same problem). On 2020-12-29 21:26, The Wanderer wrote: > On 2020-12-29 at 14:20, Romualdas Taluntis wrote: > > Newer AMD GPUs do not work at all or work randomly on some systems > with legacy BIOS since RX 500 series, > That's news to me, and I'm surprised not to have seen it better > documented; generational hardware compatibility breaks are usually made > fairly prominent, in my limited experience. Aside from a video link, any > sources / references on documenting this? > > Not that I doubt that it's true, I'm just trying to get more background, > including on why and how it managed to slip past my attention. > -- Pagarbiai Romualdas Taluntis
[toc] | [prev] | [next] | [standalone]
| From | The Wanderer <wanderer@fastmail.fm> |
|---|---|
| Date | 2020-12-29 21:20 +0100 |
| Message-ID | <BrtDk-1ZS-11@gated-at.bofh.it> |
| In reply to | #230085 |
[Multipart message — attachments visible in raw view] — view raw
On 2020-12-29 at 14:53, Romualdas Taluntis wrote: > I have no official sources on this, but if you'd do an internet > search for "new gpu legacy bios" you'd see a lot of people having the > same issue, asking for guidance on buying a new GPU for an old PC, or > wondering why they weren't better warned about such an incompatibility. It is, to be fair, a good question. > I personally also had the same problem and found a rule of thumb when > upgrading non UEFI PC with a new video card: use Nvidia (up to and > including 10 series, newer also have same problem). I have a hard "no NVIDIA" rule, based both on bad experiences dealing with their driver / etc. stack back in the day (when I built the computer *preceding* this one, i.e. sometime in the oughts), and on an objection to the proprietarian nature of their approach to Linux drivers. Last I was aware, AMD and Intel alike have Good Enough driver support on Linux via fully-open code, and I want to encourage that rather than NVIDIA's closed approach. Looks like I'm due to build a new machine! All that's left - aside from saving up the funds - is to determine the specifics. Starting with whether I acquire a new case (suitable to hold everything I've got now and potentially more), or reuse the same one (thereby making it impossible to run both computers temporarily side-by-side). I've been concerned for the past few years anyway that the secondary SATA chip on this motherboard (which brings it up to 8 ports rather than the chipset-provided 6) is tending towards failure; it takes 30+ seconds (if not over a minute) to pass its drive-detection-etc. checks during that stage of POST, unless the computer has been shut off for long enough that the components have all cooled as far as they're going to. Everything seems to work fine once things are booted, but it's still a concerning symptom. I'll also need to decide whether to go with modern Debian, or revert it to sysvinit analogously to what I currently run, or just go with Devuan outright. Or try another distro, perhaps one more suited to running what I want to run the way I want it; I've been meaning to check out Gentoo for perhaps literally decades by now... though it'd be a shame to drop the culture and environment I've gotten so used to by this point. -- The Wanderer The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. -- George Bernard Shaw
[toc] | [prev] | [next] | [standalone]
| From | The Wanderer <wanderer@fastmail.fm> |
|---|---|
| Date | 2020-12-29 21:40 +0100 |
| Message-ID | <BrtWH-262-21@gated-at.bofh.it> |
| In reply to | #230086 |
[Multipart message — attachments visible in raw view] — view raw
On 2020-12-29 at 15:34, Georgi Naplatanov wrote: > On 12/29/20 10:09 PM, The Wanderer wrote: > >> On 2020-12-29 at 14:53, Romualdas Taluntis wrote: >> >> Last I was aware, AMD and Intel alike have Good Enough driver >> support on Linux via fully-open code, and I want to encourage that >> rather than NVIDIA's closed approach. > > Be careful with AMD/ATI video cards. In the past they were dropping > support for not too old cards from their fglrx driver what I didn't > like. I used to run fglrx (on my previous AMD GPU, which I believe was on my previous computer), but haven't done so in some while. In this case I was/am planning to run the 5700 using the 'amdgpu' driver, i.e. as close as possible to a fully open stack given the hardware. I believe Phoronix has reported test results with that model based on that driver which should be more than sufficient performance for my needs. -- The Wanderer The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. -- George Bernard Shaw
[toc] | [prev] | [next] | [standalone]
| From | Georgi Naplatanov <gosho@oles.biz> |
|---|---|
| Date | 2020-12-29 21:40 +0100 |
| Message-ID | <BrtWH-262-23@gated-at.bofh.it> |
| In reply to | #230086 |
On 12/29/20 10:09 PM, The Wanderer wrote: > On 2020-12-29 at 14:53, Romualdas Taluntis wrote: > > Last I was aware, AMD and Intel alike have Good Enough driver support on > Linux via fully-open code, and I want to encourage that rather than > NVIDIA's closed approach. > Be careful with AMD/ATI video cards. In the past they were dropping support for not too old cards from their fglrx driver what I didn't like. Kind regards Georgi
[toc] | [prev] | [next] | [standalone]
| From | Michael Stone <mstone@debian.org> |
|---|---|
| Date | 2020-12-29 21:20 +0100 |
| Message-ID | <BrtDk-1ZS-17@gated-at.bofh.it> |
| In reply to | #230083 |
On Tue, Dec 29, 2020 at 02:26:27PM -0500, The Wanderer wrote: >On 2020-12-29 at 14:20, Romualdas Taluntis wrote: > >> You have an old system with first gen i7 which predates widespread >> UEFI adoption by at least 1 generation, your motherboard specs at >> https://origin-www.asus.com/Motherboards/SABERTOOTH_X58/specifications/ >> do not mention UEFI, only regular BIOS, so your system has no UEFI >> support. > >Yes, I know - I pointed out the lack of UEFI on this system in the >original post. > >> Newer AMD GPUs do not work at all or work randomly on some systems >> with legacy BIOS since RX 500 series, > >That's news to me, and I'm surprised not to have seen it better >documented; generational hardware compatibility breaks are usually made >fairly prominent, in my limited experience. Aside from a video link, any >sources / references on documenting this? > >Not that I doubt that it's true, I'm just trying to get more background, >including on why and how it managed to slip past my attention. Relatively few people want to put a new video card into a (comparatively) ancient computer. It honestly isn't a great value proposition--regardless of how much you spent on it a decade ago, a mid range ryzen from a few years ago will outperform it, let alone a current generation processor. Since nobody does this, there's no value to the manufacturer to develop the necessary BIOS interface in addition to the UEFI interface that will actually be used by basically all of their customers. They do the bare minimum necessary to let someone reconfigure their machine from traditional BIOS to UEFI, and even that is probably working more because it hasn't broken yet than because they're putting effort into it. Having played with this sort of thing before, it's really not worth it. With a big enough generational jump you run into all sorts of issues (e.g., the new hardware pushes the PCIe bus just a little harder than anything did before, and the aging hardware starts to flake out) and frustrating bottlenecks. If the hardware works, just leave it alone. If you feel the need to upgrade some of it, just upgrade the whole thing because that's where things will eventually end up anyway...but if you do it all at once you'll end up wasting a lot less time chasing gremlins.
[toc] | [prev] | [next] | [standalone]
Page 2 of 4 — ← Prev page 1 [2] 3 4 Next page →
Back to top | Article view | linux.debian.user
csiph-web