After about 12 years of service my GPU card is starting to act up. It seems to shut down so no video on my monitor, even the virtual consoles. System is up and functioning as I can access it via ssh. Sometimes this state is accompanied by the fan running at high speed though the card is not overheating.
The old card uses nvidia's 470 series of drivers while its replacement will use the 525 series.
Can I pre-install the 525 packages so they are present when I reboot after I install the new hardware? I.e. can both series be installed without consequence?
Thanks, jon
On Mon, 13 Feb 2023 09:04:39 -0500 Jon LaBadie jonfu@jgcomp.com wrote:
Can I pre-install the 525 packages so they are present when I reboot after I install the new hardware? I.e. can both series be installed without consequence?
I don't have an nvidia card, but while I think this is theoretically possible (I have both radeo and amdgpu installed in my kernel modules), I don't think it works in actual practice for nvidia. This is really a question for the rpmfusion user mailing list who package these drivers, or nvidia itself. You might find the answer in faqs at the nvidia site when they talk about installing their binary drivers.
If the new card is supported with the 470 driver then I would replace the card while the 470 driver is running.
Then do the upgrade.
Both drivers have the same name so you will not be able to have both installed and built in /lib/modules at the same time.
On Mon, Feb 13, 2023 at 8:05 AM Jon LaBadie jonfu@jgcomp.com wrote:
After about 12 years of service my GPU card is starting to act up. It seems to shut down so no video on my monitor, even the virtual consoles. System is up and functioning as I can access it via ssh. Sometimes this state is accompanied by the fan running at high speed though the card is not overheating.
The old card uses nvidia's 470 series of drivers while its replacement will use the 525 series.
Can I pre-install the 525 packages so they are present when I reboot after I install the new hardware? I.e. can both series be installed without consequence?
Thanks, jon
-- Jon H. LaBadie jonfu@jgcomp.com _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On Mon, Feb 13, 2023 at 09:27:40AM -0600, Roger Heflin wrote:
If the new card is supported with the 470 driver then I would replace the card while the 470 driver is running.
Then do the upgrade.
Ahh, good thought. I only checked the recommended drivers for the card. I didn't check the cards supported by the drivers.
Yes, the current driver I'm using, the 470 series does support both the old and the new card models. The 525 series driver only supports the new card.
Thanks for the suggestion.
Jon
Both drivers have the same name so you will not be able to have both installed and built in /lib/modules at the same time.
On Mon, Feb 13, 2023 at 8:05 AM Jon LaBadie jonfu@jgcomp.com wrote:
After about 12 years of service my GPU card is starting to act up. It seems to shut down so no video on my monitor, even the virtual consoles. System is up and functioning as I can access it via ssh. Sometimes this state is accompanied by the fan running at high speed though the card is not overheating.
The old card uses nvidia's 470 series of drivers while its replacement will use the 525 series.
Can I pre-install the 525 packages so they are present when I reboot after I install the new hardware? I.e. can both series be installed without consequence?
Thanks, jon
On 13/02/2023 15:27, Roger Heflin wrote:
If the new card is supported with the 470 driver then I would replace the card while the 470 driver is running.
Then do the upgrade.
Both drivers have the same name so you will not be able to have both installed and built in /lib/modules at the same time.
My /lib/modules/<kernel> has nvidia modules identified by
extra/nvidia-470xx/.... .XXX.ko.xz
so it should be able to make the distinction.
On Mon, Feb 13, 2023 at 8:05 AM Jon LaBadie jonfu@jgcomp.com wrote:
After about 12 years of service my GPU card is starting to act up. It seems to shut down so no video on my monitor, even the virtual consoles. System is up and functioning as I can access it via ssh. Sometimes this state is accompanied by the fan running at high speed though the card is not overheating.
The old card uses nvidia's 470 series of drivers while its replacement will use the 525 series.
Can I pre-install the 525 packages so they are present when I reboot after I install the new hardware? I.e. can both series be installed without consequence?
Thanks, jon
-- Jon H. LaBadie jonfu@jgcomp.com _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On Mon, Feb 13, 2023 at 09:27:40AM -0600, Roger Heflin wrote:
If the new card is supported with the 470 driver then I would replace the card while the 470 driver is running.
Then do the upgrade.
As I noted earlier, Roger's suggestion did show the new card was supported by both the old 470 and the 525 driver though the 525 is preferred.
So I shutdown the system, did the annual dust removal, and removed the old card -- learning that PCIe x16 slots have a latch.
Inserted the new card -- decided how to handle the switch from 2 6-pin power plugs to one 8-pin plug, and closed things up.
Press the power switch - nothing. Turns out while handling the case I flipped the rear panel switch -- whew!
System booted, graphics seem fine. Did a dnf update and rebooted. Used dnf to remove all the 470 driver bits including dependencies. I figured the in-use pieces would continue functioning, just disappear from the directory structure.
Then I added the 525 driver bits, the dependencies, and rebooted. Took quite a while (for a system that normally reboots quickly). I assume that was akmod building the driver.
But the system did come up, graphics again seem normal. System temps dropped about 2 degrees though the new graphics card runs a bit warmer, 37C vs 34C. Fan speed is lower, probably due to the dual fans vs single fan on the old card.
So thanks for all your input. Swap was essentially painless.
Jon
On 13 Feb 2023, at 14:05, Jon LaBadie jonfu@jgcomp.com wrote:
After about 12 years of service my GPU card is starting to act up. It seems to shut down so no video on my monitor, even the virtual consoles. System is up and functioning as I can access it via ssh. Sometimes this state is accompanied by the fan running at high speed though the card is not overheating.
The old card uses nvidia's 470 series of drivers while its replacement will use the 525 series.
Can I pre-install the 525 packages so they are present when I reboot after I install the new hardware? I.e. can both series be installed without consequence?
Given you have ssh access i would remove the 470 drivers. Power down Install new card Boot up and install new drivers via ssh.
You should also have console access without any driver installed I would guess.
Barry
Thanks, jon
-- Jon H. LaBadie jonfu@jgcomp.com _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On Mon, 2023-02-13 at 09:04 -0500, Jon LaBadie wrote:
After about 12 years of service my GPU card is starting to act up. It seems to shut down so no video on my monitor, even the virtual consoles. System is up and functioning as I can access it via ssh. Sometimes this state is accompanied by the fan running at high speed though the card is not overheating.
Re the "not overheating" are you judging by sensor readings, or have you felt the card to see if it's hot?
I do general maintenance on my systems once or twice a year, cleaning out the dust and fluff, reseating cards, sometimes replacing a fan.
Mind you, 12 years is a good service life in computing. And a bit surprising that you might be able to fit a modern one onto the same old board. Usually, technology changes make that difficult.
On Tue, Feb 14, 2023 at 01:10:25PM +1030, Tim via users wrote:
On Mon, 2023-02-13 at 09:04 -0500, Jon LaBadie wrote:
After about 12 years of service my GPU card is starting to act up. It seems to shut down so no video on my monitor, even the virtual consoles. System is up and functioning as I can access it via ssh. Sometimes this state is accompanied by the fan running at high speed though the card is not overheating.
Re the "not overheating" are you judging by sensor readings, or have you felt the card to see if it's hot?
from sensor readings via the nvidia-settings program. Once I suspected the video card I logged the output at 20 min intervals hoping to see the temps rise. Typically running in the low 30's (C) and reaching a max of 44 degrees. Fan speed ranged from 1200-1320 rpm.
I do general maintenance on my systems once or twice a year, cleaning out the dust and fluff, reseating cards, sometimes replacing a fan.
Plan to do the same when I swap video cards.
Mind you, 12 years is a good service life in computing. And a bit surprising that you might be able to fit a modern one onto the same old board. Usually, technology changes make that difficult.
True. My replacement, while unused, is not "current" but just a few generations newer than the failing one. Installed is a 660TI and the replacement is a 1660"super".
Tim:
Re the "not overheating" are you judging by sensor readings, or have you felt the card to see if it's hot?
Jon LaBadie:
from sensor readings via the nvidia-settings program. Once I suspected the video card I logged the output at 20 min intervals hoping to see the temps rise. Typically running in the low 30's (C) and reaching a max of 44 degrees. Fan speed ranged from 1200-1320 rpm.
It might well be worth opening it up and inspecting it. You could just have a heatsinking problem (dried out compound, not firmly attached, clogged fins).
Look at the capacitors also. The last couple of hw deaths I have all had one or more bulging capacitors. The electrolytic caps are the one component that are likely to eventually die.
On Tue, Feb 14, 2023 at 3:38 AM Tim via users users@lists.fedoraproject.org wrote:
Tim:
Re the "not overheating" are you judging by sensor readings, or have you felt the card to see if it's hot?
Jon LaBadie:
from sensor readings via the nvidia-settings program. Once I suspected the video card I logged the output at 20 min intervals hoping to see the temps rise. Typically running in the low 30's (C) and reaching a max of 44 degrees. Fan speed ranged from 1200-1320 rpm.
It might well be worth opening it up and inspecting it. You could just have a heatsinking problem (dried out compound, not firmly attached, clogged fins).
--
uname -rsvp Linux 3.10.0-1160.83.1.el7.x86_64 #1 SMP Wed Jan 25 16:41:43 UTC 2023 x86_64
Boilerplate: All unexpected mail to my mailbox is automatically deleted. I will only get to see the messages that are posted to the mailing list.
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On Mon, Feb 13, 2023 at 10:40 PM Tim via users < users@lists.fedoraproject.org> wrote:
On Mon, 2023-02-13 at 09:04 -0500, Jon LaBadie wrote:
After about 12 years of service my GPU card is starting to act up. It seems to shut down so no video on my monitor, even the virtual consoles. System is up and functioning as I can access it via ssh. Sometimes this state is accompanied by the fan running at high speed though the card is not overheating.
Re the "not overheating" are you judging by sensor readings, or have you felt the card to see if it's hot?
I do general maintenance on my systems once or twice a year, cleaning out the dust and fluff, reseating cards, sometimes replacing a fan.
Also check for bulging electrolytic capacitors (the larger cylindrical thingies with aluminum tops and marked with lines forming a cross that make it easy to identify one that isn't flat. I've seen many systems that became erratic and had one or more bulging capacitors. There was a group at my work that had the equipment and skills to replace capacitors -- only mission critical stuff warranted out-of-warranty surgery.
Sound practice. If you are near the ocean, you may also want to use "contact enhancer", now available at auto parts stores due the number of computers found in vehicles these days.
Mind you, 12 years is a good service life in computing. And a bit surprising that you might be able to fit a modern one onto the same old board. Usually, technology changes make that difficult.
For me (I do live near an ocean) few power supplies last 12 years, but longer when connected to high quality UPS. Also, newer GPU's often want more power than the originals.
On Tue, Feb 14, 2023 at 12:17:03PM -0400, George N. White III wrote:
On Mon, Feb 13, 2023 at 10:40 PM Tim via users <
.... Also, newer GPU's often want more power >than the originals.
Pleasant surprise there. While the newer, replacement card has 2.5X higher performance, max power draw is 25W lower (150W vs 125W).
Jon