Bizarre unexplained fps drop

[regnighc]

Hi, I'm at a complete loss to what is causing this issue, and I feel like i've tried absolutely everything I can think of or find online but the issue still remains.

The Issue: Sometimes (about 80% of the time) when encountering a station my fps drops massively to about 6fps and locks there until I restart my pc.

This doesnt always happen and infact I'm able to run the game under normal circumstances on high gfx at between 60-120fps.

I could be sitting in front of a station for 10 minutes at 100fps and then all of a sudden it just drops to 6fps and never recovers.

I was able to resolve the issue for a whole week by uninstalling my GFX drivers and configuration using DDU and doing a fresh reinstall but suddenly for seemingly no reason at all it returned and the same fix no longer works.

Today was the last straw and I formatted my pc and reinstalled windows and ED, the issue remained from the first login.

For this reason I know its not a conflict with another app nor does it seem to be a configuration issue, I'm leaning towards a driver or game bug with my specific hardware and im about to try a year old driver instead.

I raised a support ticket a few weeks ago but have not heard anything yet unfortunatly.

Has anyone else experienced this or have any ideas of the cause? Any help is really appreciated o7

 

[STRONTIUM DOG]

a tiny microscopic breakdown of the silca substrate on 1 of the millions of transistors in a gpu could cause such a fault where it is only partially broken
an intermittent hardware fault is hardest to detect as test equip can show no problem
what i would do RMA the card and specify an intermittent fault if still under warranty (most times they just send a new one and seldom check if old card is actually at fault)
i would run more than a few furmark burner tests in an ambient room temp of at least 25 deg c to see if that revealed anything before sending

 

[pFVD]

Assuming you can move enough to do it, can you jump to another system and does this return your fps to normal?

I've never seen it as low as 6 fps, but this happens to me 1 in 10 times at a station, more so if I disembark and return to my ship. Been this way for me for years. Jumping to another system fixes it most times, others I have to logout and in again. Never seen any correlation between anything as to why except maybe VRAM/texture size. Try lowering your texture size as I find this reduces the chances of it happening for me, especially in VR.

Good luck

 

[regnighc]

a tiny microscopic breakdown of the silca substrate on 1 of the millions of transistors in a gpu could cause such a fault where it is only partially broken
an intermittent hardware fault is hardest to detect as test equip can show no problem
what i would do RMA the card and specify an intermittent fault if still under warranty (most times they just send a new one and seldom check if old card is actually at fault)
i would run a more than a few furmark burner tests in an ambient room temp of at least 25 deg c to see if that revealed anything before sending

I should add that my card is out of warranty but I play a number of other games extensively and experience no issues, I've also run benchmarks and stress tests which all pass without issue.

 

[regnighc]

Assuming you can move enough to do it, can you jump to another system and does this return your fps to normal?

I've never seen it as low as 6 fps, but this happens to me 1 in 10 times at a station, more so if I disembark and return to my ship. Been this way for me for years. Jumping to another system fixes it most times, others I have to logout and in again. Never seen any correlation between anything as to why except maybe VRAM/texture size. Try lowering your texture size as I find this reduces the chances of it happening for me, especially in VR.

Good luck

Jumping away does increase fps to a reasonable level so it's definitely related to the stations. Yesterday I sat outside a station facing the letterbox for 10 minutes at 120fps, then suddenly around the 10 minute mark it drops to 6fps and never recovered. I've tried with every graphic setting there is, everything off or low and it is exactly the same, which leads me to believe it's unrelated to performance and more of a bug or fault.

The last action I took last night was to try a fix that was discovered a few years ago that affected my card which is to use (3DMigoto) to disable a specific shader that was found to be badly coded. Though I need to test if over a few days atleast to know if that resolves it.

 

[STRONTIUM DOG]

this game at times will stress a gpu more than a burner test it goes from having almost nothing to do to to much to do in an instant
i meant to add it will probably be fine with other games...........
try removing the card and giving it a good an proper can of air dusting is all i can suggest and using the turn on the pc with out a card method so it registers a fault
that will give it a new IRQ
if you dont really wanna remove the card....... remove the driver then disable the card in device manager then reboot which will boot to a low resolution generic desktop then reinstall the driver and enable the card again an reboot which will clean the GPU reg and give it a fresh IRQ

 

[regnighc]

this game at times will stress a gpu more than a burner test it goes from having almost nothing to do to to much to do in an instant
i meant to add it will probably be fine with other games...........
try removing the card and giving it a good an proper can of air dusting is all i can suggest and using the turn on the pc with out a card method so it registers a fault
that will give it a new IRQ
if you dont really wanna remove the card....... remove the driver then disable the card in device manager then reboot which will boot to a low resolution generic desktop then reinstall the driver and enable the card again an reboot which will clean the GPU reg and give it a fresh IRQ

Hi, thanks for the suggestions, I have already done the driver cleanup and reinstall with DDU in safe mode though. And the card is very clean I dust it regularly, also the gpu usage is not 100% when the lag happens, just to mention also that The game can run at over 100fps on high settings as it does drop to 6fps on even low settings in the exact same spot. I did consider that maybe there is one specific ship that has a texture or something causing the issue because that might explain why it suddenly drops after sitting around a station for a while. I'm convinced it's a shader issue because I can clear the nvidia shader cache and delete the GPUtable xml file from elite directory to allow it to recreate the gfx profile and shader cache and that usually clears the issue up for a short time, but it always returns

 

[STRONTIUM DOG]

it does sound like a poor gpu opmtimization on there part try contacting support with a note and a direct x report describing all you have tried to remedy it as a last resort and a link to this thread

 

[Dragsham]

How old is the gpu?
Sounds like it's throttling itself. Heat spike momentarily can cause this issue the card saving itself, drastically cuts back. Do the fan speeds ramp up just prior/during?
A good cleaning and reapplication of thermal pads on mossfats etc wouldn't hurt. This involves dismantling the gpu.
I sent mine off to be professionally done a fella in Wales.. came back works perfectly now.
Case fans full clean also. And let's not forget the psu.
Certainly worth a go.
Ohh forgot the cpu! Reapplication of thermal paste!

 

[Raptoyd]

Try This while its not a fix all it certainly helped mine. Do you monitor your temps vs % of GPU/CPU usage? Put the Nvidea app on your PC if that is the GPU you have and hit ALT-R for the overlay in the top right of screen and make sure your not throttling due to temps. I keep that info front and center on my PC panel and Nvidea overlay so if there is a problem can see it immediately.

 

[æryn]

Hi, I'm at a complete loss to what is causing this issue, and I feel like i've tried absolutely everything I can think of or find online but the issue still remains.

The Issue: Sometimes (about 80% of the time) when encountering a station my fps drops massively to about 6fps and locks there until I restart my pc.

This doesnt always happen and infact I'm able to run the game under normal circumstances on high gfx at between 60-120fps.

I could be sitting in front of a station for 10 minutes at 100fps and then all of a sudden it just drops to 6fps and never recovers.

I was able to resolve the issue for a whole week by uninstalling my GFX drivers and configuration using DDU and doing a fresh reinstall but suddenly for seemingly no reason at all it returned and the same fix no longer works.

Today was the last straw and I formatted my pc and reinstalled windows and ED, the issue remained from the first login.

For this reason I know its not a conflict with another app nor does it seem to be a configuration issue, I'm leaning towards a driver or game bug with my specific hardware and im about to try a year old driver instead.

I raised a support ticket a few weeks ago but have not heard anything yet unfortunatly.

Has anyone else experienced this or have any ideas of the cause? Any help is really appreciated o7

Have you tried opening the galaxy map when this happens?
I get frequent FPS drops/instability too (not as severe as yours though) and find that if I open the galaxy map and wait for a few seconds the FPS will stabilise again.

 

[Morbad]

Has anyone else experienced this or have any ideas of the cause? Any help is really appreciated o7

You didn't specify much of anything about your hardware or software configuration, but it sounds like a case of borderline VRAM capacity.

Knowing what's going on with CPU, GPU, and VRAM utilization when these slowdowns occur would also help diagnose the issue.

Here's an example from a VRAM capacity limited scenario on one of my systems:

Any load on GPU "Copy" (ignore the extreme spike, as that is from alt-tabbing) is probably PCI-E traffic, which, in the context of nearly complete "dedicated GPU memory" usage and most any "shared GPU memory", implies that the card is running out of local VRAM. Actively swapping data over PCI-E to system memory has a lot of overhead and even the small overflow here is enough to sent my frame rate in the scene here from 70-80 to 50-60. Much more would totally destroy performance and would require less VRAM intensive settings.

Performance in these cases can spontaneously recover if enough idle assets can be evicted, but this is hard to reproduce.

Edit, another example, where even a small increase in "shared GPU memory" results in a major performance hit:

Reducing a few settings (texture and shadow quality one step) freed up enough VRAM for performance to return to normal:

Shared GPU memory is still significant, but it's evidently simply an allocation that hasn't been freed up yet, rather than being actively used, as we can see from dedicated GPU memory.

More severe VRAM capacity excursions, especially when combined with a slower PCI-E interface (this is a 16x 4.0 card) will produce proportionally more severe performance loss.

a tiny microscopic breakdown of the silca substrate on 1 of the millions of transistors in a gpu could cause such a fault where it is only partially broken
an intermittent hardware fault is hardest to detect as test equip can show no problem

Intermittent hardware faults rarely present as intermittent performance loss.

 

[OutofSpace]

Has anyone else experienced this or have any ideas of the cause? Any help is really appreciated o7

Yes, i am experiencing random frame drops, in space, at stations, on surfaces. Not micro stutters, actual big frame drops at what appears to be random. Using msi afterburner, all seems fine, no gpu/cpu overload and all temps fine. I do get very subtle micro stutters during a jump too.
I have experienced this since updating nvidia gpu drivers of the last two, so i don't know if it is a driver issue or not. I, along with others, i believe, are having issues with what appears to be server related, so i don't know if that is related.
I have also DDU'd and cleared cache and still the random frame drops occur. The frame drops last around five seconds and i am able to continue to play, no freeze up or CTD.
o7

 

[varonica]

My old laptop started doing this, I soon figured out it was thermal throttling on the CPU, it was overheating and would drop to very low levels of FPS, have you checked CPU temp when this happens? The laptop would often run ok in space away from everything but once you get in area with a lot of geometry, like just outside a station, that just killed it.

 

[regnighc]

It

How old is the gpu?
Sounds like it's throttling itself. Heat spike momentarily can cause this issue the card saving itself, drastically cuts back. Do the fan speeds ramp up just prior/during?
A good cleaning and reapplication of thermal pads on mossfats etc wouldn't hurt. This involves dismantling the gpu.
I sent mine off to be professionally done a fella in Wales.. came back works perfectly now.
Case fans full clean also. And let's not forget the psu.
Certainly worth a go.
Ohh forgot the cpu! Reapplication of thermal paste!

A an old gpu 1060 3gb, but it's sparkling clean, as is the rest of the PC, I regularly deep clean my pc. Thermal paste was freshly replaced on both cpu and gpu about a month ago (kryonaught) and both running at very good temps. Gpu doesn't go over 58c even during the lag, there's no difference in utilisation/temps/voltages when it's running fine or at 6fps. I'm fairly certain it's not hardware related because clearing the shader cache usually fixes it temporarily for a short time .

 

[Raptoyd]

Yes I have the same experience clearing the shader runs smooth but the hitching/stutter returns which seems like it is maybe driver related or on Frontier's end as never had this issue before. Hopefully somebody figures it out as it does not seem to be affecting everyone although seems to be happening to some for quite some time now as in this thread This

 

[the100thmonkey]

It

A an old gpu 1060 3gb, but it's sparkling clean, as is the rest of the PC, I regularly deep clean my pc. Thermal paste was freshly replaced on both cpu and gpu about a month ago (kryonaught) and both running at very good temps. Gpu doesn't go over 58c even during the lag, there's no difference in utilisation/temps/voltages when it's running fine or at 6fps. I'm fairly certain it's not hardware related because clearing the shader cache usually fixes it temporarily for a short time .

This is below the minimum spec for the game, which, IIRC, asks for a 1060 6GB.

Odyssey loves VRAM. You could reduce your render resolution or turn on FSR on with the performance setting, but I suspect the only fix will be to upgrade to something with 6GB or more.

EDIT

It looks like the system requirements have been updated. 3GB is the minimum requirement. Definitely experiment with the FSR options. Consider lowering your texture quality to low and reducing the render resolution.

 

[æryn]

Yes I have the same experience clearing the shader runs smooth but the hitching/stutter returns which seems like it is maybe driver related or on Frontier's end as never had this issue before. Hopefully somebody figures it out as it does not seem to be affecting everyone although seems to be happening to some for quite some time now as in this thread This

My system has a high end CPU and 128GB memory, but the video card has 8GB.
When the FPS drops or becomes unstable VRAM usage is always hovering around 100% allocation.

Not sure what the optimal VRAM size is for various resolutions and quality settings.

 

[the100thmonkey]

My system has a high end CPU and 128GB memory, but the video card has 8GB.
When the FPS drops or becomes unstable VRAM usage is always hovering around 100% allocation.

Not sure what the optimal VRAM size is for various resolutions and quality settings.

At 3440x1440, the client will happily use >10GB at vanilla ultra settings.

 

[Markov Chaney]

My old laptop started doing this, I soon figured out it was thermal throttling on the CPU, it was overheating and would drop to very low levels of FPS, have you checked CPU temp when this happens? The laptop would often run ok in space away from everything but once you get in area with a lot of geometry, like just outside a station, that just killed it.

This might be related to something else I noticed - Ascendancy has reintroduced an issue I saw before, where the demand on the CPU from ED is somewhat weird and the CPU doesn't necessarily "notice" that there's significant demand for it so it will throttle, thinking it's safe to do so, even though the game is trying to get stuff done. On Intel CPUs you can use XTU even if you don't have a K art, and it will make the power-saving less aggressive and solve that problem. (Don't know enough about power management on AMD to comment, sorry, but I do know there was a long spell where early boards that supported the 3d parts needed a firmware update to stop similar stutters for a similar reason.)