Sudden judder or momentary freeze
- Thread starter Tyres O'Flaherty
- Start date
Which CPU do you have? (As you may be aware, the patch incoming on Wednesday will allegedly fix this.)
Much faster than my 7700K, even for single-threaded stuff. I think some other reports had the fastest CPUs out there, so it doesn't seem to be something that shows up only on old hardware...
Just sharing my experience: I'm playing in max details in FullHD with 4070 and had stutters after some playtime, graphic occassionay get corrupted with need to restart the game (usually when I was longer afk). Today, after some reading about nvidia drivers on more places, decided revert drivers from 57x to 566.36 released at 5th December 2024. No stutters, no graphics corruption, ED works perfect again. It seems like Nvidia did not had so far any luck with drivers released for 50xx cards serie.
I'm on a 14100 and I see the stutter issue.
I also see another issue where the game itself doesn't seem to tickle the CPU in the right way for the thermal management to see it as "busy." That causes a different judder and it's fixed by running XTU. Don't even have to have aggressive settings on XTU, just having it "on" is enough.
Might be slightly exacerbated by being GPU-bound as I have a 1060 but it definitely also happens in situations where the GPU is not pinned. This was one of the clues to me that there's something very odd about how Cobra offloads things to threads.
(For AMD people or people on Intel earlier than 11th / Rocket Lake - overclocking and XTU work quite differently to the earlier Intel stuff so it doesn't compare directly to the 7700k world. Rocket Lake onwards have a ridiculously complex stack of thermal management tricks so "overclocking" is now a terrible name for it; is more a case of telling it how hard to manage those thermal tricks and how little safety margin to allow.)
We're all now expecting/hoping that this issue is retired on Wednesday. That advance warning finally prompted me to pull the finger out and do what I've been thinking about for a few weeks: running Process Monitor to see wth is going on during a stuttering episode. I managed to capture one, and I reckon I learned a little bit more about what is happening. Basically, it's consistent with a lot of what I've seen proposed here and elsewhere.
NB: I think that what I'm sharing here is on the "safe" side of the line with regard to breaching the EULA (reverse-engineering and so on), but if any moderators disagree then please remove this post.
In short: the game process makes a series of TCP connections to an AWS server and downloads modest-sized chunks of data each time. For the stuttering episode I captured, the total duration was around 15 seconds, and it first became noticeable to me about 2 or 3 seconds in. During that 15-second episode, the game made 61(!) connections to the same AWS server, which is around the same as the total number of TCP connections it created in the previous 10 minutes (which were made to a few different AWS servers). Each time, it only fetched a few tens of kB (typically 20 or 30 packets), and only a few kB were sent to the server.
I can't see anything in the few tenths of a second preceding the start of the problem that would suggest that it was triggered from the server side, though that's not impossible. Certainly, none of the existing open TCP connections were used to transfer the data. It's all consistent with the game client deciding it wanted some updated data, and then fetching that, and then (crucially) doing something with that data in a way that stalls the game engine needlessly. (I'm saying "needlessly" because I simply can't conceive of any data which the game only needs to fetch periodically - minutes apart - and which yet somehow must be processed between displayed frames.)
NB: I think that what I'm sharing here is on the "safe" side of the line with regard to breaching the EULA (reverse-engineering and so on), but if any moderators disagree then please remove this post.
In short: the game process makes a series of TCP connections to an AWS server and downloads modest-sized chunks of data each time. For the stuttering episode I captured, the total duration was around 15 seconds, and it first became noticeable to me about 2 or 3 seconds in. During that 15-second episode, the game made 61(!) connections to the same AWS server, which is around the same as the total number of TCP connections it created in the previous 10 minutes (which were made to a few different AWS servers). Each time, it only fetched a few tens of kB (typically 20 or 30 packets), and only a few kB were sent to the server.
I can't see anything in the few tenths of a second preceding the start of the problem that would suggest that it was triggered from the server side, though that's not impossible. Certainly, none of the existing open TCP connections were used to transfer the data. It's all consistent with the game client deciding it wanted some updated data, and then fetching that, and then (crucially) doing something with that data in a way that stalls the game engine needlessly. (I'm saying "needlessly" because I simply can't conceive of any data which the game only needs to fetch periodically - minutes apart - and which yet somehow must be processed between displayed frames.)
Last edited:
Same for me
I'm on a 3090 running 566.36
There's a stutter on occasion but nothing game breaking
My fear is that FD will "fix" it for me...
If you select a target it can be.
Stutter on my 306012GB happens randomly and can be 5 to 6 judders, progressively getting worse till the last one. Then the frames go back up again. A few seconds later a freeze happens momentarily. This can happen at random, at any time and sometimes immediately following one I get another. I have almost crashed my explorer several times due to it. I noticed the science museum presentation also had the same stutter. Believe me when I say its like Frame..freeze...frame...freeeeeeze...frame....freeeeeeeze, until suddenly its back to normal.
Right now, its verging on game breaking. Its really disheartening to find this has been happening for months and even to the developers themselves and still no resolution. I've tried a few graphics settings with no change.
Right now, its verging on game breaking. Its really disheartening to find this has been happening for months and even to the developers themselves and still no resolution. I've tried a few graphics settings with no change.
Yup, really bad tonight, every few minutes and had to quit playing. I sure hope Wednesday's patch fixes this mess.
It hasn't
no that maybe be because tomorrow is wednesday the 30th to hazzard a guess
Then why is my client showing Elite Dangerous: Trailblazers - Update 3 when the update hasn't been pushed yet.no that maybe be because tomorrow is wednesday the 30th to hazzard a guess
Shouldn't it be showing the corsair update instead
tell that to doctor who?
a question only fd can answer via a ticket i suppose?
there was a cmdr named bright
whos speed was much faster than light
he departed earth in a relative way
and returned on the previous night
a question only fd can answer via a ticket i suppose?
there was a cmdr named bright
whos speed was much faster than light
he departed earth in a relative way
and returned on the previous night
Last edited:
because someone at Frontier can't count. Happens to the best of us. Common sense should tell you that if an update is scheduled for Wednesday the 30th, it's going to be released at that time regardless if it is called "3", "4", "42" or "Bob".
im getting every now and again a slight judder,im 26 fc jumps out heading to colonia,not game breaking though.so its game wide,but not worth loosing anysleep over it,big update weds i think,probaly clear it
He breaks the laws of time all the time!
Indeed he started out as a criminal on the run having stolen a time machine...
Perhaps FDev's next ship will be a T.A.R.D.I.S.
One of those could haul loads! Getting it through the door might problematic, however.