Odyssey: Loading time to disembark to planetside

Hi Folks,

I switched from GeforceNow to an own gaming machine (Lenovo Legion Go). Everything runs very smooth, except that I am waiting far longer now watching a black screen than before when disembarking from my ship to a planet. Before it was 5 seconds, now its full 30. Does anybody know more about what the bottlenecks could be for this ? Is it local internet connection speed, RAM, SSD performance or what else?

Thanks in advance!
 
Ok, but why is there a so much slower server response when I‘m using my own local installation compared to a GeforceNow rig?
 
Ok, sounds plausible .. so:

30 sec planet disembarking when game is running in my local home network, due to probably worse server communication.

5 sec planet disembarking via GeforceNow rig, in an excellent industrial network, probably with 1 Gb/sec speed.

What are your planet disembarking times?
Is 30 sec abysmal?
 
Servers for Live mode have been bad for months now. I reguarly see severely extended times on anything in gane that needs an instance change Jumping around (both High Wake and Low Wake), Embarkation, alighting, and also on entering the game and game mode changes. These are all things that I do reguarly on fuel rescues.

Legacy is fine as there are fewer players there.

My take on this is that FDev is that the longer jump times are during the difficulties getting a server to provide the destination instance for jumps. I strongly suspect that FDev have implemented cost cutting measures and have reduced the ability for new servers to spin up as required, or are using cheaper server VMs.

Around the time this all started, the amount of time that the Thursday morning maintenance cycle took to complete also increased dramatically. FDev have always advertised 40 mins for normal maintenance cycle. Befoe the changes, I could often gat back into the gane within 12-18 mins. Now it usually takes longer thsn the advertised 40 mins.
 
Last edited:
Yep, sounds abysmal to me. My times usually something like 5 to occasionally 10sec on home rig.

It feels abysmal for sure, black screen for half a minute totally breaks game flow. The strange thing is, all of the other instance changes are as fast as always, and e.g. disembarking from the SRV to planet side only takes the usual 5 sec.

Does the game have to load lots of stuff into the RAM when disembarking from your ship to a planet? That might be a factor, as I only have 16 gigs of RAM, and 6 of that are shared VRAM.

Servers for Live mode have been bad for months now. I reguarly see severely extended times on anything in gane that needs an instance change Jumping around (both High Wake and Low Wake), Embarkation, alighting, and also on entering the game and game mode changes. These are all things that I do reguarly on fuel rescues.

I can confirm that, it’s approximately since the increased traffic from the end phase of the Thargoid War, combined with the new ships everbody was eager for, especially the Mandalay.

Nonetheless I‘ve got the impression that there is another issue of my individual setup regarding the planet disembarkings. My device is a Legion Go BTW, maybe someone is using that too and is observing the same behavior?
 
Graphics settings can affect the load times heavily, low res textues would load faster, but all that should already be loaded in the circumstances of a disembarking.

Disembarking and SRV deploying taking so long is the most flow-breaking thing in the game.
 
There is a Embark / Disembark bug that results in a very dark view (you can just abut ee stuff) in a half disembarked state. This is not a delay - it will remain until a relog via main menu is performed (at which point your embark/disembark will have been completed. This is for both ship ans SRV but I mostly see it on SRV. As far as I know, FDev did look at this but were unable to track the cause,

The other issues (affecing all instance changes) seem to be basically server connection attempts timing out and being retried. I'm not sure if this is due to servers being busy, or failed servers being referenced. The long jumps (and other instance acquisition issues) are basically multiples retries to get a reference to the new instance. If this goes on for too many retries, the retry count is exhausted and you get a colourful snake error.

Yes it's worse at busy times of the day, but in the past, the servers coped with many more concurrent players than to do now. Hence my comments in my previous post.

The problem is not just longer jump time however. I've been on Long range rescues (plotted using Galaxy Plotter for a neutron route that takes fuel into account) where I have had retries on a jump time-out, and crashed out with a colourful snake error. On logon, I've been dumped in the system of ny destination neutron, a massive distance from the neutron that I need to use. Flying here will delay me on my rescue (and consume a small amount of fuel that might be critical for the plotted route to work) and jumping out/back i, diverting is not an option wither as thayt wil se-sync my fuel more.
 
so r u running a GTX4090 like gfn are too?
as an experiment set all gpu gfx settings to low then it can only be either processor speed or network connection speed i would have thought
is your monitor native res being used for both in game and os too?
 
Last edited:
Does the game have to load lots of stuff into the RAM when disembarking from your ship to a planet? That might be a factor, as I only have 16 gigs of RAM, and 6 of that are shared VRAM.
Shared VRAM is just plain disaster for the kind of game engine ED uses unfortunately. I'm mildly surprised Odyssey works at all...

The bottleneck is definitely RAM performance, because the game expects VRAM. That is somewhat mitigated by the fact the Go has DDR5 RAM but it still ain't VRAM, so even on raw speed, you're down, before you consider how much memory is available.

Since Ascendancy, the game is not completely comfortable on a 16GB system which ALSO has 6GB VRAM out on the GPU so yes, memory management will be the bottleneck.

so r u running a GTX4090 like gfn are too?
OP already mentioned they are using a Lenovo Legion Go. That has a mobile-focused GPU with no dedicated VRAM

The GPU itself is an AMD Radeon 780M, which benchmarks nearly ten times worse than a GTX4090.

Nonetheless I‘ve got the impression that there is another issue of my individual setup regarding the planet disembarkings. My device is a Legion Go BTW, maybe someone is using that too and is observing the same behavior?
Yep, rendering planets is very memory hungry and it depends on best of breed modern CPU<->GPU comms. Since your memory is completely off the pace and it's on a weird architecture there's no good way of making that happen.

The good news is there are lots of configurables in the graphics options and some of them are there specifically to deal with this problem so you could try turning down a lot of the planet-specific stuff to help.

By default it does a lot of preload type stuff to help performance but on your system it'll drown it before it's even started, hence the unusually long time for that phase to happen. You can turn the wick down on all of that in the options which will then make it render more as it goes long; which will mean you'll get some load-in when it has to do that, and also means you'll get lower FPS because it's spending time rendering on-demand instead of using pre-rendered stuff, but it'll probably be smoother.

Finally, seems your 780M benches a little bit worse than even my ancient 1060 so I doubt you'll get much better than FPS in the 20s on a planet surface unless you turn everything down to Lego mode. If you do get a decent FPS let me know what settings you used...

I'm not saying the Go is necessarily "bad" it's just unfortunately there are some very specific things it will never be good at and they are the exact things ED depends on most, even though other games on there will probably fly.

ETA: Specifically, Ascendancy really increased the impact on memory, both total use of it and how often data goes backwards and forwards and around.
 
Last edited:
ya unless running identical machines on the same network they wont run the same period.
its like a race between bolt an jimmy crankie the op post or machine is not exactly firing on all cylinders
 
Last edited:
Well, thanks @Markov Chaney, I gained quite some insights by your answer! What is weird still, is that the LeGo’s architecture allows for 40-60 fps in almost all situations using the „high quality“ preset. And once the planetside is loaded, it plays fluently.

I read quite much about how ok the game performs on the Steam Deck, and as the LeGo is more powerful in terms of hardware, I had little doubt that it could run it decently, which is true for everything but loading the planetside environments into the RAM.

Strangely, when I turn the quality preset to „low“, nothing improves. I get 30 seconds of loading planet surfaces on „high“ as well as on „low“ quality.
 
Finally, seems your 780M benches a little bit worse than even my ancient 1060 so I doubt you'll get much better than FPS in the 20s on a planet surface unless you turn everything down to Lego mode. If you do get a decent FPS let me know what settings you used...

Interestingly, as soon as the planet surface is loaded I can run around with constant 50 fps on high qualiry settings. Isn’t that weird? It’s just the abysmal loading time with the black screen while you just hear the footsteps that breaks the game flow. Otherwise, I honestly don’t experience any downgrade from the GFN rigs for this game.

Do you know the single settings that reduce the amount of preloading when disembarking?
 
Last edited:
Well, ya know what, that might just prove CPU<->RAM memory bandwidth IS the limiting factor on framerate once you're loaded in, because DDR5 @ 7200 is more than twice as fast as my DDR4 @ 3200. But that doesn't explain the people here that have 4k HDR running on DDR4 boxes...

If you're getting 50fps I'm a bit reluctant to suggest you mess with it now, that could easily put you back on 20...

Something's wrong though, I just found a guy on a Steam forum using integrated graphics on an Intel and he said 17 seconds for Odyssey disembark.
 
are you running any third party apps or overlays( eg steam) if so shut them all down
any instance where the sound continues but screen is blacked out is a sign of a gpu tdr
to test that run msi afterburner in the background and downclock the gpu core clock slider 50mhz but don't save the setting so that quits when the afterburner program is shutdown then see if that effects the black screen and the subsequent disembark time..............

(for the record i can run odyssey fairly successfully on as low an NIVIDIA GTX970 (with 4030mb of unshared VRAM) in an ATX tower format)
with a stripped bare bones version of win 10

any lower gpu is not advised imo

when running on any NVIDIA chipped laptop set this
nvidia control panel manage 3d settings global select cuda gpu tickbox is set to NVIDIA if you intend on gaming on it
 
Last edited:
Something's wrong though, I just found a guy on a Steam forum using integrated graphics on an Intel and he said 17 seconds for Odyssey disembark.

That would be totally ok for me, not great but ok. I have full 30 seconds every time, which is really making the game near to unplayable for exobiology.

Nothing is running in the background, just Steam. BTW I used its statistics overlay which says that VRAM is used up to 95%. Maybe I should try to increase the VRAM to 8 GB 🧐? But then only 8 GB regular RAM will remain.

I never had such a strange tech problem in my gaming history, where everything runs at high settings and FPS is good, but only one damn loading screen keeps getting in between, disregarding of how low I go with the other game settings.🤔
 
ya try it you can always set it back and as advised close all overlays they consume resources
without a stripped down version of windows the os does too
try afterburner also for a pontantial quick fix to the tdr problem which can simply be turned on off
 
Last edited:
Back
Top Bottom