Warframe Stress-Test Easter-Egg

[[DE]Glen]

Do you want to stress-test Warframe and look for leaks? This video will show you how!

(NOTE: windowed or borderless-fullscreen is recommended.)

1. Show your in-game FPS counter (so you can see how much RAM the game-code is using)
2. Quit the game and relaunch (clean-runs for science!)
3. Login as "corpus.outbreak"
4. Record the in-game RAM value and the Commit Size for Warframe in Task Manager.
5. Let it run for a fixed amount of time (30 minutes or an hour should be fine)
6. Compare memory values from the start.

Some amount of growth is fine but unbounded leaking is not. If the in-game RAM counter shows a leak, it's much easier to fix! Let us know!

If only the Commit Size in Task Manager grows, you can try a few other things:

1. Disable (or uninstall) in-game overlays or streaming tools
2. Try the latest graphics driver
3. Try an older graphics driver
4. Try changing graphics settings (PhysX Effects, Particle Quality)
5. For older graphics cards try different DirectX (launcher settings)

[DaftMeat]

Awesome

Edited March 9, 2017 by DaftMeat

[YouHaveShamedYoFamory]

cool

 

[lightdragon64]

 Neat. I'll have to try this when i get home.

 

Update: seems like it is mostly CPU usage that it messes with in the stress test. starting off it was sitting at around 20% usage and then the game alone boosted up to about 40-50% usage and fluctulated between that for the rest of the time.

 

overall interesting process to watch.

 

Specs:

GPU: Gtx 1080
CPU: intel core i7-67005 @4.00GHz
Ram: 32GB
display with:1920x1080 (have 3 monitors, may set them up to test ultra wide stress testing)

 

Edited March 10, 2017 by lightdragon64

[trunks013]

Most of the time i have the usual svchost leak but warframe didn't made leak in ages for me .

[Sannidor]

Can we have some other secret test "emails"? For science 

[Lawmonark]

I find it funny, Warframe runs better than minecraft on most computers. Minecraft currently uses about 4GB when modded.

I always use this "Fps counter" and i have only seen a few "suspicious" areas... mostly in infested areas, volumetric lighting areas...ect.

Next time i see this issue, i will screenshot the area.

 

 

Also. i can not be the only one that wants an endless mission like this. Were we must kill "X amount" of enemies each round. This outbreak makes me want it more. "Self plug", i have a thread in fan idea thread.

Edited March 9, 2017 by Krhymez

[Valiant]

Reminds me of how I kinda prefer the old loading screen. It might have been right in your face but it seems to have a better background.

Edited March 9, 2017 by Naith

[Salenstormwing]

Other overlays that folks might miss are stuff like Raptr which has some weird overlay stuff as well. I know I had to turn off some of that normally, so folks who have that might want to remember to turn that off too.

[Xekrin]

It also makes an amazing screen saver!

[Kylo.]

Gotta Test that.

[Deathbricked]

For the record, grineer.outbreak is also a thing, could find anything with sentient.outbreak or orokin.outbreak

[Pyus]

21 minutes ago, Krhymez said:

I find it funny, Warframe runs better than minecraft on most computers. Minecraft currently uses about 4GB when modded.

I always use this "Fps counter" and i have only seen a few "suspicious" areas... mostly in infested areas, volumetric lighting areas...ect.

Next time i see this issue, i will screenshot the area.

 

 

Also. i can not be the only one that wants an endless mission like this. Were we must kill "X amount" of enemies each round. This outbreak makes me want it more. "Self plug", i have a thread in fan idea thread.

MC has rendering problems with chunks. If they rendered using schrodinger's cat as a methodology it would take up much less resources. That is to say it should only render visible blocks and use averages to render as they become visible. Is the next block coal or smooth stone, it has potential to be all types until it is revealed. Then based on it's type it could alter the rule-set for nearby unrevealed types. Coal is generated in veins, so perhaps the next block revealed has a +30% chance to be a coal block. 

Another way to look at it, we perceive our 3d world as 2d, we are only able to interface with the surface of any object. We have built tools (radar, ekg, sonar) to be able to map 3d objects. If you cut an apple in half, you are now looking at the 2d surface of two halfs. If you drill into an apple and look inside, you are now peering down the internal surface of a cylinder. Even smell and sound requires surface area contact with our sensory receptors. 

 

That all said, yeah, Warframe runs exceptionally well and looks great doing it. I can provide criticisms all day on balance and content, but the sheer graphics and responsiveness of Warframe is outstanding. 

 

Edited March 9, 2017 by Pyus

[tnccs215]

Oh my.
That voice.

[Kalvorax]

hmm going to test this with my new MB and CPU once i install them.....wonder if there are other tests we as users can do.

[Wevi]

Trial loading screens?
 

[corrosivePaladin]

I can tell you right now. Ive passed out with my game on for 28 hour straight this week. Ram stays consistant at 730ish mb with Borderless Windowed. I also keep a Performance Monitor to keep watch on my cpu as I still use a really weak i3-3220 to avoid pushing it too hard.
I haven't seen a leak in at least a year.

[DaftMeat]

1 hour ago, Naith said:

Reminds me of how I kinda prefer the old loading screen. It might have been right in your face but it seems to have a better background.

Now that you mention it, I would love "Outbreak" as an active login screen. AI actively fighting in the background as Warframe loads up asking for your login info would be amazing.

Edited March 9, 2017 by DaftMeat

[Nep_Blanc]

This is cool I'll let it bake for a while but I am pretty sure that nvidia screwed up their latest drivers so I stay on an older driver.

 

Chao, The Roaring Lion

[(XBOX)Aflem1]

Jedi Lord GLEN <3

 

[SonicSonedit]

Glen posts are the best posts.

[dstrawberrygirl]

A quick question relating to fullscreen vs windowed - getting anything other than fullscreen to run correctly with GSync has proved to be pretty darn tricky (and often results in 60fps, no matter how much I try, hence I just run fullscreen these days). Are you interested in gsync test results? Are there any specific recommendations for gsync tests? I have had regular graphics card crashes on WF for the past couple of months and I'm hoping to figure out if it's the game or my system / graphics card. Done the usual DDU clean wipes and reinstall of drivers, done the reinstall the whole game thing, now looking at other factors.

[[DE]Glen]

1 hour ago, dstrawberrygirl said:

A quick question relating to fullscreen vs windowed - getting anything other than fullscreen to run correctly with GSync has proved to be pretty darn tricky (and often results in 60fps, no matter how much I try, hence I just run fullscreen these days). Are you interested in gsync test results? Are there any specific recommendations for gsync tests? I have had regular graphics card crashes on WF for the past couple of months and I'm hoping to figure out if it's the game or my system / graphics card. Done the usual DDU clean wipes and reinstall of drivers, done the reinstall the whole game thing, now looking at other factors.

The problem is the graphics driver gets kicked out when you toggle so you can't really tell if it's leaking commit size while you're full screen because you can't see task manager. You can try but it might mess up your results.

If you're really committed to testing in native fullscreen you can run the soak for an hour and then quit and check the EE.log -- there will be periodic logs that end in [Heap: 553,973,776/639,565,824 Footprint: 2,209,800,192 Handles: 757] -- the first number is the value we show in the in-game FPS value and the Footprint number is what task manager would show for Commit Size.

[[DE]Glen]

Using this tool volunteers have confirmed that there is in fact a memory leak related PhysX effects (Commit size, not heap)

We're investigating internally however there have been claims that this was introduced several driver-versions ago. It would be helpful to get data to confirm the exact driver it was introduced with.

If you've rolled back your drivers please run a test (with PhysX effects on of course) and report back with your GPU Model, Driver Version, and how much it leaked in a 30 minute run.

Also, if you could kindly keep this thread on topic I would appreciate it -- I'd like to stay focused on fixing issues and not have to dig.

Edited March 9, 2017 by [DE]Glen

[[DE]Glen]

So far I have a few data-points for the PhysX Effects leak:

378.78 on GTX 950 leaks

378.66 on GTX 950 leaks

378.48 on GTX 980Ti SLI or GTX 1080 leaks 

Edited March 9, 2017 by [DE]Glen