Framerate steadily lowers in missions, never recovers

[Jashoxu]

Specs:

OS: Windows 10

CPU: Intel i5 4590

GPU: Radeon HD 6750 Sapphire

Ram: 8 Gigs

 

Problem:

Game starts off at a smooth 120 fps, occasionally dipping to sub-90, but still far over 60. Getting into a mission steadily but surely drops my framerate down to below 30 and in some cases, into the single digits. Upon returning to the Orbiter, the framerate stays as low as it was in the mission. Nothing seems to be able to restore it besides relogging.

Things I've tried:

Running in DX9. DX10. DX11.

Running in both 32 and 64-bit modes

Verifying on Steam

Verifying / Optimizing through the launcher

Turning all graphics settings to absolute lowest

Setting graphics to default settings

Capping/Uncapping FPS

Turning VSync On/Off

Lowering Resolution

Playing in Windowed

Restarting PC

Updating Drivers

Rolling back Drivers

Selling Nova

Buying Nova Prime

Selling Nova Prime

Farming Saryn Prime

[carnaga]

On 28.3.2016 at 7:36 PM, Jashoxu said:

-snip-

Use GPU-Z and analyze what causes the problem when you play warframe and when the issue occurs. Is it a heating issue? Thermal Throttling? Voltage? GPU load? Memory Controller? and so on.

[Jashoxu]

Right, did as you asked and noticed a trend. When the game starts up, the GPU load hovers at around 60 - 70%. There's a small break when the mission loads, then once in the mission the GPU load stays as it was before. The problem occurs in the second break when the mission is over and I return to the Orbiter. The GPU load flatlines at 0% with intermittent 100% load spikes.

 

This is from a mission on Mercury, Caloris. The pattern holds true for ALL missions that I've tested, at least 3 on every planet and all tiers of the void.

Edited April 2, 2016 by Jashoxu

[carnaga]

40 minutes ago, Jashoxu said:

-snip-

I think you should contact DE support to solve the issue more efficiently cause you can attach EE.log file to the support ticket (do not post EE.log file to the forums). And if I was you: before contacting support I would reproduce the issue. When the issue occurs I would write to squad chat "dibadaba" (or use your imagination regarding the "keyword"). Your text input will be logged to the EE.log file and support can search the keyword and locate the proper section when the problem occurs. Also providing DxDiag txt file would help support to understand more about your computer rig (peripherals, driver versions and so on) . 

Also with GPU-Z I would take a complete screenshot of all bars, not just a single one - It would help a-lot to see how the rest of the graphs react (or how they do not react).

Here is a sample screen capture of my Draco run - having some performance issues there: on the middle of the "Round 2 Draco (4 ppl)" you can see a small dip with GPU Core Clock and VDDC Voltage. At the same time PrefCap is changed from "Limited by GPU utilization" (blue) to "Limited by reliability voltage" (grey).