Jump to content


Photo

Benchmarking Methodology


  • Please log in to reply
22 replies to this topic

#16 JustAnotherAntEater

JustAnotherAntEater

    Prisoner

  • Members
  • 3 posts

Posted 11 July 2015 - 03:16 AM

When I installed [Edit: the old version of] STEP, I did the benchmarking in between. But like an excited kid, I kept turning my head everywhere, admiring the new, previously unheard of level of detail.

Does that "spoil" my results? I definitely saw an increase in VRAM usage when looking around.

 

Thanks in advance!

 

Edit: Just noted that the new release differs from the old one quite drastically, so nevermind.


Edited by JustAnotherAntEater, 12 July 2015 - 03:55 AM.

  • 0

#17 Kayata

Kayata

    Prisoner

  • Members
  • 1 posts

Posted 23 October 2015 - 10:09 PM

Hi there,

 

I put my results since i finally complete the STEP Extended installation of the game.

I meticulously recorded the benchmarks for each step of the guide.

 

I generaly installed the medium quality version of the mods (ie 1024, 1k, "medium", "high rez"...) and the recomended tweaks for high quality provided by the guide.

At the "Display and Video card settings" part, i put the recomended settings, except for :

 

Ambient Occlusion setting: Performance

 

 

Here is my computer build:

502594configuration.png

 

Edit : I run SLI GTX 780M

If something is missing, please ask.

 

Attached File  benshmark_spreadsheet.xls   92KB   4 downloads

 

I hope this would help somehow...

 

Please, forgive me if i made a mistake somewhere.


Edited by Kayata, 23 October 2015 - 10:14 PM.

  • 0

#18 elwaps

elwaps

    Commander

  • Members
  • PipPipPipPip
  • 223 posts

Posted 18 November 2015 - 09:11 AM

I've just finished my installation of STEP Extended 2.2.9.1

 

Changes

- left out Dragons Shout with Voices as I like the beast like shouts

- installed Baseline everywhere except for dungeon stuff as my framerates are extremely high there

 

Rig

i7 3930K @ Stock

Gainward GTX 680 Phantom 4GB @ Stock

8GB DDR3 1333MHz @ Stock

Windows & Skyrim installed on SSD

 

Downers

I had finished most of the work already when I read about the standard deviation so unfortunately I only got min/max/avg frames and can't recalculate stuff as I haven't stored the raw frame data.

I also couldn't monitor the two memory values as GPU-Z and nV Inspector only offered one, this isn't so bad tho as I never reached full VRAM

Last but not least I didn't monitor the GPU clock

 

Combined values of four benchmarks I kept doing throughout the installation. First was the intro scene. Second was a walk from the exit of Helgen Keep to Riverwood using roads, stopping at the exit of the town. Third was a walk from Heartwood Mill to the Riften city entrace using the main road. Last was a walk through Solitude starting at the entrance, walking to the palace, walking back to the military base and back to the entrace.

stepperformance0combix5j13.png

 

Intro Scene Benchmark capped at 240 seconds by FRAPS

stepperformance1intro7zkre.png

 

I've tried to add the Excel file as attachment, unfortunately it is 108kB instead of 10.5, therefore you'll find it HERE. Don't be irritated, I'm just at installing REGS and added it to the tables, that's why it looks unfinished. STEP Extended itself is installed and benched 100%.

 

 

Combined values aswell as all four benchmarks as diagrams, you'll find the other spreadsheets in the Excel file

stepperformance0combieeus2.pngstepperformance1introssue5.pngstepperformance2riverkrunj.pngstepperformance3rift14u7v.pngstepperformance4solity9ua6.png

 

Added STEP Extended + REGS 4.0.6 (with minor adjustments as it was outdated)

regsperformance0combi9cu98.pngregsperformance1introvwufj.pngregsperformance2river3fu1q.pngregsperformance3riftcsubm.pngregsperformance4solitp4u8z.png

 

 

Another step - added Vividian ENB with the default preset aswell as the particle snow module (will have to remove it later as testing in snowy areas overloaded the script) and the cloud/fog module. Also added Climates of Tamriel, Supreme Storms and Enhanced Snow System aswell as ENB textures for Realistic Water 2, Moss Rocks and that weapon replacer mod...  :woot:

step-regs-enb_graph38lhi.png

 

 

step-regs-enb_tableyks4z.png


Edited by elwaps, 19 November 2015 - 08:34 PM.

  • 0

#19 JimBobMcKrakken

JimBobMcKrakken

    Prisoner

  • Members
  • 2 posts

Posted 24 November 2015 - 11:38 AM

Thank you for all the work you have put into the STEP project.

 

Attached is my benchmark for installing 2.2.9.1  It is missing step N, since you updated to 2.2.9.2 this mornnig, and that is where I left off. I'll try to get through the changes and re-bench everything for 2.2.9.2.

Attached Files


Edited by JimBobMcKrakken, 24 November 2015 - 11:41 AM.

  • 0

#20 elwaps

elwaps

    Commander

  • Members
  • PipPipPipPip
  • 223 posts

Posted 28 November 2015 - 06:50 PM

Just in case someone prefers putting in numbers the lazy way, without doing too much Excel stuff:

 

My Table - Clear

 

This is what the data input pages look like:

293sj0.png

 

This is what you have to do:

Use the intro scene as your benchmark and follow the STEP rules for benchmarking.

A) in the first column, just paste the content of the FRAPS file. The sheet will only use the first 235 numbers as with 240, the autosave with its framedrop would interfere.

B to E) are most comfortable to use with nV Inspector, if you're using it record Load, Clock, Temperature and Memory Usage and paste the four lines from your output file here

F) when building my setup, I was always measuring the time from clicking "launch" to the main menu showing up, ignore it if you don't care

G) same like F but for the time between clicking "new game" and the loading screen disappearing

H) describes the time needed for altering the mod order at that point and is mostly due to DynDOLOD but also the Bashed Patch, Reproccers and such.

I) is the folder size, just check it via Windows. I found it interesting to see my Skyrim grow over time and which mods contributed the most.

 

This is what the result looks like:

unbenanntxmsx5.png

 

For a new benchmark, just use the "1" tab first, then the "2" tab and so on. Give the benchmark a title on the "Benchmark" tab and feel free to rename the number tabs, the sheet will adapt.

For getting a proper chart, just hide the lines on the "Benchmark" tab (select the entire line(s), rightclick, hide) you don't yet use, else you'll get a chart with one tiny entry and 29 empty ones.


Edited by elwaps, 28 November 2015 - 06:53 PM.

  • 0

#21 Fonkin

Fonkin

    Prisoner

  • Members
  • 28 posts

Posted 29 December 2016 - 05:07 PM

Well, I'm trying to help, but I have a few questions.

 

I downloaded GPU-Z 1.16.0, but it doesn't have the options for 'Dedicated' and 'Dynamic' Memory Usage, like older versions do.  Do I need to use an older version of GPU-Z?

 

I noticed the last post to this thread was over a year ago, should I be posting my results here or elsewhere?  Is anyone still doing benchmark tests or cares about them?

 

Screen shots attached.

 

 

Attached Files


  • 0

#22 TechAngel85

TechAngel85

    Akatosh

  • Administrators
  • PipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • 12,051 posts

Posted 29 December 2016 - 07:08 PM

It's up to you whether or not you do benchmarking. It was Z's project, but he's unable to continue with it. At this point, I believe there is plenty of data collected for whatever he had in mind.

#23 Fonkin

Fonkin

    Prisoner

  • Members
  • 28 posts

Posted 29 December 2016 - 10:27 PM

Reading the STEP guide is where I found this project.  Thanks for letting me know it abandoned at the moment.


  • 0


0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users