Star Swarm - Oxide bench thread

andybird123 · 9 Feb 2014 at 13:43

Tonester0011 said:
Still an inconsistent bench.

http://forums.overclockers.co.uk/showpost.php?p=25813564&postcount=260
http://forums.overclockers.co.uk/showpost.php?p=25813990&postcount=270

Same settings, clocks etc. Using your method scores of 3956 and 3661.
This whole bench is a joke?

Better, gregs score with 3930k at 4.625 compares quite well with mine at 4930k at 4.6 e.g. should be slightly ahead
The highest fps one ends up scoring lower because its batch count is so low

Ive got results from 48 to 53 so ill do the maths on those and see what the scores look like

Tonester0011 · 9 Feb 2014 at 13:49

andybird123 said:
Better, gregs score with 3930k at 4.625 compares quite well with mine at 4930k at 4.6 e.g. should be slightly ahead
The highest fps one ends up scoring lower because its batch count is so low

Yes, but both of those scores i linked are Gregs with the same clocks and settings. Having a 10% swing on the same system is pretty...errr, yeah :eek:

Unless Greg had his Titan clocked higher on the low FPS results, then your way of calculating the scores might be valid.

andybird123 · 9 Feb 2014 at 13:52

Tonester0011 said:
Yes, but both of those scores i linked are Gregs with the same clocks and settings. Having a 10% swing on the same system is pretty...errr, yeah

Unless Greg had his Titan clocked higher on the low FPS results, then your way of calculating the scores might be valid.

It is because of the way the demo swings the camera around - we had a consistent bench set up but people complained the instructions in the Op were too complicated so Greg switched back to using the default one

Using the peak fps tells us nothing, but using some kind of scoring system does seem to help weed out the outliers

We need more people with more hardware making runs to give us a broader comparison

Tonester0011 · 9 Feb 2014 at 13:53

andybird123 said:
It is because of the way the demo swings the camera around - we had a consistent bench set up but people complained the instructions in the Op were too complicated so Greg switched back to using the default one

Using the peak fps tells us nothing, but using some kind of scoring system does seem to help weed out the outliers

ohhh, didn't know as i've not even bothered downloading let alone trying to bench it lol.

humbug · 9 Feb 2014 at 14:48

Official entr

Something wrong, i can't replicate my earlier bench, certainly not the DirectX one, i think its fair to say i got lucky with that one.

Overclocked.

CPU: Phenom II x6 1090T @ 4Ghz
GPU: 7870XT @ 1215 / 1625

DirectX 29.46
Mantle 58.63

Mantle performance increase +98%

DirectX

===========================================================
Oxide Games
Star Swarm Stress Test - ©2013
C:\Users\no one\Documents\Star Swarm\Output_14_02_09_1420.txt
Version 1.00
02/09/2014 14:20
===========================================================

== Hardware Configuration =================================
GPU: AMD Radeon HD 7800 Series
CPU: AuthenticAMD
AMD Phenom(tm) II X6 1090T Processor
Physical Cores: 6
Logical Cores: 6
Physical Memory: 8549396480
Allocatable Memory: 140737488224256
===========================================================

== Configuration ==========================================
API: DirectX
Scenario: ScenarioFollow.csv
User Input: Disabled
Resolution: 1920x1080
Fullscreen: True
GameCore Update: 16.6 ms
Bloom Quality: High
PointLight Quality: High
ToneCurve Quality: High
Glare Overdraw: 16
Shading Samples: 64
Shade Quality: Mid
Deferred Contexts: Disabled
Temporal AA Duration: 16
Temporal AA Time Slice: 2
Detailed Frame Info: Off
===========================================================

== Results ================================================
Test Duration: 360 Seconds
Total Frames: 10606

Average FPS: 29.46
Average Unit Count: 3852
Maximum Unit Count: 5512
Average Batches/MS: 398.10
Maximum Batches/MS: 878.69
Average Batch Count: 17380
Maximum Batch Count: 86385
===========================================================

Mantle

===========================================================
Oxide Games
Star Swarm Stress Test - ©2013
C:\Users\no one\Documents\Star Swarm\Output_14_02_09_1409.txt
Version 1.00
02/09/2014 14:09
===========================================================

== Hardware Configuration =================================
GPU: AMD Radeon HD 7800 Series
CPU: AuthenticAMD
AMD Phenom(tm) II X6 1090T Processor
Physical Cores: 6
Logical Cores: 6
Physical Memory: 8549396480
Allocatable Memory: 140737488224256
===========================================================

== Configuration ==========================================
API: Mantle
Scenario: ScenarioFollow.csv
User Input: Disabled
Resolution: 1920x1080
Fullscreen: True
GameCore Update: 16.6 ms
Bloom Quality: High
PointLight Quality: High
ToneCurve Quality: High
Glare Overdraw: 16
Shading Samples: 64
Shade Quality: Mid
Deferred Contexts: Disabled
Temporal AA Duration: 16
Temporal AA Time Slice: 2
Detailed Frame Info: Off
===========================================================

== Results ================================================
Test Duration: 360 Seconds
Total Frames: 21557

Average FPS: 58.63
Average Unit Count: 4507
Maximum Unit Count: 5595
Average Batches/MS: 673.28
Maximum Batches/MS: 6135.05
Average Batch Count: 19515
Maximum Batch Count: 103059
===========================================================

andybird123 · 9 Feb 2014 at 15:30

4x the number of batches going from 3.2ghz to 4 between your two mantle results
That is a 258% improvement from a 25% overclock

Hmmm

Also, why is your mantle log incomplete? No header info

humbug · 9 Feb 2014 at 15:58

Added the header to the Mantle results.

andybird123 · 9 Feb 2014 at 16:00

What is your cpu usage like running the mantle version?

humbug · 9 Feb 2014 at 16:01

I don't know.

humbug · 9 Feb 2014 at 16:25

humbug said:
Something wrong, i can't replicate my earlier bench, certainly not the DirectX one, i think its fair to say i DO NOT TRY AND OUT SMART THE SWEAR FILTER that one.

Overclocked.

CPU: Phenom II x6 1090T @ 4Ghz
GPU: 7870XT @ 1215 / 1625

DirectX 29.46
Mantle 58.63

Mantle performance increase +98%

DirectX

===========================================================
Oxide Games
Star Swarm Stress Test - ©2013
C:\Users\no one\Documents\Star Swarm\Output_14_02_09_1420.txt
Version 1.00
02/09/2014 14:20
===========================================================

== Hardware Configuration =================================
GPU: AMD Radeon HD 7800 Series
CPU: AuthenticAMD
AMD Phenom(tm) II X6 1090T Processor
Physical Cores: 6
Logical Cores: 6
Physical Memory: 8549396480
Allocatable Memory: 140737488224256
===========================================================

== Configuration ==========================================
API: DirectX
Scenario: ScenarioFollow.csv
User Input: Disabled
Resolution: 1920x1080
Fullscreen: True
GameCore Update: 16.6 ms
Bloom Quality: High
PointLight Quality: High
ToneCurve Quality: High
Glare Overdraw: 16
Shading Samples: 64
Shade Quality: Mid
Deferred Contexts: Disabled
Temporal AA Duration: 16
Temporal AA Time Slice: 2
Detailed Frame Info: Off
===========================================================

== Results ================================================
Test Duration: 360 Seconds
Total Frames: 10606

Average FPS: 29.46
Average Unit Count: 3852
Maximum Unit Count: 5512
Average Batches/MS: 398.10
Maximum Batches/MS: 878.69
Average Batch Count: 17380
Maximum Batch Count: 86385
===========================================================

Mantle

===========================================================
Oxide Games
Star Swarm Stress Test - ©2013
C:\Users\no one\Documents\Star Swarm\Output_14_02_09_1409.txt
Version 1.00
02/09/2014 14:09
===========================================================

== Hardware Configuration =================================
GPU: AMD Radeon HD 7800 Series
CPU: AuthenticAMD
AMD Phenom(tm) II X6 1090T Processor
Physical Cores: 6
Logical Cores: 6
Physical Memory: 8549396480
Allocatable Memory: 140737488224256
===========================================================

== Configuration ==========================================
API: Mantle
Scenario: ScenarioFollow.csv
User Input: Disabled
Resolution: 1920x1080
Fullscreen: True
GameCore Update: 16.6 ms
Bloom Quality: High
PointLight Quality: High
ToneCurve Quality: High
Glare Overdraw: 16
Shading Samples: 64
Shade Quality: Mid
Deferred Contexts: Disabled
Temporal AA Duration: 16
Temporal AA Time Slice: 2
Detailed Frame Info: Off
===========================================================

== Results ================================================
Test Duration: 360 Seconds
Total Frames: 21557

Average FPS: 58.63
Average Unit Count: 4507
Maximum Unit Count: 5595
Average Batches/MS: 673.28
Maximum Batches/MS: 6135.05
Average Batch Count: 19515
Maximum Batch Count: 103059
===========================================================

@ Rilot, what i meant was "fluke" as in got lucky, i miss spelt it, sorry.

Rilot · 9 Feb 2014 at 18:29

humbug said:
@ Rilot, what i meant was "fluke" as in got lucky, i miss spelt it, sorry.

Lol, apologies.
I've removed the infraction. What a mis-spell.

Geeman1979 · 9 Feb 2014 at 18:30

humbug said:
@ Rilot, what i meant was "fluke" as in got lucky, i miss spelt it, sorry.

Lol

humbug · 9 Feb 2014 at 18:31

Rilot said:
Lol, apologies.
I've removed the infraction. What a mis-spell.

Thanks

jayenh · 9 Feb 2014 at 19:45

Official Entry

CPU - 2600k @ 4.7GHz GPU R9 290X (MSI Gaming) @ 1000

Mantle

===========================================================
Oxide Games
Star Swarm Stress Test - ©2013
C:\Users\James\Documents\Star Swarm\Output_14_02_09_1914.txt
Version 1.00
02/09/2014 19:14
===========================================================

== Hardware Configuration =================================
GPU: AMD Radeon R9 200 Series
CPU: GenuineIntel
Intel(R) Core(TM) i7-2600K CPU @ 3.40GHz
Physical Cores: 4
Logical Cores: 8
Physical Memory: 8565600256
Allocatable Memory: 8796092891136
===========================================================

== Configuration ==========================================
API: Mantle
Scenario: ScenarioFollow.csv
User Input: Disabled
Resolution: 1920x1080
Fullscreen: True
GameCore Update: 16.6 ms
Bloom Quality: High
PointLight Quality: High
ToneCurve Quality: High
Glare Overdraw: 16
Shading Samples: 64
Shade Quality: Mid
Deferred Contexts: Disabled
Temporal AA Duration: 16
Temporal AA Time Slice: 2
Detailed Frame Info: Off
===========================================================

== Results ================================================
Test Duration: 360 Seconds
Total Frames: 22039

Average FPS: 61.22
Average Unit Count: 4488
Maximum Unit Count: 5899
Average Batches/MS: 1001.55
Maximum Batches/MS: 7243.31
Average Batch Count: 18730
Maximum Batch Count: 147856
===========================================================

DirectX

===========================================================
Oxide Games
Star Swarm Stress Test - ©2013
C:\Users\James\Documents\Star Swarm\Output_14_02_09_1922.txt
Version 1.00
02/09/2014 19:22
===========================================================

== Hardware Configuration =================================
GPU: AMD Radeon R9 200 Series
CPU: GenuineIntel
Intel(R) Core(TM) i7-2600K CPU @ 3.40GHz
Physical Cores: 4
Logical Cores: 8
Physical Memory: 8565600256
Allocatable Memory: 8796092891136
===========================================================

== Configuration ==========================================
API: DirectX
Scenario: ScenarioFollow.csv
User Input: Disabled
Resolution: 1920x1080
Fullscreen: True
GameCore Update: 16.6 ms
Bloom Quality: High
PointLight Quality: High
ToneCurve Quality: High
Glare Overdraw: 16
Shading Samples: 64
Shade Quality: Mid
Deferred Contexts: Disabled
Temporal AA Duration: 16
Temporal AA Time Slice: 2
Detailed Frame Info: Off
===========================================================

== Results ================================================
Test Duration: 360 Seconds
Total Frames: 15154

Average FPS: 42.09
Average Unit Count: 4476
Maximum Unit Count: 5693
Average Batches/MS: 691.77
Maximum Batches/MS: 1549.22
Average Batch Count: 18699
Maximum Batch Count: 136471
===========================================================

Mantle: 61.22
DirectX: 42.09

Thought I'd have a go at this as it's all I have with Mantle, I wanted to see how it went. I was impressed. My GPU usage was all over the place in DirectX, dropping mainly when there was a lot of stuff going on on screen. Mantle just seemed to steam through it at 100% usage the whole time.

andybird123 · 9 Feb 2014 at 20:40

I would love to test this myself on AMD hardware, only it refuses to recognise my card and insists on running on my intel integrated, despite setting both starwarm exe's to high performance

and even when Roy said it should be working

also, 7870XT 58fps, 290X 61fps and the batch count is higher on the 7870... anyone see something wrong with this picture?
are AMD basically doing themselves out of selling anything higher than a £200 GPU?

tommybhoy · 9 Feb 2014 at 23:27

andybird123 said:
7870XT 58fps, 290X 61fps and the batch count is higher on the 7870... anyone see something wrong with this picture?
are AMD basically doing themselves out of selling anything higher than a £200 GPU?

290X/3770K

== Results ===========================
Test Duration: 360 Seconds
Total Frames: 24453

Average FPS: 67.92
Average Unit Count: 4809
Maximum Unit Count: 5969
Average Batches/MS: 1229.23
Maximum Batches/MS: 3033.72
Average Batch Count: 21846
Maximum Batch Count: 119116
=====================================

V

Humbugs Phenom II x6 1090T@4Ghz/7870XT@1215/1625

== Results =============================
Test Duration: 360 Seconds
Total Frames: 21557

Average FPS: 58.63
Average Unit Count: 4507
Maximum Unit Count: 5595
Average Batches/MS: 673.28
Maximum Batches/MS: 6135.05
Average Batch Count: 19515
Maximum Batch Count: 103059
======================================

6 core Phenom v's hyperthreading???

I don't think AMD is doing themselves out of selling anything higher than a £200 GPU, I think Mantle is doing what it's designed to do-push AMD cpu's, just like games, max fps can spike, it's the average that counts.

290X pulls in 16% more fps, wonder what he would get with my 290X in his rig. :eek:

Humbug, post me your rig up to test it out.

humbug · 9 Feb 2014 at 23:57

I don't know whats going on, what i do know is Mantle gives me very constant Frame rates, i'm not getting higher frame rates in Mantle vs DX, its simply that in DX it grinds to a halt when ever there is something going on, as low as 10 FPS.
with Mantle the screen can be a wash with ships and stuff going on and the Frame rates don't take nearly as bigger hit. it powers through it no problem, it never really drops below 40.

But its not 'that' consistent, I ran it multiple times to try and get it past my original score of 57, it took a while, and there seems to be about a 10% variation if you run it enough times.

The only thing left for me to try now is to set my hardware to jihad mode, see how high i can get it.

Edit; @ tommybhoy, the x6 does do better in this than my i7, not a great deal. but certainly better, which is why i use it for this.

jayenh · 10 Feb 2014 at 09:28

Yeah, I noticed no 2 runs are the same. On one of my runs since my entry post the ship the camera follows hit, and briefly got stuck on, one of the larger ships/frigates. It's like it was trying to slide off it. This seemed the throw the whole run off and gave me a much higher average FPS because it was rarely flying towards/through any action.

smilertoo · 10 Feb 2014 at 12:00

I read somewhere that this demo was deliberately slower on non mantle than it could be, there was optimizations they could have done but it's supposed to show mantle in as good a light as possible.

Silent_Scone · 10 Feb 2014 at 12:24

smilertoo said:
I read somewhere that this demo was deliberately slower on non mantle than it could be, there was optimizations they could have done but it's supposed to show mantle in as good a light as possible.

Of course there is, you'd be naive to think it's not. It is a Mantle showcase at the end of the day, why on earth would they optimise it properly on DX

.