Benchmarking help?

[Lays]

I just got my G1 970, and I want to benchmark it vs my 290x Lightning to see which is better, can someone help me out I have a few questions.

 

 

Where in these games should I compare the cards and for how long?  (Like which scenes, which parts of the game, etc?) Should I just use fraps to record the average FPS over that time-frame?

 

I have the following games:

 

Titanfall

BF4

BF3

The Crew

FC4

FC3

Call of Duty AW

Crysis 3

Bioshock Infinite

Dirt 3

Witcher 2

Tomb Raider

 

(I have a lot more games, but those seem to be the popular ones to benchmark)

 

 

All the testing will be at 1440p, I'm going to OC the cards as far as I can on air, and write down their overclocks etc.

[terrytek]

I think FRAPS and some monitoring software, like Afterburner can help. But it wouldn't hurt to try one synthetic benchmark once in a while.

[Lays]

I think FRAPS and some monitoring software, like Afterburner can help. But it wouldn't hurt to try one synthetic benchmark once in a while.

 

 

The problem is some synthetic benchmarks seem to like certain things more, IE: a 290x may score lower in things like Unigine for having slower VRAM, but in games they will perform the same etc.

[CoolaxGaming]

Well try valley, it will give you a score. Run it 5 times and record it each.

Then do the lightning exact same settings and bench 5 times.

Why 5? Because sometimes GPU's get angry at us users and don't perform as we'll.

Valley benchmark

[terrytek]

The problem is some synthetic benchmarks seem to like certain things more, IE: a 290x may score lower in things like Unigine for having slower VRAM, but in games they will perform the same etc.

Maybe it's just the program. Try something like FurMark or 3DMax.

[CoolaxGaming]

The problem is some synthetic benchmarks seem to like certain things more, IE: a 290x may score lower in things like Unigine for having slower VRAM, but in games they will perform the same etc.

Uniting is a game, well a game map running through scenes, it simulates a little harder then bf4. Don't get the score then, record fps

[Cheddle]

I just got my G1 970, and I want to benchmark it vs my 290x Lightning to see which is better, can someone help me out I have a few questions.

 

 

Where in these games should I compare the cards and for how long?  (Like which scenes, which parts of the game, etc?) Should I just use fraps to record the average FPS over that time-frame?

 

I have the following games:

 

Titanfall

BF4

BF3

The Crew

FC4

FC3

Call of Duty AW

Crysis 3

Bioshock Infinite

Dirt 3

Witcher 2

Tomb Raider

 

(I have a lot more games, but those seem to be the popular ones to benchmark)

 

 

All the testing will be at 1440p, I'm going to OC the cards as far as I can on air, and write down their overclocks etc.

 

as far as I know

 

Bioshock Infinite - built in benchmark

Dirt 3  - built in benchmark

Tomb Raider  - built in benchmark

 

You could use FRAPS for the rest, maybe MSI AfterBurner can log the FPS to a file?

 

The hard part is not choosing which part of the game, but making each run consistant... I am very interested in your results however :-)

[dragoon20005]

Dont forget Metro and Grid AutoSport

 

Metro is as punishing as Crysis 

 

 

Grid AutoSport is all about the highest fps

[Lays]

Dont forget Metro and Grid AutoSport

 

Metro is as punishing as Crysis 

 

 

Grid AutoSport is all about the highest fps

 

 

I don't have metro, but I have grid autosport

 

I'll do all that stuff for sure.

 

 

 

Should I just make a google spreadsheet for the results?

[dragoon20005]

I don't have metro, but I have grid autosport

 

I'll do all that stuff for sure.

 

 

 

Should I just make a google spreadsheet for the results?

google docs or normal excel sheet uploaded into google drive

 

yea good luck with the data collection

[Lays]

970 appears to be a tiny bit slower, but it literally makes NO NOISE at all.

 

I ran unigine for 15 mins, max temp 64c at minimum fan speed 1418 mhz without touching the core clock at all.

[Cheddle]

970 appears to be a tiny bit slower, but it literally makes NO NOISE at all.

 

I ran unigine for 15 mins, max temp 64c at minimum fan speed 1418 mhz without touching the core clock at all.

 

my 980 is about 20% slower than my old crossfire r9 290's and about seven-thousand times quieter... 

 

oh and wont have to format your PC when you update drivers with Nvidia cards!

[grayperview]

Hitman absolution

Arkham city + Origins

Just Cause 2

Sleeping Dogs

Dirt 3 + showdown

F1 20**

 

All have built in benchmarks, try some of these if you have them

[Najuno]

Don't forget that Unigine Valley isn't a good choice in terms of benchmarking a GM204 chip. If you want more accurate benchmark results use Unigine Heaven and 3D Mark FireStrike.

[Lays]

Don't forget that Unigine Valley isn't a good choice in terms of benchmarking a GM204 chip. If you want more accurate benchmark results use Unigine Heaven and 3D Mark FireStrike.

These are my results so far

https://docs.google.com/spreadsheets/d/1bZI8G3YIwAXkiKBvbaNE-mic2Pnoab_85pslbiIYP4w/edit#gid=0

 

Lightning was at 1175/1550, G1 was at 1582/7750

[Briggsy]

Don't forget that Unigine Valley isn't a good choice in terms of benchmarking a GM204 chip. If you want more accurate benchmark results use Unigine Heaven and 3D Mark FireStrike.

 

3Dmark also relies on the CPU more for the final score. can give misleading resluts.

[Lays]

3Dmark also relies on the CPU more for the final score. can give misleading resluts.

 

 

If I'm not changing the CPU at all, that shouldn't be an issue though correct?

[Najuno]

3Dmark also relies on the CPU more for the final score. can give misleading resluts.

Actually in the very highend benchmarking the CPU power is higher Valley than in 3D Mark. With my 3way GTX980 SLI in 3D Mark ULTRA the benefit from OCing my i7 5930K from 4.4GHz to 4.7GHz were 13 points but in Valley it made a difference of almost 200 points...