Testing CPU bottlenecks using multiple CPUs and a GTX 1080ti

[(G{in}[AK)TION]]

Well this turned out to be a fiasco

Guess ill just do it the same exact was as i did before!

 

[yoadknux]

I think this is a good and interesting idea. Keep posting results

 

[(G{in}[AK)TION]]

I feel like this is taking forever but at the same time i should continue to take my time with it no matter how long it takes.

Anyone else feel the same way? or am just really slow at this? I have learned so much and the more i learn the more it makes me want to do it even more carefully.

 

[wingman99]

Thanks for taking your time it is better to know that you are bench testing the best you can, take all the time you need. Also what have you learned for more accurate testing?

 

[(G{in}[AK)TION]]

For single player, find a part of the game where you can repeat it over and over without fail. But here is where i get stuck. I can either do it twice and then find the median between both runs or just pick the lowest result i get.

for multiplayer, record gpu and cpu usage and FPS. Play a full round as you normally would.

 

[EarthDog]

Earlier it was mentioned to use an average. We run 3x here. I wouldnt take the lowest result as that isnt indicative of average performance.

 

[(G{in}[AK)TION]]

SO basically take the results from a run that stands in between the results from the other two runs done?

 

[wingman99]

For single player, find a part of the game where you can repeat it over and over without fail. But here is where i get stuck. I can either do it twice and then find the median between both runs or just pick the lowest result i get.

for multiplayer, record gpu and cpu usage and FPS. Play a full round as you normally would.

You are making very good effort to have accurate bench tests consistency. To average 3 runs. Add the 3 runs of FPS to gether, then divide by 3.

 

[EarthDog]

To average 3 runs. Add the 3 runs of FPS to gether, then divide by 3.

This. An average.

 

[(G{in}[AK)TION]]

For bf1, i am retesting at the beginning of mud and blood starting where you shows you in the tank and then ending the bench mark right before it goes to the cut scene where finch comes out of the tank.

I noticed that while during game play if i zoom at all, FPS goes all the way up to 200. Should i avoid zooming in at all in order to record actual max FPS?

 

[EarthDog]

Whatever you do, do it the same each time has been (one of) the underlying point in this thread.

That said, I suggest NOT zooming in as it seems this activity seems to skew things.

 

[(G{in}[AK)TION]]

Earth dog i did as you advised. How does it look? i also "averaged" cpu and gpu usage as well.

 

[(G{in}[AK)TION]]

I am having some mixed feelings on whether or not i should really test the core 2 quad q9650 as it is as it only has 8gb of ram instead of 16gb of ram.

Unless anyone has a kit to spare, i think i may as well test it as it is...or find another place where i can get 16gb of ram from at. i mean on newegg, its 167 dollars for a 16gb kit of ddr2-800 memory.

EDIT: WHOOPS! its actually 167 dollars for 8gb of ddr2 memory and 334 dollars for 16.

 

[Alaric]

I would think that 8 GB or 16 GB doesn't matter if you don't use all 8 GB anyway. If you only use say, 6 GB, then it doesn't matter if you have 16 or 128. The only time RAM matters is when you run out. LOL

 

[EarthDog]

Spot on, alaric.

I recall however a post here where he said he was using more in some titles. And with that, we determined the testing would be bunk already. I think Gin forgot again.

 

[(G{in}[AK)TION]]

No I did not. I reteseted the 7700k with 16gb in all games and benchmarks this time with ddr4 3200 memory. My concern with the q9650, as you may already tell, is price. Just as I was reading yournposts I was thinking of raiding 3 ssds together for raid 0 and throw the page file on there. Or I that not a good idea?

 

[wingman99]

If you can't get more ram for the 2 quad q9650 just post the games using less than 8GBs of ram.

 

[EarthDog]

No I did not. I reteseted the 7700k with 16gb in all games and benchmarks this time with ddr4 3200 memory. My concern with the q9650, as you may already tell, is price. Just as I was reading yournposts I was thinking of raiding 3 ssds together for raid 0 and throw the page file on there. Or I that not a good idea?

Not a good idea..its still exponentially slower.

You can put the data up, but make sure its listed that the result is with 8gb and paging out so people are aware the type of data it is (only good for judging performance on a system that pages oit in those titles).

 

[(G{in}[AK)TION]]

i already posted on the first post, but i will also post that in the excel spreadsheet as well.

 

[(G{in}[AK)TION]]

Back on the main computer. Resuming testing as normal.

I feel tempted to test with AMD desktop processors since there is only 1 AMD FX-8350 in there...