Jump to content

CPU / RAM/ VGA what sl likes More ? ((attempting to answer that ! ))


Nala Spires
 Share

You are about to reply to a thread that has been inactive for 2418 days.

Please take a moment to consider if this thread is worth bumping.

Recommended Posts

Ok why I am making this guide ?

Is simple really as long as I play sl there are people debating what sl benefits from

most like multiple threads / cpu speed/ ram speed / gpu.

Bellow I will try to find out what benefits more sl or at least to what degree

my system setup is this intel core i7 3770k motherboard with z77 chipset

intel 330 ssd 120 gb 8gb ram 1600 mhz fsb and Geforce gtx460 1gb @ 800mhz clocks.

And the software is Windows 7 professional 64bit firestorm Viewer v4.1.1.28744

the drivers I used are the latest and fraps for the benchmarking also the lightweight and the

heavyweight areas I stay at the same position exactly looking at the same angle .

 

I will use fraps to measure for 1 min fps on a lightweight place and then again on a really high traffic

area the lightweight area will be used so I can even spot differences that are small and high traffic will be mostly as a stress this is where you see how sl goes for you really .

Warning : I can't control high traffic areas so a difference up too 15% is more or less into the margin error and I will not really consider it important difference.

Warning 2 : My tests are not absolute … what I mean is that I might not doing it the right way since I am going pretty straight forward for it .

 

Full system baseline performance

lightweight area:

min:233

max:248

average: 240

 

high traffic area:

min:31

max:37

average:34,5

 

Ram speed from 1600 mhz under-clocked to 800 mhz fsb

lightweight area:

min:158

max:213

average: 200

 

high traffic area:

min:23

max:27

average: 25

 

Hyper-threading disabled

lightweight area:

min:236

max:252

average: 246

 

high traffic area:

min:31

max:35

average: 33

 

Geforce disabled and using igp intel 4000 instead (512 mb ram dedicated to it)

lightweight area:

min:34

max:42

average: 37

 

high traffic area:

min:16

max:19

average: 17.5

 

Geforce under-clocked from 800 – 500 mhz and ram from 1950 to 1500

lightweight area:

min:188

max:192

average: 191

 

high traffic area:

min:29

max:32

average: 31

 

 

cpu under-clocked from 3.5 down to 2.5 ghz

lightweight area:

min:128

max:169

average: 161

 

high traffic area:

min:20

max:24

average: 22

 

 

Conclusion :

Ok time for the conclusion lowering the ram to half speed looks like it had a huge negative effect to the second life second life loves memory bandwidth we suffer 20-30% less performance witch is way out of the margin of error today a 1600 kit 8 gb costs about 50 euros might worth to go for a 1866 mhz kit but ONLY if you find a good deal like 60 euros or something .

Hypperthreading disabled well I actually expected that result disabling it had no performance hit! The difference on the numbers are well inside the margin of error I can say safely that choosing 3770k or 2600k over 2500k or 3550k witch they cost 100 euros less will not bring you any clear performance advantage save your money and buy a processor from the i5 family instead of the i7 family it just not worth it even if those where absolute numbers .

Disabling the dedicated card and using the igp well the igp is a lot slower than the gtx 460 and the result on the lightweight sim (my skybox) where scary 70% proximate loss in performance I suspect thats cause of the bandwidth limitation since we are using systems memory here !

but actually the card hold her ground suffering a loss of 30-40% ONLY and I am saying only cause the difference of our dedicated card and the igp is through the space if you are really in tight budget you can count on an igp like that will not fail you!

Geforce under-clocked well I was trying to bring a category down my card giving a lot lower mhz to work with the results show that going for a super high end card probably will give you just a really really small advantage instead of going for a beast card might be better to stick with a mediocre card and go for mhz and ram speed thought the difference is clear on the lightweight sim I will say at a heavyweight sim (gor hub over 30 avatars front of me ) the difference is not noticeable and well it's even close to the error margin (( I can't control how many people are front of me they leave … more come but at any case they where more than 30 front of me enough to choke my card **Only uploaded images may be used in postings**://secondlife.i.lithium.com/i/smilies/16x16_smiley-wink.gif" border="0" alt=":smileywink:" title="Smiley Wink" /> )).

Cpu under-clocked by 1 Ghz surprised me great deal the fps really dived a lot more than I expected with up to 40% performance hit suffering I will say that don't go for the lesser clocked cpus those 20-30 extra euros will give you a good amount of performance and if you go for intel the potential of overclock on the k models looks quite strong !

 

At the end the few advices I have is that don’t go for a beast card a mediocre card like 450/460/550/560 will not give really big difference instead of going for a beast card invest on a good cpu with the head being 3550k witch have good igp for an intel if you are on tight budget you can have quiet good experience using the igp … don’t cheap out on ram going for 1333 instead of 1600 will be like 5 euros difference also if you can find ram of higher speed on a sweet deal might be a good idea to go for it …

what I'd love to test but I couldn't is SLI I know it benefits sl for some reason and I know there is a profile on nvidia drivers for it but I don't own a second card so Is impossible for me to test I hope you find my test interesting and I really hope I was able to help everyone who want new system for sl

WARNING

also my tests are not absolute , well don't build a system just for sl except if it's the only 3d application you are going to run.

 

My settings at all time where high quality with anisotropic filtering enabled 4x anti alising

enabled streaming VBO and Open gl Vertex Buffer also multi threading enabled

 

 

 

 

 

 

Bonus Lucid hydra MVP test

 

lightweight area:

min:238

max:253

average: 248

 

high traffic area:

min:30

max:36

average: 33

 

the performance is well inside the margin of error I don't see any gain from lucid hydra so just don't bother really

 

Bonus 2 Game booster test <-free program

lightweight area:

min:251

max:270

average: 263

 

high traffic area:

min:27

max:32

average: 30

even thought on heavyweight sim the performance dived worth a shot the program is free it might help you if you have problems with ram or slower machines let me know if you try it !

 

((i posted this originaly at the basic viewer  category cause i saw there most questions for hardware  but i think here is more apropriate not sure  thought ))

Link to comment
Share on other sites

  • 4 weeks later...

update now  the overclock resaults also !!!

3.5 ghz turbo core off 

min:31

max:35

average:33.5

4.3 ghz turbo core off 

min:36

max:42

average:39.7

thats 18,5% increase with 22% increase on my cpu bear in mind  +-5% margin of error i can't stop people from moving around **Only uploaded images may be used in postings**://secondlife.i.lithium.com/i/smilies/16x16_smiley-happy.gif" border="0" alt=":smileyhappy:" title="Smiley Happy" />

again fraps was used benching 1 min each time

Link to comment
Share on other sites

  • 2 years later...

Did you check to see if your vidoe card was running at 100%. A lot of the viewers are not recognized as a game so Windows runs the VC at 50% or so. Generally I have to make a game profile with the Nvidia Control Panel to get it working 100%. Try using GPU-Z and CPU-Z when you run these tests and it may give you some useful info especially where temps are concerned and usuage. Were you running Shadows and/or DoF? What was your viewer resolution? Can make a huge difference depending on your VC specs. The better the VC the less those extra features at high resolution will impact you.

Link to comment
Share on other sites

You are about to reply to a thread that has been inactive for 2418 days.

Please take a moment to consider if this thread is worth bumping.

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...