CPUs compared clock for clock. FSXmarkCPU

[ArjenVdv 190]

Those are some very interesting results. It looks like Haswell is a worthy upgrade, but I don't have the budget right now.

I am willing to do the test with my Sandy i5 2500K, however I am not willing to reinstall my FSX.

I know from my experience that most add-ons do not affect FPS. Last time I installed FSX I made that sure by checking my FPS in my test flight scenario everytime I installed an add-on. Add-on aircraft and all that don't influence FPS for the default planes.

So again, I am willing to revert my REX backup, FTX Global backup, disable UTX, disable add-on airports, default fsx.CFG, and do the test. I think this will be valid enough because I will have default scenery and textures. 

[odourboy 1,289]

This is a most worthy undertaking and I'd like to contribute. I have an i5 IB and I see that you only have one data point and the results for that group may be in question.

 

My install isn't brand new, but it's very recent - like a week - and I've documented every change and addon as I applied them, so I'm pretty sure I can restore it to equivalent to default. I have on concern though - and that is meeting the ram timing requirement: 1600MHz 9-9-9-28 1T RAM

 

I have 8 GB of g.skill ripjaws x 1866MHz - 9 10 9 28 2T. Will setting the timing as required destabilize my system? (I'd hate to have to reset my motherboard.)

[odourboy 1,289]

Okay, I have some results, but there is a caveat - I was afraid to alter my RAM timing profile, so I lowered the clock to 1600MHZ, but left the timing as 9 10 9 28 2T. I presume the impact is a possible very slight hobbling of my FPS.

 

CPU - I5 3570K; GPU GTX680

Frames, Time (ms), Min, Max, Avg

  9682,    240000,  26,  58, 40.342

  9528,    240000,  26,  57, 39.700

  9480,    240000,  27,  57, 39.500

  9429,    240000,  26,  57, 39.288

  9640,    240000,  26,  58, 40.167

  9501,    240000,  26,  57, 39.588

 

If you don't want to use the data, I understand.

[SAAB340 28]

 

Okay, I have some results, but there is a caveat - I was afraid to alter my RAM timing profile, so I lowered the clock to 1600MHZ, but left the timing as 9 10 9 28 2T. I presume the impact is a possible very slight hobbling of my FPS.

 

CPU - I5 3570K; GPU GTX680

Frames, Time (ms), Min, Max, Avg

  9682,    240000,  26,  58, 40.342

  9528,    240000,  26,  57, 39.700

  9480,    240000,  27,  57, 39.500

  9429,    240000,  26,  57, 39.288

  9640,    240000,  26,  58, 40.167

  9501,    240000,  26,  57, 39.588

 

If you don't want to use the data, I understand.

 

Thanks for running the benchmark. I won't add it to the chart as it doesn't use the correct RAM timings and as you say, the result will be slightly lower due to that.

 

You have confirmed that it was right to withdraw the previous i5 Ivy result as they were too low. Most likely with the CPU running @ stock speed. Your i5 Ivy with slightly worse RAM is slightly slower than the i7 Ivy. It looks like i5 performs the same as the i7 with Ivy as well.

 

I wouldn't be afraid to run your RAM at 9-9-9-28 1T, but I fully respect if you don't feel comfortable at doing that.

 

You see, the timing number is just a number. It has no unit. It is used together with the RAM frequency to define a latency/time that your RAM is rated for. You get that time expressed in nano seconds (ns) by multiplying the timing with 2000 and dividing with the frequency. So the rated latency of your RAM is:

9.65ns-10.72ns-9.65ns-30.01ns

The benchmark rated latency is:

11.25ns-11.25ns-11.25ns-35.00ns

As you can see, your RAM is actually rated to handle faster/shorter latencies than what CPUmark requires.