Live Performance Benchmarking Comparison for rFactor 2

Sort of but i'm really tired and not operating 100% atm. Would you mind re-writing my segment on this for me please if it's needed to address everything for those who need the info and in a clear concise manner. Thanks!

 

Basically, all IB-E and SB-E CPUs use X79 motherboards, and only X79 motherboards. Well, all X79 motherboards and all IB-E CPUs come with PCI-E 3.0 "out of the box". The SB-E CPUs, however, don't come with PCI-E "out of the box", and therefore only SB-E CPUs need the registry update.

You following me?

The only SB-E CPUs that exist are the 3960X, 3930K, and 3820K therefore only those CPUs need the registry update.

Got it?

 

Awesome - thanks for clarifying - there is no need for me to check if my combo is running @ PCIE - x3.0 since I have a i7-3770 then ... I was getting all worried

 

Yes thanks. I quote what you wrote into the OP as an additional followup to what TechAde said.

 

I see, thnx.

 

Yup. Basically your 3770K is the same "base" CPU as mine (4930K). Both are based on the Ivy Bridge architecture. The only difference is that the "E" models - with the exception of the 3820K and 4820K - have 6/12 cores/threads instead of 4/8 + the other "standard" "E-model" advantages like 40 PCI-E lanes instead of 16 (with the exception of the new 5820K [20 or 24 lanes]), quad-channel RAM, etc.

Being able to use 1 GPU + a soundcard, 2 GPUs, or 2 GPUs + a soundcard, while having the GPU/GPUs always staying at full 16x PCI-E (3.0) was one of my main priorities when i upgraded from my 2500K. The only way to achieve this is with the Intel "E" model line.

You don't even need to spend much money compared to a 2600K or 3770K because the 2600K's and 3770K's equivalent 4/8-core/thread "E"-model CPUs - the 3820K and 4820K - are the same price (give or take) as the 2600K and 3770K . What you gain, though, is 40 PCI-E 3.0 lanes instead of 16, not to mention PCI-E 3.0 (regardless of the amount of lanes) compared to only 2.0 (in the 2600K's case). Plus, a used 6/12 core/thread version is not much more expensive, if at all, than those other models previously mentioned, and can even be upgraded to at any time (don't need to change anything, not motherboard, not RAM, nothing).

 

Spinelli - it should be noted that I don't have a 3770k just a normal locked clock 3770 ... i'm not into over clocking and the 3770 was like $100 cheaper at the time (2012)

 

Min, Max, Avg
87, 126, 107.471

AMD 1100T @ 4.0GHz
GTX 770 2Gb OC
16Gb DDR3 @ 1.6GHz

 

Based on our prior conversation, i can see that your GTX 770 is underperforming but i reckon because of your cpu. Can you confirm whether or not one of your cpu cores was maxed out to 100% during the benchmark?

Also, can you check using GPU-Z if your card is running at PCI-e 2.0 x16?

 

one core maxed out and pci-e 2.0 16x


Inviato dal mio iPad utilizzando Tapatalk

 

Ok, thanks, good to know. Pretty confident it's just your CPU then.

edit: oh wait a minute. Your scoring higher than me. Huh?

Mine were:

Min, Max, Avg
77, 113, 97.240

I would ask that you double check all the settings you're using match perfectly with the screenshots in the OP and test again. Somethings not right here.

 

Yup, pretty much the same thing.

 

Didn't we mention this about the Difference between PCI 2.0 and 3.0 and you posted a video claiming no difference?

 

Sorry, by mistake I left special effects to low...
Below the results at full max settings (quite similar to yours and IMHO still very low for our GPU).

Frames, Time (ms), Min, Max, Avg
6582, 65053, 83, 118, 101.179

 

That's better but you're still performing a bit better than mine. What are your core clock and mem clocks during the benchmark? When i said stock clocks i should have said the nvidia factory clocks. But if you can tell me what your clocks were during the benchmark i can mimic them and see if i match yours.

 

Who are you talking to?

 

It's not on factory clock.
1190 gpu
7950 mems
from Asus GPU tweak

 

DrR1pper. You should do a test with your GPUs. See which one of the following makes a bigger difference to the difference between PCI-E 3.0 @ 16x and PCI-E 3.0 @ 8x / 2.0 @ 16x:

A. Less gfx strain but pushing out more fps
B. More gfx strain but pushing out less fps

Settings
1. EXACT settings for the benchmark we're already doing
2. Same settings for the benchmark we're already doing except 8xMSAA (instead of 4xMSAA), 12 AM (instead of 9AM), and Rain (instead of no rain)

P.S. Make sure you don't select 16xCSAA. In the NVidia control panel select "8x", or in Nvidia Inspector select "8xQ (8x multisample)". I'm sure you already knew that though . Finally, for rain juse set "Simple Weather" to "Raining", then underneath just set "Rain Chance" to "95%". Done


I discovered a sort of bug. The "Rain Chance" only adjusts in increments of 10. So if it's already set to 0, 30, 80 etc. (depending on the pre-set base you start with), then you can obviously selt it to 0 or 100 %, but if it's pre-set to 45, 75, etc. then the min and max you can set it to is 5 and 95 %. ISI needs to fix this (probably a 3 second fix).

 

I've tried this on a Titan, when running lowest possible in-sim gfx settings the impact of running at 2.0 rather than 3.0 is lessened quite dramatically i.e. 4% boost from 2.0 > 3.0 at lowest gfx settings, 25% boost at highest settings.

 

Ohh nice Techade, good stuff! We should really set up a new FINAL benchmark once and for all with 8xMSAA, 12 AM start time and rain (everything else should stay exactly the same as now i.e. absolutely maxed-out). The increased GPU graphics strain should make it even more obvious to people that this PCI-E problem exists, not to mention the fact that the higher graphical demands will make the benchmark relevant for a longer period of time.