Jump to content
Sign in to follow this  
Guest

3DMark Results 5960X 2x Titan X and 3960X 1x 970

Recommended Posts

Guest

I get a lot of questions about PC performance in P3D V3 and it's very hard for me to judge (close to impossible) how other systems are going to work in P3D V3.  So I thought I'd post my results using my main FS PC and my Test FS PC so folks could provide a "relative" reference point.

 

3DMark has 3 Fire Strike testing modes (base 1.1, Extreme 2K, and Ultra 4K).  I ran tests for both Base 1.1 and Ultra 4K with variou CPU and GPU OC settings.

 

Main FS PC 

5960X - 16GB 3000Mhz RAM - 2x Titan X - SSDs

 

Test FS PC

3960X - 32GB 2133 Mhz (being reported ast 3300Mhz?) - 1x GTX 970 - SSDs

 

Detail results are here with details below: http://robainscough.com/3D_Mark_Results.html

 

Basic summary:

 

e53e4d86f2ec3487d87b0e4de1ca9f3d.jpg

 

These don't necessarily translate into how well a system will run P3D, but it can be used as a generic reference point.

 

Cheers, Rob.

Share this post


Link to post
Share on other sites

Try to do a set of filés as FSMark11 for P3D V3 with some kind of STD settings from you.

Firestrike no no, gpu part only gpu CPU only CPU all cores and treads 100% and combined it say what it is.

High score firestrike 5960x and 4 way SLI

 

CPU score. 5960x 20-23k, 3960x 15-18k, 6700k 13-16k dont think a Old 3960x @4.6ghz outpeform a 5ghz Skylake as it do in Firestrike even if both run single 970.

 

In fsx FSKMark11 a 6700k @5.0ghz give avg 60-64fps avg. depending memspeed to a 5960x @4.7ghz 54-56fps avg with differnt memspeed up to 3400mhz.

Share this post


Link to post
Share on other sites
Guest

Hi Westman,

 

FSXMark11 is IMHO not a very good way to test one's PC performance (be it FSX or P3D).  It attempts to restriction resolution to 1680 x 1050 which is essentially skewing the test to be more CPU bound and will not expose the benefits or weaknesses of GPUs that don't suffer as high an FPS reduction at higher resolutions ... which is even more critical for DX11 P3D.  

 

P3D V3 has moved more of the CPU tasks over to the GPU, many P3D Settings are not represented in the FSXMark11 test, including heavy GPU tasks like Tessellation, Volumetric Fog, Various shadow options (cloud shadows), 3D waves, HDR, additional reflections, different LOD Radius, autogen changes, bathemetry, etc. ... some of these are VERY GPU intensive and options not in FSX. 

 

It also relies on the Boeing 737-800 which is not available in P3D.

 

But to be clear, I'm listing my results (and I hope other's share) as a "relative" indicator of PC performance it is NOT a relative indicator of how well one's PC will work in any given Flight Sim ... but it's a starting point.

 

I do think using a Flight Plan and autopilot with a set scenario is a good idea as the replay system has issues in P3D V3.  However, FSXMark11 would need to be "adjusted" to be useful and the screen resolutions being tested (and results) should go from a more traditional 1920 x 1080 to 3840 x 2160 as they will give a better indication of how well a GPU is doing at lower and higher resolutions.

 

Cheers, Rob.


CPU score. 5960x 20-23k, 3960x 15-18k, 6700k 13-16k dont think a Old 3960x @4.6ghz outpeform a 5ghz Skylake as it do in Firestrike even if both run single 970.
 

 

Not sure I understand what you mean here?  My results are showing 3960X and 5960X ... using 3960X as the "baseline" for % change calculations (green).  I have no idea what score results are for 6700 at 5Ghz with a 970 are using Fire Strike 1.1 and Fire Strike 1.1 Ultra.  Would be great to see someone post results with VALIDATED score so they can be cross referenced online.

 

Cheers, Rob.

Share this post


Link to post
Share on other sites

You have evrything on the futuremark ORB site.

Create a acount and start digg.

 

But i dont think Firestrike is a good indicator for a Esp 32bit based app the CPU part gone be way off .

Share this post


Link to post
Share on other sites
Guest

LM have moved ESP along considerably from it's DX9 FSX SP2 days ... V3 has transitioned as much as possible from the CPU side to GPU side ... P3D is also using CUDA libraries (3D Waves) which FSX doesn't use at all.  

 

So it would be unwise to assume ESP and P3D V3 are the same.

 

P3D is using an older CUDA library (v4.x) and I'm hoping (I think they will) LM update that to more recent CUDA libraies (5.x) ... there are some SLI issues with the older CUDA libraries.

 

But again, Fire Strike wasn't intended to be an FSXMark11 ... it's primarily to give a basic reference because it does a very good job at testing both GPU and CPU performance ... what it can do is:

 

1.  Validate your PC OC if it can complete these tests, especially the Ultra in several iterations (many people have incorrect OC's and often can't complete these tests).

 

2.  Relative scores in the test should give people a rough idea how their CPU/GPU will perform.

 

But Fire Strike's Physics test is aimed squarely at CPU performance -- see documentation here: http://s3.amazonaws.com/download-aws.futuremark.com/3DMark_Technical_Guide.pdf  page 32

 

Quoted from Physics Test documentation:

 

 

The test has three levels with different workloads. The first level is the lightest and the last is the heaviest. The purpose of the three levels is to extend the performance range for which the test is relevant.

 

The physics test is run with a fixed timestep at 30 fps. The physics test always begins with the first level and continues to the next level until either the test is finished or 90 seconds have passed. If a level did not finish completely, then it will contribute proportionally less to the final score. The final score is a weighted sum of all levels.

 

The first level of the test has 8 simulation worlds running in separate threads. Each world has one soft body with 107 vertices and 64 rigid bodies. The rigid bodies are invisible and are there to cause the blast effect to soft bodies. Additionally, there are 32 CPU simulated particle systems with about 500 particles in each.

 

The second level of the test adds 8 simulation worlds running in separate threads. Each new world has one soft body with 499 vertices and additionally 64 rigid bodies.

 

The third level of the test adds another 16 simulation worlds running in separate threads. Each new world has one soft body with 499 vertices and additionally 64 rigid bodies. Each of the 32 worlds is simulated at 30 fps and the whole test takes 16 seconds at 30 fps. The first level starts at time 0, the second level starts at time 5 and the third level at time 10.

 

All physics are computed on CPU. Soft body vertex data is updated to GPU on each frame. The Bullet Open Source Physics C++ Library version 2.83 alpha is used for physics computation.

 

So I'll have to disagree with you that this test is entirely GPU based, it's most definitely NOT and it will also thread across CPU cores to further stress multi-core CPUs.

 

Cheers, Rob.


As you can see from the details of Fire Strike test I'm hitting 60C during the Physics test with all my cores maxing out at 4.5Ghz (about 50C at 4.375Ghz) ... it's a CPU punishing test:

 

74af55d3f611817d94946f3058129089.jpg

 

At 4.375 GHz (5960X) - 50C during Physics Test:

 

f9d2bf48288aa15e92f0ef83e0acccf4.jpg

 

Cheers, Rob.

Share this post


Link to post
Share on other sites

Rob,  I cant seem to find them 2 pages you have posted in My Results,  were are they hiding.  I have looked everywhere,  what am i missing  

 

EDIT

 

Yaaaa,  I found it  


 

 

 

Share this post


Link to post
Share on other sites
Guest

Rob,  I cant seem to find them 2 pages you have posted in My Results,  were are they hiding.  I have looked everywhere,  what am i missing  

 

Are you talking about the 3DMark results?  

 

If so, when the 3DMark test completes you should have a results screen and it "may" launch a "View Online" results screen in your browser also.  In the none-browser results screen you should see details button, if you click on it that will bring up benchmark details including CPU and GPU temps (see above).

 

The first page comes from View Result Online (click it and that will log you into your account and validate online).

 

The 2nd page comes from the Details button.

 

This is the initial results window after 3DMark tests complete:

 

b37fa356711df47b46107c9122620317.jpg

 

The Graphics Score, Physics Score, Combined Score are on the View Results Online page (browser page).  I didn't include Physics Score and Combined Score in my summary spreadsheet, but I can add them if desired ... but they are available on my sites page.  

 

On a side note: I updated my web site template so my site needs some "cleanup" adjustment work.

 

Cheers, Rob.

 

EDIT: to view the images full size, select the image from my web site page and right click on it and say "open image in new tab"

CPU score. 5960x 20-23k, 3960x 15-18k, 6700k 13-16k dont think a Old 3960x @4.6ghz outpeform a 5ghz Skylake as it do in Firestrike even if both run single 970.

 

FYI, Skylake is 4/8 core CPU, 3960X is a 6/12 core CPU  ... the Physics (CPU) test operates 16 threads and will use all cores available so I would expect a 6/12 core CPU (20MB Cache) to perform better than a 4/8 core CPU (8MB Cache) even when the 6/12 core is an older gen CPU and operating 400Mhz slower.  

 

Like I've suggested in my other hardware posts, I'm a little disappointed with Skylake's performance and there is little to no value in having Intel HD Graphics 530 since in almost all cases it's never used and disabled.  But Skylake is priced at the lower end of the CPU spectrum around $350 so it's still pretty good value for money.

 

Cheers, Rob.

Share this post


Link to post
Share on other sites

Monday night is treat night,  stuffing myself with buttery jammy toast and milky coffee  :smile:

 

Sorted.  Ok so here is my rundown.  Just ran this test now  

 

 

22128292801_09b3d31347_o.png


 

 

 

Share this post


Link to post
Share on other sites

Hi Rob,

Not sure if its just me scratching my head trying to understand what you are trying to say with this thread?

 

In your initial post you start it off with "I get a lot of questions about PC performance in P3D V3 and it's very hard for me to judge (close to impossible) how other systems are going to work in P3D V3.", end with "These don't necessarily translate into how well a system will run P3D, but it can be used as a generic reference point." and in the middle you post a lot of 3D mark results at different component clock speeds.

 

I'm really struggling to understand the relevance of 3D mark scores for P3D (or FSX) as they are very much different apart from that they all put a considerable (but different) load on the computer hardware. I know you say its a generic reference point of the performance of your computer. But as you say"it is NOT a relative indicator of how well one's PC will work in any given Flight Sim" So how could they ever help answer your initial questions about PC performance in P3D V3?

 

Is it to trigger a discussion of how to design a benchmark for P3D V3 that could help answer your initial questions?

 

I'm not trying to shoot you down or discredit any of the effort you put in to this. I'm just not understanding the relevance of the 3D mark results for us flightsim users.

 

Cheers

Lars

Share this post


Link to post
Share on other sites

I agree with Lars. These synthetic benchmarks often aren't even realistic for determining performance in the newest video games. Trying to translate these results to P3d, which for the most part still is a hodgepodge of new and ten year code is a challenge.

Share this post


Link to post
Share on other sites
Guest

Hi Lars,

 

It sounds like I need to clarify.

 

Post was initiated because of two factors:

 

1.  Elaine's post of Fire Strike results in another thread ... catalyst for something I had in the back of my mind for some time now.

2.  The numerous PM's, YouTube, Skype, AVSIM threads, Prepar3D forum PMs/thread, Email, etc. I get for system recommendations and questions about "is X CPU and/or Y GPU" good for running P3D?

 

The sad reality is I can't really answer those questions with any degree of accuracy (other than obviously the components, OS, add-ons I have first hand experience with).  However, I could give a quick ballpark based on Fire Strike Scores (GPU,CPU,Combined) reported ... or better yet, the individual asking could be pointed to this thread or a thread and come to their own conclusions of where their system stands relative to others.  Fire Strike base 1.1, Fire Strike Extreme (2K res), and Fire Strike Ultra (4K res) would cover "most" (not all) of the resolutions and covers GPU, CPU, and combination.

 

FSXMark11 needs some adjustments to it's process to be meaningful for P3D (and maybe for FSX/FSX-SE).  No I'm not volunteering as I really just don't have the time, but maybe hoping someone else will take the ball and run with it or come to the same conclusion as I and realize there are just too many variables.

 

With the time I want to put into this, Fire Strike is "quick and easy" ball park answer.  To come up with a P3DMark11 would require considerable time and effort and really needs similar testing sets based on common resolutions like Fire Strike (1080p, 2K, 4K) as GPU difference when resolutions increase tend to open the performance gap considerably ... that would not otherwise be noticed at lower resolutions.  Also P3DMark11 would in a no add-on environment may miss out on revealing issues that might surface with Add-ons.  As you know, it's 3rd party add-ons (and combinations of) that start to really stress one's PC.

 

Given that 3rd party add-ons get updated frequently (and have their own configurations within themselvs), in addition to the base Sim (P3D version) it's going to be increasingly difficult to extract any meaningful results from a P3DMark11 (if FSXMark11 was updated/adjusted for P3D V3.x).

 

So in summary, I don't think a P3DMark11 (or FSXMark11) is any more or less helpful than just doing simple performance testing using Fire Strike (3DMark).

 

I'm not trying to shoot FSXMark11 down either, but I do feel it has perhaps become out dated.  Some of the issues I see in FSXMark11:

 

1.  Fixes to one screen resolution (lower) skewing ability of more recent GPUs to handle higher resolutions/textures

2.  Uses an aircraft that doesn't exist in P3D

3.  Operates with a base install only (which very few actually operate at) ... base install has no 4096 textures I'm aware off

4.  Doesn't have any awareness of additional graphics options in P3D

5.  Impossible to test with the many add-ons that have significant impacts to performance (which is where MOST users are at)

 

What I do like about FSXMark11

1.  Use saved flight and flight plan with auto so avoids "recorded flight" (which has issues)

2.  Use FRAPS to measure output to log file

 

It's a valiant effort, but times are changing in the FS world regardless of one's preferred platform (I know many do run multiple platforms, perhaps don't keep them ALL updated, but do run more than one) ... so a more generic performance test that doesn't try to take into account every possible platform and 3rd party product would IMHO be "enough" to give someone a relative ball park.

 

Cheers, Rob.

Share this post


Link to post
Share on other sites

Sorry my bad eng, agree with Lars 100%

It was my intention to write someting like that , sorry if you missunderstand me.

 

Then i dont understand wy you ar disiponted of Skylake 6700k ?

Iam opposite to you very impressed by Skylake and less impressed of the Haswell-E 5960X i have them both .

Share this post


Link to post
Share on other sites

Rob, the FSX 737-800 is running perfectly in P3Dv2/3..,

So you can use it with FSXMark11.


13900 8 cores @ 5.5-5.8 GHz / 8 cores @ 4.3 GHz (hyperthreading on) - Asus ROG Strix Gaming D4 - GSkill Ripjaws 2x 16 Gb 4266 mhz @ 3200 mhz / cas 13 -  Inno3D RTX4090 X3 iCHILL 24 Gb - 1x SSD M2 2800/1800 2TB - 1x SSD M2 2800/1800 1Tb - Sata 600 SSD 500 Mb - Thermaltake Level 10 GT case - EKWB Extreme 240 liquid cooling set push/pull - 2x 55’ Sony 4K tv's as front view and right view.

13600  6 cores @ 5.1 GHz / 8 cores @ 4.0 GHz (hypterthreading on) - Asus ROG Strix Gaming D - GSkill Trident 4x Gb 3200 MHz cas 15 - Asus TUF RTX 4080 16 Gb  - 1x SSD M2 2800/1800 2TB - 2x  Sata 600 SSD 500 Mb - Corsair D4000 Airflow case - NXT Krajen Z63 AIO liquide cooling - 1x 65” Sony 4K tv as left view.

FOV : 190 degrees

My flightsim vids :  https://www.youtube.com/user/fswidesim/videos?shelf_id=0&sort=dd&view=0

 

Share this post


Link to post
Share on other sites
Guest

Ok, so if one doesn't have FSX installed?

 

But even so, that's not the show stopper for FSXMark11 to be useful for P3D. Can use some other P3D default aircraft with an AP. It's the other issues I listed above that make FSXMark11 not useful and skewed.

Share this post


Link to post
Share on other sites

1. Run Robs recomended settings as standard , or settings you not think ar skewed up.

 

2. Run res single monitor 3840x2160 , 2560x1440,1920x1080

Res triple monitor 11520x2160, 7680x1440, 5760x1080

Do a standard setup for single and triple monitor setup , the res thats normal and not skewed up .

 

3. The 738 or some std p3d aircraft.

The saved flight and fraps was ok?

 

All above settings you can run in FSXMark11.

 

What we get compared to Firestrike?

Here is some not all, how different CPU.s react to P3D V3 ,

SLI if get a 60% increase as with Firestrike optimated SLI drivers.

SLI vs single GPU , if it is a waste of Money to get a second Gpu

If a high budget system 5960X 2xTitan X vs normal 4790k 970 is 120% faster.

 

For a normal P3D user the single monitor res of 2560x1440 is that a normal setup?

Triple monitor 5760x1080?

 

Is this res with robs settings skewed up compared to run Firestrike?

 

My eng is very bad , hope that you understand my point.

Share this post


Link to post
Share on other sites

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
Sign in to follow this  

  • Tom Allensworth,
    Founder of AVSIM Online


  • Flight Simulation's Premier Resource!

    AVSIM is a free service to the flight simulation community. AVSIM is staffed completely by volunteers and all funds donated to AVSIM go directly back to supporting the community. Your donation here helps to pay our bandwidth costs, emergency funding, and other general costs that crop up from time to time. Thank you for your support!

    Click here for more information and to see all donations year to date.
×
×
  • Create New...