Jump to content
Sign in to follow this  
pmb

News graphics adapter - same fps

Recommended Posts

Hi,

 

my system has been as follows until 2 days ago: Intel Board DZ68DB, 650 W power supply, Processor Intel i7-2600 CPU @ 3,40GHz, RAM Kingston 16 GM, 1 Intel SSD 224 GB (System), 1 Intel SSD 450 GB (Flightsim), 1 WD HD (Work), Graphics Adapter Zotac, 560Ti, 2GB VRAM, Win 7/64.

 

That's what I would call a middle-class system today. It's for 95 % work and 5 % flight simulation (would love to see it the other way round...)

 

After the fan bearing of the Zotac card became worn out I decided not to repair it, but get a new EVGA Nvidia 760, overclocked, 4 GB VRAM card.

 

Let's come to Flight simulation. I run Prepar3dv1.4 and v2.0. I'll concentrate on 2.0 here.

 

In a certain start situation (ORBX AUS Sydney on active runway) I got 18 fps before replacing the graphics adapter. This is with pretty high settings (Autogen, Scenery Density maxed - which I know ORBX does not suggest), LOD 4.5, Textures 1024, Water at maximum, a few shadows on, HDR off, Bathymetry off, Framerate unlimited, Vsync off, MSAA 4, Trilinear, Tesselation on. I use one Monitor in 1920x1080x32 for simming. No NI settings (reset game settings to default in driver a time ago).

 

Before replacing the card I de-installed the graphics drivers and deleted the shader cache and the corresponding [DISPLAY.Device] entry in the Prepar3d.ini for both simulators.

 

After inserting the new card and installing most recent Nvidia drivers I was amazed to see exactly the same 18 fps on Sydney runway. Actually, I should not complain, after takeoff fps go into the 20's, this is dense scenery, it actually looks quite stunning, as I said, settings are certainly on the high side. Overall I enjoy a quite fluent ride with very few stutters, if any at all.

 

I just expected fps to raise. The new EVGA card is reported on comparison charts as having 30-50 % more performance on essential parameters. I did not expect a fps boost of 50 % of course, but rather something like 10-20% should have shown up.

 

Obviously the graphics adapter is not the bottleneck in the system. At least Windows performance index says otherwise, though. It was 7.3 before the upgrade with the graphics adapter as the limiting component. Graphics adapter performance (and this way the combined index) rose to 7.6 with the new card.

 

Only difference I noted, I had occasional spikes/flashes with the old card (not VAS related, I checked VAS usage), maybe the adapter got too hot after some time. These seem to have been gone after replacement.

 

Anyone with a similar / different experience? Anyone to comment on this? I did not expect a fps boost in Prepar3d1.4 being CPU bound, but wasn't I told 2.0 being much more GPU oriented which should pay off?

 

Thanks for considering and kind regards, Michael

 


MSFS, Beta tester of Simdocks, SPAD.neXt, and FS-FlightControl

Intel i7-13700K / AsRock Z790 / Crucial 32 GB DDR 5 / ASUS RTX 4080OC 16GB / BeQuiet ATX 1000W / WD m.2 NVMe 2TB (System) / WD m.2 NVMe 4 TB (MSFS) / WD HDD 10 TB / XTOP+Saitek hardware panel /  LG 34UM95 3440 x 1440  / HP Reverb 1 (2160x2160 per eye) / Win 11

Share this post


Link to post

Can you please check the power etc. parameter with the EVGA Precision which I guess you have installed. What"s the difference between GTX 760 and GTX 780?

Your observations are very interesting.

Spirit

Share this post


Link to post

Hi Spirit,

 

to be more definite, this is the card I bought

 

http://eu.evga.com/products/moreInfo.asp?pn=04G-P4-3768-KR&family=GeForce 700 Series Family&uc=EUR

 

It requires a 500W power supply, mine has 650W. Power supply is Metatron 650 W which should have enough power on +12V , too.

 

Kind regards, Michael


MSFS, Beta tester of Simdocks, SPAD.neXt, and FS-FlightControl

Intel i7-13700K / AsRock Z790 / Crucial 32 GB DDR 5 / ASUS RTX 4080OC 16GB / BeQuiet ATX 1000W / WD m.2 NVMe 2TB (System) / WD m.2 NVMe 4 TB (MSFS) / WD HDD 10 TB / XTOP+Saitek hardware panel /  LG 34UM95 3440 x 1440  / HP Reverb 1 (2160x2160 per eye) / Win 11

Share this post


Link to post

I found that in Prepar3dv2 (not in 1.4) if you have a powerful GPU you can reduce the fiber frame parameter without have blurries.

It seems in this way you dedicate less cpu time in texture loading, freeing it for frames rendering, and gaining fps.

Prepar3dv2 seems use the gpu for this purpose.

 

In Prepar3d 14. instead, if you configure a too low fiber frame value you have a lot of blurries; evidently it doesn't use the GPU for this purpose.

 

Having a "modest" AMD phenom 4 core 3.4 GHz, and changing from a RADEON 5870 to a R9280X, I changed the previous parameter FIBER_FRAME_TIME_FRACTION=0.25 in 0.10 with optimal results, and I'm testing now 0.05.

Share this post


Link to post

Hi Michael,

thanks for the link. Now I see the difference between 760 and 780. Still a lot but also the price.

Do you have EVGA Precision X installed? You can use it to monitor the load, temperature and many more live. You should do this.

Spirit

Share this post


Link to post

Hi,

 

thanks.

 

@ Claudio: I had bad experience with FFTF in 1.4, but might try it in 2.0 when I find time for experiments. I think it's even suggested in LMs tuning guide. 

 

@ Spirit: yes, I have EVGA Precision installed, but did not yet explore it. Might be worth, though, to get more insight.

 

Kind regards, Michael


MSFS, Beta tester of Simdocks, SPAD.neXt, and FS-FlightControl

Intel i7-13700K / AsRock Z790 / Crucial 32 GB DDR 5 / ASUS RTX 4080OC 16GB / BeQuiet ATX 1000W / WD m.2 NVMe 2TB (System) / WD m.2 NVMe 4 TB (MSFS) / WD HDD 10 TB / XTOP+Saitek hardware panel /  LG 34UM95 3440 x 1440  / HP Reverb 1 (2160x2160 per eye) / Win 11

Share this post


Link to post

Hi Michael, sorry if I misunderstood your post but have you changed/advanced the shader cache version in your p3d.cfg? And which shader directories did you delete? 

Share this post


Link to post

Hi Dylan,

 

I deleted

 

C:\Users\(MyProfile)\AppData\Local\Lockheed Martin\Prepar3D v2\Shaders

 

and the corresponding directory for Version 1.4, This should have been enough to force simulators to rebuild shaders, I think.

 

Thanks and regards, Michael


MSFS, Beta tester of Simdocks, SPAD.neXt, and FS-FlightControl

Intel i7-13700K / AsRock Z790 / Crucial 32 GB DDR 5 / ASUS RTX 4080OC 16GB / BeQuiet ATX 1000W / WD m.2 NVMe 2TB (System) / WD m.2 NVMe 4 TB (MSFS) / WD HDD 10 TB / XTOP+Saitek hardware panel /  LG 34UM95 3440 x 1440  / HP Reverb 1 (2160x2160 per eye) / Win 11

Share this post


Link to post

I found that in Prepar3dv2 (not in 1.4) if you have a powerful GPU you can reduce the fiber frame parameter without have blurries.

It seems in this way you dedicate less cpu time in texture loading, freeing it for frames rendering, and gaining fps.

Prepar3dv2 seems use the gpu for this purpose.

 

In Prepar3d 14. instead, if you configure a too low fiber frame value you have a lot of blurries; evidently it doesn't use the GPU for this purpose.

 

Having a "modest" AMD phenom 4 core 3.4 GHz, and changing from a RADEON 5870 to a R9280X, I changed the previous parameter FIBER_FRAME_TIME_FRACTION=0.25 in 0.10 with optimal results, and I'm testing now 0.05.

Not to hijack this thread but did you seen any benefit from adding the FFTF tweak?


ATP MEL,CFI,CFII,MEI.

 

Share this post


Link to post

I didn't look it up, but I believe a 760 is a dual card, meaning there are actually 2 gpus on it, equal to 2 680's if I am right.  P3D2 does not yet use sli configurations, which I think is what you'd need to get maximum out of the 760.   You're running that card at 1/2 of it's ability.    That is why you do not yet see an improvement.  When LM releases a v2.2 update you'll probably be screaming along, and if you have more slots you can just jack that up to 4x, even 6x of what you're doing now.  You'll be glad you have 4gb too.

Share this post


Link to post

I didn't look it up, but I believe a 760 is a dual card, meaning there are actually 2 gpus on it, equal to 2 680's if I am right.  P3D2 does not yet use sli configurations, which I think is what you'd need to get maximum out of the 760.   You're running that card at 1/2 of it's ability.    That is why you do not yet see an improvement.  When LM releases a v2.2 update you'll probably be screaming along, and if you have more slots you can just jack that up to 4x, even 6x of what you're doing now.  You'll be glad you have 4gb too.

It is actually a single GPU card. Many people here are using it successfully.

 

When I upgraded my GPU, I did not see an fps increase, but I did see an increase in the distance to the right that I could put the sliders.

 

I would also look to see if in the situation described if the GPU is being utilized to its full extent. Look in the Nvidia Inspector for GPU usage.

Share this post


Link to post

@ Dave: I played around with TTFT, set it to 0,1 and to values higher than 0.33. 0.1 gave me half the fps compared to  0.33 (that's in landclass based scenery, I don't have any real photoscenery installed in Prepar3d v2).  For me, the default of 0.33 seems to fit quite well. (BTW, I made similar observations with the 560Ti before, and in 1.4 I found 0.33 to work quite well, too). I understand, different systems can give different results.

 

@ Ryan: This may indeed be the case While fps are unchanged, the system seems to better cope with higher settings. Still a bit puzzled though. I'll check GPU usage using inspect the next time I find time for flying.

 

Thanks and regards, Michael


MSFS, Beta tester of Simdocks, SPAD.neXt, and FS-FlightControl

Intel i7-13700K / AsRock Z790 / Crucial 32 GB DDR 5 / ASUS RTX 4080OC 16GB / BeQuiet ATX 1000W / WD m.2 NVMe 2TB (System) / WD m.2 NVMe 4 TB (MSFS) / WD HDD 10 TB / XTOP+Saitek hardware panel /  LG 34UM95 3440 x 1440  / HP Reverb 1 (2160x2160 per eye) / Win 11

Share this post


Link to post

I just expected fps to raise. The new EVGA card is reported on comparison charts as having 30-50 % more performance on essential parameters. I did not expect a fps boost of 50 % of course, but rather something like 10-20% should have shown up.

 

Hi Michael,

 

I'm afraid there are some situations like this that aren't predicted w/ P3D V2, which I think is largely because no one is certain just how the rendering engine works in V2.   The DX11-related effects such as HDR, tessellation, shadowing etc will be impacted predictably by the stronger GPU & VRAM.  To a lesser degree autogen will as well.  Yet there are other situations, even w/ default scenery, that demonstrate some components are still very CPU limited, and I'm thinking this is what you are looking at. For example, I have quite a strong machine, and I can sit at point blank range from a terminal at say KDEN in the QW757, and see a frame rate of maybe only 30 or so (I'm unlimited too), yet GPU utilization may only be at 55%, and that is with most sliders except autogen/vegetation maxed, i.e. all of the DX11-related sliders maxed.  Adding a more potent GPU will not I would argue have any affect on this frame rate of 30.  And yet, there are other scenarios that are markedly visually way more complex, for example at KSBA  at dusk w/ 3D lights blazing ala FTXG, I'll see a frame rate of 46 on the runway in the same QW757 where I was looking directly at a simple terminal building and nothing else at KDEN--no lights, no autogen to speak of, and saw only 30 fps.  So...adding more GPU helps some performance aspects, and very much not others.  I was surprised to learn how well FB's KFSO performs for me, whereas FSDT's KPHX really is a killer, and again I'll see only 70% GPU 'utilization' with a frame rate of only 22 there.  Hence, not always predictable.  I'm dying to see how the PMDG T7 runs in V2.  Could be nightmarishly bad, or pretty good ;o)

 

Once you get everything optimized can you maybe get some more CPU overclocking going?  Pick up an i7 SB chip maybe w/ some better headroom?  I think this is a place you will impact that part of performance that the stronger GPU isn't impacting well.

 

Also, have you verified all 4 cores are working for you, i.e. you're using a good AffinityMask value?  I think it would be 14 for your CPU. 

 

Noel


Noel

System:  9900K@5.0gHz@1.23v all cores, MSI MPG Z390M GAMING EDGE AC, Noctua NH-D15S w/ steady supply of 40-60F ambient air intake, Corsair Vengeance 32Gb LPX 3200mHz DDR4, Sabrent NVMe 2Tb x 2, RTX 4090 FE, Corsair RM 850W PSU, Win10 Pro, LG Ultra Curved Gsync Ultimate 3440x1440, TCA Boeing Edition Yoke & TQ, Cessna Trim Wheel, RTSS Framerate Limiter w/ Edge Sync for near zero Frametime Variance achieving ultra-fluid animation at lower frame rates.

Aircraft used in A Pilot's Life V2:  PMDG 738, Aerosoft CRJ700, FBW A320NX, WT 787X

 

Share this post


Link to post

Hi Noel,

 

thanks for the deliberate thoughts. I checked GPU usage, which is around 80 % typically, which I think is okay, but shows there is still headroom in this respect. I use an affinity mask of 84, as I have HT on, tried switchting it off and changing affinity mask to 14, which gave no noticeable improvement, so I switched back for the benefit of other software.

 

As you write, the bottleneck might be the processor. The present one can't be overclocked (at least not according to the specs) and I am not inclined to overclocking anyways. However at a certain point I might exchange it completely, the motherboard should allow for a more powerful Ivy Bridge one.

 

As a sidenote I observed quite good performance at FB KFSO, too,.

 

Kind regards, Michael


MSFS, Beta tester of Simdocks, SPAD.neXt, and FS-FlightControl

Intel i7-13700K / AsRock Z790 / Crucial 32 GB DDR 5 / ASUS RTX 4080OC 16GB / BeQuiet ATX 1000W / WD m.2 NVMe 2TB (System) / WD m.2 NVMe 4 TB (MSFS) / WD HDD 10 TB / XTOP+Saitek hardware panel /  LG 34UM95 3440 x 1440  / HP Reverb 1 (2160x2160 per eye) / Win 11

Share this post


Link to post

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...