Sign in to follow this  
Rob Ainscough

500MB VAS reduction and increase FPS by 22% in V3 ... need some testers

Recommended Posts

I'm looking for some volunteers that know how to do repeatable identical scenario testing ... the claim is that HT OFF will save 500MB VAS and increase FPS

 

As you know I'm extremely skeptical about "magic bullets" and have explored many of them over the last decade ... however, this "finding" came to me from Saul (LM moderator using the same CPU I have) who was reading a post by Aviator1979 here: http://www.prepar3d.com/forum/viewtopic.php?f=6312&t=114987

 

Since I had just done some VAS testing earlier in and around Aerosoft's EGLL, I thought I'd put the HT OFF claim to the test.  To my surprise I was able to confirm the 500MB VAS savings and a 22% FPS increase.

 

Here is the scenario I used:

Prepar3D V3

Aerosoft Heathrow (EGLL)
Orbx FTX Global 
Orbx OpenLC EU
PILOT's FS Global 2010 FTX Edtion Mesh
FSDT GSX ground services (active)
Active Sky Next (ASN) Weather Engine (Archive weather 10/18/2015) 
Majestic Dash-8 Q400
MyTraffic 6 - 2015 schedule
FSUIPC
GoFlight Interface Tool
 
CPU used: 5960X
 
Do NOT use any affinity tweaks (temporarily remove them if you have them).
 
Departing gate 108 at EGLL - 11:17 AM - Take note of VAS
Taxi to runway 09L (departing east)
Take off and climb to 20,000 FT - DTO to MEDWY
At MEDWY, come back to PILOK then D113B for landing
(if you miss approach just come back around and land on 09L)
Park at Gate 188 or 186
Shutdown and evoke GSX deboarding - Take note of VAS
If you have FRAPS then log time frame data to file
 
Run this scenario with HT ON and then again with HT OFF and report back and be sure to indicate what type of CPU you have.
 
I'll post my video's with my findings tomorrow (uploading them now), but would like to hear from others.
 
Cheers, Rob.

Share this post


Link to post
Help AVSIM continue to serve you!
Please donate today!

I personally get the exact same performance with or without HT on a 6700k. Haven't test how it handles VAS though.

Share this post


Link to post

I don't have all that stuff but I saved up a nasty flight with the Carenado Navajo from Renton, WA over the top of KSEA at 2000' MSL, up to the Space Needle and then over to WA05 Apex Airpark. FTX PNW is installed and NA active in FTX Central. I saved ASN weather along with the flight initially but didn't run ASN for the actual test. Heavy overcast conditions and I have all my sliders/shadows/stuff adjusted to "ridiculous", identical settings for each test. I think I'm even getting double freeway traffic from UTX v2 and PNW combined, need to investigate that further but left it as is for the tests. For all intents and purposes "unflyable" for anything other than testing, lol. No AM in the .cfg either time.

First test from a cold boot with HT on FSUIPC.log says:

928035 Sim stopped: average frame rate for last 855 secs = 10.6 fps
942137 === NOTE: not calling SimConnect_Close ...
943136 System time = 18/10/2015 23:45:01, Simulator time = 08:07:10 (15:07Z)
943136 *** FSUIPC log file being closed
Minimum frame rate was 7.2 fps, Maximum was 17.5 fps
Minimum available memory recorded was 993Mb
Average frame rate for running time of 855 secs = 10.6 fps

 

Reboot with HT off:

979265 Sim stopped: average frame rate for last 894 secs = 10.8 fps
993149 === NOTE: not calling SimConnect_Close ...
994148 System time = 19/10/2015 00:06:21, Simulator time = 08:07:49 (15:07Z)
994148 *** FSUIPC log file being closed
Minimum frame rate was 7.5 fps, Maximum was 19.8 fps
Minimum available memory recorded was 1183Mb
Average frame rate for running time of 894 secs = 10.8 fps


...so I didn't notice much difference in FPS but if "Minimum available memory recorded" is the same as "closest I came to an OOM" then it looks like HT off may have helped somewhat.

i7 2600K @ 4.4, GTX770 4Gb, 8Gb DDR3@1866mHz

Jim

Share this post


Link to post

I can't give any scientific results but I definitely get better fps with HT off on my 4790k. I can't comment on the vas though.

Share this post


Link to post

Hi Rob,

 

Only have FTXG and ASN installed, the rest is default

 

Used your settings, ironically have been doing all my own testing at EGLL.

 

Turned HT off and set my 4K monitor refresh to 30htz ? ....... assume that's what you meant?

 

I was clocking in at 2500 VAS before, using your test it dropped to between 2200 & 2250 VAS (used cloudy settings to test)

 

cheers Dave

 

ps frame rates were around 30fps using unlimited, the odd drop to 27fps

Share this post


Link to post

Hi Rob,

 

i dont have some of your add ons. But did a test flight from Cielosim KPAO over ORBX KSQL to KSFO HD + FTX NCA active in the A2A C 182. In V2.5 an OOM was guaranteed in v3 not close to. My settings are a bit lower than yours but can do that flight with your settings again.

Share this post


Link to post

I did a similar test last week.

 

Can confirm that HT OFF gives more FPS on a 4790k @ 4.5 GHz (unlimited, no AM). And of course much lower temps.

Share this post


Link to post

HT off gives me around 1-2 fps more on my I7 4790k @4.5GHz, but my VAS didn't change.

Share this post


Link to post

So oveoclock headroom increases with HTC off....interesting find for sure.

Share this post


Link to post

So oveoclock headroom increases with HTC off....interesting find for sure.

The only reason you can OC more with HT off is less heat. HT on gives more heat so you reach your OC limit a bit sooner. HT on keeps the CPU a bit cooler and so you can OC a bit more. There is no magic headroom here.

 

I always have HT on and will do my testflights asap, mainly for FPS testing: I don't have any VAS heavy addons so I won't be able to reproduce Rob's test in any way.

 

EDIT

Personal test on i7 4790K @4.5 with GTX 780 6 GB on Windows 10. Flying in default Bonanza, above Manhattan on AP, FTX Global, view looking slightly sideways. Fps at unlimited. With HT average between 24.7 and 26.7, without HT average 29.9. VAS didn't change much. So there is a small gain in fps, some 10%. However... performance seemed a little bit choppier without HT. Obviously this is hard to test because micro pauses are always there when looking sideways but I had the impression it was worse without HT.

 

Obviously this isn't a very hard stress test but it does bring fps down on my PC and it's a flight I can reproduce exactly everytime. I would have to do some more test to see if HT off really is choppier but for now I see no reason to keep it turned off.

Share this post


Link to post

I'm keeping my i3770k on 4.6ghz with HT ON, temps are close to 70C, but only on stress 100% occupancy tests, which, let's be realistic, NEVER occurs in ANY application.

Share this post


Link to post

Sorry to say this forum controls really suck, cant figure out how to cut and paste link.

 

In any case tool used for checking VAS is called ProcessExplorer

 

You need Science degree to figure out how to use the controls on the forum, would pasted you download link

Share this post


Link to post

Sorry to say this forum controls really suck, cant figure out how to cut and paste link.

 

In any case tool used for checking VAS is called ProcessExplorer

 

You need Science degree to figure out how to use the controls on the forum, would pasted you download link

Copy and paste... https://technet.microsoft.com/en-us/sysinternals/bb896653.aspx

 

I seriously don't understand why what's so hard with this...?

Share this post


Link to post

Always HT on, I'm happy.  Don't know enough about it really.   

 

Following  

Share this post


Link to post

If there really is a positive effect of disabling HT on P3Dv3, then LM has some patching to do.

Share this post


Link to post
Here's my HT ON/OFF video ... I zoomed the FPS and VAS as an overlay for the benefit of 1080p folks so they can see it since it's recorded at 4K res the text would be unreadable at 1080p.  I also tested at stock clock to make sure it was the OC that was causing this situation.
 
Anyone testing HT will need to remove and FFTF or Affinity tweaks from their Prepar3D.cfg.  For those reporting their findings please be sure to indicate your CPU type/model.
 
 
Graphics settings and VAS/FPS summary is at the end. Video description includes the Add-ons used (they are identical for each flight).

 

If there really is a positive effect of disabling HT on P3Dv3, then LM has some patching to do.

 

It could be several other factors Marc, wouldn't jump to that conclusion.  It could be an OS problem, it could be a Visual Studio C++ library problem, it could even be an Intel design issue on certain CPU types, and it could be an LM code issue.
 
Cheers, Rob.

Share this post


Link to post

Awesome video Rob - V3 does look super - thanks for taking the time

Share this post


Link to post

... the claim is that HT OFF will save 500MB VAS and increase FPS

Cheers, Rob.

 

Rob, if this is true is it perhaps because HT ON ends up loading more terrain textures per interval of runtime?  I've always concluded HT ON gets the best performance on my old SB-E machine.

Share this post


Link to post

@J van E

 

When I first joined avsim years ago I could hit ctrl v and paste is done. Now even clicking all the buttons does not result in paste. I am robotics software engineer but also do gui, and I can tell intuitive gui from a confusion that tries to get you to to do things in unintuitive way. How about a control V to paste , am not a sucker for complexity. but then I don't post much anymore.

 

If I designed the gui, Ctrl v would still work to paste, its what people expect, design GUI around expectations, what folks are used TO. That's how I make my living and that's what sells.

 

So that you don't need manual to do simple things.

Share this post


Link to post

If you don't use an Affinity Mask with HT=ON then you will likely see better performance with HT=off with no affinity mask! I think!

However, I suspect that HT=on, on a quad core processor with an affinity mask=254 will yield better performance than HT=off with No affinity Mask.

 

If I understand DX10 fixer Steve correctly that is :-)

Share this post


Link to post

Anybody here change a i7 3770k with a 6700k and what is your results?

Share this post


Link to post

Thanks for the responses ... for those that responded but didn't indicate your CPU type and OS, would be most helpful if you could include that info along with VAS increase, decrease, no change and FPS increase, decrease, no change.

 

Cheers, Rob.

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