My take on Bojote's tweaks

[Guest 413X3]

Interesting. Ok Thanks. And yes that email is me that you just received

[bojote 20]

Interesting. Ok Thanks. And yes that email is me that you just received

Sent.Sorry, I have to double check ;) I'm not sending those to any guy claiming he is 'David' ;)If anyone else is Interested FIRST send me a Private message in the forum and tell me what EMAIL you are making the request with and if you are a long stablished member and respected around here, I'll send it to you too... no first time posters please :)Also, this is PRIVATE Beta test, so no posting to the forum until all the BETA testers are satisfied with the results. Sorry it has to be this way, but... I guess you understand my reasons.

[Guest 413X3]

Thanks I'll try them tonight, first I plan on overclocking my i7 to 4ghz

[McCrash 3]

And the farther away objects are batched too ;) if you want to compensate for the 'blurriness' then increase FFTF to 0.50 and you are all set. Remeber, that FFTF won't impact performance IF you are using AffinityMask=14

What is FFTF? Thanks.

[bojote 20]

What is FFTF? Thanks.

FFTF => FIBER_FRAME_TIME_FRACTION=0.50 ( under [MAIN] )it affects the % of time fibers 'render' terrain. 0.50 means thant 50% of the time it takes to render a frame the 'fibers' will try to 'load' terrain. It's simply a % for resource allocation.Adam explained this a long long time ago :) same as Phil Taylor, so nothing new here. Aditionally, and as I explained before this 'fibers' are 'stuck' to CORE0 and they ignore the affinity mask, thats why setting AffinityMask=14 'moves' the main scheduler OFF CORE0 to CORE1 leaving the fibers with lot of room, so the 0.50 won't affect your performance one bit but will make terrain tiles load faster.Evidently, the above ONLY APPLIES to quad cores... if you do this on a DUAL core you kill it.And if you are the 'curious' type, AffinityMask=14 means:(first, use the windows calculator, and see which is the BINARY representation of the decimal '14'it is:1110Now, turn that around... swap it, as if you were looking a mirror:0111Above you see the ones represent your 'active' cores. so, 14 Will make CORE1 2 and 3 active for FSX. CORE0 will stay free BUT since fibers IGNORE this AffinityMask then they will use CORE0 effectively (and efficiently) using ALL 4 CPU's so in short, if you have a QUAD or i7 your AffinityMask should be 14 to make the most efficient use of resources.I know the answer is a bit too technical, but it allows others to grasp the concept and reply to this questions as well, so I'm not a single source of information and we all share and help others

[Paul_Yorks 97]

Above you see the ones represent your 'active' cores. so, 14 Will make CORE1 2 and 3 active for FSX. CORE0 will stay free BUT since fibers IGNORE this AffinityMask then they will use CORE0 effectively (and efficiently) using ALL 4 CPU's so in short, if you have a QUAD or i7 your AffinityMask should be 14 to make the most efficient use of resources.I know the answer is a bit too technical, but it allows others to grasp the concept and reply to this questions as well, so I'm not a single source of information and we all share and help others

J - you say here 14 for AM if you have a Quad or i7; so why, with your i7, is your AM set to 12 in the .cfg you posted?

[bojote 20]

JB - you say here 14 for AM if you have a Quad or i7; so why, with your i7, is your AM set to 12 in the .cfg you posted?

12 will make use of only ONE texture manager (instead of two) since fiber cooperatively multitask with this threads, the fewer the threads the less chance of thread collisions or 'stutters' 12 and 14 are both valid options, its just that telling people to use '12' will make them all crazy asking 'why' they need to reduce their number of CPU's when they can use all 4.Personally, I use 12. But again, could use 14 and probably won't notice a dif. performance wise is exactly the same.and to expand a little the reply.. the 'texture' managers do also async batching (autogen) you can fine tune this via:[sCENERY]MAX_ASYNC_BATCHING_JOBS=1 // Default is 1 (Should match the number of texture manager threads)by default, the above is '1' max is '4' if you want 'efficiency' then the above should match your number of texture managers. Since I usedefault (1) then, I simply set AffinityMask to 12. If I were to use Affinity Mask 14, then the above should be MAX_ASYNC_BATCHING_JOBS=2 but this 'aditional' batching can and WILL create instability IF you use BP=0This tweaks are all about 'balance' and efficient use of resources, not just 'raw' speed, and most definitely not guesswork as some like to suggest :)

[jjjallen 37]

David, I'll be sending you during the week my custom Shaders using Shader Model 3.0 for FSX so you use them with HIGH 2.0 and my custom ENBSeries config, of course, only if you want to beta test stuff ;) (and in case you are wondering, I already solved all the issues I had before running SM3.0 in FSX)I was planning on releasing them today, but I would like some people to do some private beta testing first.

Gimee, gimee, gimee them shaders.! I'll do som testing.jja

[Tabs 2,790]

To the OP - *******'s original recommendation to not use nHancer has more to do with not putting extra strain on the video card than anything else. By all means you can try the usual settings with BP=0, but the point is that it takes away from the available GPU power you have to accept data from FS through the CPU. BP=0 stops the engine from limiting what the CPU sends to the GPU and if you're running very high AA levels like 8XSQ, you're essentially setting the card's baseline for processing those calls to a lower level that what it's actually capable of, because it's using up power to do the AA. If you get a super powerful card like the GTX480, you can run all that stuff no problem without worrying about stability according to what ******* has seen in his testing.

[Paul_Yorks 97]

Thanks *******, I see what you mean re the 12/14 split. Like you, I don't see much difference, swapping between them.Am I right in thinking (and I know I'm simplifying here) that part of your approach is to allow fast texture loading, to get rid of the stutters, but then to "hide" the blurry textures that this creates, by covering everything with autogen? That seems to be how this is working, and I'm pretty happy with it, but I'd like to know if that's the general idea behind it.And is less then max LOD essential to this - I've played with the variables on the sliders, here, but see no real difference.Thanks, by the way, for your amazing contribution to the hobby! :)

[bojote 20]

Am I right in thinking (and I know I'm simplifying here) that part of your approach is to allow fast texture loading, to get rid of the stutters, but then to "hide" the blurry textures that this creates, by covering everything with autogen? That seems to be how this is working, and I'm pretty happy with it, but I'd like to know if that's the general idea behind it.

Correct, remember that the scenery LOD its a 'ring' around your A/C a LOD of 3.5 will 'batch' everything in that radius, while also decreasing the number of objects to load... anyway, the ground will be so covered with autogen (on final approach for example) that you won't notice any blurries. and if you are flying HIGH and FAST the FFTF will definitely help you! so in either case, it seems LOD 3.5 to be a very good compromise between looks and stability (when using BP=0)there is people using LOD of 6.5, 7.5 fine... but, with BP=0? forget it... it will simply not work.

[Kosta 122]

there is people using LOD of 6.5, 7.5 fine... but, with BP=0? forget it... it will simply not work.

Hm, I have pretty much success with 7.5. My only problem is rather small stuttering, which is not BAD, but its here. Not tragical at least.I have AG on Normal, and no per_cell entries in fsx.cfg.Works completely ok, no crashes.

[bojote 20]

Works completely ok, no crashes.

Try it with Manhattan X, FSDT KJFK, and Extremely Dense autogen ;)

[sargeski 16]

... anyway, the ground will be so covered with autogen (on final approach for example) that you won't notice any blurries.

People need to keep in mind that this is totally dependent of the area and/or scenery they may be flying in.

[Kosta 122]

Try it with Manhattan X, FSDT KJFK, and Extremely Dense autogen ;)

First yes, 2nd yes, but third no. I don't like Dense AG. It is nice, but I fly high, and only compensation I have is high LOD, other all mountains look very blurry. I believe the only difference is AG!