October 19, 201213 yr I couldn't decide between Gigbyte Radeon HD 7950 v.s. MSI GTX 660 Ti Power edition (priced the same). So i got them both and ran them through standard 70 second test scenario in FSX and Prepar3d (a landing on my home airport EHHV, with a somewhat demanding Katana and very demanding NL2000 photo scenery). And the results are in: FSX: 34 FPS for the 660, 32 FPS 7950 P3D: 46 FPS for the 660, 30 FPS for the 7950. So the 7950 is going back to the shop Some more details on the tests: - Yes the tests were exactly the same (same save game, same 70 second "auto landing" test). - Yes i applied some card specific optimizations (both overclocked, and AA controlled from the driver/nvidia inspector for both, but not for prepar3d). - I play windowed Some FSX.cfg details (although i don't think they are relevant for the comparison): - PoolSize=4000 (tested with 0, 1000 and 9000, but 4000 is the best for me) - DX10 (yes, DX9 performance far worse on my pc) - TEXTURE_BANDWIDTH_MULT=200 - FIBER_FRAME_TIME_FRACTION=0.01 (yes i tried higher values, and actually in P3D 0.99 performs best for me) - MAX_ASYNC_BATCHING_JOBS=3 - SmallPartRejectRadius=4.0 (i'm always open for ideas for improvement.)
October 22, 201213 yr Bert Dromer, while these are surely very interesting results, your tweaks are... sorry to say... wrong. I'm not even going to start explaining each, but I'm just gonna say that 0.01 FFTF is just plain wrong. At least all these are wrong for DX9, this is what I am based on. DX10 has never been tested, but I think the same laws apply. If you want realy improvement, check my guide (I see oyu are new here, maybe didn't see it yet): http://#####...hardware-guide/
October 23, 201213 yr Author Hi Word Not Allowed, Thanks for your reply, i always like to hear from people who are as crazy about tuning FSX as me. I bookmarked your guide, i didn't know it yet, alhtough i knew most info in it. I use dx10 (and windowed mode) simply because it's much faster on my machine then dx9. This seems to be different per pc for some reason. About the very low 0.03 FTFF that i use now: As you say in your own guide: lower FTFF gives higher FPS, in photo sceneries (much heavier then ORBX!) this is the single most important tweak. For >150knots flight i use a higer value, but for VFR this is a great balance. You say FTFF works the same in P3D, after extensive testing on my machine with the same flight/scenery/plane/settings/etc in p3d a higher (0.99) FTFF gives higher fps, i recommend retesting this. One tweak that might be interesting for your guide: Search AVsim for "0cc5b647-c1df-4637-891a-dec35c318583". This give me +3 fps. Some other remarks: Poolsize=3000 works better for modern PC's, give it a try. Very complete guide you wrote! One last advice: I always recommend changing just 1 setting at a time and then re-benchmarking using a 30 second save flight (i use a landing). I think any guide should start with a similar advice.
October 23, 201213 yr I never notice any difference with the cfg file tweaks especially FTF. I only use Highmemfix and Wideviewaspect. Each to their own and each system is different but for me the rest are useless so are left out. Interesting results though....am looking at getting the 660Ti myself, especially after BenCap almost gave himself a hernia with excitement over them :lol:
October 23, 201213 yr BenCap almost gave himself a hernia with excitement over them For the sake of airplanes....! :LMAO:
October 23, 201213 yr Author i can promise you that i7 2600k+gtx660ti will give you at least 36fps in even a heavy plane in a very heavy photoscenery (30GB nl2000) with a heavy airport with quite high settings .
October 23, 201213 yr About the very low 0.03 FTFF that i use now: As you say in your own guide: lower FTFF gives higher FPS, in photo sceneries (much heavier then ORBX!) this is the single most important tweak. For >150knots flight i use a higer value, but for VFR this is a great balance. You say FTFF works the same in P3D, after extensive testing on my machine with the same flight/scenery/plane/settings/etc in p3d a higher (0.99) FTFF gives higher fps, i recommend retesting this. FFTF is a balance between what a FSX system gives for texture loading and what for drawing of frames. Setting this too low, any time the FPS drops below locked and the CPU is running hot, close to 100%, which it usually does when FPS is below locked, you are going to see unsharp textures. I can't and I won't advise anyone (nor even post it on the forums) to go lower FFTF, as I know the downside. But you must know that many people, probably most of them, no matter what I repeatedly am saying, will simply enter the number I have set there without much thinking. Simply said, if there is a downside, make damn sure you point out way too well, so that people know what and why you are setting something. Especially something as sensitive as FFTF. You say FTFF works the same in P3D, after extensive testing on my machine with the same flight/scenery/plane/settings/etc in p3d a higher (0.99) FTFF gives higher fps, i recommend retesting this. I will return to this at some point. 0cc5b647-c1df-4637-891a-dec35c318583 Parking cores... old stuff. I know about it. I didn't include it, since I recommend disabling HT anyway and overclocking as much as possible. I can't cover anything that is out there, what I hope is that folks just set everything I suggest. Those that did really profitted much. Some other remarks: Poolsize=3000 works better for modern PC's, give it a try. Sorry, but, not in million years. I tested BP in and out, and by far, the best performance is BP=0, with some caveats. I always recommend changing just 1 setting at a time and then re-benchmarking using a 30 second save flight Definitely. The only way I do the testing. I spent more than only couple of months tweaking. I was talking with Bojote quite much, following his tweaking sessions, and doing my own, I spent virtually years tweaking until I came up with my own solutions. i can promise you that i7 2600k+gtx660ti will give you at least 36fps in even a heavy plane in a very heavy photoscenery (30GB nl2000) with a heavy airport with quite high settings . Oh and btw. photoscenery ain't that high of a requirement. Takes a century to load yes, but once loaded, it's one of the easiest things on the system. Only texture loaders are really busy tho. 2600k+660ti for: heavy plane: NGX heavy photoscenery: 30gb NL (doesn't matter how BIG, the only question is how much autogen, the photoscenery loaded coverage is always the same, if the texturesize is the same) - I would rather say ORBX, since that scenery has huge loads of autogen... And then let me see you getting 36fps in the NGX VC when running locked FPS in FSX, say at 60 (this is a key thing). Seems to me you are trying to prove me something I went up and down in tests. More than once.
Create an account or sign in to comment