Archived

This topic is now archived and is closed to further replies.

Guest Nick_N

"Holy FPS, Batman!"

Recommended Posts

See my new post "Holy FPS, Batman!" herehttp://blogs.msdn.com/ptaylor/archive/2008...fps-batman.aspxKey quote:Toms posted a OC CPU-GPU review on May 15th that included FSX-SP2 and shows a whopping 83.1 FPS at 19x12 with no AA,trilinear and Ultra Quality, and a still excellent 72.7 FPS at 19x12 with AA, Anisotropic and Ultra QualityI go into some more analysis in the rest of the post, and have links to the relevant Tom's articles. I think this is a must-read, both my post and the linked articles.

Share this post


Link to post
Share on other sites
Help AVSIM continue to serve you!
Please donate today!

Hot dog! I guess I need to break down and upgrade my HD1200 to an OC 8800 GT pretty soon... ;)

Share this post


Link to post
Share on other sites

Another mine is bigger than yours.....I would trade 50 of that fps for some stability and more addons.I am one of those unfortunate ones who has the menu crashes and keeps his fingers crossed everytime I fly.Dependent apon Nhancer to get some aa/af if at all.Yeah I see the ocassional 50+ fps with moderate settings,1920x1050 res,FEX,GEX,UTX,MytrafficX, in a default good weather.However, throw in some weather or a city and I am lucky to see 15-20.Acceptable but forget using a complex a/c.Again, its not the FPS but the overall stabilty I am after.What would really make me angry :-newburn is if I copied one of these hardware sites or the gurus setups to the T and STILL have problems.With Nvidia changing drivers every month and a new card every 3, its a real mess.I am DONE chasing the FSX carrot and awaiting FS11.

Share this post


Link to post
Share on other sites

Phil,I wonder if you might take a moment and explain to the uninitiated why we would ever want 80 frames per second.Can the human eye even perceive more than around 30?The reason that I ask, is because I specifically limit FSX to displaying 24 frames per second, on the theory that I want the simulation to use my computing power to do other things rather than redraw my screen wastefully ... things such as load texture ahead of me or recalculate mesh.Am I thinking about this incorrectly?Cheers,

Share this post


Link to post
Share on other sites

its a benchmark of systems against each other. as such, showing the high-water mark shows where the variants stand against each other and that has value for people considering a purchase.as to the "value" of any particular result, that isn't really the point. its the whole, not the parts.

Share this post


Link to post
Share on other sites

I believe, imperfections aside, that article shows:1)where 8800, 9600,and 9800 stand relative to each other.2)where CPU variants stand relative to each other.3)what an OC on both CPU and GPU can get you.and not everyone knows 3, much les 1 or 2. It is in that spirit I posted. Sorry if that is objectionable.

Share this post


Link to post
Share on other sites

This is the best relationship CPU/GPU benchmark under FSX SP2 that I've read. A must read for everyone using FSX.Have a nice evening Phil,Pierre

Share this post


Link to post
Share on other sites

Kevin:It's not about the human eye seeing 80FPS, its about having 'overhead'. If you can run max settings with default FSX then when you pile on frame rate sucking addons and it sucks 30FPS out of your system, you still have 50FPS of sexy smoothness at your disposal.If you are only chugging along at 30 FPS and your addons suck 25 FPS out of that, you no longer have a Flight Simulator, you have an Etch-a-sketch Simulator.

Share this post


Link to post
Share on other sites

Hmmm...i don't get anywhere near those frames. See sigIntel E8400 @ 4.185ghz--1860 FSB4 gig Corsair Pc2-8500 @ 930mhz 4-4-4-12Asus Maximus motherboardXFX 8800GTS 640 mb 174.74 Drivers 675 core mhz..2100 mem..1500 shader2 74gig WD 10,000 rpm Raptors Raid 0Thermaltake 750watt Toughpower PsuAcer 22" Widescreen LCDWinxp 64

Share this post


Link to post
Share on other sites

Phil:Do you know what settings were used in FSX when this benchmark was done? I obviously need to fire up FSX because I destroy his 3Dmarks of his fastest system by over 6000 and I'm most interested in 80FPS with all sliders maxed! (see my sys specs below). My overclocks are about the same, 33%, and I can take her further if I wanted. (I've had a stable 5.2ghz on cryo.)Give me this, and I will personally guarantee a Fedex'd case of anything you drink!Regards,Mike T.

Share this post


Link to post
Share on other sites

>>Can the human eye even perceive more than around 30?>Sure it can. Anyone who plays shooters knows that you need 60fps for smooth and accurate game play, which is the upper limit of what the eye can perceive. Most gamers would consider 30fps to be "unplayable", so they are surprised by the low framerate of something like FSX. Film captures motion blur so 24fps is fine for movies, but with something like a video game that does not capture motion blur, you need 60fps for total smoothness. Of course sim games are slow paced so you can get away with 25-30fps, but I certainly prefer 30+ for FSX.

Share this post


Link to post
Share on other sites

Hi,Well, I think it is about time someone did these tests, but....I can get 100 fps if I keep my autogen, weather and of course use an AC that does not have a relatively complicated and heavy memory/CPU footprint low. I run a stock qx9600 with 4gig of ddr2 and an 8800 GT overclocked. But.... with my preferred settings (high) and addons FEX, GEX, 30% traffic GA and Heavies, 25% road traffic, minimal airport utility movement, Auto Gen Very Dense, water Mid 2, at KSEA and surrounding areas using a RealAir simulations AC I can maintain 24 fps and very smooth. So, I think that the test is a good reference, but I think the community really needs to know settings,addons and the like.This is where the rubber meets the road.BG

Share this post


Link to post
Share on other sites

>Hi,>>Well, I think it is about time someone did these tests,>but....>I can get 100 fps if I keep my autogen, weather and of course>use an AC that does not have a relatively complicated and>heavy memory/CPU footprint low. I run a stock qx9600 with 4gig>of ddr2 and an 8800 GT overclocked. But.... with my preferred>settings (high) and addons FEX, GEX, 30% traffic GA and>Heavies, 25% road traffic, minimal airport utility movement,>Auto Gen Very Dense, water Mid 2, at KSEA and surrounding>areas using a RealAir simulations AC I can maintain 24 fps and>very smooth. So, I think that the test is a good reference,>but I think the community really needs to know settings,addons>and the like.This is where the rubber meets the road.>>BGGood point ... although I think the idea was to show the overhead potential with the clock/core VS the cardAt the same time Tom left out the 8800GTX/s 640/768 384bit cards but included the the newer Nv offerings and then displayed compares to the very old ones so I think there is some missing information thereThe G80 GTX 768 384bit cards have as much if not more horsepower than the 9xxxx cards... one difference is the TMU's (Texture Memory Units) and even with that Nvidia is playing games with how many they enable in G92 as the newer cards are released even though they all have 64 So for the price and if I was upgrading considering the new next gen formats and Nehalem is right around the corner.. I would rather spend less money on the 384bit card which will give me nearly the same performance knowing I will be shopping again at the end of the year for next gen parts.Of course the SLi advantage with G92 is greater than with the G80 384bit cards, G92 when compared to the 768 384bit G80 card for a single card system is not that far of a leap in performance with FSX.. something Tom left out of that list. The proc is of course important and pairing up the right team is what the report was about even with the missing data

Share this post


Link to post
Share on other sites

Yeah, great. :) But statements such as that the "DX9 version doesn't show sky and land reflections" was very inaccurate and doesn't bode well for trust in the numbers according to what ever settings, and real apples to apples facts. However it was encouraging.As for the high FPS, just remeber until we get blur etc. you need the high frames as motion picture 24-30 has blur in frames, Game does not. blah blah blah

Share this post


Link to post
Share on other sites

Yea, that was weird. Where'd that come from? First sentence in the article seemed to me to be inaccurate. Certainly set the tone in an unfortunate way.Also, anyone have any idea what the system used might cost a person?

Share this post


Link to post
Share on other sites

That's why I asked Phil if he knew what settings they used to post that article because we all know that DX9 shows reflections of both clouds and scenery in the water. We all also know that 80fps with low settings means very little. Most benchmarks I've seen with FSX in the mix have had to reduce the sliders on FSX to keep it competitive with the other games like Crysis, etc. The system they used is mid-high end (by today's standard - the Q6600 has been around for more than a year) and should cost you no more than $3,000US. Sure you can build that system for less but then you would not be able to overclock so aggressively with the lower end parts. A 33% overclock requires a high quality Mo-bo, memory, power supply and cooling at the minimum.I'm going to fire up FSX in a few minutes and try to imitate that scenario and see what type of performance I get. With over 20,000 3DMarks (vs their top end systems' 14,000) I'm very interested. Another thing I don't see in those shots is AUTOGEN. We shall see.Mike T.

Share this post


Link to post
Share on other sites

With my current settings, which are quite high (higher than Ultra high) with the exception of the water which is set to 2Xlow (Ultra High in 2Xhigh) I can get 60+fps in the Sitka mission. (And that's with AA set to 8Xs and AF=8x, I did up my res to 1920x1080x32 for this test) My settings are in this video. When water is at max I only get in the 40's, that and the comments made about the relections, lead me to think Tom's Hardware ran this test at a lower water setting.Edit: I need to make a correction I get 60's at the lower resolution 1280x720x32, and 50 at 1920x1080x32. With Max water I get 25 in either. (Don't know where I tested that 40 number from, just retested it, and can't get anything close to it with water maxed.)

http://forums.avsim.net/user_files/188805.jpg

Share this post


Link to post
Share on other sites

Well I suppose I can take a few minutes to tell of my recent excitement. I had been struggling with FSX in my past system, and as much as a I WANTED to use it...I ended up just settling to play FS9 exclusively. I have been pretty good at optimizing XP, and still couldn't get decent performance in fsx with an addon planes other than RealAir. I had an AMD FX-60 OC'd to 2.8Ghz, 2G DDR800 RAM (OCZ), an ATI x1950 Pro AGP (yes - AGP) GPU and 2 SCSI 15K Cheetahs and a 10K Raptor. But one of my cheetah's finally died (the one with the OS!) and I decided to take the oppty to do a complete system revamp. So , "ONLY" $1200 later I now have an Intel e8500 CPU, an ASUS Formula Rampage MB (finally just bit the bullet to get a great one), 4Gb G.Skillz DDR21066, an EVGA 8800GTS 512 KO, and a new WD 640G (320G/platter) drive! Everything else was re-usable...XP SP3 on the 10K Raptor and FS and Games on the 640G WD. First let me say that i am SO glad I put down big bucks on a solid motherboard. My last rig wouldn't let me OC much even tho I had good components OTHER than my Asus A8V MB. So on air cooling (Zalman 9700) alone I bumped my e8500 from 3.16 to 4.16Ghz EASILY! No crashes, BSOD, lock-ups, or errors in tests! I know I can push it more but I'm not ready to take the time necessary to do so safely. Easily OC'd my 800GTS to 750Mhz mem, 1000Mhz core and again - no errors in tests of any kind. So I spent the most of a week loading all software, going through tweaks/tunes (thanks Nick_N) to the letter. I DID load both fs9 and fsx (sp2) just for comparison. For fsx I have FEX, GEX, UTX, FSG 9.6m mesh, and a bunch of addon planes. I can now say, without a DOUBT, that FSX is here to stay! I WOULD take FS9 off except for I still have a few addons that haven't been converted, like the DF 727 or the Aeroworx B200, but honestly I may never fly those again...not unless they come back to FSX! I have NO clue what my max frate rate is...don't care. I set it at 26...smooth as silk (if you ever REALLY tweaked FS9 to be smooth as silk you probably used the TextureMaxLoad (not Texture_Max_Load) tweak and didn't lock FR above 22-26 anyway. Now the game autoconfigures most everything to Ultra-High. I have Autogen on Dense, Scenery on Extremely Dense, water at 2.0 low, Traffic on Ultra High, use 16x AF and 8xS AA, Real time weather (also Ultra High) and it is DROP DEAD GORGEOUS! In fact, I haven't even tried to squeeze more Autogen or AI Traffic out yet...I'm just so blown away that I don't really care to. I only increased my Buffer Pools to 15M, and disabled Preload...no other tweaks. So take it for what it's worth...no I haven't run a carefully constructed comparison test. Don't need to. I'm a believer. I hate that I had to drop the $$$ to get this performance, but I suppose that's the nature of the beast. Hope it helps anyone sitting on the fence. Thanks for listening to my ramblings! Back to fly...

Share this post


Link to post
Share on other sites

I just did a series of tests. FSX Antialiasing, FSX AF (Bilinear), Traffic 100%, all scenery sliders right, FPS unlocked.Dutch Harbor Alaska, weather Fair, Settings Ultra, Default C172: 209.1 FPS.Fairbanks Alaska (water view), weather Fair, Settings Ultra, Default C172:182.1 FPSDutch Harbor Alaska (mountain view, weather Fair, Settings Ultra, Default C172: 155.5London City Airport, weather Fair, Settings Ultra, Default C172:33.1 FPSSeattle-Tacoma Airport, weather Fair, Settings Ultra, Default C172:9 - 15.9 FPS depending on what objects are on screen.Now, what this shows is what FSX has always shown. If you are a Bush Pilot for this test, I believe that you will probably short out your keyboard with all the drool that will drop from your mouth. If you are a heavy pilot you will probably also short out your keyboard (as you slam it into the wall).None the less, this DOES indeed validate what was shown in the benchmark tests, and depending on what you use FSX for, you will be extremely happy...or not. Also, it should be noted at using default FSX antialiasing and AF makes FSX look pretty terrible as compared to the same being done on the video card. The difference in video quality is exacted at the cost of severe framerate loss if you want to fly around with your FSX looking like the best screenshots you see.To me this shows that high-end systems are indeed able to run FSX at the highest settings in many situations, and indeed not in others. It is probably safe to say that in another 24 months, high-end systems will probably catch up to FSX and it will be possible to run it as you can FS9 now. http://forums.avsim.net/user_files/188811.jpghttp://forums.avsim.net/user_files/188812.jpg

Share this post


Link to post
Share on other sites

Be careful about locking frames, unlike FS9, some have reported, including myself, that doing so will reduce your performance, don't know if it's a bug or by design. This didn't happen in FS9. You won't see the problem unless your in an area, where your actual FPS is at or near your target frame rate. If you have the problem, your FPS could be from 5 to 10 frames less than if you set unlimited. You can test this by setting your frames to unlimited. Record your FPS, FRAPS is best for this, since that gives you more of an average FPS than the FS frame counter. Then lock the frames at or near that frame rate. If you see lower FPS, then your system has the problem also, and you may want to consider running unlimited. I do, and don't see any negative effects. No blurries or stuttering. Not everyone reports this, so it may be system config specific.

Share this post


Link to post
Share on other sites

Are you saying, you can get that with all sliders maxed including water? I can get about 110fps sustained around Dutch harbor with those settings except water at 2xlow, but max I can only get about 70.Also when you say ultra are you referring to the Ultra High preset option? If so, that option doesn't set all sliders right.

Share this post


Link to post
Share on other sites

I used Ultra and THEN moved the scenery details further right. I also use 38m mesh, 1m scenery detail.What you see at 208.2 FPS is reflective of those settings. I used those same settins for all tests.

Share this post


Link to post
Share on other sites

Yes I found it interesting that Tom's didn't have the 8800GTX cards in the compare. Possibly they consider them as "exiting the market" parts, therefore not worth doing.Use a 6600GT/6800GT? I guess that was done to show how a very old gpu fares with a newer cpu. But most of use know how that works already.RhettFS box: E8500 (@ 3.16 ghz), AC Freezer 7 Pro, ASUS P5E3 Premium, BFG 8800GTX 756 (nVidia 169 WHQL), 4gb DDR3 1600 Patriot Cas7 7-7-7-20 (2T), PC Power 750, WD 150gb 10000rpm Raptor, Seagate 500gb, Silverstone TJ09 case, Vista Ultimate 64ASX Client: AMD 3700+ (@ 2.6 ghz), 7800GT

Share this post


Link to post
Share on other sites

I've always thought that to be by design.They're trying to control overhead that way.It seems to matter less the more cores you have. RhettFS box: E8500 (@ 3.16 ghz), AC Freezer 7 Pro, ASUS P5E3 Premium, BFG 8800GTX 756 (nVidia 169 WHQL), 4gb DDR3 1600 Patriot Cas7 7-7-7-20 (2T), PC Power 750, WD 150gb 10000rpm Raptor, Seagate 500gb, Silverstone TJ09 case, Vista Ultimate 64ASX Client: AMD 3700+ (@ 2.6 ghz), 7800GT

Share this post


Link to post
Share on other sites

Ah! Very impressive! When you say scenery details, does that include traffic, (Air and ground)? Because if not, I have some very strange results. You say you get 9-15 in Seattle, with max scenery, water 2XMax, res at 1920x1080. I get 20FPS in Seattle with max scenery and autogen, airline traffic (Large WOAI DB) at 100% No Ga or ground traffic though That will bring me down to about 12FPS if they went to 100%! MY AA is at 8Xs, and AF is 8X. With your system having a much higher, core speed, FSB, and Memory speed, there has to be something else that is causing a bottleneck! That is if you are not including ground trafic!! http://forums.avsim.net/user_files/188813.jpg

Share this post


Link to post
Share on other sites