Probably good choice mate, your I7 was a CO stepping then?

Funny enough no, its a DO model 920; when they first came out. She will do prime well at 4.0, but will have issues with linpack and fsx when over 3.8:( Tried everyting; think I got just a normal overclocker:( Almost 1.4 years old now though.

That is one heck of an overclock you have there Max, didnt notice that before! Verynice, you may be somewhat close for sure.

Depends on the application and settings but in the benchmarks I've run it's very close. At super-high res + lots of AA the 580 would likely be faster because of its additional VRAM and bandwidth.

This is FSX, kind of a known thing, overclocking yer GPU has little effect on the sims performance.

Yes, FSX is primarily CPU-bound. That just means the performance difference between 570 and 580 is virtually nil.

Higher frequencies of a 570 don't quite make up the design-related shortfall even in GPU intensive games when compared to the GTX 580

The 570 and the 580 have the same GPU. They're the same silicon, cut from the same wafer. The only GPU-specific differences are the slightly higher stock clocks of the 580 (772MHz vs. 732MHz), the additional ROP partition (48 vs. 40), the extra SM which packs an extra 4 TMUs and 32 SPs (16 vs. 15, 64 vs 60, 512 vs 480), and the extra 64-bit memory controller (384 vs 320).

however since this is FSX it IS all about the 580GTXs architecture

GTX 580's *architecture* is no different than that of GTX 570 or 480 or 470 for that matter. They're all Fermi-derived GF100 chips. The 500 series chips do not differ from the 400 series chips architecturally, they've only been tweaked to allow for higher clocks and the artificially-limited cap on FP16 performance has been raised to match the Int8 fillrate. FP16 texturing is rarely used and I'm pretty sure the ancient 3d engine of FSX doesn't make use of it so this has no impact on FSX performance. In fact, FP16 is so rarely used, virtually no games use it. The ones that do have only a few FP16 textures because most GPUs choke on them.

highly overclocked 480s dont touch the performance of a 580 in FSX

This defies logic. Being the same architecture and all, the only differences are the amount of functional units which have been enabled as well as the clock speeds of these units. Since GTX 580 only has an extra SM compared to 480, it only has an extra 6.7% functional units. 580 is clocked 10% higher than 480. All in all, this leads to a total performance difference between the cards of anywhere from 10-20%, depending on the application and settings. Knowing these *facts*, can you explain why a 20% or higher overclock of a 480 would not outperform a stock 580?

much less the weaker 570 or I wouldn't have sent mine packing.

Every review of the 570 shows it is approximately the same speed as the 480 overall. Every single one of them. The only time the 480 is faster is at high resolution with lots of AA when VRAM size and bandwidth become more important.

Not hard to figure out why if you bench FSX often (Seems Hexus graph is similar to the one I posted before, are they wrong too?580/570/480/470

There are a few mistakes (namely pixel and texture fill rates) but that's not surprising given that most reviewers just read the press guide and don't dig any deeper. It has been exposed that GF100/GF110 has a bottleneck between the ROPs and the SMs which limits throughput to only 2 pixels/clock per SM. This means that a 16 SM part like the GTX 580 can only output at most 32 pixels per clock, despite having 48 ROPs. The ROP partitions are all being fed, but up to 1/3rd of the individual ROPs sit idle depending on workload distribution. This information was made public by Damien Triolet of Hardware.fr via Beyond3d forums. Accounting for this limitation, the actual maximum pixel fillrate of the 580 is 24.7GPixels/s. The theoretical maximum would be 37GPixels/s, but this does not account for the bottleneck between the ROP partitions and the SMs. My 570 at stock speeds has a realworld pixel fillrate of 22GPixels/s and a theoretical maximum of 29.3GPixels/s. At 875MHz these rates change to 26.3GPixels/s and 35GPixels/s. Only slightly down from the GTX 580 at stock. That being said, sheer pixel fill rate hasn't been terribly important for single display setups in quite some time. With a 2 megapixel 1080P monitor you'd have to draw something like 13,000 frames per second with 4xAA or less to saturate the ROPs. You'll run into shader processing and VRAM bandwidth bottlenecks long before you'll overrun the ROPs. As for texture fill rates, the chart has the wrong rate listed for the 570, it should be 43.9GTexels/s at stock. My card is pushing 52.5GTexels/s at 875MHz.

the additional ROP partition (48 vs. 40), the extra SM which packs an extra 4 TMUs and 32 SPs (16 vs. 15, 64 vs 60, 512 vs 480), and the extra 64-bit memory controller (384 vs 320). Knowing these *facts*, can you explain why a 20% or higher overclock of a 480 would not outperform a stock 580?

Nice thorough response Max, the key to your question no doubt lies within your own answer as quoted in the specs above.I hope you can read thru this response and understand, your bound to be happy that this is probably the last post from me :) New exciting kinda life changing project just fell into my lap for me and it comes with a nice new (over bearing) NDA that takes effect rather soon and it wont be jeopardized in any way.My responses in this forum are based more on literally hundreds of hours of actual benchmarking FS with primary complex scenery add-on development, nothing else outside of FSX performance matters (why me is so passionate about FSX).As you know logic and opinions built upon actual benching in FlighSim reveals things to us not found in the white papers, charts and certainly not really revealed in any otherr benchmarks what ever they maybe they synth or actual game etc,. I dont know 100% for sure but our latest performance problems of the past few years can probably be chalked up to the relationship between FSX/chipset/display drivers and the newer WDDMs of Vista and Win7 that has been interfering or at least has changed things up with regard to stability and smooth game play in FSX, autogen and 3d libraries and complex models still seem to be at the heart of the matter.Seems those extra bits 480 vs. 285/460/470/570 and 580 vs. 570/480 are a largepart of the key to more consistent frame rates -> even when the sim slows, with the 480 and now much more so with the 580 it just hangs in there like it never will do with a lesser GPU no matter how far the overclock is.So all this talk about FSX performance is not necessarily about higher and higher FPS though that certainly does come into play, but really it is a bout smooth flight under the worst scenarios that many of us get a kick out of (Big city/hub bad weather, bush flying with gaziilion trees etc.Especially when on final decent and approach or flying fast and low. Big difference between a stuttering 16-28 FPS with an occasional drop out of 6-9 vs. a smooth 19-32 with nary any drop what so ever. (very high settings here). This is the difference in hardware, in this game not any other, they don’t apply.Bench some flights coming into various complex airports in combination with decent AI and other add-ons with the various GPUs and there is a night and day difference, ->clock per clock that no amount of overclocking can ever make up for and at a certain point in fact will make stability, stuttering and micro stuttering worse!I am sure Nick, ******* and many others on this forum that crave getting to the bottom of FSX performance can verify this for you if you don’t have the time to test for yourself.Take care Max! :)Just one more suggestion for those building or purchasing new hardware for FSX :If anyone is looking for a good solid FSX rig and you will be cranking the CPU up to near 4Ghz and more, it really is worth it to spend the extra on that 580 and rebuild your FSX.cfg, forget about anything less (based on benching just about every relevent GPU made and have a nice collection gathering dust) (perhaps a real bargain used 480 with diffrent cooling, and as hard as it is to hear the 570 can be lumped into about the same catagory as the 470 which is just too close to the 460/285s, it is not the same, forget about that SSD or wireless gaming keyboard etc, get the 580 (unless as MAX points out your on a fixed budget) It is not like it was with FS9,FS04,2K etc when the underdog chip was the way to go, something different has been at play here and the 480 and then 580 have been dealing much better with our faster overclocked systems. Flight?Hope we all hear more soon! rides off into sunset....or duststorm... .

Seems those extra bits 480 vs. 285/460/470/570 and 580 vs. 570/480 are a largepart of the key to more consistent frame rates -> even when the sim slows, with the 480 and now much more so with the 580 it just hangs in there like it never will do with a lesser GPU no matter how far the overclock is.

"those extra bits" that account for 6.7% more functional units. An overclock of 6.7% would cancel this out. Simple math. You're entitled to your opinion, but reality does not concur.

So all this talk about FSX performance is not necessarily about higher and higher FPS though that certainly does come into play, but really it is a bout smooth flight under the worst scenarios that many of us get a kick out of (Big city/hub bad weather, bush flying with gaziilion trees etc.Especially when on final decent and approach or flying fast and low. Big difference between a stuttering 16-28 FPS with an occasional drop out of 6-9 vs. a smooth 19-32 with nary any drop what so ever. (very high settings here). This is the difference in hardware, in this game not any other, they don’t apply.

If you're watching the built-in FPS counter to determine performance, you're wasting your time. The polling rate is far too low to be truly useful, you need to use a 3rd party frame rate reporting tool such as FRAPS.

Bench some flights coming into various complex airports in combination with decent AI and other add-ons with the various GPUs and there is a night and day difference, ->clock per clock that no amount of overclocking can ever make up for and at a certain point in fact will make stability, stuttering and micro stuttering worse!

If you've got such extensive benchmarking experience with FSX, why don't you share your findings with us? Surely you must have some specific numbers to report, perhaps even some handy videos which demonstrate what you state.

freebird77: You understand a lot of what you said now is the same reasoning I used against the 580 in the first place. Granted I did underestimate the cooling aspect of the new card. But that's why we have benchmarks. Good luck where ever you are off too.

"those extra bits" that account for 6.7% more functional units. An overclock of 6.7% would cancel this out. Simple math. You're entitled to your opinion, but reality does not concur. If you're watching the built-in FPS counter to determine performance, you're wasting your time. The polling rate is far too low to be truly useful, you need to use a 3rd party frame rate reporting tool such as FRAPS. If you've got such extensive benchmarking experience with FSX, why don't you share your findings with us? Surely you must have some specific numbers to report, perhaps even some handy videos which demonstrate what you state.

Last post (really)Fraps? LOL! You got me laughing so hard I have tears! Yes sir thanks for the suggestion Max, I'm a bit familiar with Fraps though... -was using long before my first decent hardware review got bits and peaces widely published, was GF3 vs, GF4 for FS back in 2002 with braging rights to the line format before it became poular ;)Doing some digging there are some remnants..Received quite a bit of milage from that first review, started off a carreer on the side (pseudonym used as common), opened doors and access to hardware and software heaven ever since...This link may take a while to load an early unedited version with all the misspellings and bad grammar GF3 TI VS.GF4 TI in FS2KRespectfully, I know what you think but you are dead flat wrong about any overclock cancleing out the differences, you just dont seem to understand and Likely never will until you really get involved with benching the sim with all options available to you, many long time users of FSX can verify that overclocking the GPU as little to no effect on FSX and will more likely cause instability and micro stuttering when stressed, it is not like the old days Max.Several times its been previously stated why more info cant be posted here and there is nothing I can do no matter I feel about it so thats a dead issue.All this digging up old work and now finding this pic brings back too many memorries of late sleepless flightsim nights and the thrill of bring your hometown to life - screenie of the old 256 adaptive color palette photoreal Rochester project, first use of 1 meter res scenery in FS! might not look like much these days but was cutting edge back when res was limited to more than 12ft/pixel! Backdoor codeing made it all possible...Love that accurate FS SDK! grey hairs I tell you!Take care everyone!

Best of luck freebird in your new endeavours, whatever those are, unless it involves nuking all the Sandy Bridge factories on earth :PNo seriously, good luck

freebird77: You understand a lot of what you said now is the same reasoning I used against the 580 in the first place. Granted I did underestimate the cooling aspect of the new card. But that's why we have benchmarks. Good luck where ever you are off too.

Hey Veeray, no not the same go back and re-read the posts, the position I took was based the exact same thing as above, correct me if I'm wrong but you laso mentioned faster clock as being the only factor etc, I know my post are long winded but I dont think people are taking the time to read thru and often trip right over the main points.I've already been told I'm not a good writer, so maybe thats the problem :)Thanks and take care!

So you posted a review in 2002 with some untitled graphs. Cool. I posted one in 2010 and my graphs were labeled ;) Your statement that functional unit deficiencies cannot be overcome by clock speed show that you simply do not know what you are talking about.

It's a 2002 MATLAB generated graph of time series data.. 2010 version pretty much looks the same.

you simply do not know what you are talking about.

Your really a class act there Maxie, just showed you I've been doing this in a very thorough way for more than ten years dedicated to this sim...can you actually do anything besides cut and paste? Just kidding LOL!Have fun everyone!

Your really a class act there Maxie, just showed you I've been doing this in a very thorough way for more than ten years dedicated to this sim...can you actually do anything besides cut and paste? Just kidding LOL!Have fun everyone!

What are you talking about? You posted some UNSOURCED, untitled graphs as proof of your vast MSFS knowledge. You failed. What are those graphs of? Where are they from? Regardless, they have no bearing on FSX performance nor do they establish any kind of credibility or knowledge on your part.

Guys, come on. Enough.This started as an interesting thread, now its purely your personal agenda. ENOUGH OF THAT!!!

Guys, come on. Enough.This started as an interesting thread, now its purely your personal agenda. ENOUGH OF THAT!!!

Yes, please guys.

Hi,Well, looks like my setup is number 3 without the overclock. Once I drop out of the top five I'll look to update. :Big Grin: