I don

The ATi card is faster in FSX because of its micro-architecture, not in spite of it. The core engine speed of 850MHz is why it can push more polygons than the GTX 285. The efficiency of the engine is near maximum, allowing it to push up to 830 million polygons per second as explained here: Beyond3D Cypress Architectural Analysis.Since GPUs are only capable of setting up at most one triangle per clock cycle we can derive the efficiency of this setup engine by using the following simple formula: triangle throughput / engine speed * 100 = efficiency in percent830/850 * 100 = 97.6% efficiency.GTX 280 only pushes 510 million polygons per second as demonstrated here: iXBT Labs GTX 280 Investigation. 510/600 = 85.0% efficiency.This is for the GTX 280 however, so let's give NV the most generous comparison possible and compare the HD 5870's stock triangle setup rate to that of an overclocked GTX 285.A user at the Beyond3D forum measured the maximum triangle throughput of his overclocked GTX 285 with a 700MHz engine speed to be at 600 million polygons per second here. If we use this figure as our (generous) baseline we arrive at the following conclusion:600/700 = 85.7% efficiencyWe have 97.6% efficiency for the ATi card with a 21.4% clockspeed advantage over an overclocked GTX 285 @ 700MHz with an efficiency of 85.7%.830/600 * 100 = 38.3% increase in triangle setup rateThis is why the HD 5870 outperforms the GTX 280/285 as per Avsim user tfm's tests as demonstrated at these very forums here: ATi 5870 vs. GTX 285 head-to-head comparison.You can see in the figures from his initial test featuring a similarly overclocked GTX 285 @ 702MHz which produces an average framerate of 36.7 FPS compared to the average framerate of a stock HD 5870 @ 850MHz producing 51.3 FPS gives an increase of 39.7%. This falls almost directly in line with increase in triangle setup rate. The variance in FPS from difference in triangle setup rate is a mere 1.4%.Nick:TMU = Texture Memory Unit and is an outdated term, coined by 3dfx. They are now referred to simply as texture units. Re: memory bus-widthIn and of itself bus-width is only a performance metric at the clock cycle level. You need to know how many memory clock cycles there are in order to derive bandwidth. Sure, GT200b in GTX 285 guise can fetch twice as much data per clock as HD 5870, but there are only (effectively) 51.7% as many clock cycles as HD 5870 (2484/4800) * 100 = 51.7%. Re: "shader-based rendering" vs. "triangle-based rendering"I hope you're just using these terms for the sake of simplification for the audience.Also, ATi's new architecture does away with dedicated texture interpolation units and instead uses the shader core (SIMDs) to perform attribute interpolation (i.e. filtering), since this is a math operation anyway, it only makes sense to get more utilization out of those 1600 ALUs than to use dedicated hardware for the task, especially since ATi's ALU utilization is notoriously low compared to NV's. ATi retains the dedicated texture fetch units (simply called texture units) and increases this count to 80 with RV870, which is the same number as GTX 280/285, and here once again we see the difference in engine speed come into play as we need to multiply the number of texture units by that of the engine speed to derive the maximum number of texture samples per second. 850 million x 80 = 68 Giga samples/s700 million x 80 = 56 Giga samples/sNot a huge difference but definitely a matter of ATi finally catching up to NV's massive texturing capabilities, and in fact surpassing them.Now I wouldn't be surprised to see all this change once again with the release of Fermi, but Fermi is not yet here and there is no solid launch date known publicly, nor do we know final product specifications so it is impossible to derive triangle setup rate and texture filtering rates so we can only work with what we have.Bottom-line: The Radeon HD 5870 is the fastest card for FSX right now, and is likely to be so for at least several months until Fermi launches, possibly longer.

Absolutely which is what I said aboveThat memory bus is one of the advantages the Nv card has over the ATi for FSX.. its been that way since the release of the 8800 series cardsPhil Taylor posted the importance of that bus speed to FSX in his blog as wellYour pipe speed is what makes that work.. GDDR doesnt mean squat if the pipe is narrow. Its 6 of 1 or 1/2 dozen of the other In reality the 5870 has a SLOWER memory bandwidth even with GDDR5 GTX285= 158.98GB/s 5870 = 153.60 GB/sbut regardless what FSX responeds to is core speed, memory bus with a good shader clock on a healthy number of Texture Multiplier UnitsYou can have a massive pixel/texture fill rate and a massive shader operation speed.. if the render engine is not 'shader' based and the bus aint gettin the data from point A to point B any faster then you wont see any advantage and the core operations have to wait for the data to get across the bus so having a higher core speed on a restricted bus isnt making the card faster to FSX either. The GTX285 still trumps it in FSX.. not by much but it still runs FSX better. Other games are different, which is why I posted its a good trade-off if you want the Crysis or other modern shader render game support

thanks for the info. Nick. that's what i thought. but at the same time Techguys' info. seems to make sense as well. i'll admit that you guys know a lot more about this than i do.thanks to all that responded as well.happy flyin, fm

I simply what I post for everyone Max and regardless of your long tech post geared at trying to make the point you know what you are talking about, the card is not what you are making it out to be to the application in questionThe only reason the card even matches the 285 and can exceed it in some areas of MSFS render is as you pointed out correctly, I'll give you credit for that part, it is of a micro-architecture and it supports a 200Mhz higher core clock as was mentioned the core speed is part of the equation,. Even with that new desing I wonder how far behind that 5870 would be if the 285 was clocked up an additional 150-200Mhz. Actually I know what would happen, but thats not important. It is also a year ahead of the 285 which if you ask me isn

thanks for the info. Nick. that's what i thought. but at the same time Techguys' info. seems to make sense as well. i'll admit that you guys know a lot more about this than i do.thanks to all that responded as well.happy flyin, fm

It is not a bad card and if you are coming from something older and intend to drive the card on a decent modern system, investing in it would not be a bad choice. What I have seen thus far in + for the 5780 is that it can match and exceed the 285 but there were no indications of any significant increase over a 285. I did see a consistent perf difference between the 2 cards (Ati trailing) under certain render load conditions in testing and verified with driver changes, again not a significant loss either. It is new on the market as well so there is room for further improvement as ATi moves through their driver dev cycle for the card and its variantsThose are the facts from my tests. Test results can vary from machine to machine. ... and user to user as wellAt this point in time there is nothing spectacular about the card over the 285 in MSFS as some would suggest

I simply what I post for everyone Max and regardless of your long tech post geared at trying to make the point you know what you are talking about, the card is not what you are making it out to be to the application in question

Words < numbers. And you damn well better know that as an engineer.

The only reason the card even matches the 285 and can exceed it in some areas of MSFS render is as you pointed out correctly, I'll give you credit for that part, it is of a micro-architecture and it supports a 200Mhz higher core clock as was mentioned the core speed is part of the equation,. Even with that new desing I wonder how far behind that 5870 would be if the 285 was clocked up an additional 150-200Mhz. Actually I know what would happen, but thats not important. It is also a year ahead of the 285 which if you ask me isn't saying much for ATi as they are still behind the 8-ball in MSFS render and may continue to be behind because as of yet the next gen of Nv card is not out for evaluation

If ifs and buts were candy and nuts... Yeah, if everyone had phase-change cooling in their cases they could run their GTX 285s at 850MHz but we all know that's not happening any time soon.

The way you have presented information Max it reads (to me) like a tech sales pitch and that the card should replace the 285 in all systems.

See my signature. I own a 285. I type this message on a system with a 9600 GT. My secondary gaming machine has an 8800 GT. Yeah, I'm definitely an ATi shill.

Although this card is the first to meet Nvidia in FSX render anyone owning a 285 would be crazy to drop it and put in that ATi card.. the perf changes where they do appear are not that great especially with a payware AC regardless of what your post is trying to suggest.

My post isn't "trying to suggest" anything, unlike yours. Numbers don't "try to suggest" things, they prove statements true or false.

The only way that may change is if later driver revsions start to display a much greater range of result past the 285 in FSX. Is that possible? Sure. Right now it is not even running 150-200Mhz faster than Nvidia on a next gen design.

39.7% higher average framerate is very certainly "150-200MHz faster than NV". The math has already been shown.

By the way.. these are people asking simple questions about video cards,. They are not a tech audience

Funny, never stopped you in the past

Once again... For those who would prefer an ATi card and/or would like to have support for the more modern games, the 5870 is a good choice and the first ATi card worth investing in a MSFS system under those conditions. Those who own a 285 would be much better served waiting for a period of time till the dust settles and Nv get their next product on the table.

I never suggested anything otherwise.

Max, please.. grow up

Max, please.. grow up

Take your own advice. I've proven mathematically why the 5870 is a better card than the GTX 285 and all you seem interested in doing is preserving your precious reputation. You have yet to offer even a shred of evidence why what I say is wrong. If you have anything useful to add to this discussion please do.

If you have anything useful to add to this discussion please do.

Yes, I do as a matter of fact39.7% higher average framerate is very certainly "150-200MHz faster than NV". The math has already been shown. Can you point me to the FSX benchmark which displays a 39.7% average increase in the frame rate as you posted above?I will even take 25%no?OK, I will take 20% MaxI would like to see that test resultor is that just another mindless sales pitch number you read somewhere and repeated that has nothing to do with the ATi card ability with respect to this thread, your post or running/rendering MSFS codeor any game for that matterIt places the rest of the sales pitch type tech blab from you into real perspective

Can you point me to the FSX benchmark which displays a 39.7% average increase in the frame rate as you posted above?

Max, NickN, since the two of you have obviously have the ability to carry on an intelligent conversation about the technical aspects of both of these GPU

Max, NickN, since the two of you have obviously have the ability to carry on an intelligent conversation about the technical aspects of both of these GPU

Max can handle it for you.. he knows what hes doing and what hes talking about

Actually just one person contributing technical observations about the comparison isn

Yes, I do as a matter of fact39.7% higher average framerate is very certainly "150-200MHz faster than NV". The math has already been shown. Can you point me to the FSX benchmark which displays a 39.7% average increase in the frame rate as you posted above?I will even take 25%no?OK, I will take 20% MaxI would like to see that test result

In my first post I provided 4 links to various websites as reference, including a link to these very forums in which a comparison between the HD 5870 and GTX 285 is carried out. The link (and thread) is titled "ATI 5870 vs. GTX 285 head-to-head comparison". Of course, if you had actually read what I posted you would know that.

or is that just another mindless sales pitch number you read somewhere and repeated that has nothing to do with the ATi card ability with respect to this thread, your post or running/rendering MSFS codeor any game for that matterIt places the rest of the sales pitch type tech blab from you into real perspective

What were you saying about "growing up" earlier?

In my first post I provided 4 links to various websites as reference, including a link to these very forums in which a comparison between the HD 5870 and GTX 285 is carried out. The link (and thread) is titled "ATI 5870 vs. GTX 285 head-to-head comparison". Of course, if you had actually read what I posted you would know that.

I did read I guess you didnt.. I am looking for a MIN of 20% increase in the frame rate in FSX head to head on payware planes Max as compared to your claim of 39.7 in which I gave you a lot of leeway to come up with.. please post thatIm not looking for 5-10% Max, I can get that out of a clock without replacing my video cardAnd I am finished playing your thread game son.Please let djt01 know exctly how to set up those benchmark flights in FSX..

Your scenario isn't a valid GPU test. Payware (i.e. complex) A/C shift the bottleneck to the CPU. Really simple stuff here...