GTX 690 or GTX Titan?

[Ferzerp]

Also, you say that 50ms = 20fps. I would really like to know how you arrived at that number

Math.

If you can't grasp that simple fact, you really don't understand enough to even have this discussion. I'm not trying to be mean, but if you don't understand that 1/.05 = 20, I'm not interested in taking the time to teach you.

 

[Rvenger]

Wow, I rather have MS than look at this thread.

 

[blackened23]

Math.

If you can't grasp that simple fact, you really don't understand enough to even have this discussion. I'm not trying to be mean, but if you don't understand that 1/.05 = 20, I'm not interested in taking the time to teach you.

Yeah, i'm aware of that and maybe you should check again. So going by this flawed and incorrect equation that would indicate a maximum framerate of 50 fps at 20ms. Is that what you're saying? The Titan has a maximum framerate of 50fps at 20ms. Alrighty.

 

[SirPauly]

Wow, I rather have MS than look at this thread.

Hehe!

 

[rgallant]

Wow, I rather have MS than look at this thread.

-me too

 

[BallaTheFeared]

Also, you say that 50ms = 20fps. I would really like to know how you arrived at that number

:whiste:

Helper chart from ABT:

 

[SirPauly]

We need to define microstutter then in the context of frametimes. Let's say SLI 680 looks like this across 6000 frames:

10ms 7ms 13ms 15ms 15ms 10ms 7ms 15ms 10ms (repeating)

While Titan is:

25ms 25ms 30ms 20ms 22ms 22ms 22ms (repeating)

That's what most of the consecutive frametime charts are showing. A couple of you stated that consecutive frametimes as charted over a graph can define microstutter - call me crazy but that would indicate that SLI has less microstutter. How do you define "microstutter" in the context of these frametime latency charts?

Also, you say that 50ms = 20fps. I would really like to know how you arrived at that number, and what you extrapolate 25-30ms to be because that's where Titan happens to land. SKYMTL of hardwarecanucks mentioned that less than 50ms is imperceptible to the human eye and I would tend to agree but regardless, I would like to know how one equates ms frametime to a framerate, i'm just curious - not being facetious.

This may help:

http://alienbabeltech.com/main/wp-content/uploads/2013/01/fps-ft.jpg

 

[SirPauly]

:whiste:

Helper chart from ABT:

We posted at the same time and posted the same thing -- LOL!

 

[BallaTheFeared]

This may help:

http://alienbabeltech.com/main/wp-content/uploads/2013/01/fps-ft.jpg

Do you know how hard that was for me to find with google? It took like 3 different search keyword attempts :$

 

[SirPauly]

It's a handy gauge though!

 

[lavaheadache]

Maybe now we can move past this issue

 

[blackened23]

This may help:

http://alienbabeltech.com/main/wp-content/uploads/2013/01/fps-ft.jpg

I'm looking for justification of the equation. I find it wrong and silly - it suggests that the maximum framerate for any frametime latency at 20ms to be 50 fps. So you're telling me that every card at 20ms (20ms is actually LOWER than what most single cards GET, remember this) that they can only produce 50 fps? Give me a break.

Break down the variables of the equation, how you arrived such an equation, and how such variables equate to framerate delivery.

You're suggesting that the baseline 1 divided to 5% (how you decided that 50ms = 5%, i'm not quite sure). You're telling me that 20ms = 50fps. 20 ms is *lower* than what most single GPU cards produce. Again, give me a break please. This equation that was developed out of thin air is apparently telling us that any single GPU can only produce 50 frames per second *if* they are at 20ms frametimes. Please. Let's not develop completely nonsensical equations that make absolutely no sense - without explanation of the variables and how they equate to frame delivery.

 

[BallaTheFeared]

I don't think he understand the concept of frames within a second.

 

[lavaheadache]

There are 1000 ms in a second. If you are displaying frames every 50 ms then this is what you come up with . 1000/50= 20. 20 frames at 50ms each in every second

 

[Ferzerp]

Bye troll. Done with you.

 

[AdamK47]

Oh Ferzerp, you be caarraazzy!

 

[Phynaz]

What was this thread about again?

 

[Larnz]

Leprechauns I think.

 

[blackened23]

Oh sure. Get mad. And don't explain the equation. And nobody explained, still, how SLI 680 has less microstutter than Titan single GPU. :awe:

Bolded part is the most applicable. Did anyone explain it? Nope. At every frame along the chart, 680 sli has lower frame latency. Carry on.

 

[BallaTheFeared]

Look blackended, I'm not sure how this is so hard for you to grasp but let me try to break it down so you can understand it in a one second example.

Let's say we have a GTX Titan and a 690, the Titan gets 30 fps, the 690 gets 60... Of course these aren't real but we want to make it simple.

Example Titan (smooth): Within 1 second, 30 frames are flashed on the screen. 1 second is 1000 MS, so how fast a video card is is based on how many frames the card can display in 1000 MS. So assuming perfect frame delivery the Titan is giving 30 frames at 33ms intervals. So we take 1000 divide that by 33, and we get 30 fps. That means within one second of time, Titan delivered 30 frames at 33ms intervals within a second of time.

The output graph would look like this --------------------------------------------------------

Example 690 (MS): Within 1 second, 60 frames are flashed on the screen. 1 second is 1000 MS, so how fast a video card is is based on how many frames the card can display in 1000 MS. So assuming poor frame delivery the 690 is giving 60 frames at alternating 33 and 8.3 MS intervals. So we take our frame average, which is 16.7 and divide that into 1000, which gives us 60 fps. That means within one second of time, 690 delivered 60 frames at alternating 33 and 8.3 MS intervals within a second of time.

The output graph would look like this /\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/


I hope that helps.

 

[sensiballfeel]

/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/


I hope that helps.

Lots of blah blah stuffs. No even running titan and post big nonsense. nvidia waterboy :biggrin:





This is your second exact comment directed towards this member.

Knock it off. If you don't having anything to add to the discussion, stay out of the thread


esquared
Anandtech Forum Director

 

[BallaTheFeared]

heh :thumbsup:

 

[rich_]

titan only outperforms the 690 option when you go balls to the wall three-way SLI.

Titan: single GPU roughly equivilent to two 670's.
690: Double GPU 680.

3x SLI = maximum amount of sli for efficient scaling.

2xSLI 690 = equivilent to SLIx4 680, gives full scale of three 680's and lesser but still present boost of performance of a fourth 680.
3xSLI TITAN = full scaling, each titan generates about two 670's of performance, so 3xSLI titan gives performance of nearly six (call it five conservatively) gtx 670's. This is better than three 680's + a little from the fourth 680gpu that you would get from SLI 690.

 

[Elfear]

Look blackended, I'm not sure how this is so hard for you to grasp but let me try to break it down so you can understand it in a one second example.

Let's say we have a GTX Titan and a 690, the Titan gets 30 fps, the 690 gets 60... Of course these aren't real but we want to make it simple.

Example Titan (smooth): Within 1 second, 30 frames are flashed on the screen. 1 second is 1000 MS, so how fast a video card is is based on how many frames the card can display in 1000 MS. So assuming perfect frame delivery the Titan is giving 30 frames at 33ms intervals. So we take 1000 divide that by 33, and we get 30 fps. That means within one second of time, Titan delivered 30 frames at 33ms intervals within a second of time.

The output graph would look like this --------------------------------------------------------

Example 690 (MS): Within 1 second, 60 frames are flashed on the screen. 1 second is 1000 MS, so how fast a video card is is based on how many frames the card can display in 1000 MS. So assuming poor frame delivery the 690 is giving 60 frames at alternating 33 and 8.3 MS intervals. So we take our frame average, which is 16.7 and divide that into 1000, which gives us 60 fps. That means within one second of time, 690 delivered 60 frames at alternating 33 and 8.3 MS intervals within a second of time.

The output graph would look like this /\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/


I hope that helps.

:thumbsup:

Good explanation Balla.

 

[aaksheytalwar]

In short:

avg is irrelevant

only standard deviation is relvevant