To SLI or not?

[Big.Wayne]

Hey Emlyn!

obviously what has just happened is the classic "switcheroo" . . . I'm not sure if your recommending hardware for me? . . or the O.P . . . or yourself! . . . I'm gonna have to take good care in the future how I word my posts!

If you have to go SLI 460s, definitely get the 1gb model #30

But then you're stuck with 2 cards with a lack of texture memory

I wouldn't buy 2 1gb GTX460's at the same time either... #64

If you're running SLI, chances are you're running larger resolutions and still wanting all the details.

I know that the 2 x 768mb 460's offer very good performance for the cash, but I'd still not buy 2 of them. #74

If you're running SLI, chances are you're running larger resolutions and still wanting all the details.

The 1920x1200 results look good . . . imagine the 1680 x 1050 gamer! . . .

1. 1680 x 1050 (17.71%)
   
2. 1280 x 1024 (16.16%)
   
3. 1920 x 1080 (14.27%)

Steam Hardware & Software Survey

A GTX470 would cost less than the 460's. That's a pro right there

Well yeah, the GTX 470 comes in at about £20 cheaper! than a pair of 768MB GTX 460's . . . but look closely at the 1920x1200 bench results in post #17 . . . looks like good gains for the little extra ££

Alien Vs Predator 4xAA
25-62
24-47

Battlefield:BC2 4xAA
45-62
35-45

Dirt2 4xAA
88-106
65-71

Far Cry 2 4xAA
66-116
56-82

Just Cause 2
43-71
29-47

Unigine v2.0 4xAA
25-58
22-38

I like budget purchases, but there are times when they are just not needed. If I was running a larger monitor I'd take the 470 any day

And then we have just a little solo GTX 460 768MB against the mighty GTX 470 . . . assuming someone went with a £223 GTX 470 that’s £100 pounds premium . . . its a bitter pill to swallow for such little gains . . . I think personally a single GTX 460 768MB would do me @1920x1200, little bit of tweaking and enjoy . . . couldn't justify the extra premium ££ . . . how about you?


11 games tested, averaged 14FPS advantage to GTX 470

Futureproof isn't really a word that has a place in PC gaming but... The 470 is the better option when looking to the future.

"Futureproof" has never worked out for me . . . ever . . . always seems like a good move at the time then fast forward 12-18 months and new cards *Hype Hype* . . . obviously you know best my good man, you are unlikely to convince me that a single GTX 470 is better than a pair of 768MB GTX 460's now although each to their own and full respect to you sir!

This isn't about filling slots on motherboards, it's just to do with what is the wisest move.

Hmmm . . . somebody pays a premium for a dual-gpu mobo with x8/x8 slots or x16/x16 slots? . . . . and leaves one of the slots empty? . . .

Why do a lot of people (not all) buy a premium mobo and only fill one of the *two* PCI-E 2.0 slots with a card costing approx (within £20) the same as a pair of cards that perform better? . . . and with that nice PSU sitting there too?

At some stage you have to admit that 768mb memory on a graphics card IS a limitation

@ 1920x1200 I totally agree . . . in about 12-18 months I reckon . .

Please check the benches in post #17 . . . look at them closely . . . and maybe search some more out and look at them . . . the 768MB SLI'ed GTX 460 are rocking!

in 12-18 months if one fancies a change when the new uBer cards are out one simply "recycles" them . . . . remove the two 768MB GTX 460's, give them a dust off, take a few pics, put them back in their *mint* unused boxes with unused accessories and sell them @ auction for "Good" money £££ probably to someone using a 1680x1050 display . . . and at that point in 12-18 months one re-evaluates the GPU choices and starts again, its seems like no-brainer for me, don't see why it isn't for anyone else too?

In a thread where SLI is mentioned, 2 low memory cards are not the way forward.

c’mon Emlyn . . . 768MB isn't low memory for everyone is it . . . we had this chat . . . you can see the benchmarks . . . plenty of games run just fine with high textures and AA with what you are calling "low memory" . . . lets not go around in circles and scare off people . . . if you want 1GB+ cards that’s fine but just bear in mind the majority of peeps still are using 512MB cards . . . Lets not be looking down our noses eh and insisting that anything below 32xAA is for the non serious gamers!

I can see you raising the price of the 768MB GeForce GTX 460

 

[Duff-Man]

Just remember I am "Devils Advocate" so you need to answer the questions and not buckle under the cross examination . . . you are making a statement delivered in a technical style that is a bit "Smokescreen & Mirrors" . . . and I am having to get you to decrypt it into normal everyday terms that can be understood by normal everyday punters . .

Seriously...

What are you trying to get at here?

I'm sorry if you find my writing style a little too technical. The fact is, I was performing an investigation for my own curiosity and to provide more information on microstutter to the technical community. I wanted data from other technically minded people, who could run the tool in different scenarios and have interest in dissecting the results. That's the main reason I made a thread at XS and on hardforums rather than on here.

. . . you are having to backup your work and I am suspecting now your conclusions may be "flawed" due to the "The Observer effect" . . . it happens and its nothing personal but I do think perhaps this whole issue has been blown totally out of proportion by a great many people . . .

I don't recall backing up anything, only trying to explain the results. As for your "observer effect" stuff, I've eliminated as many variables as possible, time and interest allowing. When I ran out of hardware configurations, I asked others for their input in circumstances as controlled as I could reasonably expect. The tool is out there, feel free to do your own subjective or objective testing, but I finished looking into this a good month ago.


If you have questions about understanding the results I collected, then I am still happy to try to explain. You can draw your own conclusions from then. But, I have no interest in trying to "prove" anything to you. I'm not some evangelist trying to prove a point...

Look, I first came across this phenomenon when I "upgraded" a pair of nvidia 6800GTXs to a 7800GTX (expecting to buy another 7800 the next month). I was playing HL2 and noticed that the gameplay seemed much smoother with the 7800, even though the framerates were around the same, or even a little lower. That's when I first got curious. A couple of years later I noticed people posting framerate variation plots and I realised that this could be the cause of the difference (I has seen the same thing, subjectively, when upgrading 7800GTX-SLI to an 8800GTX). When I got a 4870x2 I wrote a program to test out the amount of framerate variation using one and two GPUs (wasting a few hours at work ). Anyway, the results of the tool you already know about and I feel I know enough about the thing to understand that subjective difference I first noticed years ago - what it is, why it arises and how it works. I'd love to do some side-by-side pepsi-challenge type testing, but that's not going to happen unless someone wants to lend us a load of hardware.

 

[bfar]

"Microstutter" is framerate variance on the frame-by-frame level. If somebody wants to call something else microstutter, then that's fine, but it won't be the issue we're discussing.

Thats the problem dude - we don't know for sure what causes it!!
You haven't given a concrete proof that frame rate variance is the cause of Microstutter. It might be, in which case you're on to something, but it could be something else altogether. We don't even know if high variance is perceivable. You'd need a test group just to form that theory. But you're so sure, you only want to talk about frame rate variance, and not consider that there could be other issues at play.

And I definately disagree with you when you say that high variance makes it look like frame rates are lower than they are. avg fps is avg fps.

I'm sorry dude, you put some great stuff together, but it's not enough to convince a lot of people.

 

[An Exception]

 

[Peter450]

A 5870 would be better (quieter, cooler, slighty better performance) than a 470 right?

Unless your overclocking, most cards emit very low levels of fan noise at stock settings, my 470 at stock fan and speed settings could not be heard above the case fans, which are themselves generally quite quiet

I expect a 480 will be similar, and although i have not experiance ATI's newer efforts i would be very surprised if they made much noise and stock fan speeds

On the Nvidia (No idea on ATI) it's when you OC the card and bump the fan speed past 55% that it starts to make more noise than your average case fan, at the default fan speeds these cards are pretty quiet IMO

 

[An Exception]

^^^
I'v heard leakage varies considerably with GF100 and therefore heat.
Some chips apparently run pretty cool and clock well, and others not so much.

 

[An Exception]

deleted

 

[Big.Wayne]