8800GT or 9600GT

[Munky]

At least half of the reviews linked in this thread used newer drivers for the 9600gt than the 8800gt, and as such, I consider those reviews invalid. However, looking at the cards' performance in Crysis, in cases when they use the same driver version, I can see that the the difference is about 25-30%. And I'm talking about cards with reference clocks, not a highly OC'd 9600gt compared to a stock 8800gt. Here's a few examples:

http://www.behardware.com/arti...nd-radeon-hd-3650.html
http://www.computerbase.de/art...2/17/#abschnitt_crysis
http://www.legitreviews.com/article/666/10/

So you can tell yourself the difference is only 10% in games that both cards get well over 60 fps, but when a game really stresses the cards, the performance gap widens considerably.

 

[evolucion8]

Originally posted by: j0j081

Originally posted by: Azn
What is this guy's deal with 9600gt?

he is just stating the facts and supporting what I said earlier in the thread about the 8800gt being only about 10% faster than the 9600gt, opposed to 20% or more as someone else claimed.

THE POINT IS, when games are unplayable on a 9600gt, they are going to be unplayable on the 8800gt as well. Therefore if someone is on a budget and can't afford the 8800gt, the 9600gt is just about as good for a decent amount less. the 9600gt also requires less power and is great in SLI scaling.

That's a bit doubtful, both can have very little difference in performance now, but once you start cranking the Anti Aliasing and the resolution, the gap widens considerably, adn even though the 9600GT seems to be more efficient overall than the 8800GT, shader intensive games will make the 9600GT to fall behind, that's why it sometimes it gets outperformed by the HD 3850 while the 8800GT don't in some game scenarios

 

[Lithan]

Originally posted by: munky
At least half of the reviews linked in this thread used newer drivers for the 9600gt than the 8800gt, and as such, I consider those reviews invalid. However, looking at the cards' performance in Crysis, in cases when they use the same driver version, I can see that the the difference is about 25-30%. And I'm talking about cards with reference clocks, not a highly OC'd 9600gt compared to a stock 8800gt. Here's a few examples:

http://www.behardware.com/arti...nd-radeon-hd-3650.html
http://www.computerbase.de/art...2/17/#abschnitt_crysis
http://www.legitreviews.com/article/666/10/

So you can tell yourself the difference is only 10% in games that both cards get well over 60 fps, but when a game really stresses the cards, the performance gap widens considerably.

Maybe we should read my original post, Hmm?
And don't give me that driver bullshit. I've tested em. There's not any measurable difference between the 174 betas and 169's in performance with 8x cards, and the 169's are actually stable... so I'm gonna trust the guys using stable drivers rather than the ones using hacked 174's which I've tested and confirmed to crash and have performance issues.

Originally posted by: OmegaShadow
but by then wouldn't there be better cards than 8800gt and the 9600gt at price for performance?

Except the games the 3850 beats the 9600gt in, it also beats the gx2 in, you saying that the 3850 is better at shading than a gx2 evo? Also, I should tell you a story about a little function called traceroute before you start throwing your hat in with people. It's the most novel little dos function that most idiots who pretend they know comps evidently are unaware of. Tells you all sorts of fun stuff about people.

 

[Munky]

Originally posted by: Lithan

Originally posted by: munky
At least half of the reviews linked in this thread used newer drivers for the 9600gt than the 8800gt, and as such, I consider those reviews invalid. However, looking at the cards' performance in Crysis, in cases when they use the same driver version, I can see that the the difference is about 25-30%. And I'm talking about cards with reference clocks, not a highly OC'd 9600gt compared to a stock 8800gt. Here's a few examples:

http://www.behardware.com/arti...nd-radeon-hd-3650.html
http://www.computerbase.de/art...2/17/#abschnitt_crysis
http://www.legitreviews.com/article/666/10/

So you can tell yourself the difference is only 10% in games that both cards get well over 60 fps, but when a game really stresses the cards, the performance gap widens considerably.

Maybe we should read my original post, Hmm?

You mean the post where you claim the difference is not even close to 20% when I showed you situations where the difference is almost 30%?

And don't give me that driver bullshit. I've tested em. There's not any measurable difference between the 174 betas and 169's in performance with 8x cards, and the 169's are actually stable... so I'm gonna trust the guys using stable drivers rather than the ones using hacked 174's which I've tested and confirmed to crash and have performance issues.

The 169's are so stable, that when I request driver AA, it doesn't even work. I've switched to 174 betas, and not only does AA work now, but I haven't had a single crash or any performance issues.

 

[Lithan]

I was more thinking the post where I specifically noted that at higher reso's it will be 20%+. Yeah, reading is fun.
You see, a SITUATION where the difference is great does not equal an AVERAGE of greater difference, which is what we are discussing here. Hell if you wanna use situational performance the 8800gtx is 200%+ faster than the 8800gt. Most people go by the average though, as it's a much more fair comparison.

How often do you play HGL? Drops to <20fpm (yes that's frames per MINUTE) happened within 20mins of loading almost every time I loaded it up on 174's. Revert to 169's, problem gone. I don't bother with forcing driver AA on, so It may have problems in 169 I'm unaware of.

 

[AzN]

Originally posted by: Lithan

Maybe we should read my original post, Hmm?
And don't give me that driver bullshit. I've tested em. There's not any measurable difference between the 174 betas and 169's in performance with 8x cards, and the 169's are actually stable... so I'm gonna trust the guys using stable drivers rather than the ones using hacked 174's which I've tested and confirmed to crash and have performance issues.

Looks like you didn't test hard enough. I tested 169 drivers vs 174 drivers on my 8800gs I gained 3fps in Crysis high and over 10fps in medium settings alone.

edit:
Expreview found measurable performance difference (average of 6.43% increase on just the games they tested). :disgust:
http://en.expreview.com/2008/0...e-scaling-test/?page=5

 

[Munky]

Originally posted by: Lithan
I was more thinking the post where I specifically noted that at higher reso's it will be 20%+. Yeah, reading is fun.
You see, a SITUATION where the difference is great does not equal an AVERAGE of greater difference, which is what we are discussing here. Hell if you wanna use situational performance the 8800gtx is 200%+ faster than the 8800gt. Most people go by the average though, as it's a much more fair comparison.

How often do you play HGL? Drops to <20fpm (yes that's frames per MINUTE) happened within 20mins of loading almost every time I loaded it up on 174's. Revert to 169's, problem gone. I don't bother with forcing driver AA on, so It may have problems in 169 I'm unaware of.

So you basically confirmed, that at higher rez, where the stress on the gpu is greater, the performance gap will be wider. I don't particularly care how these cards perform at lower rez, because in those cases the performance is high enough to barely be noticeable, and there's not enough load on the gpu to show what it's really capable of.

 

[Lithan]

That was my point munky. If you are running @ high res and playing mostly the latest games, yes 8800gt is 20% faster, but at lower reso's it's not, so the average performance difference, as measured by people who don't claim that the 3850 is faster than the 9800gx2, is quite a bit below 20%. I'm not trying to argue that the 8800gt isn't worth the additional money. I was just correcting you for correcting someone else, when what he said was closer to accurate. When people say a card is X% faster, it makes no sense to only choose the settings where there's a difference in performance (unless you distinguish that you are only looking at those settings), like I said, using that logic you could call a 8800gtx 200% faster than 8800gt, because at certain extreme resolutions it does in fact win by that much. That's clearly an unfair portrayal of the cards capabilities though, as you are only looking at a situation where the GT is obviously out of it's league... and yes, at higher reso's the 9600gt is out of it's league... and for a brief period, the 8800gt holds up after the 9600gt falls short... the same thing happens with every card nvidia makes. The 8800gs falls short, then the 9600gt, then the 8800gt, then the 8800gts, then the 9800gtx, then the 8800gtx/ultra then the 9800gx2 (in some games). If you only look at the cases where everything that isn't multicore is getting <10fps, it gives you a pretty messed up perception of performance... and it's certainly not fair to call it an average, which is an extreme example of what you're trying to do here. Hell I OWN a 8800gts, I've owned an 8800gt and a 9600gt, and I can tell you that the 8800gt wasn't anywhere close to 20% faster than the 9600gt @ 1680x1050 in my games. It wasn't even noticably faster in most of what I ran, and I consider a 20% boost quite massive. And since OP is 22", odd's are that he's running my reso.

 

[hodgenutts]

Updated!

 

[MrPibb]

Blah blah blah, OP - go with a 512MB 8800GT, end of story.

 

[jfunk]

Yeah, I think the only confusion here is people are talking about the 256MB 8000GT for some reason. Why? The 512 8800GT can be had for less or equal to the 9600GT right now (around $150), so what gives? Whether it's 10% or 20% faster, it's still faster and it costs the same, so why choose the slower card?

 

[hodgenutts]

One reason I was looking at the slower 9600GT is because in a month or so I was going to get another and run SLI as my CURRENT power supply could handle that I have a 550W thermaltake. This would be a bit small for 2 8800GT's i think. From what i've heard 2 9600GT in SLI are guite powerful, but afterall this is hearsay

 

[hodgenutts]

That is according to the ASUS website, they reccomend a 550 Watt power supply for 2 9600GT's in SLI.

 

[myocardia]

Originally posted by: hodgenutts
One reason I was looking at the slower 9600GT is because in a month or so I was going to get another and run SLI as my CURRENT power supply could handle that I have a 550W thermaltake. This would be a bit small for 2 8800GT's i think. From what i've heard 2 9600GT in SLI are guite powerful, but afterall this is hearsay

You'd be much smarter to just get a 512MB 8800GTS to begin with.

 

[hodgenutts]

I DON'T have the money for for an 8800GTS I'm looking at what I can afford NOW. With one card I can always get another down the road when I when I have more funds.

 

[Dream Operator]

Originally posted by: myocardia

Originally posted by: hodgenutts
One reason I was looking at the slower 9600GT is because in a month or so I was going to get another and run SLI as my CURRENT power supply could handle that I have a 550W thermaltake. This would be a bit small for 2 8800GT's i think. From what i've heard 2 9600GT in SLI are guite powerful, but afterall this is hearsay

You'd be much smarter to just get a 512MB 8800GTS to begin with.

I bought one 9600GT and used it for a month before adding a second for SLI. I never got video out either card in that motherboard again (A8N-SLI premium). ASUS RMA'd the board, still no video. Tried a 9800GTX, no video. I did a lot of troubleshooting to go along with it, including putting the 9600GT's in two other motherboards w/ PCIE 1.0 slots, they worked fine. Not really sure what to say about the whole situation there, but I can tell you that I am moving on and ordered an 8800GTS 512MB to complement a P35 board, 4GBDDR2 and a Q6600. I'm looking forward to it and wish I would have stuck with buying once better card to begin with. My theory has always been to do SLI only if you can afford two of the best cards on the market at once. Not asking anybody else to follow it, but it's a safe route I think.