9600GT vs. HD3870: Which to get?

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.


I wouldn't trust Nvidia's own numbers, but we know for sure it ain't 2.0x.



Tbh in more than one of those tests it gets to 2.0x+ ....
 
To be honest, I don't know how the 80% was determined but I get the feeling it is based off something like theoretical calculations from the card specs. If any one actually knows what they are talking about wants to jump in, please do so!

This is all just my opinion or guestimations. You have to think about drivers and how they affect and inhibit the performance of a card in a certain setting or game. I know that there are theoretical calculations using the GPU specs that help them come up with performance numbers. Actual performance however, is not just hardware based but software based as well and I think that is where the deviations come from.

So basically, like I said, I have no clue if that is legit but it tells you relative, real life performances in SLI/xfire based on the single card performance.

Edit...so like if I get 40 FPS from card 1, I would want 40 FPS from card 2 as well. So in running two, I want to get 80 FPS but usually you see something like 78 FPS which says to me you aren't getting 40 FPS out of each card.
 
Its all of them put together and the average found, as simple as that...


Also this from another review:

98a58de5-91a6-4969-86e2-70763fb5e551.gif


link
 


what the hell is "double lifetime"?

your point is well taken though.
 

It's where the life time warranty still applies after selling your XFX vidcard to a second owner.
 
If it was 50mhz x 2 that would mean a 100% scaling success 😀


In fact there is 81% scaling success there... which is great for 1024x768. Idk how much AA but I guess a bit.


50mhz core doesnt account for 80% scaling.
 
Not true Evilonigiri, the computer treats sli as one card, if you have 2x 256mb cards it still treats it as 256mb instead of 512, gpu speed is the same.
 

I wasn't talking about the memory. I was talking about the frequency.

Anyways, enough SLI. I found this direct comparison from Anandtech: http://www.anandtech.com/video/showdoc.aspx?i=3235

Overall, the stock 3870 performs better than the stock 9600GT in many games, however in the games that the 9600GT does win, it wins by a large margin. Does this simply mean that the 3870 isn't performing to it's fullest?
 
Wow, that scalability is impressive!




I need your account number too. My uncle died and I need to deposit the sum of $30,000,000.00 into an account for holding so the Nigerian authorities cannot access it.
 


You do realize that not only is the GF9600 Overclocked, the HD3870X2 is underclocked to match the HD3850's speed, and in the example you pick to show, the HD3870 is scaling @ 104%, so I don't see how that particular picture supports your statement at the end of the first post about scaling.

Overall, reading more reviews shows there is a bit more variability in the cards.

Even the firingsquad review shows that while one may have the higher avg, the other may have the higher min fps. As for [H], testing 3 games with 1 resolution on a slow CPU doesn't really do much 'interesting' to look at these cards, nor is it enough to draw best playable conclusions either. C'mon after all their attacks on others 3 games is not enough to show off the cards.

Let alone start to explore the variability in shader limitations noticed in games like Oblivion.
 
Also, according to Anandtech, the 3870 is a better buy than the 9600GT if it is priced the same. Problem is, the 9600GT can be had for about $10 less, so...tough decision.
 


Links please, come on and back up your claims if you can. I hate it when posters just painout say something with no proof to back it up. :)
 



randomizer beat me to it, :kaola: its an easy choice if you like AA alot and can't play a game without it. If not using AA don't bother you the choice is even easier.


 
Well looking at the Anand pricing guide the GF9600GT is only $5 cheaper.

But like we've mentioned before, the GF8800GT is often found cruising in that price range and would easily be a better option than the GF9600GT. $20 more I'd definitely go with that if it played the games I like better than the HD3870. Because there's pretty much no situation the GF9600GT outperforms the GF8800GT, but a few that the HD3870 outperforms them both.

In my case I'd be looking at Oblivion, Crysis and COD4 results, and that first test alone would drop the GF9600 out of contention and leae the other two, and more than likely having me come to a GF8800GT decision.

So really even $5-25 more, there's too much competition in that segment to simply blindly chose and say anything is the obviously better deal.
 



One thing you can never do is compare the shader processors between Nvidia and ATI! :non: The reason is because look at the 8800GTX with only 128 stream processors compared to ATI's 320 SP's on their 2900XT, HD3870, the 8800GTX kills the two with almost half the SP's. In the end it takes two 3870's (or one 3870x2) to beat one 8800GTX and remember that ATI has 640 SP's in Crossfire with Nvidia only having 128 SP's.
 
I understand that the 8800GT is indeed the best deal, but because of the numerous issues I've seen on these forums on it, I decided to stay away from it.

However, If I can find an 8800GT that fits my extremely limited budget of $180, I would go for it.
 


Can't you just wait till next payday and add $40 to the rest you have saved up? Its funny you never hear about all the good 8800GT's just the bad ones, the good outways the bad be far.
 


Don't ever give out your account number on the internet.
 

TRENDING THREADS