Idiot Check

[ShaolinDreams]

Thanks for the help, told me everything I needed to know.

I'm gonna stick with the Phenom as it's a little bit cheaper, as is the 2x 3850s rather than an 8800GTS. Also, the 2x2G sticks only cost me $140, and leave room to upgrade to 8Gs if I want to latter on. I did consider both of those paths while picking the parts, but I have reason to go with what I plan.

Again, thanks to everyone.

you dont need 8gb and by the time you do, your pc will be obselete anyway. 4gb DDR2 with 4x1gb sticks is cheaper and faster and would be the maximum a normal person would get. ok so, with your $45 saved from that ram put that to the q6600 which will be faster then phenom. Ok ram and cpu done, and you spent the same ammount.

as for graphics well..
Newegg.com - SAPPHIRE 100226L Radeon HD 3850 512MB 256-bit GDDR3 PCI Express 2.0 HDCP Ready CrossFire Supported Video Card - Retail
2 of these..

Newegg.com - XFX PVT88PUDD4 GeForce 8800GT 256MB 256-bit GDDR3 PCI Express 2.0 HDCP Ready SLI Supported Video Card - Retail
or 1 of these

done.

 

[maroon1]

2 x 3850's will be faster
.

Check this
HD 3850 Crossfire vs. 8800GTS 512MB G92 :: TweakTown

One 8800GTS 512MB is faster than 2 x 3850 on average

 

[NosBoost300]

Check this
HD 3850 Crossfire vs. 8800GTS 512MB G92 :: TweakTown

One 8800GTS 512MB is faster than 2 x 3850 on average

and 1 card is generally more awesome to have to me... 2 cards means more heat!!!! and needing more fans and such.. though i one day hope for dual cards

 

[Apokalipse]

Check this
HD 3850 Crossfire vs. 8800GTS 512MB G92 :: TweakTown

One 8800GTS 512MB is faster than 2 x 3850 on average

I wonder what drivers they're using for both cards?

NVIDIA GeForce 8800GTS 512MB G92 Tested :: TweakTown
the G92 GTS is about 10% faster than a GT, according to that.

The benchmarks I've seen place the 512MB 3850's only a small amount behind the 3870 most of the time. And the crossfired 3870's are faster than SLI'd 8800 GT's.

I would expect two 512MB 3850's to perform roughly on par on average to two 8800 GT's in SLI, and faster than single G92 cards

the 256MB 3850's in crossfire would get beaten by two SLI'd 8800 GT's by a little bit though.

I would definitely expect the G92 8800 GTS's in SLI to perform a bit better than two 3870's in crossfire (all at stock)

but then again, I like how cheap they've made the RV670 cards. And power consumption is massively improved from R600.

 

[maroon1]

The review that shows crossfired 3870's are faster than SLI'd 8800 GT are incorrect in my opinion because in some cases two HD3870 were getting more than %100 increase than a single HD3870, which is impossible

I like to see other benchmarks that show that same result, but there is non

The driver in tweak down was using Catalyst 7.11 and Forceware 169.06. I think those are the latest drivers

 

[Apokalipse]

The review that shows crossfired 3870's are faster than SLI'd 8800 GT are incorrect in my opinion because in some cases two HD3870 were getting more than %100 increase than a single HD3870, which is impossible

I think it is.
If the video RAM is too small to run certain things, it will have to access main system memory, which slows things down.
If you start cranking up resolutions, and anti-aliasing, then even 512MB video memory can be not enough in some games.

The two video cards combined will have 1GB video memory, which will be more than enough even when you crank up anti-aliasing and resolutions.

Then there are the methods used for both video cards processing the one screen. If a method will give the cards close to 100% scaling, and couple that with the increased video memory between the cards, then I believe that can push it over 100% scaling. It just won't go much over 100%, and it won't happen often.

The driver in tweak down was using Catalyst 7.11 and Forceware 169.06. I think those are the latest drivers

I don't think the 7.11 drivers give as good performance on average as some earlier drivers.

 

[maroon1]

.
Then there are the methods used for both video cards processing the one screen. If a method will give the cards close to 100% scaling, and couple that with the increased video memory between the cards, then I believe that can push it over 100% scaling. It just won't go much over 100%, and it won't happen often.

I still don't understand how can it give more than 100%. This is like saying that Quad core can give more than 100% performance compared to a similar clocked dual core

As far as I know, the video memory between two cards are mirrored, this means that two 512MB memory won't act as 1GB memory (I'm not sure about that though)

Anyway, it is just one review. I like to see another review that shows similar results.

 

[Apokalipse]

I still don't understand how can it give more than 100%. This is like saying that Quad core can give more than 100% performance compared to a similar clocked dual core

As far as I know, the video memory between two cards are mirrored, this means that two 512MB memory won't act as 1GB memory (I'm not sure about that though)

Anyway, it is just one review. I like to see another review that shows similar results.

It could even be something to do with the drivers, that might somehow inhibit an RV670 card's performance as a single card, but doesn't when it's in crossfire.

 

[gurusan]

Yeah 2x 512mb cards in Crossfire or SLI still = 512mb memory, not 1024.

If you think about it each card is rendering the same scene, just a different frame, so they need the same textures.

 

[Apokalipse]

Yeah 2x 512mb cards in Crossfire or SLI still = 512mb memory, not 1024.

If you think about it each card is rendering the same scene, just a different frame, so they need the same textures.

that depends on the mode. In alternate frame rendering, both cards will take turns rendering frames.