Is a crossfire pair as good as one good card?

[konakid]

Hi I just bought two Asus Radeon 4650 and have them crossfired. Is this pair as powerful as one single card with 2gb of memory and 1.2ghz clock speed. The primary card is running in 16X pci-e and the second is running in 8X pci-e.

 

[mindless728]

no

1) the GPU's have to have the same info in their memory so if each is 1GB, then you still only have 1GB of memory

2) you don't add speeds like that, its more like you add the amount of shader processors then take a percentage off from crossfire (its not 100% efficient)

so your 2 low end video cards are about equal to a 4770 which is low mid range (probably not even this)

its better to buy a faster single card than 2 low end cards like that btw

 

[konakid]

Ok. I think i should have got one in the 4800 series but I got each card for $34 on newegg. I guess I wanted to try out crossfire. I does show up as 2gb of video memory windows 7 and when I "uncrossfire" them it goes back down to 1gb. Why is that?

 

[jitpublisher]

Because when you "un-crossfire" them, they work (or they are capable anyway) of working as 2 indendant cards.
It is never worth it to crossfire 2 weak cards for gaming. Your money is better spent, even if you have to add a little more cash, to get 1 faster card. There are some exceptions, but even then the cards we are talking about putting in crossfire over a single card are still considered upper midrange gaming cards themselves.

 

[konakid]

Ok..the 4850 seems to be popular for crossfire setups.But for the price difference I was under the impression that I wouldn't gain much.The memory clock is only 200Mhz higher, the GPU clock is only 40Mhz higher, the streams is 480 higher ( I'm not quite sure what that is /I believe higher is better). I have overclocked my cards to 750Mhz gpu and 1000Mhz memory(gave me .4 higher on visual index!) under full load they hit 70deg so I keep it stock normally.
Would it be worth the extra money for a card so close to mine? Or would I be better off staying with my crossfired 4650's?

 

[sabot00]

Dude, there's a things called ROPs, TMU's, SP's.

Your putting 2 Pentium 4's, overclocking them, putting them in a server mobo & comparing vs a i7 920.


4650 - 320 SP's
4850 - 800 SP's

4650 - 128-bit mem interface
4850 - 256-bit mem interface

4650 - Mostly DDR2
4850 - Mostly DDR3


1 4850 would blow the 4650's out of the water.

 

[konakid]

So what am I gaining by crossfiring these cards.Memory?Gpu Speed?Streams?
I've never bothered with my cards.(Upgraded from an ATI rage pro) until now so I'm new to this.

amd phenom 9950be smokes my p!!!

 

[sabot00]

You are gaining 2 of the same card, the efficiency is around 80-93% depending on game, cpu-limits, frame-caps, resolution, etc.

Even if you did have perfect 100% power, the 4650's would still only have 640SP's compared to 800.

 

[konakid]

I checked and my second card is running in 4X pci-e does that mean it will perform only at 1/4 as well as my primary card which runs in 16X pci-e?

 

[mindless728]

I checked and my second card is running in 4X pci-e does that mean it will perform only at 1/4 as well as my primary card which runs in 16X pci-e?

the cards are so low powered there probably isn't any affect on them

all the x4 does it limit max bandwidth which most cards only saturate an x8 set of lanes, and like i said your are low powered that it should be fine

 

[sabot00]

If it's PCI-E 2.0 it should be fine, as for PCI-E 1.0/1.1 there aren't any tests I know of that have tested.

 

[konakid]

I ran the Mark3D demo test and It benched 20,772 but when uncrossfired It benched about 12,000 which puts my setup at about 75% efficient. I did have some horrible compatibility issues with the software and windows 7 so I'm not sure how accurate it is.
It also got different readings if I ran it consecutively maybe ralated to card temps.It made my primary card sore to 90deg. once but my secondary stayed much lower so I restarted my comp. and the issue subsided. My primary card runs about 3-5deg hotter than my secondary.It may be because of different ventilation though.

 

[konakid]

Okini 55>>>Are you saying it is junk or semi decent??? or did you put your "not" in the wrong spot.

 

[jitpublisher]

I ran the Mark3D demo test and It benched 20,772 but when uncrossfired It benched about 12,000 which puts my setup at about 75% efficient. I did have some horrible compatibility issues with the software and windows 7 so I'm not sure how accurate it is.
It also got different readings if I ran it consecutively maybe ralated to card temps.It made my primary card sore to 90deg. once but my secondary stayed much lower so I restarted my comp. and the issue subsided. My primary card runs about 3-5deg hotter than my secondary.It may be because of different ventilation though.

Wow. What 3Dmark version was that, 2003, maybe 04?

 

[konakid]

It was the 2003 version.Does it make a difference on my score?My 450w psu fried>>>bummer>>>It was covered under warrenty but I think I'll get a 750w to be safe.

2x raid0 samsung f3 500gb
asus m4a78 plus
4gb corair dominater 1066mhz
amd phenom x4 9950be
2x hd4650
windows 7 pro 64bit

 

[korogui]

I think it all depends on the resolution youre going to play... Dont buy a BIIIG display and you'll be happy with the results xD

Probably your computer can run any game till 1400x900 or something like it with high detail/AA/FPS, anything higher than this with normal or low details may run well also...

Crossfire is nice, but it gives some problems with titles that likes to use data of current frame into the next one... bkz... you know, the second card wont have the data needed in the local memory to process the next frame... Anyway, i think this is what causes those daaaamn micro stutterings with games that likes blur and offscreen drawing. (Anyone with crossfire + win 7 run Guitar Hero fine? I cant with crossfire on)


BTW, i got 2 Radeon 4770's and its perfect for my 22" 1680x1050 display I can play anything @ native resolution with 4xAA. (The best bottleneck you can have is your display HAHAHAHHAHAH!) >

 

[konakid]

I can play Dirt 1 at 1680x1050 on my 22" screen and it seems liquid.Maybe some micro stutters at places.All the affects are on normal and I haven't tried changing them yet, as my psu is fried...

But need for speed pro street was a different story:it flat stuttered so bad it is nearly impossible to play.I've tried everything and as far as I can tell it isn't compatible with wind 7 yet.