1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Ip35 16 x 4 and crossfire..

Discussion in 'AMD / ATI' started by trog100, Feb 24, 2008.

  1. trog100

    Joined:
    Dec 18, 2005
    Messages:
    4,422 (1.25/day)
    Thanks Received:
    238
    toying with the idea of buying another 3870 card now the price has dropped off..

    is anybody running crossfire on a p35 board with 16 x 4.. there is a performance loss but information on exactly how much is hard to find..

    also how will it all work out.. will a slightly slower clocked card in the x 4 slot pull the other one down or not..??

    i have two no three basic choices.. buy a 3870 x 2 card.. buy another 3870 card.. or not bother at all.. ???

    performance wise the x 2 will do better but another 3870 card even at x 4 will be cheaper.. how much performance loss thow is what i need to find out..

    any ideas guys..

    nooo noo noo to buying yet another mobo.. he he he

    trog
     
  2. ShadowFold

    ShadowFold New Member

    Joined:
    Dec 23, 2007
    Messages:
    16,921 (6.02/day)
    Thanks Received:
    1,644
    Location:
    Omaha, NE
    Are you talking about a 16x slot running in 4x mode? If so ATi cards only lose maybe 5 fps in 4x mode. Your IP35PRO has a second slot running at 4x mode so you should be good.
     
  3. kylew

    kylew New Member

    Joined:
    Nov 19, 2007
    Messages:
    604 (0.21/day)
    Thanks Received:
    70
    Location:
    Liverpool
    Some people report a pretty much 0% penalty when running in 16x and 4x. I reckon you'll be fine
     
  4. trog100

    Joined:
    Dec 18, 2005
    Messages:
    4,422 (1.25/day)
    Thanks Received:
    238
    yes one would be x 16 the other would be x 4.. how would it all meld in.. i have read that once the data is loaded into the cards memory after that the bus dosnt do that much.. the performance loss might not be that great..

    i cant find any exact comparisons thow..

    trog
     
  5. kylew

    kylew New Member

    Joined:
    Nov 19, 2007
    Messages:
    604 (0.21/day)
    Thanks Received:
    70
    Location:
    Liverpool
    I think you'll find this useful: Link :)
     
    trog100 says thanks.
  6. ShadowFold

    ShadowFold New Member

    Joined:
    Dec 23, 2007
    Messages:
    16,921 (6.02/day)
    Thanks Received:
    1,644
    Location:
    Omaha, NE
    Random Murderer says thanks.
  7. trog100

    Joined:
    Dec 18, 2005
    Messages:
    4,422 (1.25/day)
    Thanks Received:
    238
    it seems to suggest a possible 5% performance drop.. which would relate to 2.5 frames at 50 fps.. or if it follows thru 500 points off a 20000 2006 score..

    what i would really like to see is a straight side by side comparison between 8 x 8 or 16 x 16 compared to 16 x 4 in a 2006 run..

    anybody seen such a thing.. ???

    thanks for the help so far dudes..

    trog
     
  8. Darknova

    Darknova

    Joined:
    Nov 8, 2006
    Messages:
    5,037 (1.56/day)
    Thanks Received:
    535
    Location:
    Manchester, United Kingdom
    I can't find the link, but I remember reading an article comparing 16 x 16, 8 x 8 and 16 x 4.

    8 x 8 and 16 x 4 performed within 1% of each other in all tests. 16 x 16 was about 10-15% better than the previous 2.
     
  9. Duffman

    Duffman New Member

    Joined:
    Dec 5, 2007
    Messages:
    1,011 (0.36/day)
    Thanks Received:
    67
    Location:
    Mt. Pocono, PA
    i switched from the ip35pro to the maximus formula to get 16x2. I'd have to go dig for my 3dmark scores to try and give you a comparo. I think i was around 16k and now my best score is 17644
     
  10. trog100

    Joined:
    Dec 18, 2005
    Messages:
    4,422 (1.25/day)
    Thanks Received:
    238
    assuming your cpu speeds to be the same thats a bit over a 10% gain roughly going from 16 x 4 to 16 x 16..

    50 fps to 55 fps or so.. so far its looking in favour of saving some money and getting the other card.. course i could play silly buggers and bung a x2 card in along side the single card in the x 4 slot.. that would be an interesting combo.. he he

    trog
     
  11. King Wookie

    King Wookie New Member

    Joined:
    Jan 1, 2008
    Messages:
    1,197 (0.43/day)
    Thanks Received:
    277
    Location:
    Johannesburg, South Africa
    I can't remember where, but I did recently read a review where they tried that combination.
    I think they were using a beta version of catalyst, but the gain was around 60 to 70 % in most games. (Crysis and ETQW being the usual spoilers)
     
  12. KainXS

    KainXS

    Joined:
    Sep 25, 2007
    Messages:
    5,629 (1.94/day)
    Thanks Received:
    516
    I know that a 3850@4x will get its performance cut neaely in HALF if you run it at 4x with one card, but I don't know about crossfire, I'm thinking of buying a crossfire board myself but don't know which one to buy for hybrid crossfire support
     
  13. trog100

    Joined:
    Dec 18, 2005
    Messages:
    4,422 (1.25/day)
    Thanks Received:
    238
    i dont think the cut by half for a 3850 is correct.. i have read x 4 pcie is the same as agp x 8.. an agp 3850 dosnt suffer a lot..

    when i selected the abit ip35 pro i had decided not to go crossfire.. perhaps the 3870 x 2.. but bang for buck it still seems the extra card is the best option.. i am just trying to pin down exactly what performance loss i will have to take compared to a 3870 x 2..

    now the 3xxx series cards have gone down in price the x 2 card isnt so atractive..

    it figures that when the green teams x 2 card arrives the red x 2 will drop thow..

    as it is now a x 2 card is £290.. another x 1 card is £111.. UK prices..

    the performance between normal crossfile and the single x 2 card is about the same..

    i already had two 3870 cards a few weeks back then sold one.. he he

    trog
     
  14. magibeg

    magibeg

    Joined:
    May 9, 2006
    Messages:
    2,035 (0.60/day)
    Thanks Received:
    229
    Well trog, you could always do what toms did and start taping across your board until it hits x4, do a couple benchmarks at that to find out what the performance loss is. Then you could take it one additional step to look up reviews of the crossfire scores for those same games/benchmarks and subtract the performance hit for running 4x with the efficency of crossfire. :-O



    edit - http://www.tomshardware.com/2008/01/23/crossfire_meets_pci_express/index.html

    a little more information on the subject.
     
    Last edited: Feb 24, 2008
  15. trog100

    Joined:
    Dec 18, 2005
    Messages:
    4,422 (1.25/day)
    Thanks Received:
    238
    seeing what one card does at x 4 isnt quite the same as a crossfire 16 x 4 combo..

    but i do find it surprising nobody has run a crossfire benching comparison useing the p35 16 x 4 chipset.. there must be plenty of them out there..

    i have to be pretty desperate to ask a forum question.. i always google around plenty first and usually find an answer.. not with this one..

    trog
     

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page