Crossfire on 16x 4x

Meistrup · 13 Feb 2015 at 20:21

Hey all

i just tried running crossfire with a mates 290x on my Gigabyte GA-P67A-D3-B3

but it appears that one card is faster than using 2 in crossfire.

Any one have experience running crossfire on 16x 4x boards ?

Pc specs

2600K @ 4,5 Ghz
Gigabyte GA-P67A-D3-B3
16gb 1600 Mhz ram
and 2 290x in crossfire
Psu super flower 1000w gold

shankly1985 · 13 Feb 2015 at 20:24

Pci 4x 2.0 will massively effect the performance of a gpu.. But I don't think it would effect it that much for one to be faster than two.

Puppetmaster · 13 Feb 2015 at 21:21

What games were you trying?

Meistrup · 13 Feb 2015 at 21:37

Puppetmaster said:
What games were you trying?

Mainly bf4 at the moment ill try fidle some more with it...

Just tried 3dmark is this score normal ? www.3dmark.com/fs/4063748

P4Clock · 13 Feb 2015 at 21:58

the info on that board say when you run an x16 card in slot 1, slot 4 will run at upto x1

so i would say there is no point running 2 cards, it will just use more power and do sod all for performance#

When the PCIEX1_2 or PCIEX1_3 slot is populated with an expansion card, the PCIEX4 slot will operate at up to x1 mode

Confused Stu · 13 Feb 2015 at 22:19

P4Clock said:
the info on that board say when you run an x16 card in slot 1, slot 4 will run at upto x1

so i would say there is no point running 2 cards, it will just use more power and do sod all for performance#

When the PCIEX1_2 or PCIEX1_3 slot is populated with an expansion card, the PCIEX4 slot will operate at up to x1 mode

That board can actually run two GPUs at 16x/4x. The second PCI-Ex16 slot is only limited to x1 speed if the PCI-Ex1 slots are used, otherwise it will run at 4x speed. Not ideal, but two cards should still be faster than a single card in any Crossfire supported game.

TaKeN · 13 Feb 2015 at 22:32

Years ago I had two 5770 cards in Crossfire on a board (Abit IP35 Pro) that did 16x on one slot and 4x on the other, it was a stutter fest due to the speed difference.

shankly1985 · 13 Feb 2015 at 22:32

Meistrup said:
Mainly bf4 at the moment ill try fidle some more with it...

Just tried 3dmark is this score normal ? www.3dmark.com/fs/4063748

This is mine one 290x @1030/1250 - One 290 @977/1250 - i7 2600k @4.4
16x-8x
http://www.3dmark.com/fs/3938588

Meistrup · 13 Feb 2015 at 22:45

shankly1985 said:
This is mine one 290x @1030/1250 - One 290 @977/1250 - i7 2600k @4.4
16x-8x
http://www.3dmark.com/fs/3938588

You get a bit better graphics score hmm

Meistrup · 13 Feb 2015 at 22:48

Confused Stu said:
That board can actually run two GPUs at 16x/4x. The second PCI-Ex16 slot is only limited to x1 speed if the PCI-Ex1 slots are used, otherwise it will run at 4x speed. Not ideal, but two cards should still be faster than a single card in any Crossfire supported game.

So if i only run gfx cards and nothing else in the pci-ex it should work ?

at the moment i got an Intel network card in pci-ex1

thx for all your sugestions guys rly appreciate it..

shankly1985 · 13 Feb 2015 at 23:21

Do you have Tomb Raider? I could bench a a game and compare are scores.

We both have quite same systems 16GB Ram, i7 2600k close clocks and 290x+290 for and 2x 290x for you..

I guess the 4x will take about 10-15fps hit

Meaker · 14 Feb 2015 at 06:30

You picked the worst card to xfire on that board with since they use the PCI express bus exclusively to communicate. You are running 1/8th the ideal bandwidth.

Meistrup · 14 Feb 2015 at 10:22

shankly1985 said:
Do you have Tomb Raider? I could bench a a game and compare are scores.

We both have quite same systems 16GB Ram, i7 2600k close clocks and 290x+290 for and 2x 290x for you..

I guess the 4x will take about 10-15fps hit[/QUOTE

What settings should i run the bench in ?

Meistrup · 14 Feb 2015 at 10:23

Meaker said:
You picked the worst card to xfire on that board with since they use the PCI express bus exclusively to communicate. You are running 1/8th the ideal bandwidth.

So a new motherboard with 8x 8x will make a big difference ?

shankly1985 · 14 Feb 2015 at 12:19

Meistrup said:
So a new motherboard with 8x 8x will make a big difference ?

Use the settings from here. I'll also do same 1080p also.
http://forums.overclockers.co.uk/showthread.php?t=18654946

shankly1985 · 14 Feb 2015 at 13:28

This is what I got!

i7 2600k @4.4
Crossfire 290x @1030/1250 - 290@977/1250
1080p
Driver 14.12

Meistrup · 14 Feb 2015 at 14:07

shankly1985 said:
This is what I got!

i7 2600k @4.4
Crossfire 290x @1030/1250 - 290@977/1250
1080p
Driver 14.12

Just tried it got some weird minimums :confused:

and tried Bf4 lot of stuttering and again lower or nearly equal scores to single card 45 % gpu usage on both cards :confused:

i7 2600k @4.5
Crossfire 290x @1050/1350 - 290x@1050/1350
1080p
Driver 14.12

shankly1985 · 14 Feb 2015 at 14:39

Meistrup said:
Just tried it got some weird minimums

and tried Bf4 lot of stuttering and again lower or nearly equal scores to single card 45 % gpu usage on both cards

i7 2600k @4.5
Crossfire 290x @1050/1350 - 290x@1050/1350
1080p
Driver 14.12

You overclocking them GPUs?? you beat me on the avg FPS so 4x isn't really doing what i was expecting there..

Meistrup · 14 Feb 2015 at 17:42

shankly1985 said:
You overclocking them GPUs?? you beat me on the avg FPS so 4x isn't really doing what i was expecting there..

Its factory overclocked 290x pcs + card and one reference clocked like the pcs+
hmm so Tomb Raider crossfire works but whats wrong with that minimum .. :mad:

Alexrose1uk · 14 Feb 2015 at 18:00

If you're seeing the same in other titles, I would suggest those minimums are the results of the stutters caused by the bandwidth reduction to the second card.

I have 7850 Crossfire on 16x/4x and even that isn't getting the best out of both cards, and thats WITH a crossfire bridge.
As your 290s don't even use a crossfire bridge, it can't even make up any lost bandwidth there and am guessing it's hitting some pretty hard bottlenecks

Do you know anyone else with a motherboard better cut out for Crossfire whom might let you test the cards in thier rig to compare?