My CrossFire Experience

[DragonTamer1]

So a little over a year go I bought an RX 480 as an upgrade for my GTX 760. I had never owned an AMD card and was getting irritated with some of Nvidia's BS regarding shadowplay (hiding it behind an account wall after I had been using it for free for so long). I switched to AMD and have been happy ever since.

 

About 3 months ago Vega was all the news and I was curious about more GPU power. Vega was and still is a rarity and a bit overpriced from some outlets with no current AIB partner cards. I found a used RX 480 that matched my card for MSRP (unheard of during the mining craze) so I grabbed it up.

 

Lots of people have talked about how multi-GPU configurations are not a good idea because of stability issues and poor performance in games and it is an unsupported technology. The problem is that I've experienced minimal issues, most of which were crashing AMD drivers while simply sitting on the desktop. These still happen and I'm unsure if it was because the card was used for mining (no evidence but highly likely) or if it still driver issues. At the same time I have had no stability issues in game except for one instance that was caused by disabling unified GPU monitoring in MSI afterburner, turning it back on helped. Many of the games that I have played support scaling to some degree with some games offering very large gains. The problem is that in quite a few of these games I only have 60 Hz monitor so I don't get the most use out of the cards, however it does work to keep the temps of each card down for a nice quiet gaming experience. In the Radeon software I believe there was an option to turn off the second card if a game doesn't support it, so in games that don't support the dual cards (or even the desktop) the second never even comes on.

 

I just wanted to share my experience so far with my setup and how it is working. I wanted to know peoples feedback on the topic as well as experiences that others have had with multi card systems. I have never used more than one Nvidia card so I am also interested in hearing about that as well.

[Tz000]

22 hours ago, DragonTamer1 said:

So a little over a year go I bought an RX 480 as an upgrade for my GTX 760. I had never owned an AMD card and was getting irritated with some of Nvidia's BS regarding shadowplay (hiding it behind an account wall after I had been using it for free for so long). I switched to AMD and have been happy ever since.

 

About 3 months ago Vega was all the news and I was curious about more GPU power. Vega was and still is a rarity and a bit overpriced from some outlets with no current AIB partner cards. I found a used RX 480 that matched my card for MSRP (unheard of during the mining craze) so I grabbed it up.

 

Lots of people have talked about how multi-GPU configurations are not a good idea because of stability issues and poor performance in games and it is an unsupported technology. The problem is that I've experienced minimal issues, most of which were crashing AMD drivers while simply sitting on the desktop. These still happen and I'm unsure if it was because the card was used for mining (no evidence but highly likely) or if it still driver issues. At the same time I have had no stability issues in game except for one instance that was caused by disabling unified GPU monitoring in MSI afterburner, turning it back on helped. Many of the games that I have played support scaling to some degree with some games offering very large gains. The problem is that in quite a few of these games I only have 60 Hz monitor so I don't get the most use out of the cards, however it does work to keep the temps of each card down for a nice quiet gaming experience. In the Radeon software I believe there was an option to turn off the second card if a game doesn't support it, so in games that don't support the dual cards (or even the desktop) the second never even comes on.

 

I just wanted to share my experience so far with my setup and how it is working. I wanted to know peoples feedback on the topic as well as experiences that others have had with multi card systems. I have never used more than one Nvidia card so I am also interested in hearing about that as well.

Do you have any real benchmark numbers from before and after?  I have an RX 580 8GB and would be curious how they would perform in Crossfire even tho I know most games are now being optimized for a single gpu.

[DragonTamer1]

7 hours ago, Tz000 said:

Do you have any real benchmark numbers from before and after?  I have an RX 580 8GB and would be curious how they would perform in Crossfire even tho I know most games are now being optimized for a single gpu.

Unfortunately, I am working all weekend but I will try to test some of the games when able. Don't expect a huge in depth analysis.

[DragonTamer1]

On 12/14/2017 at 6:05 PM, Tz000 said:

Do you have any real benchmark numbers from before and after?  I have an RX 580 8GB and would be curious how they would perform in Crossfire even tho I know most games are now being optimized for a single gpu.

Sorry it took so long, I have the numbers. The system in question is in my sig, the Xeon is basically a 3930K. Fraps was used for the frame times, but refused to work for ROTR for some reason. Just used the in game results for this. The other 2 games that I have that support CF are multiplayer only or extremely old (as in the game hits its frame cap before the cards even hit 10% load). The multiplayer game is Arma 3 and it is very nice when the server is running smoothly. Performance can change at the drop of a hat though so even a multiple test run with one setting would yield different results. The old game is Assassin's creed 2. I could test Sniper Elite 3 if you want but I don't know how well it scales.

 

CF Off

Assetto Corsa:

avg - 111

1% - 89

.1% - 78

 

GTA 5:

avg - 69

1% - 28

.1% - 25

 

ROTR:

Peak;

avg 113

max 185

min 74

 

Syria;

avg 87

max 117

min 31

 

Geo;

avg 85

max 105

min 64

 

CF On

Assetto Corsa:

avg - 164

1% - 118

.1% - 106

 

GTA 5:

avg - 93

1% - 44

.1% - 40

 

ROTR:

Peak;

avg - 194

max - 275

min - 77 (this was caused by a low dip, the dip was not repeatable)

 

Syria;

avg - 89

max - 158

min - 6

 

Geo;

avg - 140

max - 173

min - 108

[Tz000]

On 12/19/2017 at 9:01 PM, DragonTamer1 said:

Sorry it took so long, I have the numbers. The system in question is in my sig, the Xeon is basically a 3930K. Fraps was used for the frame times, but refused to work for ROTR for some reason. Just used the in game results for this. The other 2 games that I have that support CF are multiplayer only or extremely old (as in the game hits its frame cap before the cards even hit 10% load). The multiplayer game is Arma 3 and it is very nice when the server is running smoothly. Performance can change at the drop of a hat though so even a multiple test run with one setting would yield different results. The old game is Assassin's creed 2. I could test Sniper Elite 3 if you want but I don't know how well it scales.

 

CF Off

Assetto Corsa:

avg - 111

1% - 89

.1% - 78

 

GTA 5:

avg - 69

1% - 28

.1% - 25

 

ROTR:

Peak;

avg 113

max 185

min 74

 

Syria;

avg 87

max 117

min 31

 

Geo;

avg 85

max 105

min 64

 

CF On

Assetto Corsa:

avg - 164

1% - 118

.1% - 106

 

GTA 5:

avg - 93

1% - 44

.1% - 40

 

ROTR:

Peak;

avg - 194

max - 275

min - 77 (this was caused by a low dip, the dip was not repeatable)

 

Syria;

avg - 89

max - 158

min - 6

 

Geo;

avg - 140

max - 173

min - 108

Excellent, thanks for going through all that trouble much appreciated.   

 

Cheers!