Anyone have DX12 capable cross fire GPU's to test on Ryzen?

K

Kato1144

Limp Gawd
Joined
Oct 18, 2007
Messages
354
I watch this review of Ryzen gaming performance on Rise of the tomb Raider, in the review he tested a 7700k and a 1800x on a overclocked MSI GTX 1070, he said he got his 1070 close to 1080 speed as close as he could and he did get results about what you would expect in DX11 and DX12, were the DX12 suffers hard especially on Ryzen. After that he throws in 2 MSI RX 480's in a crossfire setup and as me and any one who has ever used crossfire in the past, he excepted it do worse then the single 1070 and yes in DX11 it sucked buttttt in DX12 it was pretty amazing, the 1800x has caught up to the 7700k in the DX12 Crossfire setup, even surpassed the DX11 1070 results in some places.

To say the least I'm surprised this could be something most of use have over looked when looking at DX12 results on Ryzen, this is mostly down to AMD not having a top tier GFX card so no one wants to test with the lowly RX 480 that almost guarantees a GPU bottleneck let alone in crossfire witch does not always scale well or at all. Now I don't know any more then what i have said already, I would love to see if this is possible in other games and recreate this guys test but I only have a GTX 1070 and I can not go out and buy 2 RX480's, so I ask you the [H] members who have access to DX12 capable GPU's and two of them, lets see if the scaling is possible in DX12, AMD made many clams about the CPU scaling in DX12 and we have yet to truly see that, this may be telling of how Vega will preform.

Now I'm not doing this to start "RYZEN IS WINNER WITH AMD GFX RULZ!!!!" crap or "VEGA BEATS 1080TI CONFIRMED!!!" bs, this is neat and I want to explore it further, if it is real that a RX480 crossfire can scale as such, well the future may be very bright for us all.

Now I'm going to link the video, I'm honestly a little apprehensive as I'm sure some will judge me for watching this guys vids, just like everything take this with a grain of salt and if you can recreate the benchmarks he is doing and see if your numbers are with in margin of error, ok here is the link:

 
Last edited:
This is very interesting. It's a good point when reviews use similar setups (for example, almost all using Nvidia cards) and not considering how that could affect the results.

I have 480 CF in my machine. In games with good CF support, the performance is nearly at 1080 Ti levels, for example in GTA V. I benched the same system with a 1080 Ti I had briefly, and it was good, but seeing as the 480s held up in some games I ended up putting them back in and selling the Ti.

In RoTR specifically, I saw around 45 - 50 fps at 4K High settings with both 1080 Ti and 480 CF. I did test in DX12, but didn't see a huge benefit. Since I couldn't hit 60 fps I didn't really test much further but this was also at 4K which is really intensive for games like this.

If Vega is 1080-level performance, I think 2 Vegas in CF are going to kill it and look forward to upgrading as soon as they release.
 
cybereality said:
This is very interesting. It's a good point when reviews use similar setups (for example, almost all using Nvidia cards) and not considering how that could affect the results.

I have 480 CF in my machine. In games with good CF support, the performance is nearly at 1080 Ti levels, for example in GTA V. I benched the same system with a 1080 Ti I had briefly, and it was good, but seeing as the 480s held up in some games I ended up putting them back in and selling the Ti.

In RoTR specifically, I saw around 45 - 50 fps at 4K High settings with both 1080 Ti and 480 CF. I did test in DX12, but didn't see a huge benefit. Since I couldn't hit 60 fps I didn't really test much further but this was also at 4K which is really intensive for games like this.

If Vega is 1080-level performance, I think 2 Vegas in CF are going to kill it and look forward to upgrading as soon as they release.
Click to expand...

You are still pretty much bound to DX12 (Vulkan?) titles for that performance boost.
 
In most games, absolutely not. The 1080 Ti is a monster of a card, and (in general) leagues above the 480 (even in Crossfire). However, specifically in GTA V I can get a solid 60 - 70 FPS all max settings in 4K (except for advanced options) with either setup. I was very surprised by this, but I put in several hours benching this game and it's true. I don't believe it's a CPU bottleneck because the GPUs were hitting 100% and the CPU usage was probably around 50% split on all the cores. I think GTA may be on outlier, there were lots of games with no (or poor) CF support and the performance was not playable at 4K (even on lower settings). But the fact that it can work in one game shows the potential can be there.
 
cybereality said:
I don't believe it's a CPU bottleneck because the GPUs were hitting 100% and the CPU usage was probably around 50% split on all the cores. I
Click to expand...
50% split on all the cores as in all the cores were at ~50%? Because if it was, then it was CPU bottlenecked most likely, due to specifics of Windows load balancing algorithm.

But whatever, that's a curious case to keep in mind for future, thanks.
 
I have a ryzen 1700 and crossfire 480s. Don't have tomb raider though...
 
lolfail9001 said:
50% split on all the cores as in all the cores were at ~50%? Because if it was, then it was CPU bottlenecked most likely, due to specifics of Windows load balancing algorithm.

But whatever, that's a curious case to keep in mind for future, thanks.
Click to expand...

Yeah cool, what system are you using to test Ryzen / 1080 Ti / 480 CF on I am curious? Thanks.
 


Watchdogs Ryzen with a GTX 1060 and a RX 480 720p low settings. Results are interesting to say the least, looks like Nvidia has fallen a bit behind in the driver department.
 
buttons said:
I have a ryzen 1700 and crossfire 480s. Don't have tomb raider though...
Click to expand...
do you have access to any other DX11/DX12 bench marking titles, like Hitman, Dues Ex MKD, or AotS I can't think of any others right now... it will be interesting to see how AMD GPU's and CPU's will scale in DX12, kinda wish i had 2 RX480's to test with my 1070 although I would annoy the crap out of my brothers.... hopefully Vega is solid in DX11 and glorious in DX12, that would be a nice change of pace
 
Kato1144 said:
do you have access to any other DX11/DX12 bench marking titles, like Hitman, Dues Ex MKD, or AotS I can't think of any others right now... it will be interesting to see how AMD GPU's and CPU's will scale in DX12, kinda wish i had 2 RX480's to test with my 1070 although I would annoy the crap out of my brothers.... hopefully Vega is solid in DX11 and glorious in DX12, that would be a nice change of pace
Click to expand...

I have overwatch and division the rest of my games are older stuff and or VR.
 
I wonder if Ryzen's two CCX inherently work well with crossfire or dual GPU cores, or is TRotTR just weird?

Check this out:
 
sabbilon said:
I wonder if Ryzen's two CCX inherently work well with crossfire or dual GPU cores, or is TRotTR just weird?

Check this out:
Click to expand...


I think it's more of a issue with the Nvidia driver not working well with RotT and requiring more single thread performance in DX12 then the AMD driver and the AMD DX12 driver is just working very well in this scenario compared to the Nvidia driver.


as more look in to the CF on Ryzen scenario the picture will become clearer, right now we need to wait for more results from trusted sources as the AMD DX12 numbers could just be a fluke with RotT and the Division and might not mean really anything otherwise only time will tell.
 
I remember some video about how well some developer said Radeon crossfire GPUs scaled (think it was 480s), almost 100%. I can't find it at the moment nor remember more details. In my search a found another interesting nugget.

It certainly would help with the smoothness described and indicates that TRotTR was on the mind ot the driver developers
 
You must log in or register to reply here.
Back
Top