• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Rise Of The Tomb Raider PC Patch Adds DX12 Multi-GPU & ASync Compute Support & More

D.P. said:
DX11 got 114.72FPS
DX12 got 150.13

31% improvement by switching to DX12.Looks pretty good to me.
Click to expand...

Thats for SLI - could simply be SLI improvements.

Need to see what DX12 / async update brings on a single card to see whether Nvidia apparent support for async with Pascal actually materialises.
 
Was this game developed with dx12 or has this just been implemented after? I mean could rockstar add dx12 support to GTA to give a few extra FPS?
 
sprite- said:
Thats for SLI - could simply be SLI improvements.

Need to see what DX12 / async update brings on a single card to see whether Nvidia apparent support for async with Pascal actually materialises.
Click to expand...

But that is probably one of the few scenarios where you might see DX12 make a difference because you are more likely to be CPU bound in SLI.

The Single card DX11 performance shows that there really isn't any CPU bottle necks for Pascal in DX11 due to low driver overhead.
 
980ti SLI user here. I'll do some testing tomorrow and post some results for some maxwell info.

One thing I need to emphasise though is that for SLI and rise of the tombraider, I've found that the original SLI fix from months ago using tweaked SLI bits using nvidia inspector still yields miles better results for SLI than any of the recent drivers (this fix http://www.dsogaming.com/news/rise-...d-offering-95-sli-scaling-on-two-nvidia-gpus/)

So this is definitely something to bear in mind. And does nvidia inspector even work with dx12? If not, which I suspect it doesn't, then we can't do a proper comparison.
 
layte said:
I'll be the first to hold my hands up. Please accept my apology.

I wonder why the developer and the AMD rep here said it was being used at the time?
Click to expand...

I knew it wasn't included as i checked internally after, but it wasn't my place to talk about it as i wasn't sure if it was public knowledge due to that blog link you shared.
 
3440 x 1440 / Very High Preset / MSAA

DX11 (Min / Max)
Mountain Pass: 142.87 fps (43.44 / 220.51)
Syria: 111.04 fps (54.43 / 146.13)
Geothermal: 90.58 fps (45.02 / 140.59)
Overall: 115.04

DX12 (Min / Max)
Mountain Pass: 160.99 fps (79.67 / 243.00)
Syria: 125.40 fps (49.21 / 172.67)
Geothermal: 125.89 fps (10.72 / 184.42)
Overall: 138.12

A decent bump up overall, though not sure why DX12 mins were so low at points, think when scene initially loaded, but overall was much smoother.
 
Well I just checked with my 980ti SLI setup. DX11 was still faster for me (by a couple of frames), even know it said there were SLI improvements in DX12 (or were they only for Pascal?). I tested at 4k with ultra settings, no AA. I was getting very good scaling before the patch anyway.
 
2560x1440 / High Preset / no AA

DX11
Mountain Peak: 64.55 fps (min: 27.59, max: 142.52)
Syria: 49.77 fps (min: 13.31, max: 71.82)
Geothermal Valley: 50.14 fps (min: 14.19, max: 60.93)
Overall: 55.11

DX12
Mountain Peak: 67.34 fps (min: 37.09, max: 94.41)
Syria: 52.28 fps (min: 28.20, max: 61.52)
Geothermal Valley: 52.91 fps (min: 40.62, max: 64.91)
Overall: 57.82
 
I did 2 runs of each test back to back to remove the texture in memory issues previously reported, Interestingly the only time it mad a difference was with the 1st run of DX11 where I saw stutter, It wasn't continuous it was a single judder and it only happened once or twice on the last 2 maps, The preceding tests ran smooth with no stutter at all.

The PC used: A 4790k @ 4.5, 8gb's of ddr3, Fury Tri-x @ 1040/500, Windows 10 pro O/S.

The game settings: Every setting was manually put to it's highest with the exception of Texture which was put down one setting to high because of the 4gb of memory limitation. Running at 1080p

Results:

DX11 run 1 with triple-buffered V-sync
Mountain Peak: 55.42 fps (min: 27.62, max: 109.74)
Syria: 58.79 fps (min: 8.12, max: 72.40)
Geothermal Valley: 54.63 fps (min: 3.38, max: 70.46)
Overall: 56.13

DX11 run 2 with triple-buffered V-sync
Mountain Peak: 57.9 fps (min: 28.02, max: 90.12)
Syria: 58.99 fps (min: 12.28, max: 69.79)
Geothermal Valley: 56.05 fps (min: 13.30, max: 67.57)
Overall: 57.57


DX12 run 1 with triple-buffered V-sync
Mountain Peak: 57.79 fps (min: 32.28, max: 91.32)
Syria: 59.89 fps (min: 36.00, max: 78.86)
Geothermal Valley: 54.69 fps (min: 28.68, max: 64.03)
Overall: 57.31


DX12 run 2 with triple-buffered V-sync
Mountain Peak: 57.05 fps (min: 34.89, max: 80.04)
Syria: 59.70 fps (min: 34.21, max: 67.79)
Geothermal Valley: 54.48 fps (min: 27.45, max: 69.08)
Overall: 56.92

IMPORTANT EDIT:

Being the retard I am I just realised all the above results were done using triple-buffered V-sync so I just turned v-sync off and got the following results showing some significant differences, If actual gameplay follows the trend set in the benchmarks while DX11 often had better averages the higher DX12 minimums would make for a much preferable experience

DX11 V-sync off
Mountain Peak: 81.04 fps (min: 43.00, max: 149.95)
Syria: 62.49 fps (min: 10.24, max: 96.53)
Geothermal Valley: 59.11 fps (min: 13.47, max: 75.65)
Overall: 67.90


DX12 V-sync off
Mountain Peak: 76.16 fps (min: 44.01, max: 100.16)
Syria: 64.92 fps (min: 51.93, max: 76.69)
Geothermal Valley: 56.28 fps (min: 38.68, max: 71.93)
Overall: 65.83

What a dope :confused:
 
Last edited:
Very glad I went for a 1070 rathern than a 980ti - Maxwell's being left behind in the dust after all.

Does indeed seem like NVIDIA gave them the go ahead to release this async patch once Pascal was ready, NVIDIA have such massive market power after all.
 
You must log in or register to reply here.
Back
Top Bottom