Fall Creator Update | 20% MORE Gaming Performance

Phixsator · 20 Oct 2017 at 12:17

from the limited testing with game mode on and off in BF1 MP 64 player full servers i see no definitive difference. BUT!, i dont know if game mode works like that or if i have to restart the game after i change it to on or off so my findings are not accurate at this point.

EDIT: Gave wildlands a go and initially same results however after turning game mode on and then restarting the game i noticed higher CPU util in the benchmark while FPS results was the same. Further light testing ingame in a very problematic area (small town named Buena Vida) i noticed i was able to hold an fps of minimum 55(would normally dip to the low 40ish) and the frame times seemed smoother. Normally it feels plain wrong at less than 60 but its not as bad as it use to be with gsync on. Hmm.. Cant figure out yet if i am just a victim of wishful thinking or there might actually be some credibility to this whole performance gain story.

40Four · 20 Oct 2017 at 12:37

I don't think i'll bother to be honest. I imagine it helps most with systems that have a smaller amount of RAM than say 4GB. Windows uses enough on its own without adding games to the mix. How i wish for an viable alternative for gaming.

nashathedog · 20 Oct 2017 at 12:38

Mr Shanks posted link's to Jokers video on the new update in the Vega review thread last night so I tested a few things last night and then installed the update and redid the tests this morning.

My hardware is all at it's out of the box clocks and the tests are at 3440 x 1440.

Ryzen 5, 1600x,
Two 8gb sticks of 3200mhz DDR4 (8 pack Team Pro) and a
GTX 1080 with an out of the box overclock of 1797 boost with stock memory. (MSI 1080 Armor)

Game tested, API, min/avg/max

ROTTR DX11 Before Update

MP- 12.52/32.12/69.97
S-- 4.26/21/05/38.04
GV 9.64/22.11/41.16
Overall 25.13

ROTTR DX11 After Update

MP-- 24.28/39.89/77.14
S-- 6.99/26.17/47.71 v
GV-- 16.73/26.54/39.53
Overall 31.14

ROTTR DX12 Before Update

MP-- 22.81/40.42/60.22
S-- 2.50/22.76/68.16
GV-- 1.19/24.34/32.76
Overall 29.49

ROTTR DX12 After Update

MP-- 21.59/40.08/61.31
S-- 18.42/25.26/28.39
GV-- 6.63/24.72/32.47
Overall 30.29

F1 2016 Before Update

Clear Weather 55/75/95

Heavy rain 58/74/89

F1 2016 After Update

Clear Weather 61/78/96

Heavy Rain 58/74/88

Unigine Valley Before Update

25/53.8/115.9
Score 2252

Unigine Valley After Update

25.4/53.9/115.7
Score 2255

Everything shows a margin of error sized improvement except Rise of the Tomb Raider in DX11 which has a respectable increase of a 6 frame overall average and much better minimums.
Nothing to write home about but definitely an improvement for Pascal, If Joker's Vega results are repeatable that's a good win for AMD.

Rroff · 20 Oct 2017 at 12:42

nashathedog said:
If Joker's Vega results are repeatable that's a good win for AMD

I'm beginning to think his results are comparing game mode before and after or something as his before results are mostly down on the general ballpark for my 1070 and what I've usually seen of the Vega 56 though its difficult to match up with his testing scenarios exactly.

40Four · 20 Oct 2017 at 12:42

nashathedog said:
Mr Shanks posted link's to Jokers video on the new update in the Vega review thread last night so I tested a few things last night and then installed the update and redid the tests this morning.

My hardware is all at it's out of the box clocks and the tests are at 3440 x 1440.

Ryzen 5, 1600x,
Two 8gb sticks of 3200mhz DDR4, (8 pack Team Pro) and a
GTX 1080 with an out of the box overclock of 1797 boost with stock memory. (MSI 1080 Armor)

Game tested, API, min/avg/max

ROTTR DX11 Before Update

MP- 12.52/32.12/69.97
S-- 4.26/21/05/38.04
GV 9.64/22.11/41.16
Overall 25.13

ROTTR DX11 After Update

MP-- 24.28/39.89/77.14
S-- 6.99/26.17/47.71 v
GV-- 16.73/26.54/39.53
Overall 31.14

ROTTR DX12 Before Update

MP-- 22.81/40.42/60.22
S-- 2.50/22.76/68.16
GV-- 1.19/24.34/32.76
Overall 29.49

ROTTR DX12 After Update

MP-- 21.59/40.08/61.31
S-- 18.42/25.26/28.39
GV-- 6.63/24.72/32.47
Overall 30.29

F1 2016 Before Update

Clear Weather 55/75/95

Heavy rain 58/74/89

F1 2016 After Update

Clear Weather 61/78/96

Heavy Rain 58/74/88

Unigine Valley Before Update

25/53.8/115.9
Score 2252

Unigine Valley After Update

25.4/53.9/115.7
Score 2255

Everything has a margin of error size improvement except Rise of the Tomb Raider in DX11 which has a respectable increase with a 6 frame overall average increase and much better minimums.
Nothing to write home about but definitely an improvement for Pascal, If Joker's Vega results are repeatable that's a good win for AMD.

Thats interesting. From what i've seen so far it seems that it's mainly CPU bound games seeing the biggest improvements.

I would also like to see if it's Ryzen CPU's seeing the biggest gains.

nashathedog · 20 Oct 2017 at 12:47

40Four said:
Thats interesting. From what i've seen so far it seems that it's mainly CPU bound games seeing the biggest improvements.

I'd like to see focused CPU testing with this update, Let's hope someone does it.

nashathedog · 20 Oct 2017 at 12:52

Rroff said:
I'm beginning to think his results are comparing game mode before and after or something as his before results are mostly down on the general ballpark for my 1070 and what I've usually seen of the Vega 56 though its difficult to match up with his testing scenarios exactly.

We'll have to wait and see how other peoples testing goes. Hopefully the likes of Digital Foundry and the Aussie fella will do video's on it.

LoadsaMoney · 20 Oct 2017 at 13:01

Windows runs better on AMD, LOL

opethdisciple · 20 Oct 2017 at 13:18

So out of the box, is game mode on or off?

Do I have to enable it per game?

Panos · 20 Oct 2017 at 13:23

LoadsaMoney said:
Windows runs better on AMD, LOL

Expected. They fixed how to move around the processes in relation to the NUMA cores topology. I bet you similar perf gains should apply to SkylakeX CPUs and their topology. Which is similar to Ryzen

And those 10-20% gains on CPU bound games, is what separates Ryzen and CoffeeLake performances.....

Time for Nvidia optimise their stupid drivers for Ryzen and SkylakeX

enigma27 · 20 Oct 2017 at 13:48

opethdisciple said:
So out of the box, is game mode on or off?

Do I have to enable it per game?

Off.. you have to turn it on per game using the game bar.

Also this update does not work well with rivatuner.. getting eratic FPS with it running. Turned it off and is smooth as butter

opethdisciple · 20 Oct 2017 at 13:50

enigma27 said:
Off.. you have to turn it on per game using the game bar.

Also this update does not work well with rivatuner.. getting eratic FPS with it running. Turned it off and is smooth as butter

Alriggghty then.

Makes me wonder why they haven't turned it on by default.

Any one noticed any adverse effects?

Also you turn it on per game and it stays on every time you run the game?

40Four · 20 Oct 2017 at 14:04

opethdisciple said:
Alriggghty then.

Makes me wonder why they haven't turned it on by default.

Any one noticed any adverse effects?

Also you turn it on per game and it stays on every time you run the game?

I think they leave it off by default becasue it kills not critical processes and restarts them when you exit the game. I honestly doubt that very many on this forum will notice anything at all. The main benefit will be on systems with smaller amounts of system ram and weaker CPU's.

gupsterg · 20 Oct 2017 at 14:04

40Four said:
Thats a good increase for Vega there. Would be good to know what CPU was used and whether the same gains can be seen on Nvidia cards.

shankly1985 said:

I know it was Ryzen not sure what chip though.

Click to expand...

I think it was i7 7700K @ 5GHz, stated at ~2min into video.

LoadsaMoney · 20 Oct 2017 at 14:33

Everybody better rush out and grab a Vega, as AMD have just taken the gaming market from Nvidia in one hit!

CAT-THE-FIFTH · 20 Oct 2017 at 14:37

LoadsaMoney said:
Everybody better rush out and grab a Vega, as AMD have just taken the gaming market from Nvidia in one hit, as games run on Windows

AFAIK,Nvidia cards also get a boost?? Joker even said it in the comments.

CAT-THE-FIFTH · 20 Oct 2017 at 14:39

nashathedog said:
Mr Shanks posted link's to Jokers video on the new update in the Vega review thread last night so I tested a few things last night and then installed the update and redid the tests this morning.

My hardware is all at it's out of the box clocks and the tests are at 3440 x 1440.

Ryzen 5, 1600x,
Two 8gb sticks of 3200mhz DDR4 (8 pack Team Pro) and a
GTX 1080 with an out of the box overclock of 1797 boost with stock memory. (MSI 1080 Armor)

Game tested, API, min/avg/max

ROTTR DX11 Before Update

MP- 12.52/32.12/69.97
S-- 4.26/21/05/38.04
GV 9.64/22.11/41.16
Overall 25.13

ROTTR DX11 After Update

MP-- 24.28/39.89/77.14
S-- 6.99/26.17/47.71 v
GV-- 16.73/26.54/39.53
Overall 31.14

ROTTR DX12 Before Update

MP-- 22.81/40.42/60.22
S-- 2.50/22.76/68.16
GV-- 1.19/24.34/32.76
Overall 29.49

ROTTR DX12 After Update

MP-- 21.59/40.08/61.31
S-- 18.42/25.26/28.39
GV-- 6.63/24.72/32.47
Overall 30.29

F1 2016 Before Update

Clear Weather 55/75/95

Heavy rain 58/74/89

F1 2016 After Update

Clear Weather 61/78/96

Heavy Rain 58/74/88

Unigine Valley Before Update

25/53.8/115.9
Score 2252

Unigine Valley After Update

25.4/53.9/115.7
Score 2255

Everything shows a margin of error sized improvement except Rise of the Tomb Raider in DX11 which has a respectable increase of a 6 frame overall average and much better minimums.
Nothing to write home about but definitely an improvement for Pascal, If Joker's Vega results are repeatable that's a good win for AMD.

Look at the minimums in the TR bench though!! The same goes with F1 2016.

shankly1985 · 20 Oct 2017 at 14:56

gupsterg said:
I think it was i7 7700K @ 5GHz, stated at ~2min into video.

Ah thanks. Silly me count cores, for get about hyper threading.

nashathedog · 20 Oct 2017 at 15:19

opethdisciple said:
So out of the box, is game mode on or off?

Do I have to enable it per game?

Game mode?

CAT-THE-FIFTH said:
Look at the minimums in the TR bench though!! The same goes with F1 2016.

It's always good see the minimums go up.

CAT-THE-FIFTH · 20 Oct 2017 at 15:21

nashathedog said:
Game mode?

It's always good see the minimums go up.

Enable game mode by using Win+G. Depending on the game it can add a few extra FPS.