Extreme 4-Way Sli Tuning

GREEDFALL - 4-WAY SLI - FINAL CONCLUSION #2 - 2/5/20:

PLEASE HEAD OVER TO THE LAST POST ON PAGE 20, TO THE PREVIOUS POST FOR THE FIRST PART BEFORE CONTINUING, ALL PERFORMANCE READINGS ARE ON THAT POST, THANK YOU.

And the most important of all, the Actual Custom made bits themselves, here they are BUT FIRST:

ABSOLUTE FAIR WARNING ABOUT MY CUSTOM MADE BITS:

ONLY, REPEAT, ONLY FOR THE NVIDIA GTX 970 FAMILY, IT HAS NOT BEEN TESTED WITH ANYTHING ELSE, AND ONLY BEEN TESTED AND 110% GUARANTEE TO WORK WITH WIN 10X64 (15063-V.1703(RS2/RTM)ONLY, OTHER WISE I WILL GUARANTEE 0%, PLEASE HAVE BOTH OF THESE BEFORE CONTINUING, FAILURE TO DO SO, I WILL HOLD ABSOLUTELY NO RESPONSIBILITIES IF ANYTHING GOES WRONG, SO KEEP THIS IN MIND AT ALL TIMES BEFORE CONTINUING, 4-WAY IS NOTHING TO BE PLAYED AROUND WITH.

Custom Bits:

First grab a copy of this:

https://github.com/DeadManWalkingTO/NVidiaProfileInspectorDmW

Make sure you are using Nvidia Quadro Drivers 431.94 with modded .inf

Open up the nvidia inspector profile program and make the following changes below:

UNDER - 1 COMPATIBILITY:

AMBIENT OCCLUSION COMPATIBILITY: 0x0000001E (Need for Speed: Shift)

SLI BROADCAST (DX1x) BITS: 0x000008BA

SLI COMPATIBILITY BITS (DX10+DX11): 0x8A9D25F5

SLI COPY ENGINE: 0x00000002 FORCE_CE_BLIT - Force using CE blitter even if CE perfstrategy says otherwise.

SLI SPECIFIC HACKS (DX1x): 0xA0C00B6B

5 - COMMON:

AMBIENT OCCLUSION SETTING: PERFORMANCE

AMBIENT OCCLUSION USAGE: ENABLED

CUDA - P2 FORCE STATE: ON

MULTI-DISPLAY/MIXED-GPU ACCELERATION: SINGLE DISPLAY PERFORMANCE MODE

OPTIMIZE FOR COMPUTE PERFORMANCE: OFF

6 - SLI:

ANTIALIASING - SLI AA: ENABLED

# OF GPU'S TO USE SLI RENDERING MODE: FOUR
NVIDIA PREDEFINED # OF GPU'S TO USE IN SLI RENDERING MODE: AUTOSELECT
NVIDIA PREDEFINED # OF GPU'S TO USE IN SLI RENDERING MODE(DX1x): FOUR
NVIDIA SLI RENDERING MODE: AUTOSELECT
NVIDIA SLI RENDERING MODE IN (DX1x): SFR
SLI RENDERING MODE: SFR

Head over to 8 - extra and make the following changes below:

MCLINKINSTANCED_LIMIT: 0x00000064 Default - Threshold in megabytes for using of RT/ZB allocations as LinkInstanced (SLI DX9/10).

SYSMEMCACHEDCOPYBEHAVIOR: 0x00000002 ALLOW_FOR_TEXTURES_WITH_SUBRESOURCES - Allow for textures with mips.

And the very last piece of this puzzle, the UNKNOWN id settings, if you don't see these, I have another alternative.

First, click on the magnify glass on top of the nvidia inspector program(after launching the program) and scroll down to where you see UNKNOWN or UNKNOWN predefined nvidia settings and make the following changes below, be careful while doing it:


0x00ACBCC1: (1 Profile) 0x00000011 (Star Wars: Battlefront II (2017))
0x00C9F6F1: (4 Profiles) 0x0000000F (Battlefield 1, Plants vs Zombies Garden,etc)
0x00CBE005: (1 Profile): 0x00000001 (Space Junkies)
0x809D5F60: (399 Profiles): 0x08000001 (Battlefield 1, Sid Meier's Civilization VI,etc)

Another Alternative around this if you don't see these UNKNOWN id settings, and also, this other alternative will import all of bits themselves(whatever I used to tune with).

First of all, Open the Nvidia Inspector program:

Then under Export:

Drop down to where you see:

Export all driver Profiles(NVIDIA TEXT FORMAT)

Then open up the exported document with something like notepad, search for Greedfall, and delete all the stuff there and add this, MAKE SURE you make proper spacing where you see the setting ID:

Profile "Greedfall" UserSpecified=true
ShowOn All
ProfileType Application
Executable "greedfall.exe" UserSpecified=true
Setting ID_0x0007fbb6 = 0x00000000 UserSpecified=true
Setting ID_0x0013fec2 = 0x00000000 UserSpecified=true
Setting ID_0x002c7f45 = 0x0000001e UserSpecified=true
Setting ID_0x002ecaf2 = 0x00000001 UserSpecified=true
Setting ID_0x00313536 = 0x00000000 UserSpecified=true
Setting ID_0x0043ed70 = 0x00000003 UserSpecified=true
Setting ID_0x005a375c = 0x99941284 UserSpecified=true
Setting ID_0x0064b541 = 0x00000001 UserSpecified=true
Setting ID_0x00664339 = 0x00000001 UserSpecified=true
Setting ID_0x00667329 = 0x00000001 UserSpecified=true
Setting ID_0x00738e8f = 0xfffffffb UserSpecified=true
Setting ID_0x0078d9d9 = 0x00000000 UserSpecified=true
Setting ID_0x009cfec8 = 0x00000002 UserSpecified=true
Setting ID_0x00a06746 = 0xa0c00b6b UserSpecified=true
Setting ID_0x00a06946 = 0x8a9d25f5 UserSpecified=true
Setting ID_0x00a0694b = 0x000008ba UserSpecified=true
Setting ID_0x00a34323 = 0x00000000 UserSpecified=true
Setting ID_0x00acbcc1 = 0x00000011 UserSpecified=true
Setting ID_0x00b21c03 = 0x00000000 UserSpecified=true
Setting ID_0x00c96f61 = 0x0000000f UserSpecified=true
Setting ID_0x00cbe005 = 0x00000001 UserSpecified=true
Setting ID_0x00d98f93 = 0x00000000 UserSpecified=true
Setting ID_0x00db3853 = 0x00000002 UserSpecified=true
Setting ID_0x00db3857 = 0x00000064 UserSpecified=true
Setting ID_0x1033cec2 = 0x00000004 UserSpecified=true
Setting ID_0x1033ced1 = 0x00000004 UserSpecified=true
Setting ID_0x1033dcd1 = 0x00000004 UserSpecified=true
Setting ID_0x1033dcd3 = 0x00000004 UserSpecified=true
Setting ID_0x107afc5b = 0x00000001 UserSpecified=true
Setting ID_0x1094f157 = 0x00000000 UserSpecified=true
Setting ID_0x1094f1f7 = 0x00000000 UserSpecified=true
Setting ID_0x10a879ac = 0x00000001 UserSpecified=true
Setting ID_0x10a879cf = 0x00000001 UserSpecified=true
Setting ID_0x10c158ad = 0x00000000 UserSpecified=true
Setting ID_0x10fc2d9c = 0x00000004 UserSpecified=true
Setting ID_0x1194f158 = 0x00000000 UserSpecified=true
Setting ID_0x2089bf6c = 0x00000023 UserSpecified=true
Setting ID_0x20c1221e = 0x00000001 UserSpecified=true
Setting ID_0x702244b7 = 0x00000000 UserSpecified=true
Setting ID_0x809d5f60 = 0x08000001 UserSpecified=true
EndProfile

EVERYTHING that I have done with this game while tuning should all be there, Including the UNKNOWN ID settings.

----------------------------------------------------------------------------------------------------------------------------------------------------------

So in conclusion to all about this game.

If your NOT going to be using SSAO and prefer to stay away from its graphical anomaly issues, then 3-WAY SLI will be the very best option at this point, if in-fact you can can play using SSAO and do NOT care about the blur/trail while moving the camera, then I would Very highly recommend using 4-Way, but if you do, take a break every now and then, not very good for your eyes in this condition.

I mean, I don't mind playing it in this condition but I wouldn't play it for long hrs, maybe for a few hrs then take a good break.

So Final Conclusion:

My recommendation under 4-Way SLI???

ONLY if you use SSAO, to further the stress the card down and bring excellent potential to the Gpu's, otherwise, stick with 3-Way Sli.

------------------------------------------------------------------------------------------------------------------------------------------------------

GRAPHIC DRIVERS USED: ALL THESE ARE MODIFIED:

NVIDIA QUADRO DRIVERS: VERSION 431.94

GPU CORE/MEMORY TESTED: 1 TO 4-WAY SLI MODE = 1443MHZ(CORE)/7600+(MEMORY)

SYSTEM RAM USAGE: ABOUT 14GB+

OS USED: WIN 10 x64 (15063.2108/VERSION: 1703/RS2/RTM)

STORAGE USED: 2 SAMSUNG 950 PRO(512GB) IN STRIPPED(RAID 0) MODE.

MOTHERBOARD USED: SUPERMICRO X10drg-q

MOTHERBOARD BIOS USED: 3.1

INTEL CHIPSET DRIVER USED: 10.1.18010.8141+10.1.18263.8193 update

TOTAL TUNING/PERFORMANCE MONITORING/TESTING TIME/OVERALL STABILITY CHECKS: ABOUT 30-40+ HRS EASY.

THANK YOU.

 

CURRENT GAMES TO BE WORKED ON - 4-WAY SLI - UPDATE - 2/6/20:

Going to go back and put more work on the following game or further improve performance.

COD: Infinite Warfare(Scales Excellent, just need some work)

Ghost of a Tale(Worked on it months ago, see if I can help further improve performance)

Darksiders 3(Only been on it for a bit, needs more work)

And a new game never played before:

Project Cars 2

Also looking into Dark Souls III, I know I worked on this in the past and I already hit the fps cap wall of 60fps.

But its been a long time since I been on it, I'll re-check everything again.

But going to take a look on Project Cars 2, Looks very good, I like some car games every now
and then.

List of ALL DX10 games:

https://www.pcgamingwiki.com/wiki/List_of_DirectX_10_games

Here are list of ALL DX11 games:

https://www.pcgamingwiki.com/wiki/List_of_DirectX_11_games

If anyone have a personal request for a game to be looked into, I will gladly do so without no hesitation whatsoever, so everyone, I am more then open to anything at this point, Whatever games you all want me to look into, wether personal request or otherwise, doesn't matter, more then welcome, thanks.

Be sure to check back for more updates in regards to this, thank you for reading.

 

PROJECT CARS 2 - 4-WAY SLI - 1ST UPDATE AND STAGGERING UPDATE - 2/8/20:

Hello everyone, once again, continuing my efforts on my 4-Way Sli Project.

Got some Absolute damning news to everyone...

I checked one gpu performance, messed around with some supersampling and other AA, to figure out which one would be right for me, then All I did was make some quick changes within the SLI Compatibility bits themselves, no other changes, and rendering mode of choice again is non other then SFR, you heard me right, SFR, might shock alot of people here, SFR is really not being used nowadays at all, at least, not that I know of.

holy SWEET baby JaMes!!!

Did quick test, everything was maxed out(EXCEPT TEXTURE), Supersampling set to Medium, NO MSAA, NO Post AA, Supersampling takes tremendous hit folks so bare that in mind please.

1 GPU gave me around 12 fps or so, and guess what 4 cards gave me? SHAKES HEAD(Like Can't believe it).

About 51-53 FPS and I am not kiddin', Very serious about this one.

All running at a Staggering 3840x2400 with a Gtx 970!

Although the overall scaling with supersampling turned on = Unbelievable scaling, if I use other AA methods such as MSAA or SMAA for that matter, even at their max respective settings, the overall scaling is no where near good as the supersampling, supersampling takes absolute tremendous hit, the performance cost is HUGE.

So trying to work on the overall scaling with other AA methods as they perform quite a bit better then its Supersampling counterpart.

Just to make it perfectly clear, this isn't the only game that over-performs under 4-Way Sli, Hitman Absolution(Enhanced edition) is very well known for it and other few games, Hitman being the absolute most craziest one out there, one area gave me 8 fps using 1 gpu, 4 gave me over 50 and I kid you not, might want to back track to where I was doing hitman absolution, its been months ago, all have a gander when you all get a chance but that was with SGSSAA+MSAA though Nvidia Inspector, so there will be an absolute tremendous hit on performance.

The only Explanation I might have, especially in Hitman's Case, I have to Re-Observe the situation to actually pin-point it, but I believe it could be that since its putting alot of stress with just 1 card, Well in Hitman's case, Over stressing the card, the MS(Frametime) is very very very very high, thus, explains the low fps, It could be that, the card, 1 gpu mode, it wants to run its desired fps, but since the MS(Frametime) is super super high, it could be I lost some fps while I was at it and so while running in MGPU mode, the MS(Frametime) drastically goes down, and fps goes up, This is my only explanation, could be the MS(Frametime) playing a major role here, not exactly to sure, its a very good possibility, Alot more testing on this needs to be conducted to be absolutely sure for the fact, so again, do NOT take this for granted just as of yet, I will be conducting a full investigation on this issue.

Better then perfect scaling.

Also, it seems the AA quality is Exactly the same as it looks while under 1 gpu mode, Great news.

Thank goodness for being a DX11 title.

And I just got on to the game, so this is my first time playing the game and first initial impression under 4-Way SLI.

There you have it.

Going to play long time to check further stability, but initial impression is Absolute Damning Magnificent...

In conclusion to my experience on all of this 4-Way Sli this far, if anyone has been paying absolute close attn. From beginning of this project up an till now, you can clearly see that there is a lot more pro's then cons, just takes tons of time and a very strong ambition to do so.

Sli is not officially 100% dead yet, long as I am still around, it will continue to pounce, the Rebirth of SLI, .

Thank you all for reading.

 

UPDATE ON 3 DRIVERS - 431.98 | 442.19 | 450.12 - 4-WAY SLI - 2/1/19:

Just heads up on all of these 3 drivers I personally tested...

First one being the 450.12, and as expected, just like the 44x.xx themselves = X.

Next Driver I used: Modified Quadro Drivers 442.19, again, I expected the same results as the 450.12 = DUD.

And save best for last:

First time I ever used these drivers:

Modified Quadro Drivers: 431.98, and it works "PERFECTLY".

So now I can confirm, the absolute latest drivers for 4-Way that's 100% functioning correctly is the 431.98(Quadro) and Vulkan 436.65(Beta) Drivers.

Again, as a remainder, anyone with similar setup as mine, DO "NOT" USE any drivers from 44x.xx and 45x.xx until further test is conducted...

Although ALL the 44x.xx and the 450.12 drivers that I personally tested seems to be working 3/4 of the way, I did some bit tuning on 442.19 drivers to see if I can change anything, Absolutely no go, so do NOT use.

Thanks for reading.

 

PROJECT CARS 2 - 4-WAY SLI - UPDATE - 2/10/20:

Progress report thus far in this game:

So far, everything is looking very good.

Couple of things here:

Supersampling takes Tremendous hit on this game, I can set it to medium + SMAA to ultra but @ 2400p but the texture's themselves MUST be set to medium otherwise you will see decrease scaling potential, Medium will fix this issue.

Although Supersampling @ medium settings takes Tremendous punch on vram, Although my Vram is completely wiped out, My physical ram utilization is in the low 20gb, despite the scaling is absolutely incredible, you will experience kinda jerky motion when moving the camera left and right, the transition through the frames won't be smooth, despite being over 60 fps, this is due to kinda high MS(Frametime).

I am now currently trying to get the most out 2400p, Texture @ high, SMAA @ Ultra, and SSAA set to low.

Although, some tracks, scaling is an absolute beast, some areas of the tracks, scaling is not quite perfect, its good, but not like SSAA @ medium setting.

MS(Frametime), the tuning I am been doing lately, seems to be very good, about 6ms-12ms.

Trying to keep it within that range.

Far as weather goes, Heavy Rain and Blizzard, there is some kinda high MS(Frametime), over 14ms-15ms+, so there will be jerky motion while moving your car left and right(background speaking), despite fps being solid.

So far, runs Very Very Very well, there is one more that I must mention here:

Shadow Lod Fade,

There seems to be a problem in this game, just like GTA V, but the distance is much further out in this game then GTA V but it is there, no biggie though...

So Far, looks really good, just trying to improve overall scaling in certain tracks, for the most part, Runs Excellent, with no visible graphic anomalies of any sort, All running under SFR(SPLIT-Frame-Rendering) Mode.

Always continue to check back later as this game is currently:

----------WORK IN PROGRESS----------

 

GHOST IN A TALE - SFR MODE UPDATE - 2/20/20:

As stated before, Wanted to go back to this game and double check its performance as I haven't been on this game in quite some time.

Wanted to see how SFR would scale and as expected, Scales like Incredible Hulk.

Pretty much the same as I previously did before under AFR mode, MS(Frametime) seems roughly the same.

I stated earlier that this scales decent, let me re-phrase, it scales more then TOP DOG.

Resolution @ 2400p(4k/16:10 Aspect Ratio), Everything Set to Max(Except Texture/half)

I can Set the Texture to normal but there might be some slight hang here and there, usually in the area where new visiting area is being loaded, seems texture set to half helps this issue a bit.

Even at half texture resolution, seems every bit as good for the most part, looks real nice.

I did one movement pattern test, a quick one:

1 GPU mode:

MIN FPS: 18

MAX FPS: 23

4 GPU mode:

MIN FPS: 72-73

MAX FPS: 89

MS(FRAMETIME) is 10ms - low 13ms.

This is just heads up on SFR mode in general...

I found some weak scaling spots but they are very MINOR, in these weak scaling spots/area/specific camera angle/view, they are not very serious at all, but trying to improve this situation a bit, but nothing to serious.

Also, another heads up, there is some Slight Scaling drag spots, some not to bad, at times it hits real hard, same exact spot, depending where you are in the game, but its always in the same spot/location, this ALSO pertains to 1 gpu mode so it does not matter.

Next, I tried lowering down all the other settings, even to quarter half texture resolution, same thing.

Other then that, Overall Scaling Potential = A+ all the way.

In regards about Scaling Drag issues in Certain Minute Spots:

Yakuza Kimawi 2 and DQ11 had the same exact problems as well, Yakuza 2 being the Worst...

There was some in RE2 remake as well but not to bad, NOT ALL game have this, need to make it perfectly clear, only a few games have this problem.

Also, Anti-Aliasing in this game, Wether @ native Resolution or Custom Resolution(Greater then Native), seems to be working perfectly.

I know games in the past, Like Mass Effect 2, SGSSAA+MSAA only works @ native resolution or under, not anything higher, This pertains only in 4-Way Sli, 2x2 SS works in that game though so it all pays off.

I would like to further conduct performance testing for this game as I have not before, but also currently working on Project Cars 2 and that's coming along Incredibly well...

I just changed My Sli compatibility bits around to work with SFR, and yes, Beast mode.

Gonna be a bit before any performance write-up is done.

So check back later on this, it will be conducted later, thanks for reading once again.

----------------------------------------------------------------------

CRASH UPDATE UPON LOADING SAVE GAME:

----------------------------------------------------------------------

Just another heads up on this game on what I found...

I found some Crash error upon loading up saved game, usually it will quit about half way to 3/4 of the way, might get an error report or just simply would stay there and quit.

How this all happened???

I was trying something within this game, what I did is trying to cap the fps to 60, to see if the scaling drag would go away, upon loading up saved game, game quits but the loading screen is still up.

Next what I did is trying to force vsync within nvidia inspector, same thing...

So I then reverted everything back to normal where I originally had it, guess what? Same thing again...

So what I did next is reset my Profile within Nvidia Inspector, Put back everything I once had before, Restarted my System, loaded up the game, this time with FPS LOCK set to none, loaded up saved game once more, this time it goes through.

So just heads up on this, people with similar setup, keep this in mind, thanks.

UPDATE ON 4.5K TESTS & OVERALL STABILITY CHECKS.

Did some 4.5k(4434x2771) tests, and its putting more further load to the spots that's has slightly low scaling potential, Game Scales Absolutely incredible.

For best scaling, @ least 2400p or higher(Would not go any higher then 4.5k).

For the most part, very smooth, most of the time over 60fps, I have seen some slower.

Far as overall game stability is concerned, Rock Solid stable during game play, played the game long enough to know the overall stability is excellent, game has never cut me off during gameplay.

There is some slight graphical blur anomaly while your hiding in the barrels, when you have motion blur turned on, you can see an overlay blur over the barrel while your in it, to fix this issue, disable motion blur, other then that, good stuff.

Tuning Sensitivity = Top Notch, absolutely nothing to be concern about.

Longs your overclocks are 100% stable, game is rock hard stable...

Going to do some jot down some performance reading off this game, Incredible scaling here folks, your not kiddin'.

 

PROJECT CARS 2 & GHOST IN A TALE - 4-WAY SLI - UPDATE - 2/12/20:

Final Performance write-up for both games will be coming soon, as for now, here are some trailers, Get ready as both games Scales like there is no tommorrow:

Project Cars 2:





Ghost in a Tale(Game Created only hand full of people):

 

GHOST IN A TALE - 4-WAY SLI - REVISITED UNDER SFR MODE - FINAL CONCLUSION - 2/12/20:

I present to you Ghost of a Tale, Under SFR(Split-Frame-Rendering) Mode...

Although I worked on this game Long time ago, I wanted to go back on the game and see how SFR would turn out, and as I originally Expected, This sucker scales like there is absolutely no Tomorrow, oh my goodness and my Gracious!!!

As promised, I wanted conduct my own personal performance testing and here they are...

Far as tuning the bits themselves, they were not much work at all under SFR mode, I just what I thought it would work and walla.

Before I present you the Bits, here are the results that everyone has been waiting for.

Before I start, All the Static(Idle in one particular location/camera view) and Movement Pattern, all are @ 4434x2771(4.5k/16:10 Aspect ratio) Resolution.

Reason for this, I notice very few minute spots that could beef up the overall scaling potential, although far from being bad, just wanted a booster and does it help.

To make good use under 4-Way with the 970's, AT LEAST 2400p(4k/16:10 Aspect) Must be used, although 2160p(4k/16:9 Aspect Ratio) could be used, you won't get the extra potential out of it.

If anyone needs some list down for their 16:9 monitors, here they are:

https://pacoup.com/2011/06/12/list-of-true-169-resolutions/

---------------------------------------------------------------------------------------------------------------------------------------------

First lets go over some STATIC performance shall we???

ALL TEST CONDUCTED FROM HERE ON OUT, 4434x2771(4.5K), EVERYTHING SET TO MAXIMUM EXCEPT TEXTURE(HALF). ALL STATIC PERFORMANCE TEST WERE RECORDED AT THEIR MAXIMUM FPS SHOWN.

STATIC #1

1 GPU: 22 FPS

4 GPU's: 84 FPS

STATIC #2:

1 GPU: 15 FPS

4 GPU's: 59 FPS

STATIC #3:

1 GPU: 17 FPS

4 GPU's: 66 FPS

STATIC #4:

1 GPU: 20 FPS

4 GPU's: 80 FPS

STATIC #5:

1 GPU: 14 FPS

4 GPU's: 52 FPS

STATIC #6:

1 GPU: 16 FPS

4 GPU's: 60 FPS:

-----------------------------------------------------------------

MOVEMENT PATTERN PERFORMANCE TEST:

-----------------------------------------------------------------

MOVEMENT PATTERN #1:

1 GPU:

MIN FPS: 18//AVG FPS: 20//MAX FPS: 23//1% LOW: 17 FPS//.1% LOW: 17 FPS

4 GPU's:

MIN FPS: 67//AVG FPS: 80//MAX FPS: 88//1% LOW: 58 FPS//.1% LOW: 14 FPS

MOVEMENT PATTERN #2:

1 GPU:

MIN FPS: 14//AVG FPS: 15//MAX FPS: 17//1% LOW: 11//.1% LOW: 7 FPS

4 GPU's:

MIN FPS: 54//AVG FPS: 59//MAX FPS: 68//1% LOW: 46 FPS//.1% LOW: 13 FPS

MOVEMENT PATTERN #3

MIN FPS: 14//AVG FPS: 16//MAX FPS: 17//1% LOW: 14 FPS//.1% LOW: 14 FPS

4 GPU's:

MIN FPS: 57//AVG FPS: 61//MAX FPS: 65//1% LOW: 51 FPS//.1% LOW: 43 FPS

MOVEMENT PATTERN #4:

1 GPU:

MIN FPS: 13//AVG FPS: 17//MAX FPS: 20//1% LOW: 13 FPS//.1% LOW 8 FPS

4 GPU's:

MIN FPS: 54//AVG FPS: 68//MAX FPS: 83//1% LOW: 53 FPS//.1% LOW: 38 FPS

---------------------------------------------------------------------------------------------------------------

TEMP RESULTS:

---------------------------------------------------------------------------------------------------------------

ALL GPU'S WERE @ 99% SCALING USAGE, EXCEPT FOR GPU # 3(COMMON), ABOUT 95-96%+

Ran for good 20-30 mins or so give or take in a Static(idle) Position, in other words, in specific camera angle/view while not moving your main charc around.

In a mid 70 degree Fahrenheit room, with room to the door wide open, windows closed.

And the Results are as follows:

GPU #1: 60-61c @ 1.168V
GPU #2: 65-66c @ 1.175V
GPU #3: 66-67c @ 1.187V
GPU #4: 63c @ 1.1206V

---------------------------------------------------------------------------------------------------------------------------

CUSTOM MADE BITS:

Here they are BUT FIRST:

WARNING ABOUT MY CUSTOM MADE BITS:

110% GUARANTEE TO WORK, ONLY, REPEAT, ONLY FOR THE NVIDIA GTX 970 FAMILY, ALONG WITH WIN 10 X64(15063.V1703/RS2-RTM) ONLY, FAILURE TO MEET THESE REQUIREMENTS, I WILL GUARANTEE 0%, SO MAKE SURE OF THIS BEFORE YOU CONTINUE.

First grab this version of Nvidia Inspector:

https://github.com/DeadManWalkingTO/NVidiaProfileInspectorDmW

Then open up the program and make the following changes below:

1 - COMPATIBILITY

SLI BROADCAST (DX1x) BITS: NOT USED OR 0x00000000

SLI COMPATIBILITY BITS (DX10+DX11): 0x8A9815F5

SLI COPY ENGINE: 0x00000001 On - Enable the copy engine blitter.

SLI SPECIFIC HACKS: 0x00000100 FORCE_REPORT_NO_SLI_SUPPORT_FROM_NVAPI - If applition has buggy its own SLI support we need to do SLI magic in the driver and force app think it's run on single GPU.

2 - SYNC AND REFRESH:

ALL G-SYNC SETTINGS: TURNED OFF

MAXIMUM PRE-RENDERED FRAMES: 4

PREFERRED REFRESH RATE: HIGHEST AVAILABLE

VERTICAL SYNC: OFF

VERTICAL SYNC TEAR CONTROL: ADAPTIVE

4 - TEXTURE FILTERING

TEXTURE FILTERING - LOD BIAS (DX): 0.625

TEXTURE FILTERING - QUALITY: HIGH QUALITY

5 - COMMON

AMBIENT OCCLUSION SETTING: PERFORMANCE

AMBIENT OCCLUSION USAGE: ENABLED

MULTI-DISPLAY/MIXED GPU-ACCELERATION: 0X00000006

OPTIMIZE FOR COMPUTE PERFORMANCE: OFF

THREADED OPTIMIZATION: ON

6 - SLI

ANTIALIASING - SLI AA: ENABLED
# OF GPU's TO USE SLI RENDERING MODE: FOUR
NVIDIA PREDEFINED # OF GPU's TO USE IN SLI RENDERING MODE: AUTOSELECT
NVIDIA PREDEFINED # OF GPU's TO USE IN SLI RENDERRING MODE(DX1x): FOUR
NVIDIA PREDEFINED SLI MODE: AUTOSELECT
NVIDIA PREDEFINED SLI MONE IN DX1x: SFR
SLI RENDERING MODE: SFR

----------------------------------------------------------------------------------------------------------------------------

FINAL WRAP UP ABOUT THIS GAME:

Most of my time is checking the overall stability of this game under SFR mode, as Just did a quick change to the SLI Compatibility bits, ran it through SFR and wham.

So great deal of my time was checking overall stability and doing performance test...

Overall Stability for this game is Absolutely superb.

There is some Scaling Drag issues when you hit certain spot in a specific location, similar to my experience with DQ11 and Yakuza Kiwami 2, but in this game, it pertains to 1 gpu mode as well, so it doesn't matter in the end.

This game is very playable with Textures set to high but, other then those infamous drag spots, you will experiences a bit more hitching and such IF textures set to high, for this reason, I absolutely strongly recommend leaving the texture set to half, Still looks real nice...

Overall scaling potential, if played in the right resolution, is holy living hell, incredible to say at least.

My Final Conclusion under 4-Way Sli with this game?

4-WAY SLI FOR GHOST IN A TALE???

100% RECOMMENDED, 2 Royal thumbs up...

---------------------------------------------------------------------------------------------------------------------------

GPU USED:

4 GIGABYTE GTX 970'S WINDFORCE

GPU SPEEDS: 1443mhz(CORE)/7.5K+mhz(MEMORY)

GRAPHIC DRIVER USED: MODIFIED QUADRO DRIVERS 431.98

MOTHERBOARD USED: SUPERMICRO X10drg-q

MOTHERBOARD BIOS USED: 3.1

POWER SUPPLIES USED: 2 FSP 1200W AUROM pt

SYSTEM RAM USAGE: UP TO 20GB(SAFE)

OS USED: WIN 10 x64 (15063.2108/VERSION: 1703/RS2/RTM)

STORAGE USED: 2 SAMSUNG 950 PRO(512GB) IN STRIPPED(RAID 0) MODE.

INTEL CHIPSET DRIVER USED: 10.1.18010.8141+10.1.18263.8193 update

TOTAL TUNING/PERFORMANCE MONITORING/TESTING TIME/OVERALL STABILITY CHECKS + WHATEVER I DID MONTHS AGO WITH THIS GAME: OVER 20-30 HRS.

SEE YOU ALL IN THE FINAL CONCLUSION TO PROJECT CARS 2, BE SURE TO LOOK FOR IT.

THANK YOU.

 

PROJECT CARS 2 - 4-WAY SLI - FINAL CONCLUSION - 2/13/20:

Final Conclusion to Project Cars 2 is now here:

First, if anyone wants to listen to the OST while viewing:



Tuning wasn't all that hard for me from the get go, it was the tuning of the in-game settings and the amount of cars while playing the game(VERY IMPORTANT).

The amount of cars shown in the background while you play, plays an extremely important role on how your overall SLI consistency will react.

For instance, in one track, I can increase it to 15 cars along with blizzard weather, in other tracks, I can only put it to like 7 cars.

Every track and weather along with it is going to play different.

As Promised, I jotted down some performance in regards about this game.

I did some unusual testing here, all because for the scaling to be top notch, along with semi-good MS(Frametime), and etc.

Overall, Game scales Extremely well, all depends on certain weather/cars/situation,etc.

I might add more in this testing, so you all can continue to check back more later, but I think this is good enough for now.

Several different test's, I used different resolution all together, all because to fine tune the Sli Consistency and its Overall Usage, Alot of it is how many cars being displayed while playing, plays a big role here.

Settings are:

Texture Quality - Medium
Smaa - Ultra
SuperSampling - Low
Reflections - High
Environmental Map - High
Car Detail - High
Track Detail - Medium
Pit Crew - All
Shadow Quality - High
Enhanced Mirror - Yes
Motion Blur - High
Max Frames Render Ahead - 4
Grass Density - High
Particle - High
Particle Density - Ultra

Also "ALL" Visual FX is turned on.

For the FIRST 5 performance readings:

Resolution is set to 4352x2720, 0 cars on ALL FIRST 5 TRACKS.

For TRACK #1:

Algarve(0 cars used) @ 14:00 PM/(Medium Cloud):

1 GPU:

MIN FPS: 23//AVG FPS: 27//MAX FPS: 30//1% LOW: 22 FPS//.1% LOW: 21 FPS

4 GPU's:

MIN FPS: 80//AVG FPS: 93//MAX FPS: 108//1% LOW: 62 FPS//.1% LOW: 49 FPS

Algarve(0 cars used) @ 14:00 PM/(Thunder Storm):

1 GPU:

MIN FPS: 15//AVG FPS: 17//MAX FPS: 19//1% LOW: 15 FPS//.1% LOW: 15 FPS

4 GPU's:

MIN FPS: 58//AVG FPS: 65//MAX FPS: 73//1% LOW: 55 FPS//.1% LOW: 46 FPS

Algarve(0 cars used) @ 21:00 PM/(Blizzard/Head lamps used):

1 GPU:

MIN FPS: 16//AVG FPS: 18//MAX FPS: 23//1% LOW: 16 FPS//.1% LOW: 15 FPS

4 GPU's:

MIN FPS: 61//AVG FPS: 69//MAX FPS: 81//1% LOW: 59 FPS//.1% LOW: 53 FPS

Track #2:

Bathurst(0 cars used) @ 14:00 PM/(Medium Clouds):

1 GPU:

MIN FPS: 20//AVG FPS: 25//MAX FPS: 29//1% LOW: 21 FPS//.1% LOW: 20 FPS

4 GPU's:

MIN FPS: 77//AVG FPS: 92//MAX FPS: 106//1% LOW: 75 FPS//.1% LOW: 57 FPS

Bathurst(0 cars used) @ 14:00 PM/(Thunder Storm):

1 GPU:

MIN FPS: 14//AVG FPS: 16//MAX FPS: 20//1% LOW: 14 FPS//.1% LOW: 14 FPS

4 GPU's:

MIN FPS: 53//AVG FPS: 61//MAX FPS: 74//1% LOW: 52 FPS//.1% LOW: 47 FPS

Bathurst(0 cars used) @ 21:00 PM/(Blizzard):

1 GPU:

MIN FPS: 16//AVG FPS: 19//MAX FPS: 25//1% LOW: 16 FPS//.1% LOW: 15 FPS

4 GPU's:

MIN FPS: 57//AVG FPS: 68//MAX FPS: 87//1% LOW: 58 FPS//.1% LOW: 54 FPS

TRACK #3:

Circuit of Americas(0 cars used) @ 14:00 PM/(Medium Clouds):

1 GPU:

MIN FPS: 24//AVG FPS: 27//MAX FPS: 30//1% LOW: 24 FPS//.1% LOW: 22 FPS

4 GPU's:

MIN FPS: 81//AVG FPS: 92//MAX FPS: 106//1% LOW: 64 FPS//.1% LOW: 50 FPS

Circuit of Americas(0 cars used) @ 14:00 PM/(Thunder Storms):

1 GPU:

MIN FPS: 15//AVG FPS: 17//MAX FPS: 20//1% LOW: 15 FPS//.1% LOW: 15 FPS

4 GPU's:

MIN FPS: 56//AVG FPS: 63//MAX FPS: 75//1% LOW: 48 FPS//.1% LOW: 39 FPS

Circuit of Americas(0 cars used) @ 21:00 PM/(Blizzard):

1 GPU:

MIN FPS: 16//AVG FPS: 19//MAX FPS: 24//1% LOW: 17 FPS//.1% LOW: 16 FPS

4 GPU's:

MIN FPS: 62//AVG FPS: 70//MAX FPS: 87//1% LOW: 56 FPS//.1% LOW: 45 FPS

TRACK #4:

Bryno(0 cars used) @ 11:00 AM/(HAZY):

1 GPU:

MIN FPS: 22//AVG FPS: 25//MAX FPS: 29//1% LOW: 22 FPS//.1% LOW: 21 FPS

4 GPU's:

MIN FPS: 81//AVG FPS: 93//MAX FPS: 102//1% LOW: 80 FPS//.1% LOW: 75 FPS

Bryno(0 cars used) @ 11:00 AM(Heavy Fog+Rain):

1 GPU:

MIN FPS: 13//AVG FPS: 15//MAX FPS: 19//1% LOW: 13 FPS//.1% LOW: 12 FPS

4 GPU's:

MIN FPS: 49//AVG FPS: 56//MAX FPS: 72//1% LOW: 48 FPS//.1% LOW: 45 FPS

Byrno(0 cars used) @ 8:00 AM/(Blizzard/Snow Season)

1 GPU:

MIN FPS: 14//AVG FPS: 15//MAX FPS: 19//1% LOW: 14 FPS//.1% LOW: 13 FPS

4 GPU's:

MIN FPS: 53//AVG FPS: 58//MAX FPS: 67//1% LOW: 57 FPS//.1% LOW: 48 FPS

TRACK #5:

Fuji GP(0 cars used)/(Summer Weather/Overcast @ 11:00 AM):

1 GPU:

MIN FPS: 20//AVG FPS: 21//MAX FPS: 24//1% LOW: 20 FPS//.1% LOW: 18 FPS

4 GPU's:

MIN FPS: 69-70//AVG FPS: 75//MAX FPS: 82//1% LOW: 67 FPS://.1% LOW: 48 FPS

------------------------------------------------------------------------------------------------------------------------

THIS SECTION COVERS WITH CARS @ 2559p(4095x2559)...

Fuji GP(15 cars in total on track @ 11:00 AM)/(SNOW SEASON/Heavy Snow Weather):

1 GPU:

MIN FPS: 19//AVG FPS: 21//MAX FPS: 24//1% LOW: 20 FPS//.%1 LOW: 19 FPS

4 GPU's:

MIN FPS: 71//AVG FPS: 81//MAX FPS: 92//1% LOW: 57 FPS//.1% LOW: 47 FPS

Fuji GP(3 cars in total on track @ 11:00 AM)/Heavy Rain + Fog):

1 GPU:

MIN FPS: 14//AVG FPS: 16//MAX FPS: 19//1% LOW: 14 FPS//.1% LOW: 13 FPS

4 GPU's:

MIN FPS: 56 FPS//AVG FPS: 63//MAX FPS: 87//1% LOW: 55 FPS//.1% LOW: 26 FPS

NurBurgring GP(5 cars total on track @ 11:00 AM)/(Heavy Fog+Raid):

1 GPU:

MIN FPS: 12//AVG FPS: 14//MAX FPS: 19//1% LOW: 13 FPS//.1% LOW: 12 FPS

4 GPU's:

MIN FPS: 54//AVG FPS: 73//MAX FPS: 93//1% LOW: 54 FPS//.1% LOW: 24 FPS

NurBurgring GP(7 cars total on track)/(Blizzard Weather conditions):

1 GPU:

MIN FPS: 17//AVG FPS: 19//MAX FPS: 22//1% LOW: 18 FPS//.1% LOW: 17 FPS

4 GPU's:

MIN FPS: 64//AVG FPS: 71//MAX FPS: 83//1% LOW: 54 FPS//.1% LOW: 37 FPS

---------------------------------------------------------------------------------------------------------------

TEMPS:

About good 30+ mins of game play, Absolute brutal scaling potential on all 4 gpu's.

Ambient Temp: low 70 degree Fahrenheit, Windows Tightly Shut, room to the door: WIDE OPEN:

Results are as follows:

GPU #1: 63c @ 1.168V
GPU #2: 69-70c @ 1.175V
GPU #3: 69-70c @ 1.187V
GPU #4: 66-67c @ 1.1206V

----------------------------------------------------------------------------------------------------------------

I would like to include more Performance tests, if so, I will revise the above in a later time.

I have not went through all the tracks with all the weather, but played a good deal to know whats going on in the game.

Personally, if your going to play, such as in practice mode with 0 cars on the track, you might want to beef up the resolution to: 4352x2720.

If your going to add cars to your tracks, Try reducing it to: 4095x2559.

Reason being is that, if your adding cars @ 2720p resolution, you will get some abnormal reactions on the scaling usage, so you MUST bring it down some.

If you bring the resolution down to 2400p, guess what happens? The overall scaling potential would still be good but it will not come close to the above performance reading.

All this is about getting games to scale at its peakest potential, whatever it takes, while keeping the MS(Frametime) at a reasonable rate.

If you want to play your game with High Quality Texture setting, try reducing your res to 2400p.

Do NOT under any circumstances, Going higher with the Track Detail setting, leave it at medium setting, going any higher, will lessen the overall scaling potential, keep it medium or to low for best performance.

If your doing some tracks with cars in general, and having problems with the Sli Consistency, for instance if gpu 2 and 3 is like saying 90% then all then sudden drops down to 79%, this usually means that certain settings are set to high, in this case, the amount of vehicles on screen, you MUST toy around with the amount of cars on screen, and along with the weather and the track itself, Every track and weather in general, you MUST toy around with the amount of vehicle displayed on the track, try to find the best one the keeps good scaling usage and overall consistency, try finding one that scales over 90% of the time, nothing under.

---------------------------------------------------------------------------------------------------------------------

FINAL CONCLUSION:

This has been an interesting game to tune. Had a good time with it.

Overall scaling consistency, if your using cars on the tracks, it MUST be toyed around with to find the very best SLI scaling Consistency.

If find the right one, along with the weather conditions and such, the Scaling is Absolutely Superb.

There is one downfall while under weather with the above mention resolutions:

On the background itself, you will experience kinda jittering movement, while moving the camera left and right, the overall transition won't be as smooth due to the extreme weather conditions(Due to MS(Frametime) being a lil high, really doesn't take it away really, since the weather is real bad out there, you probably won't even notice it.

I have NOT tried this this game with vsync enabled to see if this situation goes away, I will give it a try some other time, but like I said, its really not that to be concern about.

RECOMMEND UNDER 4-WAY???

If proper amount of car's, weather, etc is being used, then: ABSOLUTELY YES!!!.

Despite some issues here and there, Overall scaling potential ranges from good to Excellent.

---------------------------------------------------------------------------------------------------------------------------

MY CUSTOM MADE BITS:

My Custom made bits, here they are BUT FIRST:

WARNING ABOUT MY CUSTOM BITS:

"ONLY" FOR THE NVIDIA GTX 970 FAMILY, IT HAS NOT BEEN TESTED ON ANY OTHER GPU'S, AND ONLY BEEN TESTED ON WIN 10 X64(15063.V1703-RS2/RTM) ONLY.
FAILURE TO MEET BOTH OF THESE REQUIREMENT, I WILL GUARANTEE 0%, SO MAKE SURE OF THIS BEFORE CONTINUING.

Going to make this Real short and snappy.

First Grab this version of nvidia inspector:

https://github.com/DeadManWalkingTO/NVidiaProfileInspectorDmW

Then on top of the program, drag your cursor to where it says:

Export user defined profiles, click on that, then on the drop list menu, click on the 4th one down saying:

Export all driver profiles(NVIDIA TEXT FORMAT).

After it has been exported, open up file with a program like notepad, then search for:

Project cars 2

After you find it, underneath where it says:

Profile "Project Cars 2"

Delete everything, then copy and paste this:

ShowOn GeForce
ProfileType Application
Executable "pcars2gld.exe"
Executable "pcars2avx.exe"
Executable "pcars2.exe"
Setting ID_0x002ecaf2 = 0x00000001 UserSpecified=true
Setting ID_0x00313536 = 0x00000008 UserSpecified=true
Setting ID_0x005a375c = 0x99941284 UserSpecified=true
Setting ID_0x0064b541 = 0x00000001 UserSpecified=true
Setting ID_0x00664339 = 0x00000001 UserSpecified=true
Setting ID_0x00667329 = 0x00000001 UserSpecified=true
Setting ID_0x00738e8f = 0xfffffffb UserSpecified=true
Setting ID_0x0078d9d9 = 0x00000000 UserSpecified=true
Setting ID_0x00854cab = 0x00000038 UserSpecified=true
Setting ID_0x0098c1ac = 0x00000000 UserSpecified=true
Setting ID_0x009cfec8 = 0x00000004 UserSpecified=true
Setting ID_0x00a06746 = 0xa0000060 UserSpecified=true
Setting ID_0x00a06946 = 0x000000f5
Setting ID_0x00a06946 = 0x8b9825f5 UserSpecified=true
Setting ID_0x00a34323 = 0x00000001 UserSpecified=true
Setting ID_0x00e32f8a = 0x00000000 UserSpecified=true
Setting ID_0x1033cec2 = 0x00000002
Setting ID_0x1033cec2 = 0x00000004 UserSpecified=true
Setting ID_0x1033ced1 = 0x00000004 UserSpecified=true
Setting ID_0x1033dcd1 = 0x00000004 UserSpecified=true
Setting ID_0x1033dcd3 = 0x00000004
Setting ID_0x105e2a1d = 0x00000004
Setting ID_0x107afc5b = 0x00000001 UserSpecified=true
Setting ID_0x107efc5b = 0x00000000 UserSpecified=true
Setting ID_0x1094f157 = 0x00000000 UserSpecified=true
Setting ID_0x1094f1f7 = 0x00000000 UserSpecified=true
Setting ID_0x10a879ac = 0x00000001 UserSpecified=true
Setting ID_0x10a879cf = 0x00000001 UserSpecified=true
Setting ID_0x10d48a85 = 0x00000000 UserSpecified=true
Setting ID_0x10d773d2 = 0x00000000 UserSpecified=true
Setting ID_0x10fc2d9c = 0x00000004 UserSpecified=true
Setting ID_0x1194f158 = 0x00000000 UserSpecified=true
Setting ID_0x2089bf6c = 0x00000023 UserSpecified=true
Setting ID_0x70092d4a = 0xb19c3535 InternalSettingFlag=V0
Setting ID_0x701eb457 = 0x2241ab21 InternalSettingFlag=V0
Setting ID_0x702442fc = 0x1c22fe24 InternalSettingFlag=V0
SettingString ID_0x7049c7ec = "웭ꑌ" InternalSettingFlag=V0
SettingString ID_0x704d456e = "쪧⃟쁽�㓵債婙簐鹩瀴딘ᤪ取�븋ᱲ뼜褰籉謠⑒贿㱇矺粐ꗽ燕胥㖵뫥뚙ᯉ쑎냬锺䭬乄솼᝵ᣢା譗壮嫠̞" InternalSettingFlag=V0
SettingString ID_0x7051e5f5 = "籭鸙" InternalSettingFlag=V0
Setting ID_0x708db8c5 = 0x5db7b8be InternalSettingFlag=V0
Setting ID_0x709a1ddf = 0x4b1cd968 InternalSettingFlag=V0
Setting ID_0x709adadb = 0x00000000 UserSpecified=true
Setting ID_0x709d3ad2 = 0x00000002
Setting ID_0x70a4995c = 0x00000000 UserSpecified=true
SettingString ID_0x70b5603f = "榛鳈⏙ꢗ" InternalSettingFlag=V0
Setting ID_0x70f8e408 = 0x80b671f5 InternalSettingFlag=V0
Setting ID_0x80857a28 = 0x00000001
EndProfile

After where it says: Profile "Project cars 2", make SURE you have proper spacing, about 4 spaces, or check the other game profiles for reference.

Next, save the document.

And lastly, open up your nvidia inspector program, then drag your mouse cursor to:

Import user defined profiles, then click on the 2nd one from the drop down menu, should say:

Import (replace) all driver profiles(NVIDIA TEXT FORMAT)

My profile should be there, All the tuning within nvidia inspector will be up there, including the Unknown ID Settings.

---------------------------------------------------------------------------------------------------------------------------

GPU USED:

4 GIGABYTE GTX 970'S WINDFORCE

GPU SPEEDS: 1443mhz(CORE)/7.5K+mhz(MEMORY)

GRAPHIC DRIVER USED: MODIFIED QUADRO DRIVERS 431.98

MOTHERBOARD USED: SUPERMICRO X10drg-q

CPU UTILITY PROGRAM USED: THROTTLE STOP

MOTHERBOARD BIOS USED: 3.1

POWER SUPPLIES USED: 2 FSP 1200W AUROM pt

SYSTEM RAM USAGE: OVER 16GB+ EASY

OS USED: WIN 10 x64 (15063.2108/VERSION: 1703/RS2/RTM)

STORAGE USED: 2 SAMSUNG 950 PRO(512GB) IN STRIPPED(RAID 0) MODE.

INTEL CHIPSET DRIVER USED: 10.1.18010.8141+10.1.18263.8193 update

TOTAL TUNING/PERFORMANCE MONITORING/TESTING TIME/OVERALL STABILITY CHECKS: CLOSE TO 20 HRS.

 

MY NEXT GAME - GTX 970 4-WAY SLI - UPDATE - 2/14/20:

Here's my next game of interest, from the original creators of XCOM:

I bring you: PHOENIX POINT(Unity Engine):

[







GAME IS RUNNING ON UNITY ENGINE, SHOULD SEE SOME GOOD RESULTS IN THE COMING FUTURE, SEE YOU ALL ON THE GAME IN THE COMING UPDATES, YOU ALL DON'T WANT TO MISS THIS ONE.

 

PHOENIX POINT - GTX 970 - 4-WAY SLI - 1ST UPDATE - 2/15/20:

Alright for first 1st update, very very very impressive results we have here but there is some very interesting issues here and yes I even brought it up to snapshot games attn.

First lets talk about overall scaling potential, how well does it scale? Absolutely Incredible but some problems here:

The fps cap issue:

I looked up some stuff on this issue and some people have this issue with the Vysnc, if you enable it, it removes the cap, it DOES NOT WORK.

Also some say it even removes the cap ONLY IF you set your preset to high or to very high, again, it does NOT WORK, Also, while messing the graphic settings in general, lets say right off the back on high preset, you got good 30 fps on tutorial mode, then you change different graphic settings, then back to where you had it originally, the fps does not stay the same as it was originally, instead, there is drastic reduction in performance, and logging off the game and re-launching the program doesn't seem to correct the issue, the only way around this is to actually restart your whole system then start over.

This also pertains to 1 gpu mode as well.

Battletech had very similar situation but different, like every time you load your saved game, fps could be 22, next it could say 18, not being consistent, Sekiro: Shadow dies twice as well.

For example here:

Lets say, on max settings, texture to medium @ 4.5k(4434x2771), on 1 gpu mode, I get:

11-17 fps.

Next, I mess with graphic settings, resolution and such, going back to where I had it originally, It will drop by a huge margin, have to do a system restart to correct the issue.

However, during this time of bug or whatever you want to call it, I check with 4 gpu's, and yes, It scales like perfect, so scaling is there, Under AFR conditions, it seems to perform quite a bit better but game needs to flicker alot to do so, you can get away with SFR but certain bits needs to be changed in order to do so, Depending on situation in the game, seems AFR has better overall potential, SFR in other hands, seems gpu 3-4 says 99%, the 1st 2, kinda lower, so currently fixing the issue under SFR mode, so right now trying to get rid of the fps cap.

Any graphic anomalies?

There is one while tuning but no need as bit digit 2 fixes the issue and works just as good a the bit digit 1, only pertains only under SFR mode.

under SLI Compatibility bits:

0x00000<<~~ this one.

NO Sli broadcast or Specific hack bits are needed at this point, all the unknown settings ID are at default, just the SLI Compatibility is needed at this point.

Main primary focus of Scaling rendering = SFR(Split-Frame Rendering), which very few people will even use nowadays, and also looking into AFR mode as well.

So in the end, yes it does Scale like mad crazy.

Always check back later as this game is now currently:

----------WORK IN PROGRESS----------

UPDATE ON DECREASED PERFORMANCE: 2/15/20:

Ok I got some relieving news here.

I mentioned above that you will see a major significant cut in performance after changing some graphic settings and or with resolution change then going back to where you originally had it, Found a quick fix instead of rebooting your system.

Sign off windows then log back in, re-launch the program, don't change anything, load up the game, everything should run at its peakest potential, thanks again.

UPDATE ON DECREASED PEFORMANCE #2: 2/15/20:

Further investigation into this matter:

Its a coin toss, signing off = it may or may not fix the issue, restarting will defin't 100% will fix it, can confirm it.

Very strange game indeed, as a remainder, this applies to 1 gpu mode as well so keep this in mind at all times, be absolutely weary of changing your in-game graphic settings while in the game, you might want to change it, tweak it a bit, then restart your system and then re-launch the program, this will ensure in the end the game will run at its peakest potential at all times.

So, signing off and then back on, may or may not correct the issue, restarting will, thanks.

 

PHOENIX POINT - GTX 970 - 4-WAY SLI - STAGGERING UPDATE - 2/16/20:

Got some very damning updates in regards to this game.


I just went back to Nvidia Inspector, used the same bits for SFR mode, although scaling is good, GPU 1 and 2 had kinda low scaling, while 3 and 4 had Excellent scaling usage.

So, just for kicks, I tried turning on optimize for compute performance and Holy living hell, Wow, Perfect F'in Scaling.

As a remainder to all and absolute fair warning for those who turns on the setting called:

Optimize for compute performance...

I had also tested this particular setting in the past on my 2 1080ti's.

There is, far's I remember, only 2 game that works with this setting, and I have personally tested MANY games with this particular setting, all others games, you will experience decrease performance and or rise of overall scaling usage(WITH NO EXTRA PERFORMANCE BENEFIT), and those 2 games are:

The Division(1st) one, tested with 2 1080ti's, and yes, good amount of boost in performance,

and the second one, and it is REQUIRED for 4-way sli to work:

LORDS OF FALLEN.

But back to phoenix point, Although, the MS(Frametime) is rather high, in the low 20's, thus, some jerkin', Jittering can be seen while moving your camera around, in other words, the transition in-between frames while moving your camera, won't be as smooth, despite showing 60 fps.

I had the preset under very high but it was actually custom, I put everything to maximum except texture(medium), and at 4.5k(4434x2771).

So going to toy around to bring the MS(Frametime) down a bit so camera movement and its transition in-between each frame will be much smoother, usually high MS(Frametime) is an indication that the gpu is putting an overwhelming amount of stress that its just not fast enough, thus, very high MS(Frametime), just heads up, 1 gpu, the MS(Frametime) is EXTREMELY HIGH, lol, over 80ms.

Just to let everyone know, the driver of choice is non-other then:

Modified Quadro Drivers: 431.98.

Absolute Crazy scaling here folks.

I'll continue to work on the Frametime issue and its settings.

That being said, I only been mainly focusing in the tutorial mode area, as there was no point of passing that if scaling wasn't working properly, so now its fully functioning and far's I can tell, no known graphic anomalies, I will see how well game scales in general afterwards, if any real significant changes occur, I will announce it right here ONLY on guru3d, that's right, only on guru3d.com

So be sure to further check updates as this game is still currently:

----------WORK IN PROGRESS----------\

QUICK UPDATE DURING FIRST PERSON MODE AND PRESETS.

Just right now, I never really went into first person mode while playing, during this time, some stuff scales great, some stuff no, so currently working on these issues as we speak...

in regards about preset, it seems if you put preset let's say on medium but textures to high, next time you launch the program, the preset will go according to your texture settings, in this case, high, so those who tuning or playing this game in general, be very strongly aware of this situation.

UPDATE ON OPTIMIZE FOR COMPUTE PERFORMANCE - 2/18/20:

Ok, almost 8 hrs in tuning, big difference on tuning now, it seems from much closer investigation while using this particular setting:

Optimize for compute performance, is really not needed, it seems there is no difference, however, in certain locations/spots during normal view mode, when the scaling is very good, I notice like a 10 fps drop with this setting turned on, so like before, this setting is really not needed, it may change in the future, but for now, I have it off, it did made a huge difference in the beginning, but it seems its not needed now.

Thanks for reading.

 

OVERALL SYSTEM UPDATE ON NEW BUILD - 2/17/20:

So its been a month or so ever since my last Asus board went out and took my psu with it, due to a short on the board.

Just wanted to give heads up on how the overall system stability is ever since I changed to the Supermicro X10drg-q motherboard,ram and the same cpu I had on the last asus board, As Expected, Everything works 100% perfect, with no signs of crashing,bsod,etc, Absolutely rock solid stable thus far.

Ever since I first booted the system, Not a one single minute problem.

I have tried prime95 test for a bit, not one problem, Ran Excellent.

I have not ran memtest, I will though eventually but everything seems perfectly stable, far now that is.

Far's I can tell, Excellent Motherboard.

Thanks for reading.

 

PHOENIX POINT - GTX 970 - 4-WAY SLI - UPDATE - 2/17/20:

Found some very useful stuff for those who's tuning this game or just doing one gpu in general.

This has only been confirmed on a Nvidia GTX 970 family, I have NOT tried it with any other gpu's, so I can not confirm if it will effect on the performance.

This what I have Learned so far:

I was able to re-create the degraded performance situation as stated before, this time I can confirm 100%, that this is the cause:

Lets say for starters, base resolution is set @ your native resolution, for this example: 1600p(2560x1600), then all the sudded you beef up to 4k, then want to revert back to your original native resolution, guess what happens next? You all guess it, degraded Performance by a large margin.

Before = 52 fps

After changing resolution then reverting back to original = 30 fps

In order to correct this you MUST put your resolution to a lower resolution, like 1200p(1920x1200), then afterwards, switch it back to the resolution you originally wanted to play on, comes back to normal, Already confirmed it several times over.

Now, that being said, I have NOT tried it with under 4-Way Sli mode, I know the degraded performance does effect Sli mode in general, but I'll re-check and confirm to see if this method will work or not, it will save LOTS and LOTS of downtime, especially what I am doing, massive amounts of tuning, restarting your system all the time = MAJOR league downtime.

The developers of the game from Snapshot games, has already replied to me back and currently investigating the issue as we speak and hopes to implement something in the next patch.

Again, If I can confirm this corrects the issue under 4-Way Sli mode, I'll just put a update headline underneath this post, thanks again.

This game is still currently:

-----WORK IN PROGRESS-----

UPDATE:

Yes, I can confirm now 100%, under 4-Way Sli mode, switching to low a lower resolution and back to original, will correct the issue, just like 1 gpu mode.

Quick update of gpu bus usage

At times, the gpu bus usage acts normal and other times it can go haywire, despite of unchanged in performance, currently investigating the issue as we speak.

 

PHOENIX POINT - GTX 970 - 4-WAY SLI - UPDATE ON AFR AND SFR MODES - 5/17/20:

Found some stuff that some might find interested while tuning this game.

Under the same SLI Compatibility DX1x Bits I am currently using, I can switch back to AFR and SFR modes with ease and whats very interesting is this:

It seems that If I disable shadows and its distance all together, it will improve overall scaling potential under AFR mode whereas if you use Shadows and its distance, it doesn't like it, it reduces overall scaling performance by quite a bit, and if you use SFR??? its the exact opposite, one like shadows and the other does not.

Far as overall fps performance is concerned, it seems AFR provides best overall fps in general but far from being done of course.

Game is still:

------WORK IN PROGRESS-----

UPDATE ON SFR PERFORMANCE WITHOUT SHADOWS:

LOL, forgot to turn off shadows and re-observe performance, I can confirm now that its on the same level of performance as AFR.

Totally forgot to shut off shadows completely to see performance differences, yup, pretty much the same.

 

PHOENIX POINT - GTX 970 - 4-WAY SLI - UPDATE ON VSYNC AND FPS CAP - 2/18/20:

Finally got the fps to be uncapped:

First of all, within Rivatuner Statistics server program, under setup, make SURE you have frame rate limiter to off.

Next, within nvidia inspector program, you MUST put the Vsync to: FORCE OFF, and not on use application settings, it will NOT work period.

3rd: while on your game, make sure the Vsync toggle box is switch to on, again, I know it doesn't make any sense here, but already confirmed 100%, that this check box MUST be toggled on, otherwise it will not work, others had the same issue where the Vsync toggle in the actual game, must be turned on.

Also, at times when logging in the game, your fps might be capped to 60, even with the above method, to fix this issue, all you got to do is toggle the Vsync off and back to on, fps is now uncapped.

Several other things I found that will make the FPS cap is the following nvidia inspector settings:

Under 5 - Common:

Multi-Display/Mixed Gpu Acceleration: MUST BE SET to single display mode/Compatibility mode, do NOT set it to 0x00000006(Quadro Gpu's) for this game, it will make the fps capped, Laser locked.

And lastly, while tuning this game, another thing that will make this game capped(FPS) is:

Under 8 - extra:

ASYNC10_NVAPI_MODE: 0x00000002 ALLOW_NOBLOCKING_SLISTATE - Allow specific non-blocking optimizations for GetCurrentSliState NVAPI.

So make sure you do NOT use these settings.

Far as overall game performance after the change, It made some, but not to much, I was expecting much difference, not really, so more tuning will be required on my side, absolutely no exceptions.

Also another thing I mentioned to snapshot game is this:

After very close observation while under battle mode, when your view shifts to first person kinda view, I notice that at times your gpu usage, it will have a reduction in performance, for instance, @ 1600p resolution, it may say 42-45 fps tops, then back to normal view then back to the battle view, it will say like 35fps, it may take like up to 5 times to correct the issue while viewing the first person mode, the gpu usage % bounces up and down.

This pertains to Both single gpu and Multi-gpu(Sli) Modes, so be absolutely aware of this situation.

Again, they are aware of this issue as we speak...

If anyone has this issue on a video card that I am not using currently, please let me know what your experiences are on this, thanks.

Thanks for reading and once again this game is still currently:

--WORK IN PROGRESS--

OTHER RECOMMENDATIONS ON FPS CAP / UPDATE: 2/18/20:

Those who are really tuning this game or playing the game in general and want the best possible performance right off the back, do the following:

Start at a PISS POOR resolution, such as: 1600x900, while your Vsync is toggle is unchecked, to make life easier, create a shortcut for the actual config document file to your desktop for much easier access, just in case you can save it and no need to dig for it in the future, but back to fps cap, yes, make sure your resolution is set to like 1600x900 and that Vsync off, then the moment your on the game(Title screen), go to option, then set it to your desired resolution and the Vsync checked on then load up your game, game should be now at its peakest potential for the most part and the fps uncapped, thanks for reading.

 

PHOENIX POINT - GTX 970 - 4-WAY SLI - DUAL XEON's(2 CPU's) AND RAM UTILIZATION UPDATE - 2/18/19:

Very interesting game we have here.

Also to let everyone know, this is by far, the most CPU Utilization I have ever seen in any game I have tuned thus far, Even at GPU bound @ 4k or slightly higher, the CPU usage is the highest I have ever seen at this resolution.

I checked with Afterburner osd and looked at my Task Manager performance cpu window, both reads roughly the same.

During normal view, it peaks out over 50% utilization, while under battle mode view, I have seen high as 63%(peak).

As a Remainder, Using Dual Xeon, 12 core(24 thread), So together:

24 cores/48 threads.

So in reality, Its using over 12 cores/24 threads already.

This one very heavy duty game for Cpu or Cpu's for that matter, remember, this is at GPU bound resolution such as 2400p or higher, that is insane, but not surprised either as this is a turn based tactic game.

Far as Main System Physical ram is concerned, while tuning:

I have seen much as 22-23GB+ utilization.

So make sure you have plenty of cores to spare for this game, it seems to love it, and very important, if possible, make sure your ram is at a bare minimum of 16gb to be safe, if slightly higher then 4k, you might want to beef it up to 20gb+, personally I would take it high as 24gb as I have seen getting near that ballpark, so to be utmost safe, 24-32gb should be good. below 16gb is NOT RECOMMENDED if similar setup as mine is used, thanks.

 

PHOENIX POINT - GTX 970 - 4-WAY SLI - UPDATE ON CURRENT TUNING PROGRESS - 2/19/20:

Just up on current stats on this game thus far:

In some areas, there seems to be some very serious frame skipping issues(Only on certain locations and stuff being viewed), Tons of Jerkin' movements while moving the camera around in the normal view mode, whereas, other areas, no.

Even when only 1 gpu is being used, very high MS(Frametime), the jerkin' movement while moving your camera around, is tons better then doing sli mode in general.

I also tried in 2-way sli mode to see if the same reaction is there in these certain locations, and yes its still there but not as bad but yes still there.

In these tough tuning locations, while viewing into first person mode, depending on what you are viewing, it seems the overall scaling goes absolute haywire, this also implies to AFR mode and 2 and 3-way sli mode, already double checked.

I am about roughly over 12 hrs+ into this game so far, very very hard game to get everything going.

how is the overall scaling, some minute spots seems to scale alright, but for the most part from what I can tell thus far, up to 3-way at best, even then...

I'll do my best to see what I can do but its looking pretty tough.

Again, there is lots of problems in this game.

Sli Compatibility bits is pretty much all done.

8 - Extra is all done, I did went through some Unknown id settings, seems it made 0 difference...

Not an easy game folks we have here, not at all, probably just as hard or harder then Battletech.

I am now going to be tuning both the Sli broadcast and specific hack bits, so always continue to check back later for latest updates.

Also, I personally tried:

AFR_OF_SFR__FALLBACK_3AFR, Which I believe is AFR+SFR combo, and it seems to work just as good as SFR itself, Although I really don't use it but it works just as good.

Check back later as this game is still currently:

--WORK IN PROGRESS--

 

PHOENIX POINT - GTX 970 - 4-WAY SLI - UPDATE - 2/21/20:

Haven't been on the game for almost couple days, I just went back on it and this time I tried another trick that 100% works in Hellblade and to my knowledge, only game that works...

Geforce Experience/Shadow Play recording method.

If you all look back on my hellblade tuning, you will see that the game really benefit in the increase performance while using shadow play but you have to put the exact recording playback as while you are in windows.

You must hit record button prior to getting on the game and then switch to your desired resolution, if you don't, it won't work, in that case, hellblade

I tried it just out of curiosity sake, and kinda expected the results in the end, NO DIFFERENCE...

It was certainly worth the try...

So again as a remainder, out of all the games I have tuned, Shadow play will boost performance big time in hellblade and only with that game, all others will NOT WORK.

Check back later for more updates, game is still: ---WORK IN PROGRESS---

ANOTHER HEADS UP:

On the continuing efforts of tuning this game, I found one particular setting that is very useful for those who are currently tuning this game as I am.

Open up your nvidia inspector program and head down to:

8 - Extra:

Look for: ColorCompressEnable

And switch it to: 0x25558997 On

I really never use this setting at all, only a few times in the past, gave it a try, and yes, gave me about 5 fps boost in performance @ 2400p, But that was just done in one location...

This game is extremely hard folks, your not kiddin', There is some places that likes Screen space Reflections, other areas, if you have it on, guess what, it scales down to 1 gpu performance, it doesn't like it, again, all part of the tuning process, Very very very common...

But here's another down fall, lets say your in an area already that doesn't like Screen Space Reflections, and you checked it on and then you switched it back off, guess what, fps stays the same, you must restart your game at a much lower resolution as before with vsync turned off then when you come back to your game, switch it back to your desired resolution along with vsync on(Developers are very aware of the issue) for the fps to be uncapped, now, performance should be at its peakest.

Remember, try to reduce your resolution extremely low before logging off the game, like 1600x900 along with Vsync off, then when you get back on the game, switch everything then and load up your game.

You all might want to make the config. file shortcut to your desktop for easy access, just in case you logged off the game, NOT in the lower resolution, you can change it there, save the document then re-launch the game.

There might be a chance where you log off the game @ 4k then re-launch the program once more, you won't lose fps, that's IF you didn't change any settings, but either way, changing settings in the game or not, Please before logging off the game, lower your resolution drastically along with Vsync off, then back on the game switch everything.

Overall scaling thus far: Total Mixed Bag, Some stuff scales 4-way EXCELLENT, some no, some 3-way. Lots and Lots of performance checking, Tuning, Etc, JESUS!!!(Shakes Head).

In about 14+ hrs in, Far and I mean FAAAAAR from being finished.

Thanks for reading.

 

THE TECHNOMANCER - GTX 970 - 4-WAY SLI - 1ST UPDATE - 2/22/20:

QUICK REMINDER: MORE THEN 8 CORES = WILL NOT LAUNCH, MAKE SURE TO REDUCE THE AMOUNT OF CORES BEFORE LAUNCHING THE PROGRAM.

I did mention this game in the past, And I have said only a very quick test to see if it scales and yes it does, this time, I did more further test, as a remainder, My full attn./focus is on Phoenix Point, just want give this game a spin, jotted down some quick tests.

TAA must be turned off under Sli mode as it will leave a Deja Vu like blur trail while moving your camera, Best to keep it on FXAA.

Everything set to maximum @ 2400p(4k/16:10 aspect Ratio)/8 core(Had to/would not launch otherwise).

Rendering mode of choice:

AFR of SFR(Hybrid/Combo)

For those who need more in depth on this, please go to this link for better explanation:

https://docs.nvidia.com/gameworks/content/technologies/desktop/sli.htm

4 quick tests here:

On performance test #2, I purposely waited just a slight during the cutscene during 4-way as there was a slight burp on the 1% and .1% low's, wanted to keep it high and stable:

Also as a remainder, on Movement Pattern Test #2, I had some problems with the Max Fps, some of these movements was near some stuff where there was kinda no load on it, so my scaling kinda went down a bit, but like I stated way before, I have NOT did any real hardcore tuning at all, all I did was make some very quick adjustments within the Sli Compatibility bits and that's it, no other was changed, it will in a future time, but these are just rough down.

Movement pattern 1:

1 GPU:

MIN FPS: 15//AVG FPS: 16//MAX FPS: 17//1% LOW: 15//.1% LOW: 14 FPS

4 GPU's:

MIN FPS: 58//AVG FPS: 62//MAX FPS: 66//1% LOW: 52//.1% LOW: 45 FPS

Cutscene #1:

1 GPU:

MIN FPS: 8//AVG FPS: 14//MAX FPS: 20//1% LOW: 8 FPS//.1% LOW: 5 FPS

4 GPU's:

MIN FPS: 30//AVG FPS: 56//MAX FPS: 72//1% LOW 24: FPS//.1% LOW: 12 FPS

Cutscene #2:

1 GPU:

MIN FPS: 11//AVG FPS: 15//MAX FPS: 20//1% LOW: 10 FPS//.1% LOW: 9 FPS

4 GPU's:

MIN FPS: 42//AVG FPS: 58//MAX FPS: 69//1% LOW: 37 FPS//.1% LOW: 30 FPS

Movement Pattern #2:

1 GPU:

MIN FPS: 14//AVG FPS 18//MAX FPS: 31//1% LOW: 14 FPS//.1% LOW: 13 FPS

4 GPU's:

MIN FPS: 55//AVG FPS: 69//MAX FPS: 88//1$ LOW: 50 FPS//.1% LOW: 35 FPS

As you all can see, very impressive results we have here, especially when only Slight Changes was down to the Sli Compatibility bits, no other bits was changed...

I will improve the scaling on this game later.

I will be working on this after Phoenix Point.

Thanks for reading.