Overclock.net › Forums › Industry News › Video Game News › [WCCF]Witcher 3 Benchmarked, GTX Titan Does 30FPS At 1080P And Ultra Settings
New Posts  All Forums:Forum Nav:

[WCCF]Witcher 3 Benchmarked, GTX Titan Does 30FPS At 1080P And Ultra Settings - Page 26

post #251 of 379
Damn I guess this 7970 still kicks butt then biggrin.gif
post #252 of 379
Quote:
Originally Posted by moustang View Post

Strange...

I'm running SLI 4GB GTX 770s.

For the most part I'm running Ultra settings, including in-game AA and HBAO+, running the game in 1080p resolution.
I have disabled Motion Blur, Vignette, Chromatic Aberration, and Hairworks.
Every other graphics setting is maxed out. I have not made any changes to the config file.
I am running Nvidia 352.86 drives.

My average frame rate is right around 60fps. In very graphically intense areas and situations I get drops into the high 40s, but probably 70% of the time I'm at 60fps or higher.

I get some slight stuttering during cutscenes, but none during gameplay.

Hairworks lowers my frame rates by 10fps or so. Maybe a bit more in scenes with several wolves or other hairy monsters. All other graphics effects which I have disabled combined only cost me about 5fps. I've turned these off because I prefer the graphics with them off, not for performance reasons.


To those complaining of stuttering during gameplay, have you selected the Hardware Cursor option? This disables mouse acceleration which may be contributing to your stuttering. Also I've found that if I cap the frame rate at 60FPS the majority of the time the game will run in the 56-59fps range rather than staying at a steady 60FPS, even though removing the game and disabling vsynch in the same area will let the game run in the 70fps range. Also, how many of you have Hyperthreading DISABLED? This game makes really good use of Hyperthreading if you have it enabled.



EDIT>>>

Also, I think some of you Kepler owners are not giving Maxwell credit where credit is due. Maxwell does several things IN HARDWARE that Kepler lacks the hardware to do. Tessellation is around 30% faster on Maxwell because of hardware. Maxwell also supports Raster Ordered Views and Conservative Raster in hardware, which Kepler does not. Kepler can do these things at the driver level (software) but it's much more costly in performance. No amount of optimization for Kepler is going to make Kepler equal to Maxwell when these features are used, and The Witcher 3 does use them. It's not a matter of driver optimization, it's a matter of trying to do things in software on Kepler that are done in hardware on Maxwell, or in the case of Hairworks Tessellation, simply not being able to do anything about the 30% advantage Maxwell hardware gives over Kepler hardware. Driver optimization will only get you so far, at some point you'll have to accept the hardware difference is just too much to overcome.

And on that note, I fully expect Maxwell to have the same problem when compared to Pascal when it's released. Nvidia could make massive efforts to optimize drivers for Maxwell, but it's never going to overcome the hardware advantages that Pascal is going to introduce.

Okay what about amd? Beating a 780 with a 7970 or even gtx 960 beating nVidia top of the line predecessor
Turcoman Rig
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7 6800K Rampage V ED10 EVGA FE GTX 1080 Corsair 3000 c15 32gb 
Hard DriveHard DriveHard DriveOptical Drive
Western Digital Blue Seagate Barracuda 7200.14 Samsung 830 128 Sony Blu Ray 
CoolingOSMonitorKeyboard
Corsair h110 8+7+XP BenQ Xl2420T razer blackwidow ultimate stealth 2013 
PowerCaseMouseMouse Pad
Corsair AX1200W Corsair 900D Razer Mamba elite 2012 steelseries QcK+ heavy 
AudioOther
2.1 Speakers Nvidia 3D Vision 2 Kit 
  hide details  
Reply
Turcoman Rig
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7 6800K Rampage V ED10 EVGA FE GTX 1080 Corsair 3000 c15 32gb 
Hard DriveHard DriveHard DriveOptical Drive
Western Digital Blue Seagate Barracuda 7200.14 Samsung 830 128 Sony Blu Ray 
CoolingOSMonitorKeyboard
Corsair h110 8+7+XP BenQ Xl2420T razer blackwidow ultimate stealth 2013 
PowerCaseMouseMouse Pad
Corsair AX1200W Corsair 900D Razer Mamba elite 2012 steelseries QcK+ heavy 
AudioOther
2.1 Speakers Nvidia 3D Vision 2 Kit 
  hide details  
Reply
post #253 of 379
Quote:
Originally Posted by MerkageTurk View Post

Okay what about amd? Beating a 780 with a 7970 or even gtx 960 beating nVidia top of the line predecessor

What about a card with a bunch of Nvidia specific features disabled and a bunch of other features lowered, beating a card with the same features enabled and all settings maxed out?

Apples to oranges comparison my friend.


As far as a 960 beating a "top of the line predecessor", what exactly are you talking about?

http://www.overclock.net/t/1556367/wccf-witcher-3-benchmarked-gtx-titan-does-30fps-at-1080p-and-ultra-settings/120#post_23930774

Sure looks to me like the 780 easily beats the 960 in every resolution. And that's not even the 780ti. In fact the 960 falls exactly in the middle between the 760 and the 780. Right about where a single 770 would be.


My SLI 770s are just about equal with a 980 on the same settings, give or take a couple of frames per second. That's pretty much what I would expect on a game that uses DX11.2 features. In games that are pre-DX11.1 I'm usually 10% or more faster than the 980. I think most Kepler owners are overlooking the DX11.1 and DX11.2 hardware features of Maxwell and simply thinking that since their cards were closer in performance last year with standard DX11 games they should maintain the same performance gap now.
post #254 of 379
Project cars? Does that ring a bell?
Turcoman Rig
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7 6800K Rampage V ED10 EVGA FE GTX 1080 Corsair 3000 c15 32gb 
Hard DriveHard DriveHard DriveOptical Drive
Western Digital Blue Seagate Barracuda 7200.14 Samsung 830 128 Sony Blu Ray 
CoolingOSMonitorKeyboard
Corsair h110 8+7+XP BenQ Xl2420T razer blackwidow ultimate stealth 2013 
PowerCaseMouseMouse Pad
Corsair AX1200W Corsair 900D Razer Mamba elite 2012 steelseries QcK+ heavy 
AudioOther
2.1 Speakers Nvidia 3D Vision 2 Kit 
  hide details  
Reply
Turcoman Rig
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7 6800K Rampage V ED10 EVGA FE GTX 1080 Corsair 3000 c15 32gb 
Hard DriveHard DriveHard DriveOptical Drive
Western Digital Blue Seagate Barracuda 7200.14 Samsung 830 128 Sony Blu Ray 
CoolingOSMonitorKeyboard
Corsair h110 8+7+XP BenQ Xl2420T razer blackwidow ultimate stealth 2013 
PowerCaseMouseMouse Pad
Corsair AX1200W Corsair 900D Razer Mamba elite 2012 steelseries QcK+ heavy 
AudioOther
2.1 Speakers Nvidia 3D Vision 2 Kit 
  hide details  
Reply
post #255 of 379
Thread Starter 
Quote:
Originally Posted by moustang View Post

What about a card with a bunch of Nvidia specific features disabled and a bunch of other features lowered, beating a card with the same features enabled and all settings maxed out?

Apples to oranges comparison my friend.


As far as a 960 beating a "top of the line predecessor", what exactly are you talking about?

http://www.overclock.net/t/1556367/wccf-witcher-3-benchmarked-gtx-titan-does-30fps-at-1080p-and-ultra-settings/120#post_23930774

Sure looks to me like the 780 easily beats the 960 in every resolution. And that's not even the 780ti. In fact the 960 falls exactly in the middle between the 760 and the 780. Right about where a single 770 would be.


My SLI 770s are just about equal with a 980 on the same settings, give or take a couple of frames per second. That's pretty much what I would expect on a game that uses DX11.2 features. In games that are pre-DX11.1 I'm usually 10% or more faster than the 980. I think most Kepler owners are overlooking the DX11.1 and DX11.2 hardware features of Maxwell and simply thinking that since their cards were closer in performance last year with standard DX11 games they should maintain the same performance gap now.

So, it´s better to go AMD, right? Considering their old cards can still compete with Maxwell cards? Remember those days when a 290X = GTX TITAN? Now they are 50% faster in newer games! Is that normal? Really? rolleyes.gif

Care to see how your 770 performs against a 280x? biggrin.gif
post #256 of 379
Quote:
Originally Posted by Devnant View Post

So, it´s better to go AMD, right? Considering their old cards can still compete with Maxwell cards? Remember those days when a 290X = GTX TITAN? Now they are 50% faster in newer games! Is that normal? Really? rolleyes.gif

Titan = DX11.1 hardware
290 = DX11.2 hardware

Hidden Surface Removal and Tile Rendering in hardware makes a difference.

Quote:
Care to see how your 770 performs against a 280x? biggrin.gif

I already know how it performs. Obviously much better than you do. I know what features each card supports, and I know what features are found in the API that is used by this game, again obviously better than you do.

Like I said, I'm about equal with a single 980 in The Witcher 3. That's right about what I would expect from SLI 770s in this game. In Crysis 3 maxed out I would be an easy 10+ FPS faster than the 980, again, matching my expectations.

See....

770 SLI = 62fps
http://www.guru3d.com/articles-pages/geforce-gtx-770-sli-review,20.html

Single 980 = 48fps
http://www.guru3d.com/articles-pages/geforce-gtx-980-sli-review,19.html



You're the one with unrealistic expectations because you don't understand the API nor the hardware features that these cards use.

As you said, in the old days 290X = Titan... Well, now what do you think happens if you then support the hardware hidden surface removal and tile rendering on the 290X but you do not support those features on the Titan? Still think they would be =? That's precisely what The Witcher 3 does. It supports DX11.2, and both Maxwell and the R9 series cards have specific hardware features that are supported by DX11.2 that Kepler lacks.



Want to see something funny? Go benchmark all of these cards with the exact same settings under Windows 7 and see how radically the results change. Watch in amazement as Kepler suddenly closes the gap as you strip the DX11.2 feature support from both Maxwell and the R9s. Win7 only supports DX11.1.
Edited by moustang - 5/22/15 at 10:46am
post #257 of 379
Thread Starter 
Quote:
Originally Posted by moustang View Post

Titan = DX11.1 hardware
290 = DX11.2 hardware

Hidden Surface Removal and Tile Rendering in hardware makes a difference.
I already know how it performs. Obviously much better than you do. I know what features each card supports, and I know what features are found in the API that is used by this game, again obviously better than you do.

Like I said, I'm about equal with a single 980 in The Witcher 3. That's right about what I would expect from SLI 770s in this game. In Crysis 3 maxed out I would be an easy 10+ FPS faster than the 980, again, matching my expectations.

You're the one with unrealistic expectations because you don't understand the API nor the hardware features that these cards use.

As you said, in the old days 290X = Titan... Well, now what do you think happens if you then support the hardware hidden surface removal and tile rendering on the 290X but you do not support those features on the Titan? Still think they would be =? That's precisely what The Witcher 3 does. It supports DX11.2, and both Maxwell and the R9 series cards have specific hardware features that are supported by DX11.2 that Kepler lacks.



Want to see something funny? Go benchmark all of these cards with the exact same settings under Windows 7 and see how radically the results change. Watch in amazement as Kepler suddenly closes the gap as you strip the DX11.2 feature support from both Maxwell and the R9s. Win7 only supports DX11.1.

Okay, I´m humbled. Thanks for the technical explanation, though. thumb.gif

Seems like the real problem was that I purchased a really overpriced GPU which lacked critical features.
post #258 of 379
Yes which is another reason not to buy nvidia as nvidia knows people will still keep buying their gpus even if outdated.

Nvidia also does not support dx12 full. Which is another another nail in the coffin ow and they also stopped supporting overclocks on laptops, what's next gpus in computers?
Turcoman Rig
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7 6800K Rampage V ED10 EVGA FE GTX 1080 Corsair 3000 c15 32gb 
Hard DriveHard DriveHard DriveOptical Drive
Western Digital Blue Seagate Barracuda 7200.14 Samsung 830 128 Sony Blu Ray 
CoolingOSMonitorKeyboard
Corsair h110 8+7+XP BenQ Xl2420T razer blackwidow ultimate stealth 2013 
PowerCaseMouseMouse Pad
Corsair AX1200W Corsair 900D Razer Mamba elite 2012 steelseries QcK+ heavy 
AudioOther
2.1 Speakers Nvidia 3D Vision 2 Kit 
  hide details  
Reply
Turcoman Rig
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7 6800K Rampage V ED10 EVGA FE GTX 1080 Corsair 3000 c15 32gb 
Hard DriveHard DriveHard DriveOptical Drive
Western Digital Blue Seagate Barracuda 7200.14 Samsung 830 128 Sony Blu Ray 
CoolingOSMonitorKeyboard
Corsair h110 8+7+XP BenQ Xl2420T razer blackwidow ultimate stealth 2013 
PowerCaseMouseMouse Pad
Corsair AX1200W Corsair 900D Razer Mamba elite 2012 steelseries QcK+ heavy 
AudioOther
2.1 Speakers Nvidia 3D Vision 2 Kit 
  hide details  
Reply
post #259 of 379
Quote:
Originally Posted by Devnant View Post

Okay, I´m humbled. Thanks for the technical explanation, though. thumb.gif

Seems like the real problem was that I purchased a really overpriced GPU which lacked critical features.

To add a gorgeous thing about GCN it's completely programmable so it should support whatever DX variant is to ever come so long as the hardware needed to utilize the features are there physically. A lot of GPUs in the past had to have constantly added software side features to the device. GCN however is completely programmable. So long as no additional new hardware features are required for the next after DX12 GCN will remain compatible.

This is great because it makes hardware compete against itself, not the software.
Power Tower
(22 items)
 
SteamBox
(9 items)
 
Doge Miner
(7 items)
 
CPUMotherboardGraphicsRAM
Ryzen 1700X AX370-Gaming 5 AMD Radeon R9 200 Series G.Skill DDR4-2400 
RAMRAMRAMHard Drive
G.Skill DDR4-2400 G.Skill DDR4-2400 G.Skill DDR4-2400 Samsung 840 Pro 
Hard DriveHard DriveHard DriveHard Drive
CX300 Crucial 480GB Toshiba 4TB Toshbia 4TB Western Digital Black 1TB 
CoolingOSMonitorMonitor
h110i Windows 10 42" LG TV 20" Digitizer ASUS 
KeyboardPowerCaseMouse
Corsair Vengeance Mechanical Keyboard  850watt Vampire Gold Rated NZXT S340 Elite Corsair RGB FPS Mouse 
Mouse PadAudio
Borderlands Mousepad Realtek HD 
  hide details  
Reply
Power Tower
(22 items)
 
SteamBox
(9 items)
 
Doge Miner
(7 items)
 
CPUMotherboardGraphicsRAM
Ryzen 1700X AX370-Gaming 5 AMD Radeon R9 200 Series G.Skill DDR4-2400 
RAMRAMRAMHard Drive
G.Skill DDR4-2400 G.Skill DDR4-2400 G.Skill DDR4-2400 Samsung 840 Pro 
Hard DriveHard DriveHard DriveHard Drive
CX300 Crucial 480GB Toshiba 4TB Toshbia 4TB Western Digital Black 1TB 
CoolingOSMonitorMonitor
h110i Windows 10 42" LG TV 20" Digitizer ASUS 
KeyboardPowerCaseMouse
Corsair Vengeance Mechanical Keyboard  850watt Vampire Gold Rated NZXT S340 Elite Corsair RGB FPS Mouse 
Mouse PadAudio
Borderlands Mousepad Realtek HD 
  hide details  
Reply
post #260 of 379
Thread Starter 
Quote:
Originally Posted by moustang View Post

What about a card with a bunch of Nvidia specific features disabled and a bunch of other features lowered, beating a card with the same features enabled and all settings maxed out?

Apples to oranges comparison my friend.


As far as a 960 beating a "top of the line predecessor", what exactly are you talking about?

http://www.overclock.net/t/1556367/wccf-witcher-3-benchmarked-gtx-titan-does-30fps-at-1080p-and-ultra-settings/120#post_23930774

Sure looks to me like the 780 easily beats the 960 in every resolution. And that's not even the 780ti. In fact the 960 falls exactly in the middle between the 760 and the 780. Right about where a single 770 would be.


My SLI 770s are just about equal with a 980 on the same settings, give or take a couple of frames per second. That's pretty much what I would expect on a game that uses DX11.2 features. In games that are pre-DX11.1 I'm usually 10% or more faster than the 980. I think most Kepler owners are overlooking the DX11.1 and DX11.2 hardware features of Maxwell and simply thinking that since their cards were closer in performance last year with standard DX11 games they should maintain the same performance gap now.

Okay, I've seen some Windows 7 benchmarks, and Kepler still has unsatisfactory performance, far behind Maxwell (though not that far from AMD this time):
http://www.techpowerup.com/mobile/reviews/Performance_Analysis/The_Witcher_3/3.html

Do you have a source that confirms TW3 uses Dx11.2? I couldn't find that info anywhere. Thanks!
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Video Game News
Overclock.net › Forums › Industry News › Video Game News › [WCCF]Witcher 3 Benchmarked, GTX Titan Does 30FPS At 1080P And Ultra Settings