Overclock.net › Forums › Graphics Cards › NVIDIA › NVIDIA Drivers and Overclocking Software › Driver battle: 331.82 WHQL vs. 337.50 BETA (12 benchmarks)
New Posts  All Forums:Forum Nav:

Driver battle: 331.82 WHQL vs. 337.50 BETA (12 benchmarks)

post #1 of 13
Thread Starter 
Since I had a rather vast collection of benchmark results with my system using the 331.82 driver, I decided to have a look at the overall performance gain of the new 337.50 driver.

This isn't an exact apples to apples since the 331.82 driver was tested on a fresh Win 8.1 install whereas 337.50 was tested on my daily 8.1 install. Still, I don't think there is much difference.

Test setup:

MSI Geforce GTX 780 (reference model) @1176MHz/1725MHz
Asus P9X79
i7-4820k @4.6GHz
16GB DDR3 2133MHz
Rosewill Armor Evo (3x140mm & 1x230mm fans)
1920x1080 resolution

Win 8.1 Pro
GeForce 331.82 WHQL & Afterburner 3.0.0 beta 17
GeForce 337.50 BETA & Precision X v4.2.1

Heaven Benchmark
331.82: 61.1 FPS Warning: Spoiler! (Click to show)
337.50: 60.9 FPS (-0.3%) Warning: Spoiler! (Click to show)

Sleeping Dogs
331.82: 72.8 FPS Warning: Spoiler! (Click to show)
337.50: 74.7 FPS (+2.6%) Warning: Spoiler! (Click to show)

Sniper Elite v2
331.82: 85.3 FPS Warning: Spoiler! (Click to show)
337.50: 86.2 FPS (+1.1%) Warning: Spoiler! (Click to show)

Metro 2033
331.82: 53.24 FPS Warning: Spoiler! (Click to show)
337.50: 52.60 FPS (-1.2%) Warning: Spoiler! (Click to show)

Hitman Absolution
331.82: 53.95 FPS Warning: Spoiler! (Click to show)
337.50: 72.69 FPS (+34.7%) Warning: Spoiler! (Click to show)

Aliens vs Predator
331.82: 101.0 FPS Warning: Spoiler! (Click to show)
337.50: 102.1 FPS (+1.1%) Warning: Spoiler! (Click to show)

STALKER: Call of Pripyat Sun Shafts benchmark
331.82: 95.0 FPS Warning: Spoiler! (Click to show)
337.50: 94.0 FPS (-1.1%) Warning: Spoiler! (Click to show)

Final Fantasy XIV: A Realm Reborn Benchmark score
331.82: 15976 Warning: Spoiler! (Click to show)
337.50: 16114 (+0.8%) Warning: Spoiler! (Click to show)

Mafia II built-in benchmark
331.82: 138.4 FPS Warning: Spoiler! (Click to show)
337.50: 141.1 FPS (+2.0%) Warning: Spoiler! (Click to show)

Just Cause 2 Concrete Jungle benchmark
331.82: 87.88 FPS Warning: Spoiler! (Click to show)
337.50: 110.73 FPS (+26.0%) Warning: Spoiler! (Click to show)

Resident Evil 6 benchmark score
331.82: 15848 Warning: Spoiler! (Click to show)
337.50: 16216 (+2.3%) Warning: Spoiler! (Click to show)

Passion Leads Army benchmark
331.82: 143.0 FPS Warning: Spoiler! (Click to show)
337.50: 145.1 FPS (+1.5%) Warning: Spoiler! (Click to show)

So the gains are not global. Some games benefit a little, some by a very large margin and some even go down a bit. The average performance gain is still +5.8% which is nothing to sneeze at IMHO. After all, a few percentage performance difference between GPUs is often enough to make or brake a product.

I came across a couple of bugs during the tests. 3DMark wouldn't even complete the first Ice Storm test, it would crash to desktop each time at the very same place. The sun shafts in STALKER benchmark were also buggy, the global lighting would flash from time to time. Perhaps the WHQL will fix these.
The Air Tunnel
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7-4820K (4500MHz@1.28V) P9X79 Sapphire R9 290 Tri-X New Edition Team Group Vulcan 4x4GB 2133MHz 
Hard DriveCoolingOSPower
2x240GB SSD @RAID 0 Noctua NH-D14 SE2011 Windows 10 Pro Corsair AX750 
CaseAudio
Rosewill Armor EVO hiFace+AudioGD+DT770Pro 
  hide details  
Reply
The Air Tunnel
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7-4820K (4500MHz@1.28V) P9X79 Sapphire R9 290 Tri-X New Edition Team Group Vulcan 4x4GB 2133MHz 
Hard DriveCoolingOSPower
2x240GB SSD @RAID 0 Noctua NH-D14 SE2011 Windows 10 Pro Corsair AX750 
CaseAudio
Rosewill Armor EVO hiFace+AudioGD+DT770Pro 
  hide details  
Reply
post #2 of 13
Why are people testing GPU benchmarks that have no CPU overhead, when the point of this driver is to reduce CPU overhead lol? Also, the ones with +/- less than 1-5% are within margin of error.
post #3 of 13
Quote:
Originally Posted by mercs213 View Post

Why are people testing GPU benchmarks that have no CPU overhead, when the point of this driver is to reduce CPU overhead lol? Also, the ones with +/- less than 1-5% are within margin of error.

I second this.

+rep to the OP though
post #4 of 13
Thread Starter 
Quote:
Originally Posted by mercs213 View Post

Why are people testing GPU benchmarks that have no CPU overhead, when the point of this driver is to reduce CPU overhead lol? Also, the ones with +/- less than 1-5% are within margin of error.

Maybe because people actually play games that are mainly limited by GPUs?

If you consider any difference below 5% threshold within margin of error, I suppose you don't get much out of any review between, say, GTX 780 and R9 290? All within the margin of error, you know.
The Air Tunnel
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7-4820K (4500MHz@1.28V) P9X79 Sapphire R9 290 Tri-X New Edition Team Group Vulcan 4x4GB 2133MHz 
Hard DriveCoolingOSPower
2x240GB SSD @RAID 0 Noctua NH-D14 SE2011 Windows 10 Pro Corsair AX750 
CaseAudio
Rosewill Armor EVO hiFace+AudioGD+DT770Pro 
  hide details  
Reply
The Air Tunnel
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7-4820K (4500MHz@1.28V) P9X79 Sapphire R9 290 Tri-X New Edition Team Group Vulcan 4x4GB 2133MHz 
Hard DriveCoolingOSPower
2x240GB SSD @RAID 0 Noctua NH-D14 SE2011 Windows 10 Pro Corsair AX750 
CaseAudio
Rosewill Armor EVO hiFace+AudioGD+DT770Pro 
  hide details  
Reply
post #5 of 13
FF xiv is not cpu limited rolleyes.gif

Thanks by the way for the mini review:thumb:
Workstation
(4 items)
 
  
CPUMotherboardGraphicsMonitor
Xeon E5-2690 Supermicro 2011 Nvidia GP100/ Vega FE Dell ultrasharp 4k 
  hide details  
Reply
Workstation
(4 items)
 
  
CPUMotherboardGraphicsMonitor
Xeon E5-2690 Supermicro 2011 Nvidia GP100/ Vega FE Dell ultrasharp 4k 
  hide details  
Reply
post #6 of 13
Quote:
Originally Posted by mercs213 View Post

Why are people testing GPU benchmarks that have no CPU overhead, when the point of this driver is to reduce CPU overhead lol? Also, the ones with +/- less than 1-5% are within margin of error.

If you look at the GPU usage you will know some of the games are a bit CPU bound. Metro 2033 benchmark is a CPU bound benchmark even though slightly but with no gains in that game. JC2 is considered a CPU intensive game these days and saw a huge increase. So yeah I think those benchmarks are fairly representative...

Thanks for the benchmarks OP . Rep+ ...
Desktop
(16 items)
 
Laptop
(10 items)
 
 
CPUMotherboardGraphicsRAM
Intel Core i7-5820K @ 4.40 GHz Asus Rampage V Extreme/U3.1 MSI GTX 980 Ti Gaming 6GB  Corsair Vengeance LPX 16GB (4x4) DDR4-3000 
Hard DriveHard DriveCoolingOS
Samsung 850 Pro 1TB Samsung 950 Pro NVMe 512GB NZXT Kraken X62 Windows 10 Pro 
MonitorMonitorKeyboardPower
Asus VG236H 23" 120 Sony XBR43X800D 43" 4K TV CoolerMaster Octane Storm + Razer Turret EVGA SuperNOVA 1000 G3 
CaseMouseMouse PadAudio
Corsair Graphite 760T Black Logitech G502 Proteus Spectrum + Razer Turret Razer Goliathus Asus Xonar Essence STX II 
CPUMotherboardGraphicsRAM
Intel Core i7-4720HQ @ 2.60 GHz Asus G751JY Asus GTX 980M 4GB SK hynix HMT41GS6BFR8A-PB 16GB (2x8) DDR3L-1600 
Hard DriveHard DriveOptical DriveOS
SanDisk SD7SB3Q128G1002  HGST Travelstar 7K1000 1TB HL-DT-ST DVDRAM GUC0N Windows 10 Home 64-bit 
MonitorAudio
LG Philips LP173WF4-SPD1 + G-SYNC @ 75 Hz Realtek ALC668 
  hide details  
Reply
Desktop
(16 items)
 
Laptop
(10 items)
 
 
CPUMotherboardGraphicsRAM
Intel Core i7-5820K @ 4.40 GHz Asus Rampage V Extreme/U3.1 MSI GTX 980 Ti Gaming 6GB  Corsair Vengeance LPX 16GB (4x4) DDR4-3000 
Hard DriveHard DriveCoolingOS
Samsung 850 Pro 1TB Samsung 950 Pro NVMe 512GB NZXT Kraken X62 Windows 10 Pro 
MonitorMonitorKeyboardPower
Asus VG236H 23" 120 Sony XBR43X800D 43" 4K TV CoolerMaster Octane Storm + Razer Turret EVGA SuperNOVA 1000 G3 
CaseMouseMouse PadAudio
Corsair Graphite 760T Black Logitech G502 Proteus Spectrum + Razer Turret Razer Goliathus Asus Xonar Essence STX II 
CPUMotherboardGraphicsRAM
Intel Core i7-4720HQ @ 2.60 GHz Asus G751JY Asus GTX 980M 4GB SK hynix HMT41GS6BFR8A-PB 16GB (2x8) DDR3L-1600 
Hard DriveHard DriveOptical DriveOS
SanDisk SD7SB3Q128G1002  HGST Travelstar 7K1000 1TB HL-DT-ST DVDRAM GUC0N Windows 10 Home 64-bit 
MonitorAudio
LG Philips LP173WF4-SPD1 + G-SYNC @ 75 Hz Realtek ALC668 
  hide details  
Reply
post #7 of 13
Always good to have another person testing systematically. We don't expect to see improvements in GPU limited benches, but it's good to see the drivers didn't introduce any accidental performance regressions either.
BattleStation
(18 items)
 
TV Buddy
(15 items)
 
 
CPUMotherboardGraphicsGraphics
i7 3770k Gigabyte Sniper G1 Sniper 3 Gigabyte Gaming G1 GTX 970 Gigabyte Gaming G1 GTX 970 
RAMHard DriveHard DriveOptical Drive
Corsair Vengeance Pro 16 GB Crucial MX 100 WD Green 3TB Asus Blu-Ray 
CoolingOSMonitorKeyboard
CM Hyper 212 Evo Windows 10 Pro Acer Predator XB270HU CM Quickfire TK 
PowerCaseMouseMouse Pad
Seasonic X-1050 Antec 1100 Logitech G500s Logitech G440 
AudioAudio
Logitech G35 headset StudioSpares Powered Studio Monitors 
CPUMotherboardGraphicsRAM
i5-4690 Asus Z97-A EVGA GTX 670 FTW Crucial Ballistix 4GB x 2 
Hard DriveHard DriveOptical DriveCooling
Western Digital Caviar Green 3 TB Mushkin Chronos Deluxe 240 GB SSD Pioneer BDC-207DBK Blu-Ray Reader, DVD/CD Writer Noctua NH-L12 
CoolingOSKeyboardPower
Scythe Gentle Typhoon 37.0 CFM 120mm x 2 Windows 8.1 Pro Logitech K400 SeaSonic G550 
CaseOtherOther
Silverstone GD09B FLIRC USB Dongle for Media Centre Harmony 650 Remote 
  hide details  
Reply
BattleStation
(18 items)
 
TV Buddy
(15 items)
 
 
CPUMotherboardGraphicsGraphics
i7 3770k Gigabyte Sniper G1 Sniper 3 Gigabyte Gaming G1 GTX 970 Gigabyte Gaming G1 GTX 970 
RAMHard DriveHard DriveOptical Drive
Corsair Vengeance Pro 16 GB Crucial MX 100 WD Green 3TB Asus Blu-Ray 
CoolingOSMonitorKeyboard
CM Hyper 212 Evo Windows 10 Pro Acer Predator XB270HU CM Quickfire TK 
PowerCaseMouseMouse Pad
Seasonic X-1050 Antec 1100 Logitech G500s Logitech G440 
AudioAudio
Logitech G35 headset StudioSpares Powered Studio Monitors 
CPUMotherboardGraphicsRAM
i5-4690 Asus Z97-A EVGA GTX 670 FTW Crucial Ballistix 4GB x 2 
Hard DriveHard DriveOptical DriveCooling
Western Digital Caviar Green 3 TB Mushkin Chronos Deluxe 240 GB SSD Pioneer BDC-207DBK Blu-Ray Reader, DVD/CD Writer Noctua NH-L12 
CoolingOSKeyboardPower
Scythe Gentle Typhoon 37.0 CFM 120mm x 2 Windows 8.1 Pro Logitech K400 SeaSonic G550 
CaseOtherOther
Silverstone GD09B FLIRC USB Dongle for Media Centre Harmony 650 Remote 
  hide details  
Reply
post #8 of 13
Thread Starter 
Even though I got some really nice results, there were a few titles there that I was expecting a bit more from. Especially Sleeping Dogs, where Nvidia claimed an improvement of 13.7% going from 335.23 to 337.50 using the exact same settings. I got 2.6% going from 331.82 to 337.50. But then again, Hitman was quoted at 2.1% and JC2 wasn't even in the release notes...
The Air Tunnel
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7-4820K (4500MHz@1.28V) P9X79 Sapphire R9 290 Tri-X New Edition Team Group Vulcan 4x4GB 2133MHz 
Hard DriveCoolingOSPower
2x240GB SSD @RAID 0 Noctua NH-D14 SE2011 Windows 10 Pro Corsair AX750 
CaseAudio
Rosewill Armor EVO hiFace+AudioGD+DT770Pro 
  hide details  
Reply
The Air Tunnel
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7-4820K (4500MHz@1.28V) P9X79 Sapphire R9 290 Tri-X New Edition Team Group Vulcan 4x4GB 2133MHz 
Hard DriveCoolingOSPower
2x240GB SSD @RAID 0 Noctua NH-D14 SE2011 Windows 10 Pro Corsair AX750 
CaseAudio
Rosewill Armor EVO hiFace+AudioGD+DT770Pro 
  hide details  
Reply
post #9 of 13
Wasnt nvidia talking about much more improvement on SLI systems?
post #10 of 13
Thread Starter 
Quote:
Originally Posted by RnRollie View Post

Wasnt nvidia talking about much more improvement on SLI systems?

Actually some of their single GPU claims were higher than the SLI numbers. They gave a lot of very specific numbers with different setups in this article:

http://www.geforce.com/whats-new/articles/nvidia-geforce-337-50-beta-performance-driver
The Air Tunnel
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7-4820K (4500MHz@1.28V) P9X79 Sapphire R9 290 Tri-X New Edition Team Group Vulcan 4x4GB 2133MHz 
Hard DriveCoolingOSPower
2x240GB SSD @RAID 0 Noctua NH-D14 SE2011 Windows 10 Pro Corsair AX750 
CaseAudio
Rosewill Armor EVO hiFace+AudioGD+DT770Pro 
  hide details  
Reply
The Air Tunnel
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7-4820K (4500MHz@1.28V) P9X79 Sapphire R9 290 Tri-X New Edition Team Group Vulcan 4x4GB 2133MHz 
Hard DriveCoolingOSPower
2x240GB SSD @RAID 0 Noctua NH-D14 SE2011 Windows 10 Pro Corsair AX750 
CaseAudio
Rosewill Armor EVO hiFace+AudioGD+DT770Pro 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
Overclock.net › Forums › Graphics Cards › NVIDIA › NVIDIA Drivers and Overclocking Software › Driver battle: 331.82 WHQL vs. 337.50 BETA (12 benchmarks)