Hacked iTEMS TAGGED WITH 3DMARK
Download C and 3DMark Vantage Pro 1.02 Free. If the 3DMark results above reflect gaming performance you're correct. Share and compare benchmark scores from 3DMark, PCMark and VRMark benchmarks.
'Patch' in Full Version Software Free Download Crack with
Serial Number Download - Smart Serials. Each benchmark produces metrics on graphics. Monitor network activity and block potential threats with a powerful firewall.
Key 3dmark vantage crack
Simplifying your search should return more download results. Crack & Product Serial Key Results for "Futuremark 3DMark 11" [HOST], Page: 1 - for most relevant results include both Software Title+Version number in your search. Gaming Gift Cards Discover all.
Keygen free 3dmark 188.8.131.52_3dmark 1.0.5 download Download
The newly launched AMD Radeon RX 6800 XT is already a world record beater, as verified by the HWBot overclocking records website. Any questions regarding the Graphics Amplifier, how to use it, tweak it, repair it, should also be presented in this thread. To improve search results for 3dmark 11 Pro V try to exclude using words such as: serial, code, keygen, hacked, patch, warez, etc.
3dmark 11 crack 1 0 3 ser Download 2020
Basic Edition – the gamer's benchmark for DirectX 11 3DMark 11 is the latest version of the world's most popular benchmark for. Steam Gift Card 5 USD - Steam Key - For. Vitek 07: 09-03-2020 18: 24 1875636: danik0411, /.
Benchmark Results: 3DMark 11 - AMD Radeon HD 7970 GHz
Nvidia's graphics cards boast a range of exclusive and cutting-edge features. Base Loader Downloader: 05 3DMARK KEYGEN continue reading this. We created Recover Keys to help you save your time and money.
- Buy & Sell Online: PC Games, Software, Gift Cards and More
- SystemInfo installer issue with Steam version: : 3DMark
- NVIDIA GeForce GTX 970 video card benchmark ...
- US$299 HP Omen Accelerator TB3 enclosure/dock
- Yokecat Software: Download 3DMark 11 and 3DMark Vantage
- 3dmark 11 Free Download Full Version
- OMEN Accelerator Review - HP's External GPU Voodoo Doll
- Intel graphics drivers employ questionable 3DMark Vantage
- Parallel programming: Introduction to GPU architecture
- Ashampoo Internet Accelerator 3 Cracked
- Free 3dmark Downloads
- PPU's help 3Dmark 06?
- Which software activation codes can Recover Keys retrieve
- 3DMark 11 Pro 1.0.1 crack - marquitta4575's blog
- 3DMark 11 Professional Edition v1.0.1 patch - peggys's blog
Serial number 3DMark 11 Advanced Edition 1.0.4 For Gamer - Sofware Ku
DirectX 11 features including tessellation, compute shaders and multi-threading. Your PC's performance is determined by the interactions between different hardware components, the operating system and the type and amount of software running. Students pay US$19.99/mo the first year, and US$29.99/mo after Ashampoo.
3DMark 11 Free Download
3dmark 11 keygen accelerator. With Turing GPU cores, complex physics simulations are carried out using PhysX to simulate realistic water, particles, and debris effects in-game. By combining DirectX8.1 support with completely new graphics, it continues to provide benchmark results that empower you to make informed hardware assessments.
Hack nVIDIA GeForce GTX 1070 Ti 3DMark performance
At the beginning, supported software list of RK was much shorter (only a few programs) but we keep adding new supported software and features (now Recover Keys can retrieve activation codes for 10, 004 products). Smart Serials: Your serial numbers database https://vipweb-portal.ru/download/?file=1140. This has been created in cooperation with the major 3D accelerator and processor manufacturers to provide you with a reliable set of diagnostic tools.
Download 3DMark 11 Update
Sometimes, the best things in life are free, and that's especially true in the software world. DirectX* and OpenGL* ES Gaming performance using stand-alone benchmarks: Time Spy (DX 12), Night Raid (DX 12), Fire Strike (DX 11), Sky Diver (DX 11), Cloud Gate (DX 10), Sling Shot (OGL ES 3.0) and Ice Storm (DX 9 / OGL ES 2.0). Fl Studio 10 serials maker.
Activity code download Benchmarking Tools for Windows
Radeon R9 390X is - 4.7% faster in 3DMark 11 Performance GPU* - 6.1% faster in 3DMark 11 Extreme GPU - 4.8-6.3% faster in 3DMark Fire Strike Performance GPU** - 4.5-10.1% faster in 3DMark Fire Strike Ultra GPU** *KitGuru (VCZ result for R9 390X is lower than their R9 390. OS: Windows Vista/7 Download Homepage Screenshot Rating: 5/5 - 282Mb Shareware $19.95 3DMark 11 is the latest version of the world's most popular benchmark. Announcing end of support for 3DMark 11, PCMark 7, Powermark, 3DMark Cloud Gate, and 3DMark Ice Storm December 19, 2020.
Patch aMD Polaris Thread: Radeon RX 480, RX 470 & RX 460
ZdswklBlog: 3DMark 11 Pro full version free download official website. Intel demonstrated their Quad Core CPU with this brand new benchmark from. Basic Edition V. Use the menu at the top to select the amount of /10().
Serial key graphics test freeware download - 3DMark 2020 freeware
Crack Full Version With Patch is one of the best tools. Windows XP Professional 32-bit en-US - Black Edition v2010.12.18 patch 10021 Windows Xp. I scored C0 in 3DMark 11 Custom https://vipweb-portal.ru/download/?file=1138.
3DMark test to measure "pure raytracing performance
Intel Core 2 Quad Processor Q8200, NVIDIA GeForce GT 1030 x 1, 4096 MB, 64-bit Windows 7}. AMD Instinct MI100 Accelerator Unveiled 2020/11/16 @ 17: 56: 50. PowerColor Unveils Radeon RX 6800 XT Red Devil – Triple-Fan & Triple-Slot Cooling With 16 Phase Custom PCB; AMD Radeon RX 6800 XT Scores 3DMark Fire Strike World Record at 2.65.
|1||3dmark and pcmark scores for p4pe 2.53b - Page 2||43%|
|2||IObit Game Booster 184.108.40.206.1 Download||54%|
|3||3DMark 11 - New Update Released||36%|
|4||Intel Graphics Media Accelerator (GMA) 3600||69%|
|5||Zona Perangkat Komputer: 3DMark 11Free to try||64%|
|6||EGPU.io - Your Go-to Destination for All Things eGPU||16%|
|7||WELCOME TO MY SITE: 1000 Serial Number||26%|
|8||AMD Radeon R5 (Stoney Ridge) - NotebookCheck.net Tech||53%|
|9||AMD Announces Worlds Fastest MI100 HPC Accelerator for||20%|
Longitudinal Benchmarking of Game Ready WHQL Display Drivers for Pascal GPUs (Updated 2017.9.26)
INTRODUCTIONThe following is a longitudinal benchmarking of the graphical performance of a wide and representative selection of 13 NVIDIA driver versions on a mid-to-high-end rig with a graphics card Gigabyte GeForce GTX 1070 G1 Gaming.
Is there a perfect version of a graphics driver?If I had to answer quickly I would say that obviously Not because drivers are in continuous correction, development and optimization based on the advances and innovations of both hardware and graphics software. The best driver will depend on the situation and particular configuration of each rig, so what seems optimal for me does not have to be optimal for you.
However, I consider that there are ways to find a driver version which is better in performance and which fix more numbers of reported issues in a longitudinal and representative series of versions under a certain setup. The latter is precisely what I tried to find and to show you when doing this benchmarking of NVIDIA Game Ready WHQL Display Driver versions.
Driver versions testedThe WHQL versions which have been selected in this analysis are as follows:
- The first version that supported a Pascal GPU (v368.39).
- The last build of each main version from the one mentioned in the previous point (versions 368.81, 372.90, 373.06, 375.95, 376.33, 378.92, 381.89, 382.53, 384.94, 385.69).
- The latest version released at the time of publishing this analysis (v387.92).
- The driver build immediately preceding the last version mentioned in the previous point (v385.69).
- One extra version which had quite a few good reports and which I personally remembered as pretty solid and stable (v382.33).
- 368.39 (2016.6.7)
- 368.81 (2016.7.14)
- 372.90 (2016.9.21)
- 373.06 (2016.10.6)
- 375.95 (2016.11.18)
- 376.33 (2016.12.14)
- 378.92 (2017.3.20)
- 381.89 (2017.4.25)
- 382.33 (2017.5.22)
- 382.53 (2017.6.9)
- 384.94 (2017.7.24)
- 385.69 (2017.9.21)
- 387.92 (2017.10.9)
- Case: NZXT Noctis 450 Black
- Case cooling: Front In 2x140 | Rear Ex 1x140 | Top Ex 2x140 (All Cougar Black HB CFD14)
- CPU: Intel Core i7-6700
- CPU Cooling: Cooler Master Hyper 212X EU
- Motherboard: MSI Z170A GAMING M7 (MS-7976)
- RAM: 32 GB (2x16 GB) DDR4-2133 Kingston HyperX Fury
- GPU: Gigabyte GeForce GTX 1070 G1 Gaming (Factory OC)
- HDD: Seagate ST2000DX001 SSHD 2TB SATA 3.1
- PSU: Corsair RM750x
- Monitor: Samsung S24D300HS 24" @ 1080p / 60 Hz
- Win 10 Pro 64 bits (v1703 / Build 15063.632)
- Game Mode, Game Bar and Game DVR OFF
NVCP Global Settings:
- Multi-Display/Mixed GPU Acceleration = Single Display Performance Mode (if available)
- Texture filtering - Quality = High Quality
- Texture filtering - Negative LOD Bias = Clamp
- DSR Factors = 2.00x
NVCP Program Settings:
- Power Management Mode = Prefer maximum performance
Always DDU old driver in safe mode, clean & restart.
FINAL BENCHMARKING STATISTIC SHEETImgur
Legend of the statistical sheet:
- Values in black and highlighted in bold = best result in the benchmark throughout the series of versions.
- Values in red and highlighted in bold = worst result in the benchmark throughout the series of versions.
Custom formulas used for estimating Driver PerformanceDriver Performance Index (Pi):
Pi=a-b Where: a = number of best series results in each benchmark (in a driver). b = number of worst series results in each benchmark (in a driver).Driver Performance Score (Ps):
Ps=(d*c)/z Where: d=Pi+z z = number of benchmarks used. c = 100 (a constant value which indicates neutral driver performance level).According to this custom formula the driver performance score range goes from a minimum value of 0 to a maximum of 200 (0-200).
These two custom formulas are used in the final section ('Conclusions') of our analysis for giving you a tentative recommendation of the best driver version in both performance and level of fixes.
SYNTHETIC BENCHMARKSThe following 3DMark synthetic benchmarks have been used:
- Fire Strike (Overall & Graphics)
- Time Spy (Overall & Graphics)
- API Overhead Feature Test (DX11 ST, DX11 MT, DX12 & Vulkan)
Fire Strike Benchmarks
Fire Strike OverallImgur
This chart shows an upward trend in overall performance in this DX11 synthetic benchmark from driver 368.39 to 376.33, the latter reaching the maximum score in the entire series (15147). From this point there is an average decrease of the overall performance in this test and there are different ups and downs, with the lowest point in the series corresponding to the score of version 384.94. From version 385.69 to the latest 387.92 this latter situation improves and 387.92 offers an acceptable and notably better score (15007).
Fire Strike GraphicsImgur
In this case we see how the graphics performance in the test has different ups and downs, so the progression is not always upward or constant. The maximum performance point corresponds to version 381.89 (19045) and the minimum to the most recent version 384.94 (18562). The last two versions of the series (385.69 & 387.92) improve this situation importantly (18718 & 18832 respectively) although they are still far from offering scores close to those of 381.89 in this benchmark.
Time Spy Benchmarks
Time Spy OverallImgur
Here we can see an upward progression of the global performance in this test although in this upward trend we find some slight ups and downs. The most remarkable improvement is observed when moving from version 376.33 to 378.92 (a jump of 79 points from one version to the other). The maximum peak corresponds to version 382.33 (5882) although the latest version 387.92 offers a fairly close value (5872).
Time Spy GraphicsImgur
In this case we can observe a progression of the graphical performance analogous or equivalent to the previous overall score although with the following difference: here the maximum peak of performance corresponds to the latest version 387.92 (6170).
API Overhead Feature Test
DirectX 11 - Single ThreadImgur
This single thread test shows an irregular performance of the API throughout the series of versions. Here the maximum performance peak was at v382.53 (2.23M) and the minimum was at v381.89 (2.09M). The latest version 387.92 improves that value and it is near the avg value between the mentioned max/min values.
DirectX 11 - Multi ThreadImgur
Here the DX11 multi-thread test also shows an irregular performance development of the API throughout the series of versions. The best score in the series corresponds to version 373.06 (3.27M) while the worst one corresponds to the later and intermediate version 378.92 (3.01M). The latest version 387.92 ([3.23M) is close to the referred maximum.
DirectX 12 API TestImgur
Although with slight ups and downs, in this chart we can see that the performance of the API in this benchmark has been quite solid throughout most of the series of versions. It should be noted [b]two noteworthy performance drops: in version 375.95 (22.74M) and later in 385.69 (22.88M). There is major recovery and improvement in the latest version 387.92 (24.31M). The maximum performance is achieved with version 382.53 (24.43M).
Vulkan API TestImgur
In this case an ideal situation is observed: a steady progression of the API performance in this benchmark. We went from a first minimum point in the first version of the series (368.39 scoring 18.4M) to a maximum point in version 385.69 scoring 22.15M. These results give a good indication of the nice job that would be done to progressively optimize Vulkan API performance over the series.
NOTE: Vulkan test fails due to a program related mismatching with some latest (1.0.61+) Vulkan specifications. Until this problem is fixed by 3DMark programmers I will not include the result of this test, indicating this situation with N/A symbol.
NON-SYNTHETIC BENCHMARKSThe following UNIGINE benchmark suites have been used:
- Unigine Heaven (FPS Avg & Score)
- Unigine Valley (idem)
- Unigine Superposition (DX: FPS Avg & Score / OpenGL: idem)
UNIGINE Heaven BenchmarkSettings are as follows:
- 1920x1080 / DX11 / Ultra / Tessellation Extreme / 3D Disabled /Multi-Monitor Disabled / AA x8 / Full Screen
Unigine Heaven (DX) FPS & ScoreImgur
The chart shows an upward trend with some slight ups and downs from version 368.39 to a peak performance point at version 381.89 (94.3 FPS avg scoring 2376). There was an acute and notable drop in performance at version 382.33 (92.5 FPS avg scoring 2329) and it also represents the minimum graphical performance point in the series in this test. From this minimum peak to the latest driver version 387.92 a progressive restoration of an upward trend in graphical performance in this benchmark is observed.
UNIGINE Valley BenchmarkSettings are as follows:
- 1920x1080 / DX11 / Ultra / Stereo 3D Disabled / Monitors Single / AA x8 / Full Screen
Unigine Valley (DX) FPS & ScoreImgur
In this benchmark the series returns to show a rather irregular development of the graphical performance: we can observe important ups and downs in performance at different points of the series. The maximum peak performance is in version 381.89 (91 FPS avg scoring 3809) and the minimum peak in the later version 382.53 (88.8 FPS avg scoring 3715). On the other hand, from driver 382.33 to the latest the graphical performance in this test are situated at values slightly lower than the average of the values of the series. In addition there is a little improvement in performance in version 387.92 compared to the immediately previous version 385.69.
UNIGINE Superposition BenchmarksSettings are as follows:
- 1080p Extreme Preset
Unigine Superposition (DX) FPS & ScoreImgur
With version 372.90 we can see an outstanding jump (0.67 FPS and 89 points) in graphic performance in this test. The maximum performance point corresponds to versions 376.33 (26.48 FPS avg scoring 3541). The graphical performance of the latest version 387.92 is noteworthy lower than the mentioned maximum performance peak in the series, performing slightly worst than its immediately previous version [b]385.69[/b].
Unigine Superposition (OpenGL) FPS & ScoreImgur
There is a solid upward progression of the graphical performance in this test throughout the series of versions. It is worth noting the important performance improvement that occurs at version 381.89 (23.07 FPS avg scoring 3804) with respect to the previous ones. From that peak point the performance shows very similar values till the end of the series.
BUILT-IN GAME BENCHMARKSThe following built-in game benchmarks have been used:
- Batman: Arkham Knight
- Metro: Last Light Redux
- Deus Ex: Mankind Divided (DX11 & DX12)
- Hitman (2016) DX11 & DX12
- Tom Clancy's Ghost Recon Wildlands
- Rise of the Tomb Raider (DX11 & DX12)
- Grand Theft Auto V
Batman: Arkham Knight
Batman: Arkham Knight FPSIn-game settings are like this:
- Full Screen / 1920x1080 / V-Sync OFF / Texture High / Shadow High / Detail High /Motion Blur ON / AA ON / Chromatic Aberration ON / Film Grain ON /Texture Filtering TRILINEAR / NVIDIA GameWorks All OFF
The peak performance is in both versions 375.95 and 376.33 (both 134 FPS avg) and the minimum peak is in versions 382.33 (130 FPS avg). Over the series there are not big game performance differences in graphics in this built-in benchmark. However, there is a relative noteworthy performance drop in the latest version 387.92 compared to the previous 385.69 (it goes from 133 to 131 FPS avg).
Metro: Last Light ReduxBuilt-in benchmark settings are as follows:
- Full Screen / 1920x1080 / Quality Very High /SSAA ON / AF 16x / Motion Blur OFF / Tessellation Very-High / V-Sync OFF / Advanced PhysX ON
Metro: Last Light Redux FPSImgur
The graphical performance in this built-in benchmark seems irregular along the series. The minimum performance peak corresponds to versions 368.39 and 382.53 (both 70.67 FPS avg) and the latest version 387.92 shows an slightly lower performance (71 FPS avg.) than it previous version 385.69, being the latter at a non-exclusive maximum value of 71.33 FPS on average.
Deus Ex: Mankind DividedIn-game settings for both DX11 and DX12 are like this:
- Full Screen / Exclusive Full Screen / 1920x1080 / MSAA OFF / V-Sync OFF / Stereo 3D OFF / Texture Quality Ultra / AF 16x / Shadow Very High / AO Very High /Contact hardening Shadows OFF / Parallax Occlusion Mapping High / Detail Very High / Volumetric Lighting Ultra / Screen Space Reflections Ultra / TAA / Motion Blur ON / Sharpen ON / Bloom ON / Lens Flares ON / Cloth PhysX OFF / Subsurface Scattering OFF / Chromatic Aberration ON /Tessellation ON
Deus Ex: Mankind Divided (DX11) FPSImgur
This chart shows an irregular progression of graphic performance in this built-in game benchmark as we can see a series with repeated ups and downs at different points. The peak performance is in version 384.94 (68.2 FPS avg) and the minimum performance point is in the far version 368.81 (66.1 FPS avg). This shows that despite the different ups and downs, the progression of the graphic performance in the series in this test goes up. Also noteworthy is the fact that the average framerate under the last versions 385.69 and 387.92 is the second highest (67.2) in this benchmark.
Deus Ex: Mankind Divided (DX12) FPSImgur
Here is another example of an ideal situation of an almost constant and upward progression of graphic performance throughout the series in a built-in game benchmark. The peak performance is shared by versions 384.94 and 385.69 (both 67.6 FPS avg) and the minimum performance point is situated in the second version of the series 368.81 (66.3 FPS avg.). I said almost constant and an upwards performance progression due to the remarkable drop observed in the latest version 387.92 (66.5 FPS avg).
Hitman (2016) DX11 FPSIn-game settings are as follows:
- 2715x1527 (DSR) / Exclusive Full Screen / V-Sync OFF /V-Sync Interval 1 (100% FPS) / HDR OFF / Super Sampling 1.00 / Detail Ultra / AA SMAA / Texture High / AF 16x / SSAO / Shadow Ultra / Shadow Res High / Override Mem Safeguards OFF
We can see an important improvement in performance from version 378.92 (~10 FPS on average of difference from previous version 376.33) and from this point the performance in the series keeps steady around 79 FPS on average. Performance in the latest version 387.92 (79 FPS avg) is slightly lower than in the previous driver 385.69 (79.39 FPS avg).
Hitman (2016) DX12 FPSIn-game settings are like this:
- 2715x1527 (DSR) / Exclusive Full Screen / V-Sync OFF / V-Sync Interval 1 (100% FPS) / HDR OFF / Super Sampling 1.00 / Detail Ultra / AA SMAA / Texture High / AF 16x / SSAO / Shadow Ultra / Shadow Res High / Override Mem Safeguards OFF / Render Target Reuse Auto / Multi GPU OFF
As in the previous chart here there is a big improvement in graphical performance under this built-in game benchmark at version 378.92 (here the difference from the immediately previous version 376.33 is about 9 FPS on average) and from this version the performance is maintained on average although with a slight upward progression. The peak performance lies in version 385.69 (80.97 FPS avg). Should be noted that latest version 387.92 (78.56 FPS avg) shows a little performance drop with respect to its previous version 385.69 (80.97 FPS avg.)
Tom Clancy's Ghost Recon WildlandsIn-game settings are as follows:
- Full Screen / 1920x1080 / Res Scaling 1.00 / V-Sync OFF / Framerate Limit OFF / Extended FOV 0% / TAA / HBAO+ / Draw Distance Very High / Detail Ultra / Texture Ultra / AF 16x / Shadow Very High / Terrain Ultra / Vegetation Very High / Turf ON / Motion Blur On / Iron Sights DOF ON / HQ DOF ON / Bloom ON / God Rays Better / Subsurface Scattering ON / Lens Flares ON / Long Range Shadows OFF
Ghost Recon Wildlands FPSImgur
In this case an irregular progression of the graphical performance is observed in this game benchmark in the version series. The peak performance here is at v384.94 (61.59 FPS avg) and the minimum peak is situated in version 368.81 (60.5 FPS avg). With the latest driver 387.92 there is a little performance drop (61.27 FPS avg) with respecto to its previous 385.69 (61.27 FPS avg.), being the latter slightly better in performance and not far from the referred maximum.
Rise of the Tomb Raider
Rise of the Tomb Raider (DX11)In-game settings are like this:
- Full Screen / Exclusive Full Screen / Stereo 3D OFF / 1920x1080 / V-Sync OFF / FXAA / Texture Quality Very High / AF 16x / Shadow Very High / Sun Soft Shadows Very High / DOF Very High / Detail Very High / Dynamic Foliage High / VXAO / PureHair Very High / Specular Reflection Quality Very High / Vignette Blur ON / Motion Blur ON / Bloom ON / Tessellation ON / Screen Space Reflections ON / Lens Flares ON /Screen Effects ON /Film Grain ON
This chart shows an irregular graphical performance with a peak performance in version 376.33 (90.36 FPS avg). The minimum performance point is in the distant version 368.81 (88.1 FPS avg). Here the performance of the latest version 387.92 (89.29 FPS avg) is slightly lower than the average of the series (~89.4 FPS avg).
Rise of the Tomb Raider (DX12)In-game settings are like this:
- Full Screen / Exclusive Full Screen / Stereo 3D OFF / 1920x1080 / V-Sync OFF / FXAA / Texture Quality Very High / AF 16x / Shadow Very High / Sun Soft Shadows Very High / DOF Very High / Detail Very High / Dynamic Foliage High / HBAO+ / PureHair Very High / Specular Reflection Quality Very High / Vignette Blur ON / Motion Blur ON / Bloom ON / Tessellation ON / Screen Space Reflections ON / Lens Flares ON / Screen Effects ON / Film Grain ON
Here the progression of the graphical performance is ascending although with ups and downs from version 368.39 to 382.33 but is down from 382.53 to the latest driver of the series. The peak performance is in version 382.33 (103.87 FPS avg) and the minimum corresponds to the first version of the series 368.39 (100.21 FPS avg.). The performance of the latest version 387.92 is slightly lower than the average of the series.
Grand Theft Auto VIn-game settings are as follows:
- Full Screen / 1920x1080 / Turf Ultra / Particles Very High / Reflection Very High / Shadows Very High / Shader Very High / Texture Auto / Water Very High / Population Density Max. / Extended Shadows Distance Max. / In-Game DOF Effects ON / Distance Scaling Max. / Extended Distance Scaling Max. / AF x16 / FXAA / MSAA x2 / Reflection MSAA x4 / TXAA OFF / AO Normal / Post-FX Ultra / High Detail Streaming While Flaying ON / Long Shadows ON / High Res Shadows OFF / Soft Shadows NVIDIA PCSS / Tessellation Very High / Population Variety Max.
Grand Theft Auto VImgur
In the above chart we can see an almost graphically stable performance throughout the series even though there are some ups and downs. However, there is a significant performance drop observed in latest version 387.92 (76.44 FPS avg.) which also corresponds to the minimum peak of the series. On the other hand it should be noted that the performance of version 385.69 (84.93 FPS avg.) is very close to the peak performance of the series which is located in version 375.95 (84.98 FPS avg).
Here ends our drivers benchmarking and from here we move to the final section of conclusions of this article.
CautionsSurely many of you will not be interested or you will not have time to read and consider all the above analysis and what you may really be interested in is the final recommendation announced in the introduction of this article. This is entirely logical since this part is precisely the goal of this benchmarking of driver versions.
For those who have followed the development of the previous analysis is likely that you have already been able to guess which may be the best candidates to be the best driver of this series under an equal or similar rig to the one used for this benchmarking. So the results we are going to present below may not surprise you.
It should be clear that the following final results and the recommendation derived from them can in no case be absolute or definitive as it is based on a specific sample of the population of versions studied and even thought it can be considered representative, is logically subject to change over time. Therefore any conclusion and recommendation that we make should be considered provisional or tentative and should be considered valid only for configurations with same or similar specifications to the one we have used in this analysis (at least for systems that include a Pascal GPU).
How to identify our best NVIDIA driver version
- I have tried to formalize and standardize somehow these final results to offer a more accurate and non-intuitive recommendation of the best driver in the series. For this I have created two mathematical formulas with their indexes and corresponding parameters that allow us to estimate the overall performance of each driver version analyzed.
- The first formula results in a positive or negative variable value that we will call here Performance Index (Pi). This index will act as a modifier of the second formula which will result in a overall Performance Score (Ps) for each driver version.
- To finish and share our current recommended NVIDIA driver in terms of performance for Pascal GPUs we will introduce a second qualitative criterion, which is the version Level of Fixes (Lf). This level corresponds to the number of technical issues reported that are solved in the corresponding driver version. In this case the higher and more recent version would be the one that will result in a higher level of fixes.
- Finally the combination of the Ps value and the qualitative Lf of each driver version will give us an adequate way for estimating which is our current recommended driver.
Final Results & Analysis
Final Results SheetImgur
According to the final statistical sheet and the charts above our best driver version in terms of performance it is still clearly 381.89 because it shows a Ps of 123.08 which is the highest score of the whole version series. However, when we combine this numerical criterion with the qualitative Lf we can not consider it our best driver since that version is far from recent, reason why here many reported issues would not be solved.
That is why we should consider the next driver version with better Ps and with a higher version iteration (build) as the candidate for our best driver. In this case it is still version 385.69 which shows a Pi equal to +2 and a Ps of 107.69 points in our custom scale.
OUR RECOMMENDED NVIDIA DRIVER
For all of the above and according with our analysis, driver version 385.69 has to be currently considered the best of the series and therefore 385.69 is our current recommended driver for Pascal GPUs.
Maxed out Mac Pro (dual Vega II Duo) benchmarks
Specs as configured:-Intel Xeon W-3275M 28-core @ 2.5GHz
-12x 128GB DDR4 ECC 2933MHz RAM (6-channel, 1.5TB)
-4TB SSD (AP4096)
-Two AMD Radeon Pro Vega II Duo
-Afterburner accelerator (ProRes and ProRes RAW codecs)
Benchmarks:3DMark Time Spy: 8,618
-Graphics score: 8,537
-CPU score: 9,113
3DMark Fire Strike Extreme: 11,644
-Graphics score: 12,700
-CPU score: 23,237
No multi GPU support
VRMark Orange Room: 10,238
-CPU: 34,074 samples
-GPU: 232 mpaths (used CPU, did not detect GPU on macOS or Windows)
Cinebench R20: 9,705
-BMW CPU: 1:11 (slower in Windows)
-BMW GPU: did not detect GPUs in macOS, detected in Windows but forgot to log time because it was ~7 minutes. Possible driver issue?
-Classroom CPU: 3:25
-Gooseberry CPU: 7:54
-CPU single: 1151
-CPU multicore: 19650
-GPU metal: 82,192
-GPU OpenCL: 78,238
Blackmagic disk speed test:
-Write: 3010 MB/s
-Read: 2710 MB/s
Blackmagic RAW speed test (8K BRAW playback):
-CPU: 93 FPS
-GPU (metal): 261 FPS
1080p high: 12,031
Games (antialiasing, vsync and motion blur off):Shadow of the Tomb Raider:
-4K ultra 50 fps
-4K high 65 fps
-1080p ultra 128 fps
-1080p high 142 fps
-1080p OpenGL ultra 100 (90-120 while moving, 180 while standing still)
-1080p vulkan ultra 170
-4K Vulkan low 52FPS (4K Vulkan = CPU bottleneck?)
-4K Vulkan med 52FPS
-4K Vulkan high 52FPS
-4K Vulkan ultra 52FPS
-1080p ultra 132FPS
-4K ultra 56fps
-4K high 56 FPS
-4K med 60fps
Team Fortress 2 (dodgeball mode)
-4K 190-210 FPS
Counter Strike Global Offensive (offline practice with bots, Dust II)
-1080p 240-290 fps
-1080p enhanced 160fps
-1440p enhanced 163 fps
-4K enhanced 116 fps
-1080p ULTRA 73 FPS 13.6ms
-1440p ultra 58fps 17.12 ms
-4K ultra 34.41 FPS 29.06 ms
Deus Ex Mankind Divided
-1080p ULTRA 84fps
-1440p ultra 75.4 FPS
-4K ultra 40.8 FPS
Ashes of the Singularity (DirectX 12, utilizing 2 of 4 GPUs):
-1080p extreme 87.3 FPS (11.5ms)
-1440p extreme 89.3 FPS 11.2ms
-4K extreme 78.4 FPS 12.8 ms
"Crazy" graphics setting (max setting, one step higher than extreme)
-1080p crazy 63.3 FPS 15.8 ms
-1440p crazy 60.2 FPS 16.6 ms
-4K crazy 48.5 FPS 20.6ms
1080p extreme (GPU bottleneck)
-Normal batch 89.9% GPU bound
-Medium batch 77.1% GPU bound
-Heavy batch 57.8% GPU bound
Notes: -macOS does not recognize the Vega II Duo, nor dual Vega II/Duo as a single graphics card. Applications still only use 1 of 4 Vega II GPUs even under Metal. Only benchmark here that utilized all four GPUs was Blackmagic RAW speed test. -Windows also sees the two Vega II Duos as four separate graphics cards, and Ashes of the Singularity is the only game that supports Explicit Multi GPU in DirectX 12 that utilizes multiple graphics cards through the motherboard, allowing you to combine completely different cards like NVIDIA and AMD together. Even then, it only used two of the four Vega II GPUs.
I have read conflicting info regarding whether the Vega II silicon is the same as the Radeon VII, where the VII has 4 of its 64 CUs disabled and half the VRAM as the Vega II. Does anyone know if this is true?