Microsoft Kernel Patch CPU Before and After Benchmarks Thread


Recommended Posts

Here is a thread where users can post benchmarks of their CPU before and after the update from Microsft.

 

Intel Core i7 477R CPU 3.20GHz 

Brix Pro

 

Before patch 

 

39488119491_30104f8c4e_b.jpg

 

After

 

25618130438_d94ae4538f_b.jpg

 

Benchmark software 

PerformanceTest

https://www.passmark.com/products/pt.htm

  • Like 3

Not sure why I am getting result that is opposite of what I expected. :/

I am running Windows 10 1709 (16229.192) after patched, downloaded from this link:

https://support.microsoft.com/en-us/help/4056892

 

 

 

Lenovo Laptop

 

SuperPi Mod 1.9 WP
==================
Before (1M): 14 seconds
Before (8M): 157 seconds

 

After (1M): 14 seconds
After (8M): 153 seconds


PassMark PerformanceTest (CPU)
==============================
Before: 3797, 3812, 3823 (Average: 3810)
After : 3844, 3837, 3846 (Average: 3842)


Geekbench 4 (CPU)
=================
Before (Single-Core): 3139, 3146, 3146 (Average: 3144)
After  (Single-Core): 3156, 3172, 3188 (Average: 3172)

 

Before (Multi-Core): 5588, 5602, 5574 (Average: 5588)
After  (Multi-Core): 5556, 5556, 5576 (Average: 5563)

 

(I lost the links for before patched)

 

Score link for after patched:
https://browser.geekbench.com/v4/cpu/6101200
https://browser.geekbench.com/v4/cpu/6101313
https://browser.geekbench.com/v4/cpu/6101448


Jetstream 1.1
=============
Before: 165.50
After : 164.57

 

 

 

HP Laptop

 

SuperPi Mod 1.9 WP
==================
Before (1M): 9 seconds
Before (8M): 112 seconds

 

After (1M): 9 seconds
After (8M): 112 seconds


PassMark PerformanceTest (CPU)
==============================
Before: 5955, 5961, 5875 (Average: 5930)
After : 6197, 6099, 5797 (Average: 6031)


Geekbench 4 (CPU)
=================
Before (Single-Core): 4668, 4672, 4644 (Average: 4661)
After  (Single-Core): 4661, 4679, 4700 (Average: 4680)

 

Before (Multi-Core): 9209, 9180, 9168 (Average: 9186)
After  (Multi-Core): 8755, 8743, 9275 (Average: 8924)

 

Score link for before patched:
https://browser.geekbench.com/v4/cpu/6101071
https://browser.geekbench.com/v4/cpu/6101179
https://browser.geekbench.com/v4/cpu/6101271

 

Score link for after patched:
https://browser.geekbench.com/v4/cpu/6102483
https://browser.geekbench.com/v4/cpu/6102564
https://browser.geekbench.com/v4/cpu/6102847


Jetstream 1.1
=============
Before: 264.45
After : 253.10

  • Like 2

Great idea Warwagon:)

 

here is the start of mine 

 

Benchtools used on sig rig @1080p

 

  • Cinebench 4D
  • 3D Mark (Timespy Extreme)
  • Novabench
  • Rog RealBench
  • Passmark

Before Patch 

 

Passmark CPU test BEFORE

image.thumb.png.2a0caaf441a5705f451f4729244e8af9.png

 

Passmark CPU AFTER

image.thumb.png.22faf2114646503b195c9f2f77df1aae.png

 

Cinebench 4D BEFORE

  • CPU Benchmark = 966CB
  • OpenGL Bench = 140 FPS

AFTER

  • CPU Benchmark = 975CB ???
  • OpenGl Bench = 149.95 FPS ???
     

3D Mark timespy extreme BEFORE

  • 3998

AFTER

  • 3997 ????

Novabench 

  • 2713

After

  • 2703
     

RealBench BEFORE

  • Image Editing 212,486/ Time 25.0745
  • Encoding 93,501/time 56.9829
  • OpenCL 113,529/ Ksamples per sec 20877
  • Heavy Multitasking 110,647
  • Time 68.9761
  • System Score 132,540

RealBench AFTER

  • Image Editing 208.830/Time 25.5135
  • Encoding 92,841/time 57.3881
  • OpenCL 113,529/ Ksamples/sec 20922
  • Heavy Multitasking 103,867/Time 73.4779
  • System Score 129,766

 

Overall slightly slower with some benches, Cinema 4D benches.....anyones guess! :p

 

before and after had all the same processes running and apps, what i have noticed however, Asus AI Suite 3 wont run :p

 

btw Webroot Secure anywhere working no problem at all post patch,

  • Like 2

Tried some differnet benchmark software which didn't give the greatest info, but you can see the results.

 

Intel Pentium N3700 1.60GHz

 

Before

 

38793771664_5533e6e52b_b.jpg

 

After

 

38793771734_fb1cd59f13_b.jpg

 

  • Like 2
  On 04/01/2018 at 12:44, alexander3133 said:

Not sure why I am getting result that is opposite of what I expected. :/

 

Expand  

It’s in the realm of measurement accuracy/error. 

  On 04/01/2018 at 18:58, Elliot B. said:

So, no difference.

 

The whole thing was overblown.

Expand  

Maybe for Windows. It’s real in Linux. 

ok 

 

Lenovo X201 Test. (old system)

 

Core i7 ULV M620@2.67ghz

 

Benches used 

  • Cinema4D
  • Passmark

 

Passmark Before 

5a4e8383b9af7_lappypre.thumb.jpg.6024137879040880573249a58fc44833.jpg

 

AFTER 

5a4e958691f73_lappypost.thumb.jpg.259bfe9816feb65a582df57b81d53235.jpg

 

Cinema 4D

CPU = 204cb

OpenGL= 0.78 FPs (LMFAO)

 

AFTER

CPU= 207cb ????

OpenGL= 0.79

 

Hardly thorough benchmarking but both higher than before patch....go figure!

 

I would like to see some benchies submitted with AMD chips for a more rounded dataset tbh.

 

Teacup, meet storm ;) 

 

 

  On 04/01/2018 at 21:36, Mockingbird said:

I installed the update and it turns out that kernel VA shadow is off.

 

No wonder it's not having any performance impact.

Expand  

good info to know mate, thank you.

The reason why older CPUs see a larger performance hit is the absence of PCID. PCID mostly mititgates the performance hit. Not sure how far it goes back (haswell maybe?), but seventh and 8th gen Intel CPUs should not feel the burn.

  On 05/01/2018 at 00:28, adrynalyne said:

The reason why older CPUs see a larger performance hit is the absence of PCID. PCID mostly mititgates the performance hit. Not sure how far it goes back (haswell maybe?), but seventh and 8th gen Intel CPUs should not feel the burn.

Expand  

what about my stellar perf on that X201? 

 

M620 i7 ULV? 

  On 05/01/2018 at 00:35, Mando said:

what about my stellar perf on that X201? 

 

M620 i7 ULV? 

Expand  

Further research and discussion with co-workers on my end  has shown that unless the process/app/whatever is making sys calls, it’s not going to make much of a difference. 

 

 

  On 05/01/2018 at 00:37, adrynalyne said:

Further research and discussion with co-workers on my end  has shown that unless the process/app/whatever is making sys calls, it’s not going to make much of a difference. 

 

 

Expand  

I also read something similar, the performance hit is only when making sys calls, so a VM would potentially have a performance decline, correct?

  On 05/01/2018 at 00:40, Circaflex said:

I also read something similar, the performance hit is only when making sys calls, so a VM would potentially have a performance decline, correct?

Expand  

ill get back to you on that tomms mate, I set up a W10 VM and 2k16 server VM for testing at work before leaving today. On Lenovo X Series hosts running Esxi 6. (12 Xeons per host)

 

corpwide were assessing and deciding if its worth pushing the patch out, Sec dir above us reckons while its a stinker, it doesnt pose any real world threat to our global estate at this point in time. for one the systems need to be compromised before this is even a "thing", games already lost at that point.

  • Like 3
  On 05/01/2018 at 00:40, Circaflex said:

I also read something similar, the performance hit is only when making sys calls, so a VM would potentially have a performance decline, correct?

Expand  

Unknown. I’m way out of my comfort zone with the complexity of this. 

  On 05/01/2018 at 00:43, adrynalyne said:

Unknown. I’m way out of my comfort zone with the complexity of this. 

Expand  

yep, im the same, grasping some of it, but bowing to my Sec teams superior knowledge in this field, im just helping them spread the workload, cos im nice like that (and im whoring myself to Global Corp Sec team for a new role with them from home :P)

 

im sick of going into an office and dealing with retarded end losers directly as well as server/site adm work (which i love!), im getting too old to enjoy juggling multiple roles at a single site, and my boss is an inept A-hole :p but thats another story!

 

might even post a soapbox article on todays events re: NTP settings and synching issues, i was without words when i discovered what had been setup. put it this way, in general Default GPO holds every policy estate wide....no just no!! /faceplant! thats a starter for 10 in what i discovered!

  On 04/01/2018 at 18:28, Mando said:

 

before and after had all the same processes running and apps, what i have noticed however, Asus AI Suite 3 wont run :p

 

Expand  

Im having the same issue with Asus AI Suite 2, were you able to find a fix ?

  On 05/01/2018 at 00:56, jamester64 said:

Im having the same issue with Asus AI Suite 2, were you able to find a fix ?

Expand  

Nah mate, tried reinstalling and no dice. no new version atm either. meantime ive set it all up in bios manually, inc fan tuning.

 

Ive emailed Asus support to notify them of my findings. 

  On 05/01/2018 at 00:58, Mando said:

Nah mate, tried reinstalling and no dice. no new version atm either. meantime ive set it all up in bios manually, inc fan tuning.

 

Ive emailed Asus support to notify them of my findings. 

Expand  

Strange that AsusFanControlService is still running in the background so at least I got that :-) I just cant get the GUI to open :-(

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Posts

    • No operating system works well with programs that modify its interface. There will always be bugs, whether it's MacOS or Linux. Windows too, you'll only suffer from bugs using this crap software.
    • KDE brings UI improvements, bug fixes and more to Plasma 6.4 as stable release draws near by David Uzondu With less than a month to the release of Plasma 6.4, the KDE dev team has shared what it has been working on in the latest issue of its weekly roundup. The update shows a heavy focus on user interface polish and a whole slew of bug fixes as the June 17 release date gets closer. The team has pushed a number of UI refinements for the upcoming version. On the System Settings page for Wi-Fi, the network list can now be fully navigated with a keyboard. KDE also disabled the ability to drag and drop displays on top of one another in the monitor settings. This was done because it could create unsupported arrangements that triggered a cascade of strange bugs throughout the system. Waking up a sleeping computer by pressing the power button no longer causes the bizarre logout screen to appear after you unlock it, which is a relief. Alignment issues in the settings page for the Digital Clock widget were also resolved. The list of bug fixes for 6.4 is extensive. The development team has fixed the most common crash affecting the System Monitor and squashed another one related to a divide-by-zero error. For users with multiple monitors, a long-awaited fix has landed that prevents windows from disappearing when the screen they are on gets disconnected. Even the humble Sticky Notes widget received attention; it will no longer freeze the Plasma shell if you place it on a very thick panel. Discover, the software center, also had a bug patched that caused it to crash if closed too quickly after launch. Here's the full list of improvements: Putting a Sticky Note widget on a very thick panel can no longer cause Plasma to freeze; now, you can use a thick panel with a sticky note on it as a notes sidebar. Fixed the most common System Monitor crash. Fixed another crash in System Monitor, this time a divide-by-zero. Fixed a case where xdg-desktop-portal-kde could crash after you choose a video source to start streaming. Fixed a bug that caused Discover to crash if you close it immediately after it launched. Fixed multiple subtle bugs with the screen chooser widget and OSD that caused it to do the wrong thing on rotated screens or when mirroring screens. Fixed a bug that caused the System Settings’ search field not to be focused properly when pressing Ctrl+F while any UI elements in a settings page already had focus. Fixed the root cause of multiple issues involving windows disappearing when you disconnect the screen they’re on. Clicking a button on a desktop widget that opens a menu no longer inappropriately makes the widget enter Widget Edit Mode. Files with a # or ? character in their name or full path are no longer unexpectedly missing from the history lists in Kicker/Kickoff/etc launchers. Fixed a bug that caused the screen chooser window to sometimes not appear as expected when OBS was launched. Fixed a bug that caused tiled windows on a multi-screen setup to lose their tiling settings when the system went to sleep and woke up again. The "Move window to [activity]" feature now works properly when invoked from the Task Manager widget. Fixed a bug in the Kicker Application Menu that caused keyboard navigation not to work if the popup opened with an item under the pointer. In the Overview effect’s grid view, dragging windows tiled on one virtual desktop over to a different virtual desktop now keeps them tiled as expected. The focus may be on the imminent 6.4 release, but work on what comes next never really stops. Looking ahead, development on Plasma 6.5 is already well underway. Just last week, the KDE team brought several performance improvements to Plasma 6.5.0. This week, the team fixed a nagging bug that sometimes caused the Networks widget to think a hotspot was still enabled after disconnecting from Wi-Fi. On the features side, the System Settings' Fonts page now prevents you from breaking your system entirely by setting fonts below 4pt. In addition to that, switching time spans in the Info Center’s energy page now features smooth graph animations. Performance-wise, kwriteconfig should also be faster, allowing changes you made to the keyboard layout using the tool to reflect immediately.
    • Display Driver Uninstaller (DDU) 18.1.1.4 by Razvan Serea Display Driver Uninstaller (DDU) is a utility for completely removing AMD/NVIDIA/INTEL graphics drivers and related packages from your system, attempting to eliminate all leftovers (including registry entries, folders and files, driver store). Though AMD/NVIDIA/INTEL drivers can usually be removed via the Windows Control Panel, this uninstaller tool was created for situations where standard uninstall fails, or when you need to fully remove NVIDIA or ATI graphics card drivers. After using this driver cleaner, your system will behave as though it’s the first time you’re installing a new driver—similar to a fresh Windows installation. As with all such tools, we recommend creating a restore point beforehand, allowing you to undo changes if issues arise. If you're having trouble installing an older or newer driver, try it—there are reports that it resolves such problems. Recommended usage: The tool can be used in Normal mode but for absolute stability when using DDU, Safemode is always the best. Make a backup or a system restore (but it should normally be pretty safe). It is best to exclude the DDU folder completely from any security software to avoid issues. You do NOT need to uninstall the driver prior using DDU. Requirements: .NET Framework 4.8 Compatible with Windows 7, 8, 8.1, 10, and 11 (32-bit or 64-bit) Note: Using on Insider Preview builds is at your own risk. Display Driver Uninstaller (DDU) 18.1.1.4 changelog: Intel: Added NPU presence detection before removing shared DLL files (these were previously left to prevent potential NPU-related issues). Intel: Added optional NPU removal Improved "Extension" driver removal process. Updated several translations. Download: Display Driver Uninstaller 18.1.1.4 | 1.7 MB (Freeware) Download: DDU Portable | 1.2 MB Links: Display Driver Uninstaller Home Page | Screenshot | Forum Get alerted to all of our Software updates on Twitter at @NeowinSoftware
    • That is what I also think to a point. How is 26 necessarily different than just another iteration like 18, 19, 20, etc? At first I think it would be better for Apple to use 2026, 2027, and then maybe truncate it to just 28, 29, and so on. Granted, I also think it makes more sense to use the year when it was released (2025 or 25), not 26/2026. Maybe Apple's thinking is as stated by Aditya that the bulk of a release being the current release is during the following year (2026, for example) after it is released (2025, for example). There are other examples of these sorts of things. In the NBA, the season has always started in the Fall and ends in the Spring. When the season ends is how the season is named. So, this is the 2025 NBA season and the Pacers and Thunder are playing to be the 2025 NBA Champions. The NFL starts in the Fall and the end of the season is always in the beginning of the next year as well. The Super Bowl was played on February 9, 2025, but that was the end of the 2024 NFL season. So, contrary to the NBA, the NFL names it season based on when it starts, not when it ends. Maybe that is because more of the NFL season is played at the end of the year (2024 in the most recent example) whereas most of the NBA season is played in the first half of the following year (2025 in the current example).
  • Recent Achievements

    • Week One Done
      abortretryfail earned a badge
      Week One Done
    • First Post
      Mr bot earned a badge
      First Post
    • First Post
      Bkl211 earned a badge
      First Post
    • One Year In
      Mido gaber earned a badge
      One Year In
    • One Year In
      Vladimir Migunov earned a badge
      One Year In
  • Popular Contributors

    1. 1
      +primortal
      495
    2. 2
      snowy owl
      252
    3. 3
      +FloatingFatMan
      251
    4. 4
      ATLien_0
      228
    5. 5
      +Edouard
      191
  • Tell a friend

    Love Neowin? Tell a friend!