Microsoft Kernel Patch CPU Before and After Benchmarks Thread


Recommended Posts

  On 09/01/2018 at 05:36, adrynalyne said:

Maybe? This isn't my forte. As it was explained to me, it matters if it is x86 code instructions being called with sys calls. The GPU isn't x86 as far as I know.

Expand  

This certainly is not a problem how processor executes x86 code. It is a problem here how user mode app communicates with the OS kernel and how processor handles that matter. All requests for hardware (graphics, disk, memory access) must be passed to the OS which then executes them. Apps do not write directly to GPU . They use OS to handle these calls which then calls graphics driver (at the OS kernel level) which then writes data and commands to the GPU. Therefore I think there is a great chances that these fixes affect 2D graphics as well as other data intensive tasks.

 

Here is the security problem (as far as I understand): cpu allowed user mode app to gain access to OS kernel data, which it should never do. CPU does not handle OS and user mode isolation securely enough.

 

Intel (and others) have used a "shortcut" to speed up things and created a security flaw which is severe.

 

But I must say I was amazed that the effect of this fix was so huge at my Lenovo Thinkpad T440s laptop. Did not expect that much. Maybe Lenovo will provide fixes to the drivers which will help here.

  On 09/01/2018 at 09:25, Jampe said:

Here are my results. FIrst test was right before update. After update I ran tests 3 times since I could not believe my eyes. Results are consistent. PC is Lenovo thinkpad t440s model 20ARS04W00

PerfRes1.gif

Expand  

Disk and 2D Graphics really suffer. I hope neither Microsoft nor Intel will stop optimising their patches. As they are now, I won't be updating. In fact, I will try to find updated Windows 10 ISOs with December patches and archive them if I ever need to reinstall Windows.

On the Azure front we're seeing slowdowns on our legacy Azure VM's running SQL Server. We're having to scale up the VM's to cope with the slowdowns.

 

Obviously this is all being logged and the extra costs will be argued.

 

Not seeing the same impact on Azure SQL Servers, there is an impact but probably only 10% extra as opposed to the ~25% on the VM's.

 

To make matters worse, MS scheduled production VM's for updates then threw the date out the window and upgraded when they felt like it. We're talking multiple end clients productions systems going down for anywhere from 30 mins to 5 hours without warning. A cluster###### if ever there was one.

 

Read some of the comments on this Azure status. I won't say which is my real life account...

https://azure.microsoft.com/en-us/blog/securing-azure-customers-from-cpu-vulnerability/

 

Don't forget that these patches are only part of the fix. According to the articles I've read, a BIOS update will also be required to fully patch the fix. I presume this is where a lot of the slowdowns will come from.

  On 09/01/2018 at 10:54, Daedroth said:

Don't forget that these patches are only part of the fix. According to the articles I've read, a BIOS update will also be required to fully patch the fix. I presume this is where a lot of the slowdowns will come from.

Expand  

the bulk of dell systems should get a bios fix today.

guys the following CERT advisory has links to lots of affected platforms and products, including timescales for fixes from vendors.

 

https://www.us-cert.gov/ncas/alerts/TA18-004A

 

we are currently assessing the entire global enterprise (20k endpoints, 8k+ servers) and we have reached out to NetApp, Cisco and other suppliers of enterprise kit.

 

Were using the tack, if it has a CPU, its potentially susceptible until we are told otherwise (called due diligence).

 

NetApp have confirmed their SAN arrays are  not affected. They dont permit execution of 3rd party code. Awaiting confirmation about Cisco kit.

 

*UPDATE** Cisco have responded. A mixed bad, some products immune but the bulk are affected.

 

https://tools.cisco.com/security/center/content/CiscoSecurityAdvisory/cisco-sa-20180104-cpusidechannel

 

 

Asus have released the microcode fix/BIOS update for my S270 Strix H. 

 

https://www.asus.com/Motherboards/ROG-STRIX-Z270H-GAMING/HelpDesk_BIOS/

 

STRIX Z270H GAMING BIOS 1203
Improved DRAM compatibility.
Fixed SAMSUNG M.2 device issue.
Fixed SanDisk M.2 device issue.
Fixed AMD SSD device issue.
Fixed SteelSeries Keyboard issue.
Update CPU MicroCode.

 

Cinebench 4D BEFORE Win patch

CPU Benchmark = 966CB

OpenGL Bench = 140 FPS

AFTER

CPU Benchmark = 975CB ???

OpenGl Bench = 149.95 FPS ???

NVidia Driver update

CPU Benchmark = 962CB

OpenGL Bench =147.64FPS

Post Bios UPD

CPU Benchmark=942CB

OpenGL Bench=147.36

 

3D Mark timespy extreme BEFORE

3998

AFTER

3997 ????

NVidia Driver Update

3971

Post Bios UPD

3955

 

Novabench 

2713

After

2703

NVidia Driver update

2677

Post Bios UPD

2618

 

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

Post NVidia 

Image Editing 108,788/ Time 48.9757

Encoding 84,939/time 62.7273

OpenCL 113,529/ Ksamples per sec 20948

Heavy Multitasking 99,848

Time 76.4357

System Score 101,776 

Post Bios UPD

Image Editing 91,665/ Time 58.1242

Encoding 80,259/time 66.3843

OpenCL 112,996/ Ksamples per sec 20798

Heavy Multitasking 94,712

Time 80.5809

System Score 94,908

 

Hmm taking it with a pinch of salt, as i noticed post BIOS upd i left the OC as auto, so its at 4.37Ghz, not 4.5Ghz I benched with before...DOH

 

ill redo it tomorrow and update.

  On 09/01/2018 at 22:47, Mando said:

Asus have released the microcode fix/BIOS update for my S270 Strix H. 

 

https://www.asus.com/Motherboards/ROG-STRIX-Z270H-GAMING/HelpDesk_BIOS/

 

STRIX Z270H GAMING BIOS 1203
Improved DRAM compatibility.
Fixed SAMSUNG M.2 device issue.
Fixed SanDisk M.2 device issue.
Fixed AMD SSD device issue.
Fixed SteelSeries Keyboard issue.
Update CPU MicroCode.

 

Cinebench 4D BEFORE Win patch

CPU Benchmark = 966CB

OpenGL Bench = 140 FPS

AFTER

CPU Benchmark = 975CB ???

OpenGl Bench = 149.95 FPS ???

NVidia Driver update

CPU Benchmark = 962CB

OpenGL Bench =147.64FPS

Post Bios UPD

CPU Benchmark=942CB

OpenGL Bench=147.36

 

3D Mark timespy extreme BEFORE

3998

AFTER

3997 ????

NVidia Driver Update

3971

Post Bios UPD

3955

 

Novabench 

2713

After

2703

NVidia Driver update

2677

Post Bios UPD

2618

 

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

Post NVidia 

Image Editing 108,788/ Time 48.9757

Encoding 84,939/time 62.7273

OpenCL 113,529/ Ksamples per sec 20948

Heavy Multitasking 99,848

Time 76.4357

System Score 101,776 

Post Bios UPD

Image Editing 91,665/ Time 58.1242

Encoding 80,259/time 66.3843

OpenCL 112,996/ Ksamples per sec 20798

Heavy Multitasking 94,712

Time 80.5809

System Score 94,908

 

Hmm taking it with a pinch of salt, as i noticed post BIOS upd i left the OC as auto, so its at 4.37Ghz, not 4.5Ghz I benched with before...DOH

 

ill redo it tomorrow and update.

Expand  

Mando, is RealBench only for ASUS gaming boards bench utility? Thx for the update, btw :)

  On 10/01/2018 at 13:44, Yogurth said:

Mando, is RealBench only for ASUS gaming boards bench utility? Thx for the update, btw :)

Expand  

not 100% its via ROG aspect.

 

give it a whirl, it may work

http://dlcdnmkt.asus.com/rog/RealBench_v2.56.zip?_ga=2.161564219.660284626.1515594169-1054940240.1499777541

 

 

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

    • No registered users viewing this page.
  • Posts

    • It makes no sense to name it iOS 26 after iOS 18 but logic has left the building a long time ago (remember Microsoft and their Windows 8, 8.1 and 10 or Windows 7 that was actually 6.2)? Not to mention the stupid trend of "rapid releases" where minor releases with some bug fixes and UI changes (basically unnoticeable) masquerade as major releases...
    • Have you tried setting your font to Aptos? Its specifically designed to display better.
    • WebChangeMonitor 25.06 by Razvan Serea Monitors allows you to quickly check a number of web pages and tracks changes based on the content of the web pages. Allows to monitor several protocols, including HTTP and HTTPS. Allows to view and record differences. Available for Win7/10, Linux and others. WebChangeMonitor features: Allows monitoring of web pages and informs about content changes Indication of states of currently monitored items in the tool and taskbar Reporting as sound and/or email as well as log file or HTML log Several configuration / filter options Support all protocols, e.g. http, https Multi-threaded, running in the background Bulk-import and bulk-export of items (from/to CSV) to monitor Export of results to CSV file for further processing Allows running command on items states and/or showing diff (changes) of content with preferred diff-tool ...and many more! Open Source (C++, wxWidgets) Cross platform for Windows (7/10), Linux, RPi and Mac (if self-compiled) WebChangeMonitor 25.06 release notes: Updates several libraries to stay up-to-date with latest security fixes. Few bug-fixes took place related to ignore items being no displayed correctly sometimes in the UI. especially, the sqlite library was updated to hopefully also fix a recent (but very rare and random) crash bug. Download: WebChangeMonitor 64-bit | 8.9 MB (Open Source) Download: WebChangeMonitor 32-bit | 8.2 MB View: WebChangeMonitor Website | Other Operating Systems | Screenshot Get alerted to all of our Software updates on Twitter at @NeowinSoftware
    • KEF Q Concerto Meta, Polk Reserve R200 Limited Edition speakers are lowest priced by Sayan Sen Nowadays, soundbars with wireless subwoofer systems offer people a convenient way to experience decent-quality audio; systems such as the Samsung Q900F, Q800F, and Q800F, which are available at their lowest price levels. However, if you are someone who is looking for a smaller bookshelf speaker for more sound accuracy and can live without louder or deeper bass notes, then take a look at KEF and Polk Audio's Q Concerto Meta and Reserve R200 speakers, respectively, as both of them are up for sale at their lowest ever prices (purchase links under the specs lists below). First up, we have the KEF Q Concerto Meta, which is a three-way stereo system despite being designed within a small bookshelf footprint. That is because the KEF has a coaxial Uni-Q driver array that packs both the mid-range and tweeter. The woofer is a 6.5-inch driver, and thus it does not go below 48 Hz, so you will need a separate subwoofer if you want bass frequencies below that. This is a passive speaker with two terminals and no other fancy connectivity features. The technical specs of the KEF Q Concerto Meta are given below: Drive Units: Uni‑Q Driver Array: HF: 19 mm (0.75 in.) vented aluminium dome with MAT™ MF: 100 mm (4 in.) aluminium cone Bass Unit (LF): 165 mm (6.5 in.) hybrid aluminium cone Frequency Response: (-6 dB): 40 Hz – 20 kHz (±3 dB): 48 Hz – 20 kHz Typical In-Room Bass Response (at -6 dB): 36 Hz Crossover Frequencies: 430 Hz and 2.9 kHz Amplifier Requirements: 15 – 180 W Sensitivity (2.83V/1m): 85 dB Harmonic Distortion (at 90 dB, 1m): Less than 2% above 37 Hz Less than 1% between 91 Hz – 20 kHz Maximum Output: 108 dB Impedance: 4 Ω (minimum 3.2 Ω) Connectivity: Wired connection only (no wireless options) Get the KEF Q Concerto Meta at the link below: KEF Q Concerto Meta Three-Way Bookshelf Speaker (White, Pair): $1199 (Amazon US) Up next, we have the Polk Audio Reserve R200 50th Anniversary model. This is a special edition speaker and the company says that only 1000 pairs were produced. Unlike the KEF, this is a two-way design and so does not have a mid-range driver. So you do sacrifice some of the mids. However, it should still be great for listening to music as vocals are mainly necessary for movie-watching. The technical specs of the Polk Reserve R200 are given below: 1" Ring Radiator Tweeter 6½" Turbine Cone Woofer Minimum Impedance 3.8Ω Sensitivity 86dB Recommended Amplifier Power 30–200W Frequency Response (-3dB) 51Hz–38kHz Get the Polk Reserve R200 limited edition at the link below: Polk Reserve R200 50th Anniversary Limited Edition Bookshelf Speaker - Dolby Atmos & IMAX Enhanced, Cherry: $797.97 (Amazon US) This Amazon deal is US-specific and not available in other regions unless specified. If you don't like it or want to look at more options, check out the Amazon US deals page here. Get Prime (SNAP), Prime Video, Audible Plus or Kindle / Music Unlimited. Free for 30 days. As an Amazon Associate, we earn from qualifying purchases.
  • 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
      488
    2. 2
      +FloatingFatMan
      257
    3. 3
      snowy owl
      247
    4. 4
      ATLien_0
      222
    5. 5
      +Edouard
      191
  • Tell a friend

    Love Neowin? Tell a friend!