SlideShare a Scribd company logo
1 of 30
Data Classification Hardware Acceleration




Ivan Manuylov
Senior Software Developer, DLP Research
Kaspersky Lab
Security Analyst Summit 2013, Puerto Rico
Speed




PAGE 2   Security Analyst Summit 2013, Puerto Rico
Speed




PAGE 3   Security Analyst Summit 2013, Puerto Rico
Speed?




PAGE 4   Security Analyst Summit 2013, Puerto Rico
Speed?




PAGE 5   Security Analyst Summit 2013, Puerto Rico
Speed?




PAGE 6   Security Analyst Summit 2013, Puerto Rico
Speed?




PAGE 7   Security Analyst Summit 2013, Puerto Rico
Hardware
Cores

  Who has access to 1000+ cores?




PAGE 9     Security Analyst Summit 2013, Puerto Rico
Purpose of GPU

General Purpose computation on GPU




PAGE 10    Security Analyst Summit 2013, Puerto Rico
How to use GPGPU



                                                               Direct Compute
 Cross platform           Cross platform                      Windows
 Multi vendor             NVIDIA                              Multi vendor
 C kernel                 C/C++ kernel                        HLSL kernel
 Languages:               Languages:                          Languages:
    C/C++                    C/C++                               C++
    .NET                     .NET                                C++ AMP
    Java                     Java                                .NET
    Ruby                     Fortran
    Python                   Perl
                              Python
PAGE 11            Security Analyst Summit 2013, Puerto Rico
Kernel code




PAGE 12       Security Analyst Summit 2013, Puerto Rico
Using of Hardware Acceleration
Using of Hardware Acceleration




PAGE 14     Security Analyst Summit 2013, Puerto Rico
Research
Hardware


CPU                               GPU
Intel Core i5 – 750              NVIDIA GeForce GTS 450
2.66 GHz                         1.6 GHz
4 cores                          192 cores
DDR3:                            GDDR5:
  667 MHz                          1.8 GHz
  12 Gb/s                          58 Gb/s



PAGE 16         Security Analyst Summit 2013, Puerto Rico
Fingerprinting
Fingerprinting




PAGE 18     Security Analyst Summit 2013, Puerto Rico
Fingerprinting




                                             100 000 hashes


     100 Kb


PAGE 19       Security Analyst Summit 2013, Puerto Rico
Fingerprinting

            2x faster




PAGE 20     Security Analyst Summit 2013, Puerto Rico
Fingerprinting
      1800
µs
      1600

      1400
               CPU          GPU
               is           is
      1200
                                                                              2x
      1000     better       better
       800                                                                               CPU
                                                                                         GPU
       600

       400

       200

          0
               0        5              10                  15            20         25
                                                                                   Kb


     PAGE 21                 Security Analyst Summit 2013, Puerto Rico
Keywords
Keywords




PAGE 23    Security Analyst Summit 2013, Puerto Rico
Keywords
          What is the performance boost?




PAGE 24           Security Analyst Summit 2013, Puerto Rico
Keywords
          What is the performance boost?



                10+
               times
PAGE 25           Security Analyst Summit 2013, Puerto Rico
Keywords
      60000
µs                                            1000 keywords

      50000

                       GPU
      40000
                       is
      30000
                       better
                                                                             10X              CPU
                                                                                              GPU
      20000



      10000



               0
                   0    10      20            30            40              50     60    70
                                                                                        Kb

     PAGE 26                    Security Analyst Summit 2013, Puerto Rico
Epilogue
Summary



          Fingerprints search                           2x

          Keywords search                               10x

PAGE 28     Security Analyst Summit 2013, Puerto Rico
We can do it!




PAGE 29         Security Analyst Summit 2013, Puerto Rico
Data Classification Hardware Acceleration

                          Thank you!



Ivan Manuylov
Senior Software Developer, DLP Research,
Kaspersky Lab
Ivan.Manuylov@kaspersky.com
Security Analyst Summit 2013, Puerto Rico

More Related Content

Viewers also liked

Formative assessment copy
Formative assessment copyFormative assessment copy
Formative assessment copy
klbattle
 
Formative assessment copy
Formative assessment copyFormative assessment copy
Formative assessment copy
klbattle
 
Challenges in building a mobile apps platform
Challenges in building a mobile apps platformChallenges in building a mobile apps platform
Challenges in building a mobile apps platform
Avi Wortzel
 
Who were-the-most-influential-persons-in-the-history-of-mankind-1197931534304...
Who were-the-most-influential-persons-in-the-history-of-mankind-1197931534304...Who were-the-most-influential-persons-in-the-history-of-mankind-1197931534304...
Who were-the-most-influential-persons-in-the-history-of-mankind-1197931534304...
Lukman O
 
40interestingwaystouseqrcodesinthecla 110724032613-phpapp02
40interestingwaystouseqrcodesinthecla 110724032613-phpapp0240interestingwaystouseqrcodesinthecla 110724032613-phpapp02
40interestingwaystouseqrcodesinthecla 110724032613-phpapp02
Amy Slefendorfas
 
The Past perfect and Present perfect
The Past perfect  and Present perfectThe Past perfect  and Present perfect
The Past perfect and Present perfect
Carlos Es
 
Proposal lomba pramuka
Proposal lomba pramukaProposal lomba pramuka
Proposal lomba pramuka
Akrom Alhaq
 
1st Round Athletics Deck (1)
1st Round Athletics Deck (1)1st Round Athletics Deck (1)
1st Round Athletics Deck (1)
Sterling Brewster
 

Viewers also liked (14)

Formative assessment copy
Formative assessment copyFormative assessment copy
Formative assessment copy
 
Formative assessment copy
Formative assessment copyFormative assessment copy
Formative assessment copy
 
Maths space
Maths spaceMaths space
Maths space
 
Challenges in building a mobile apps platform
Challenges in building a mobile apps platformChallenges in building a mobile apps platform
Challenges in building a mobile apps platform
 
Who am i
Who am iWho am i
Who am i
 
Preso
PresoPreso
Preso
 
Scaling up your R&D group
Scaling up your R&D groupScaling up your R&D group
Scaling up your R&D group
 
Who were-the-most-influential-persons-in-the-history-of-mankind-1197931534304...
Who were-the-most-influential-persons-in-the-history-of-mankind-1197931534304...Who were-the-most-influential-persons-in-the-history-of-mankind-1197931534304...
Who were-the-most-influential-persons-in-the-history-of-mankind-1197931534304...
 
40interestingwaystouseqrcodesinthecla 110724032613-phpapp02
40interestingwaystouseqrcodesinthecla 110724032613-phpapp0240interestingwaystouseqrcodesinthecla 110724032613-phpapp02
40interestingwaystouseqrcodesinthecla 110724032613-phpapp02
 
The Past perfect and Present perfect
The Past perfect  and Present perfectThe Past perfect  and Present perfect
The Past perfect and Present perfect
 
Proposal lomba pramuka
Proposal lomba pramukaProposal lomba pramuka
Proposal lomba pramuka
 
Croydon property sale_1
Croydon property sale_1Croydon property sale_1
Croydon property sale_1
 
1st Round Athletics Deck (1)
1st Round Athletics Deck (1)1st Round Athletics Deck (1)
1st Round Athletics Deck (1)
 
Mooroolbark property sale
Mooroolbark property saleMooroolbark property sale
Mooroolbark property sale
 

Similar to Hardware acceleration.kl.dlpr.sas.2013 final

Quotation & Comparison to Purchase 5 Laptops
Quotation & Comparison to Purchase 5 LaptopsQuotation & Comparison to Purchase 5 Laptops
Quotation & Comparison to Purchase 5 Laptops
Hl Henry
 
Gpu Compute
Gpu ComputeGpu Compute
Gpu Compute
jworth
 
Future Cloud Infrastructure
Future Cloud InfrastructureFuture Cloud Infrastructure
Future Cloud Infrastructure
exponential-inc
 

Similar to Hardware acceleration.kl.dlpr.sas.2013 final (20)

Monitoring of GPU Usage with Tensorflow Models Using Prometheus
Monitoring of GPU Usage with Tensorflow Models Using PrometheusMonitoring of GPU Usage with Tensorflow Models Using Prometheus
Monitoring of GPU Usage with Tensorflow Models Using Prometheus
 
Silicom Ventures Talk Aug 2013 - GPUs and Parallel Programming create new opp...
Silicom Ventures Talk Aug 2013 - GPUs and Parallel Programming create new opp...Silicom Ventures Talk Aug 2013 - GPUs and Parallel Programming create new opp...
Silicom Ventures Talk Aug 2013 - GPUs and Parallel Programming create new opp...
 
TiECon Florida keynote - New opportunities for entrepreneurs using GPU & CUDA
TiECon Florida keynote - New opportunities for entrepreneurs using GPU & CUDATiECon Florida keynote - New opportunities for entrepreneurs using GPU & CUDA
TiECon Florida keynote - New opportunities for entrepreneurs using GPU & CUDA
 
NVIDIA vGPU - Introduction to NVIDIA Virtual GPU
NVIDIA vGPU - Introduction to NVIDIA Virtual GPUNVIDIA vGPU - Introduction to NVIDIA Virtual GPU
NVIDIA vGPU - Introduction to NVIDIA Virtual GPU
 
GPU Cloud Server in India
GPU Cloud Server in IndiaGPU Cloud Server in India
GPU Cloud Server in India
 
Bhargav_PPTgrgregregrgregregregregregregre.pptx
Bhargav_PPTgrgregregrgregregregregregregre.pptxBhargav_PPTgrgregregrgregregregregregregre.pptx
Bhargav_PPTgrgregregrgregregregregregregre.pptx
 
Scalable AI Solution cross AI platforms
Scalable AI Solution cross AI platformsScalable AI Solution cross AI platforms
Scalable AI Solution cross AI platforms
 
Big Data LDN 2017: BI Converges with AI - GPUs for Fast Data
Big Data LDN 2017: BI Converges with AI - GPUs for Fast DataBig Data LDN 2017: BI Converges with AI - GPUs for Fast Data
Big Data LDN 2017: BI Converges with AI - GPUs for Fast Data
 
Quotation & Comparison to Purchase 5 Laptops
Quotation & Comparison to Purchase 5 LaptopsQuotation & Comparison to Purchase 5 Laptops
Quotation & Comparison to Purchase 5 Laptops
 
Hi Ahmed
Hi AhmedHi Ahmed
Hi Ahmed
 
Gpu Compute
Gpu ComputeGpu Compute
Gpu Compute
 
Enabling innovation in mobile SoCs
Enabling innovation in mobile SoCsEnabling innovation in mobile SoCs
Enabling innovation in mobile SoCs
 
Tim McDonough Presentation for Qualcomm Snapdragon 820
Tim McDonough Presentation for Qualcomm Snapdragon 820Tim McDonough Presentation for Qualcomm Snapdragon 820
Tim McDonough Presentation for Qualcomm Snapdragon 820
 
組み込みから HPC まで ARM コアで実現するエコシステム
組み込みから HPC まで ARM コアで実現するエコシステム組み込みから HPC まで ARM コアで実現するエコシステム
組み込みから HPC まで ARM コアで実現するエコシステム
 
GPU-Virtualization-in-openSUSE.pdf
GPU-Virtualization-in-openSUSE.pdfGPU-Virtualization-in-openSUSE.pdf
GPU-Virtualization-in-openSUSE.pdf
 
Future Cloud Infrastructure
Future Cloud InfrastructureFuture Cloud Infrastructure
Future Cloud Infrastructure
 
Raspberry Pi 2 + Windows 10 IoT Core + Node.js
Raspberry Pi 2 + Windows 10 IoT Core + Node.jsRaspberry Pi 2 + Windows 10 IoT Core + Node.js
Raspberry Pi 2 + Windows 10 IoT Core + Node.js
 
BladeCenter GPU Expansion Blade (BGE) - Client Presentation
BladeCenter GPU Expansion Blade (BGE) - Client PresentationBladeCenter GPU Expansion Blade (BGE) - Client Presentation
BladeCenter GPU Expansion Blade (BGE) - Client Presentation
 
NVIDIA Keynote #GTC21
NVIDIA Keynote #GTC21 NVIDIA Keynote #GTC21
NVIDIA Keynote #GTC21
 
GTC Taiwan 2017 主題演說
GTC Taiwan 2017 主題演說GTC Taiwan 2017 主題演說
GTC Taiwan 2017 主題演說
 

Hardware acceleration.kl.dlpr.sas.2013 final

  • 1. Data Classification Hardware Acceleration Ivan Manuylov Senior Software Developer, DLP Research Kaspersky Lab Security Analyst Summit 2013, Puerto Rico
  • 2. Speed PAGE 2 Security Analyst Summit 2013, Puerto Rico
  • 3. Speed PAGE 3 Security Analyst Summit 2013, Puerto Rico
  • 4. Speed? PAGE 4 Security Analyst Summit 2013, Puerto Rico
  • 5. Speed? PAGE 5 Security Analyst Summit 2013, Puerto Rico
  • 6. Speed? PAGE 6 Security Analyst Summit 2013, Puerto Rico
  • 7. Speed? PAGE 7 Security Analyst Summit 2013, Puerto Rico
  • 9. Cores Who has access to 1000+ cores? PAGE 9 Security Analyst Summit 2013, Puerto Rico
  • 10. Purpose of GPU General Purpose computation on GPU PAGE 10 Security Analyst Summit 2013, Puerto Rico
  • 11. How to use GPGPU Direct Compute  Cross platform  Cross platform  Windows  Multi vendor  NVIDIA  Multi vendor  C kernel  C/C++ kernel  HLSL kernel  Languages:  Languages:  Languages:  C/C++  C/C++  C++  .NET  .NET  C++ AMP  Java  Java  .NET  Ruby  Fortran  Python  Perl  Python PAGE 11 Security Analyst Summit 2013, Puerto Rico
  • 12. Kernel code PAGE 12 Security Analyst Summit 2013, Puerto Rico
  • 13. Using of Hardware Acceleration
  • 14. Using of Hardware Acceleration PAGE 14 Security Analyst Summit 2013, Puerto Rico
  • 16. Hardware CPU GPU Intel Core i5 – 750 NVIDIA GeForce GTS 450 2.66 GHz 1.6 GHz 4 cores 192 cores DDR3: GDDR5: 667 MHz 1.8 GHz 12 Gb/s 58 Gb/s PAGE 16 Security Analyst Summit 2013, Puerto Rico
  • 18. Fingerprinting PAGE 18 Security Analyst Summit 2013, Puerto Rico
  • 19. Fingerprinting 100 000 hashes 100 Kb PAGE 19 Security Analyst Summit 2013, Puerto Rico
  • 20. Fingerprinting 2x faster PAGE 20 Security Analyst Summit 2013, Puerto Rico
  • 21. Fingerprinting 1800 µs 1600 1400 CPU GPU is is 1200 2x 1000 better better 800 CPU GPU 600 400 200 0 0 5 10 15 20 25 Kb PAGE 21 Security Analyst Summit 2013, Puerto Rico
  • 23. Keywords PAGE 23 Security Analyst Summit 2013, Puerto Rico
  • 24. Keywords What is the performance boost? PAGE 24 Security Analyst Summit 2013, Puerto Rico
  • 25. Keywords What is the performance boost? 10+ times PAGE 25 Security Analyst Summit 2013, Puerto Rico
  • 26. Keywords 60000 µs 1000 keywords 50000 GPU 40000 is 30000 better 10X CPU GPU 20000 10000 0 0 10 20 30 40 50 60 70 Kb PAGE 26 Security Analyst Summit 2013, Puerto Rico
  • 28. Summary Fingerprints search 2x Keywords search 10x PAGE 28 Security Analyst Summit 2013, Puerto Rico
  • 29. We can do it! PAGE 29 Security Analyst Summit 2013, Puerto Rico
  • 30. Data Classification Hardware Acceleration Thank you! Ivan Manuylov Senior Software Developer, DLP Research, Kaspersky Lab Ivan.Manuylov@kaspersky.com Security Analyst Summit 2013, Puerto Rico