SiSoft Sandra 20/21 download v31.66

Generic Benchmarks 55 Updated by Hilbert Hagedoorn

About this file
File Size
Operating Systems

We are pleased to announce the launch of SiSoftware Sandra 201, the latest version of our award-winning utility, which includes remote analysis, benchmarking and diagnostic features for PCs, servers, mobile devices and networks.

We have expanded our portfolio with brand-new CPU and GPGPU Image Processing benchmarks (testing common filters) that support all modern instruction sets (FMA, AVX, SSE2) as well as GPGPU interfaces (CUDA, OpenCL, DirectX 10/11 ComputeShader). Since Compute is now supported in the very latest versions of OpenGL (4.3+) we have also ported the GPGPU benchmarks to this new interface.

With the public release of Windows 10 RTM we have transitioned to the brand-new tools in order to use the very latest technologies, including future DirectX 12 (both shader and compute).

As SiSoftware operates a "just-in-time" release cycle, some features were introduced in Sandra service packs: in Sandra they have been updated and enhanced based on all the feedback received.

  • New Benchmark Module: GPGPU Image Processing (common filters: blur, sharpen, sobel, median/de-noise) supporting all modern interfaces (CUDA, OpenCL, DirectX ComputeShader)
  • New Benchmark Module: CPU Image Processing (common filters: blur, sharpen, sobel, median/de-noise) supporting all modern vectorised SIMD instruction sets (FMA, AVX, SSE2)
  • New OpenGL Compute Support: Ported GPGPU benchmarks to OpenGL (4.3+) Compute Shader (Fractals, Crypto, Image Processing)
  • New GPU Precision: FP16/half-float precision benchmarks (Financial, Scientific)
  • New CPU Test: 64-bit Integer Dhrystone measuring 64-bit integer workload performance.
  • New Transcode Test: HEVC/H.265 media transcode test, brand-new high-bitrate master AVC1 media file 1080p and UHD/4K (commercial versions) for UHD/4K, 3K, 1440p transcoding benchmarking.
  • Updated Benchmark: Updated Overall Score  by adding new benchmarks to the index.
  • New Operating System Support: Full support for Windows 10 , 8.1, 8, 7 as well as Server 2016*, 2012/R2, 2008/R2.
  • New Hardware Support: Modern and future hardware support

While each benchmark measures the performance of a specific device (CPU, Memory, (GP)GPU, Storage, etc.), there is a real need for a benchmark to evaluate the overall computer performance: this new benchmark is a weighted average of the individual scores of the existing benchmarks:

  • Native CPU Arithmetic, Cryptographic, Multi-Media (SIMD), Financial and Scientific: measures native processing performance using the very latest instruction sets (AVX2, FMA3, AVX, SSE2)
  • .Net/Java Arithmetic: measures software virtual machine performance (e.g. for .Net WPF/Silverlight/Modern applications)
  • Memory and Cache Bandwidth and Latency: measures memory and caches performance
  • File System/Storage Bandwidth and I/O: measures storage performance
  • GP (General Processing) / HC (Heterogonous Compute) (GPU/APU) Arithmetic, Cryptographic, Financial, Scientific: measures (GP)GPU/APU processing performance
  • GP (General Processing) / HC (Heterogonous Compute) (GPU/APU) Memory Bandwidth and Latency: measures (GP)GPU/APU memory performance).

Change Info:
New Benchmarks:

* Crypto-Processor (TPM) Benchmark

Benchmark the cryptographic performance of the crypto-processor (TPM). Measure the random number generator, hashing (SHA, etc.), encryption/decryption (AES) bandwidth.

* Internet DNS Benchmark

Benchmark the performance of the DNS service. Measure the latency of both cached and un-cached DNS queries to local and remote DNS servers.

* Overall Software VM (.Net/Java) Score

A combined performance index all software VM engine benchmarks, .Net (CLR) Arithmetic, Multi-Media and Java (JVM) Arimetic and Multi-Media.

Updated Benchmarks:

* Processor Multi-Core Efficiency Benchmark

Separate reporting of different type latencies (inter-thread/core/module/node/die/package/socket), e.g. when testing multi-module (CCX on AMD) processors:

Inter-Thread (same Core) Latency : (ns)
Inter-Core (same Module) Latency : (ns)
Inter-big/LITTLE Core (same Module, different Core Type) Latency : (ns)
Inter-Module (same Package) Latency : (ns)
Inter-Node/Die (same Package) Latency : (ns)
Inter-Package (different Socket) Latency: (ns)

Updated Client (GUI):

Brand-new "dark-mode" using native operating system (Windows 10+, Server 2019+) theme and colours, with automatic "light/dark" switch based on schedule. The Official SiSoftware Ranker has also been "dark-mode" enabled with theme selected based on web-browser setting.

* Brand-new hardware support and optimisations for brand-new CPU & GP-GPU architectures:

- AMD Ryzen CPU 5000-series (Zen3)
- AMD Radeon RX 6000 GP-GPU (RDNA2, Navi2)
- Hygon/AMD CPU (Zen)
- All the fixes in the very latest Sandra 20/20

- Intel
- Preliminary Intel Core Gen 11 ("RocketLake" RKL), Gen 12 ("Alderlake" ADL)
- Intel Core Gen 11 ("TigerLake" TGL), Iris Xe (Gen12) graphics
- Intel DG1 (Iris Xe Max) dedicated graphics
- All the fixes in the very latest Sandra 20/20

- nVidia
- nVidia GeForce RTX 3000-series (Ampere) CUDA 8.x

- Zhaoxin/Centaur
- ZHAOXIN KaiXian KX-6000, KX-5000

Share this content
Twitter Facebook Reddit WhatsApp Email Print