AMD Ryzen Threadripper 1950X 16c 32t Processor Review

Testing and Methodology

Test Procedure

Here at eTeknix, we endeavor to disclose vital information regarding the benchmarking process so that readers can quantify the results and attempt to replicate them using their hardware. When it comes to CPU reviews, the benchmarks are pretty self-explanatory although there are a few exceptions.

Rember that your choice of motherboard, the silicon lottery, and other factors can yield different numbers, and there’s always a margin for error when using the software. Therefore, your experience may vary. Each benchmark runs three times, and the average figure is taken to try to reduce the effect of hardware variation. Of course, any relevant details regarding the parameters will be listed below.

Test Systems

X399

• Motherboard – MSI X399 Gaming Pro Carbon AC
• RAM – Crucial Ballistix 32GB (4x8GB) 2400MHz
• CPU Cooler – CM Master Liquid 120mm (will be retested with Noctua D15S with dual fans in coming weeks)
• Graphics Card – Nvidia GeForce GTX 1080 Ti
• Power Supply – BeQuiet Dark Power Pro 850 Watt
• Main Storage Drive – Toshiba OCZ VX500 500GB
• Chassis – Lian Li T80 Test Bench
• Operating System – Windows 10 Pro 64 Bit

X299

• Motherboard – Gigabyte AORUS Gaming 9
• RAM – Crucial DDR4 2400MHz Quad-Channel 32GB
• CPU Cooler – Noctua D15S with dual fans
• Graphics Card – Nvidia GeForce GTX 1080 Ti
• Power Supply – BeQuiet Dark Power Pro 850 Watt
• Main Storage Drive – Toshiba OCZ VX500 500GB
• Chassis – Lian Li T80 Test Bench
• Operating System – Windows 10 Pro 64 Bit

AM4 Ryzen R3, R5 and R7

• Motherboard – MSI XPower Titanium X370
• RAM – GeIL  DDR4 2977MHz Dual Channel 16GB
• CPU Cooler – Noctua D15S with dual fans
• Graphics Card – Nvidia GeForce GTX 1080 Ti
• Power Supply – BeQuiet Dark Power Pro 850 Watt
• Main Storage Drive – Toshiba OCZ VX500 500GB
• Chassis – Lian Li T80 Test Bench
• Operating System – Windows 10 Pro 64 Bit

X99 – Our GPU Review Test Bench

• Motherboard – Asus Rampage V Extreme X99
• RAM – Crucial 16GB (2 x 8GB) 2666 MHz
• CPU Cooler – Noctua D15S with dual fans
• Graphics Card – Nvidia GeForce GTX 1080 Ti
• Power Supply – BeQuiet Dark Power Pro 850 Watt
• Main Storage Drive – Toshiba OCZ VX500 500GB
• Chassis – Lian Li T80 Open Air Test Bench
• Operating System – Windows 10 64-bit

Games Used

All games are tested at both 1920 x 1080 and 2160 x 3840 (4K) on a 60 Hz display with V-Sync off for all tests. Previously we would use “extreme” presets, but these have now been adjusted to “Medium” or equivalent to better test the capabilities of the CPU, not the GPU.

• Rise of the Tomb Raider (Steam)
  • DX12 Medium Preset
  • Pure Hair Off
• Deus Ex (Steam)
  • DX12 Medium Preset
• Ghost Recon: Wildlands (Uplay)
  • Medium Preset
  • Turf Effects Off
• Far Cry Primal (Uplay)
  • Normal Preset

Software Used

• 3DMark Fire Strike (download)
  • FireStrike (1080p) Benchmark
• Unigine Superposition (download)
  • 1080p Extreme Benchmark
• PCMark 10 Professional (download)
  • Express Benchmark
• WPrime (download)
  • 32M and 1024M
  • Power usage recorded at 1024M 50% complete, thermals recorded at 75%
• CineBench R15 (download)
  • CPU Multi
  • CPU Single
• Handbrake (download)
  • Custom MP4 to MKV 4K conversion (details below)
• AIDA64 Engineer (download)
• CPU-Z (download)
• HWMonitor (download)

Handbrake

To stress processors to their absolute limit and accurately judge their performance in video editing workloads, we transcode a 7.7GB compilation of gaming footage; this particular file is freely available from here. The captured footage is 22 minutes and 12 seconds long, it has a bit rate of 50.1 Mbps and it uses the Advanced Video Codec. Additionally, the video runs at a constant 30 frames-per-second and opts for a 3820 x 2140 (4K) resolution. Once loaded into Handbrake, we then transcode the 4K MP4 to an MKV file using the “normal” profile.

Other Notes

A rest period of 2-5 minutes is observed between each piece of software allowing the system to return to its idle power usage and temperatures. Background services like Windows Update are checked to not be running during the testing period by setting WiFi to Metered Connection.

Thanks

Thank you to Noctua, Crucial, ASUS, Gigabyte, Lian-Li, be quiet!, OCZ, for providing the hardware that helps makes these tests possible!