MR² CPU-Mark 2.0.1 Beta 64-Bit Features: CPU- Speedtest Testing the arithmetic and numeric performance, as well as the efficiency of transcendental operationens. Numerical deviations are neglected. Cache - Speedtest Matrix multiplication is a cache-intensive application. PI - Speedtest The number PI is calculated based on the Bailey-Borwein-Plouffe-Formula. Which means it's also applicable on the GPU. MEM -Speedtest Testing the load and store-performance of the underlying architecture. All tests are Single-, Multithreaded and OpenCL Fixed time for calculations Requirements: at least XP 64Bit at least .NET 4.0 Framework x64-Processor Notes: Tested with AMD GCN, Nvidia Kepler+ and Intel Onboard HD5000+ Older GPU architectures weren't tested yet. Update 2.0.1 More than one OpenCL device can now be benchmarked at the same time.(SLI,Crossfire,IGPU+Dedicated GPU etc.) Results from the older 2.0 version aren't comparable with this version. Download: MR² CPU-Mark 2.0.1 Beta 64-Bit Beta Disclaimer: This right here is a beta version. The purpose of this software release is to optimise the reliability and quality until the final version is finished. Original Thread on 3DCenter (German)
Submitting 3930k @ 4.7GHz & waiting on approval by U.S.A., Inc. I will post some results later & thanks K-putt for the share.
@4.6GHz atm tried to run openCL @ igpu4600 but app crashed & stopped responding. EDIT: ok newer intel gpu driver passed, but its slooow :nerd: Although its a bit inconsistent by multicore.. This is the max I got 972, next run 962, another 930 again.. Maybe its a win10 thing too.. dunno..
Don't have enough post to embed an image, but the scores for my 4930K @ 4.5 Ghz break down as follows: CPU: 209/1524/10741 Cache: 275/1568/7813 Pi: 165/1125/1253 Mem: 277/1058/3764 Singe:226 Multi: 1299 OpenCL: 4460 http : //s15.postimg.org/sczdtroe3/ascd.png (remove spaces)
Could you add an iterative load balancer so my r7 240 can help my r7870 ? It wouldnt be so hard to implement one that balances in just 5-6 iterations depending on the timings of each gpu.