Passmark releases PerformanceTest v10 benchmark, adds AVX512 - Receives Critique for favoring Intel

And all of the sudden the benchmarks are in favor of Intel again. Last week Passmark 10 was released, and the CPU benchmarks invike an AVX 512 test, which AMD does not support. However, it jacks up the scores significantly specific for Intel processors.

Some examples, a Core i9 9900K is now listed somewhere 2954 to 2540 points, Ryzen 9 3900X scores only 86% of that. However before the update took place, the AMD was 2% faster than that intel processor as the Ryzen scored 2939 points , and the i9 9900K reached 2889 points. AMD does not support AVX512 and therefore sees a significant impact on performance Technology critics around the globe calls it 'bribe from Intel' , since the Intel chips are now considerably better. 

Not only Intel CPUs now have pushed all AMD CPUs out of the top positions, the ranking does not make sense even anymore. Interesting is that the AVX512 command extension cannot explain the good performance for Coffee Lake-S processors, as they do not have an AVX512.

Passmark issued a statement, and explains why the new scores are higher. However, oddities as a laptop apu (the Ryzen 3 4300U) being the fastest AMD processor, are not explained.

We released a new version of PerformanceTest a few days ago, version 10. Improvements in the benchmark test algorithms & using a more modern compiler resulted the single threaded test performing a much higher number of operations per second. These changes should push the CPU harder and use modern CPU features (out of order execution and multiple pipelines) better. The result was roughly 3x times more operations per second being performed, compared to PerformanceTest V9.

Yesterday we started to switch over the graphs on the web site to start to use results from PerformanceTest V10. This accounts for the change in the results in the graphs. However, in hindsight we think have done the wrong thing. We should had scaled down the PT10 single threaded result to match the PT9 results for the single threaded test. This single threaded test was already an average of values from several different single threaded algorithms. So additional scaling wouldn't have changed the significance of the value.

On Monday (9th March 2020) we plan to patch the version 10 release to scale the single threaded value back to the PT9 results. Things should then be back to normal. In the meantime we have reverted the single threaded graph on the web site to use only PT9 results.

As we collect more PT10 results we expect PT10 to perform better on modern CPUs compared to older ones (relative to PT9). So overtime there might be a spreading out of the single threaded results, with the newer hardware pulling away from the older hardware a bit more. Sorry for any confusion all this has caused.

-- Passmark Press release--  We are excited to inform you that PassMark Software have just launched Performance Test V10. PassMark have developed V10 to provide a more comprehensive analysis of PC systems through advancements in almost every aspect of our testing. Some of they key PerformanceTest 10 features include:

 CPU Benchmark Changes:

• Revised CPU testing to use real world algorithms. Including Zip compression, AES encryption, Bullet Physics & SHA256 hashing.
• Introduction of a cross platform score that will allow meaningful comparison of CPU performance on PC, Android & Apple devices.
• Algorithm optimization allowing for better out of order instruction execution & introduction of new CPU instructions (including AVX512)
• Updated compiler to Visual Studio 2019
• In short, better utilization of modern CPU features.

2D Graphics Benchmark Changes:

• Addition of a SVG vector image rendering test
• Addition of a PDF rendering test
• More use of DirectX to get accelerated 2D rendering

3D Benchmark Test Changes:

• Increased default resolutions across tests to add load to GPU. 1080p is now minimum recommended resolution, 4K is optimal.
• More comprehensive GPU compute tests
• Better VM support

RAM Benchmark Changes:

• Rewrote database test to use SQLite V3.31
• Added support for processor groups and thread affinity
• Latency test now samples several different memory ranges to get better average figure

Disk Test Changes:

• Relabeled “Random seek” test to “32KQD20”. 32KB block size and Queue depth of 20
• Added new 4KQD1 test (4KB block size and queue depth of 1)
• Removed the CD/DVD test

Advanced Tests:

• Network tests can now use Windows RIO (registered input / output) and extensive use of threading. Testing 10Gb+ networks is now possible.
• Added temperature collection to the disk test to allow checking for thermal throttling
• New advanced CPU test. Allowing threading bottlenecks to be tested.
• New advanced internet speed test

For the full list, see here: https://www.passmark.com/products/performancetest/history.php