Technology

Figure 8: Novabench Home Page (source: www.novabench.com)

Figure SEQ Figure \* ARABIC 8 ???

Figure SEQ Figure \* ARABIC 8 ???

There have been many kinds of research and testing using a benchmark in the IT world. The following are some articles that relate the most to this report’s case study.

In his research, Burk, (1993) used a simple benchmark program, bench*c, to test the speeds of various methods of storing instance data namely GetWindowWord(), GetProp() using a string and GetProp() using an atom; in comparison to SendMessage() and PostMessage() under both Windows 3.0 and 3.1 versions. In Windows 3.1, on an 80486 33MHz clone running in an enhanced mode, GetProp() with an atom is three times faster than GetWindowWord(), the speed of which is now slower than in Windows 3.0. The goal was to compare the efficiencies of these function calls from the user's perspective, by calculating the overhead in SendMessage() and PostMessage(). These results indicate that GetProp() is more effective under Windows 3.1. Moreover, using GetProp() with an atom was five times faster under Windows 3.1, and nearly three times faster than using GetWindowWord(). However, in Windows 3.1, GetWindowWord() became slower. The speed of SendMessage() and PostMessage() stayed the same under both Windows 3.0 and Windows 3.1 (Burk, 1993).

In his experiment, Kennedy (2008) started by examining Windows 7's kernel and other low-level structures, then subsystems behavior and application runtime characteristics in terms of improved performance comparing to its predecessor Windows Vista. Kennedy used a combination of the Windows Performance Monitor utility and some reference data for Windows Vista and XP. Then, he compared the runtime structure and composition of various OS processes and services. He found that Windows 7 and Vista have the same number of threads (97 to 100) in the system process. Windows 7 M3's System process used a similar amount of memory (3.5MB to 4.5MB). The similarities were a fact for key subsystems, including the Desktop Window Manager (DWM) and Client/Server Runtime (CSRSS). In short, all in all, the OSs made little difference. As a benchmark tool, Kennedy used a combination of the Clarity Studio's ADO (ActiveX Data Objects), MAPI (Messaging Application Programming Interface), and WMP (Windows Media Player) Stress workload objects, to simulate a complex multi-process-workload under Windows 7 consisting of client/server database, workflow, and streaming media playback tasks. Analysis of the system and process indicators collected during the test shows that, in terms of performance, Windows 7 M3 is a virtual twin of Vista (Kennedy, 2008).

In their thesis, Bienia & Li (2011), developed a method to design effective benchmark suites. They demonstrated the method’s effectiveness by deploying it on evaluating multiprocessors, using a parallel benchmark suite called PARSEC for evaluating modern shared-memory multiprocessors. The tool has been adopted by many architecture groups in both research and industry (Bienia & Li, 2011).

In 2017, Microsoft planned to release ARM-based laptops, powered by Qualcomm's Snapdragon 835 CPU, from different Original Equipment Manufacturers (OEMs), including ASUS, Lenovo, and HP ("Benchmarks for ASUS laptop powered by Windows 10 on ARM leaked," 2017). Microsoft, in collaboration with Qualcomm Technologies Inc, released the Windows 10 laptops built around Qualcomm's Snapdragon 835 processor (CPU). The two companies promised that battery life will not have any negative impact on performance. However, a benchmark test using Geekbench comparing the Qualcomm processor (CLS) to an Intel Core i3 attributes a single-core score of 1,202, to CLS and a score of 3,692 to the Intel processor. For multicore results, the CLS received 4,068, while the Intel processor got an 11,860 score ("Leaked benchmarks for new Microsoft Windows 10 laptops show meager performance," 2017), which is three times the performance in both cases. Before the hardware was released, the early benchmark score had leaked. It is not certain whether the benchmark test includes an ARM emulator, which can help the CLS processor work faster. The used OS was Windows 10 Pro (32-bit), a 4096 MB memory, and a base frequency of 2.21 GHz.

Figure 13: UserBenchmark Testing Results Visualization for MS Windows 10 Home

Microsoft Windows 10 Pro 10.0.17763.

According to Novabench results on Microsoft Windows 10, Pro 10.0.17763 CPU is better than the average of the people’s systems (CPU peak number is 264). However, the GPU seems to be a bit below the average. (The lowest number is 154). The test results of Novabench on Windows pro show that its speed is higher than that of Prime95. The higher the number, the faster the speed in Novabench. In prime95, the smaller the time value, the higher the performance of the computer. The test also showed that the RAM speed in the Novabench test is relatively faster than the RAM speed in the prime95 test.

Figure 14: Novabench Testing Results Visualization for MS Windows 10 Pro

Figure 15: Novabench Testing Results for MS Windows 10 Pro