Novabench Cpu Score
Sandra Grady
The overall score - the largest number at the top of the results page - is an impression of your computer's performance relative to other computers scored by Novabench. It is the sum of weighted sub-scores for each major part of your computer: CPU, GPU, RAM, and Disk.
A Novabench Score has no upper limit - a higher score is better. Scores are designed to be linearly comparable; a CPU that has a score of 2x another is roughly twice as fast, for the common workloads that Novabench tests.
Under the Online Comparison tab of your test result, quick links are provided to search the Novabench online parts database for your CPU and GPU. The search results will show you the average Novabench score for your part. Click the part to open it on Novabench.com, for more detailed information.
The part page shows the distribution of benchmark scores for a this CPU or GPU. Many factors influence how the part performs. A part may test below average due to thermal or power limits, high activity by other running programs, or misconfiguration. The higher end of the curve may be overclocked parts. Examine the score distribution for your part, and see how your score fits within it. If there's room for improvement, read our guide to tuning your CPU and GPU.
The percentile ranking of your CPU or GPU indicates how many tests results for the same part fell below yours. For example, if your CPU scores at the 99th percentile, 99% of the results for the same CPU are lower than years. A percentile ranking is also given for all parts benchmarked in the same year that your part was released. This is helpful for comparing your part to its contemporaries, to assess the relative performance and longevity of your hardware.
For your CPU and GPU, a box plot is shown for the recorded temperature ranges of your CPU and GPU, as tested by other users, for high and low loads. The box represents the interquartile range (the middle 50% of temperatures), while the line inside the box represents the median temperature. The "whiskers" (lines extending from the box) indicate the overall range of temperatures, excluding outliers.
Your system's temperatures: Your system's temperature readings are displayed above the box plot. This allows you to easily compare your temperature against the overall range of temperatures. If the temperature displayed is above the median, it may indicate an issue with your cooling solution. Conversely, if the temperature displayed is closer to the lower whisker of the box plot or beyond, it suggests that your system is well-cooled.
Remember, temperature is a crucial aspect of system performance. Overheating can lead to reduced component lifespan, throttled performance, and in extreme cases, system failure. If your system's temperatures are consistently high, it may be worth exploring more efficient cooling solutions or investigating potential issues with your current setup.
Now, what about performance tests between the two? First I wanted to see how the external and internal internet connectivity was performing, so no big surprise, IONOS way outperformed Azure by a factor of 3, which is to be expected given the infrastructure back end design running on InfiniBand and the datacentre interconnects.
The Azure instance had a low score for its CPU benchmark which makes sense as the CPU is a shared resource with other instances being hosted on that Hyper-V cluster node within the Azure cloud, the RAM score was also low with a throughput of 3929 MB/s, but what was noticeable was that the disk read performance was good with a throughput of 163 MB/s but write speeds were a complete polar opposite.
The CPU performance was 385% that of the CPU in Azure and for Azure to achieve a similar score an additional 3 CPUs would have to be added to maintain the same CPU score. The RAM speed also was way beyond that of Azure and achieved 19318 MB/s a factor of 3 times faster, the disk read & write performance both outperformed Azure, it did maintain an equal throughput for both write and read speeds with writes outperforming by 18 times that of Azure. Just a note here that I used a standard HDD as the storage medium and could have used an SSD instead which would have increased the performance even more.
Finally, I configured another instance in IONOS Enterprise Cloud using an AMD Opteron 62xx 2.8Ghz processor to see it that could match the Intel-based Azure instance and for much of the benchmark scores it was comparable to the Azure instance, even better the cost of the instance was 31.52 a month giving a saving 368.16 over the year. It should be mentioned that IONOS Enterprise Clouds let you configure cores and storage at will in the most granular way possible: core by core and Gigabyte by Gigabyte.
For Azure to catch up to similar performance of that of IONOS Enterprise Cloud the Azure instance would need to be reconfigured to a A4_v2 size this is 4 times the resources of the IONOS Instance which would increase the monthly cost to 182.44 which would equate to 2210.64 for the year of which 1599.12 would be for the cost of an equal performance instance of that of the IONOS instance.
Can you really justify that type of expense of spending an additional 1600 per year for the same performance? IONOS Enterprise Cloud employs KVM based virtualisation making extensive use of hardware virtualisation and maps the CPU power of a real core to a vCPU and provides dedicated memory so it is surely the way to go.
As described in (1 2) other posts, I was looking at VirtualBox 5.2.6 and VMware Workstation Player 14.1.1 as possible ways to run certain software in a Windows 7 guest virtual machine (VM) on a Windows 10 host computer. The question considered in this post is whether one had visibly better performance than the other.
This was a rough, first-cut exploration of numerous factors related to that question. As with other long posts in this blog, I may write a more concise version at some point in the future, if I revisit this question. Another post does update some aspects of VM benchmarking.
This post explores benchmarking programs, to find the best way of comparing VirtualBox and VMware. It turned out that the best benchmarkers required unspecified Windows updates. So there was a choice of whether to install a lot of updates, which might themselves affect performance, or just settle for the relatively limited benchmarkers that would run without Windows updates.
Windows updates could have a performance impact by imposing clutter; by increasing exposure to Windows 7 instabilities, introduced by Microsoft starting in 2016, to push users toward Windows 10; and by requiring antivirus software, which itself could impose a major performance hit. It seemed that I might avoid the need for antivirus software by keeping my VMs offline. That could work: I only needed these VMs to run programs like Adobe Premiere Elements and Microsoft Office. Those did not need Internet access.
Nonetheless, I did go ahead, download and install the full supply of Windows updates, and run a variety of benchmarking programs. I found that most of these programs were component-oriented (measuring the performance of e.g., CPU or RAM), but PCMark10 provided a nice alternative, with a set of eight practical elements (e.g., spreadsheets, photo editing). On a machine with the requisite updates, I found that PassMark PerformanceTest 9, SiSoftware Sandra Lite 2017, and CrystalMark all provided relatively sophisticated general-purpose comparisons, though CrystalMark was only 32-bit.
I took a brief look at the subjective difference between host and VM, but decided not to pursue that. Specifically, there were early signs that video editing in Adobe Premiere Elements would feel somewhat slower in a VM, but it would take hours of experience before I could offer more specific comments on that.
I also took a brief look at the question of how many CPUs to allocate to the VM. It tentatively appeared that the best approach would be to assign half of the total number of CPUs available (e.g., four of eight) to the VM. Some contended that it was possible to get better performance without instability, at least in VirtualBox, by assigning all CPUs to the VM and setting them at a maximum 80-90% execution cap.
My impression, based on this and other recent explorations, plus several mostly satisfactory years of using VMware Workstation circa 2009, was that VMware Player would provide a more stable, organized, and user-friendly feel, while VirtualBox offered more options, more complexity, and probably somewhat faster performance.
The computer used for this performance comparison was running Windows 10 Pro x64 on an ASUSTeK H97-PLUS motherboard with an Intel Core i7-4790, with 24GB DDR3 RAM and a 4GB NVIDIA GeForce GT 730. The Windows 10 installation was on an M.2 SSD; the VMs were on a PNY CS1311 SSD.
I made similar performance adjustments on each VM. These included two CPU cores and 8GB RAM allocated; 256MB graphics RAM; 2D and 3D acceleration, if available; no other significant tasks running on the host; antivirus on the host configured not to scan VM program or storage directories; and other settings described in another post. I verified that VirtualBox had Guest Additions installed, and that VMware had VMware Tools installed.
In this comparison, I was interested in general-purpose, real-life performance. I was not a gamer. Video editing tended to be my most demanding task. I was also not an academic researcher, concerned with precise and reproducible results. To the contrary, I expected to be using my chosen VM in a system with other stuff running, so I did not feel particularly compelled to develop a test machine that would eliminate all other processes. Rather, I wanted to see VM performance in a system where assorted things would always be going on. So I had various programs open (e.g., web browsers, Microsoft Excel), although not demanding significant system resources (e.g., not doing calculations) nor changing their demands (e.g., not starting or closing programs) while these tests were underway.
ff7609af8f