
I have a few CPU reviews on here and I also have a few articles on overclocking CPUs. In all of these I refer to certain benchmarks such as SuperPI and the RealTemp XS Score. But how do various CPU compare in these? Well here’s a short comparison to help you read the other articles in a clearer light…
Introduction
It’s really of little use having me tell you that my latest Q9550 E0 stepping CPU can complete the SuperPI 1M benchmark in around 16 seconds if you have no reference point for how good or bad this is. Sure you can google around and find various comparisons but it would be much nicer to be able to find a graph here on mods-n-clocks wouldn’t it?
I have access to a variety of PCs running a variety of CPUs and the example of 1M SuperPI only takes a few seconds to complete so I have decided to test all the different PCs that I stumble across to see how they compare including my own rigs of course!
SuperPI 1 Million Benchmark
As a quick recap for those few who aren’t yet familiar with SuperPI. The SuperPI benchmark application uses a single core of the CPU to calculate the value of PI as quickly as possible to a pre-set number of decimal places. There are a variety of tests available in the software menu depending on how many decimal places you want to use. The most common test in the PC enthusiast community is the 1Million decimal place test and that is what I am using here. This is a good test for estimating the raw power of the CPU for single threaded requirements
SuperPI 1M (Lower is better)
Excellent, I hope you found your CPU amongst all that or at least found the chart to be of interest! I don’t really want to comment too much on the various results, obviously it’s important to note that the lower the value the better after all these are in seconds! Also it may be worth pointing out that the AMD CPUs seen here are cheap, low end and low power CPUs and so it should not be read from this graph that Intel beat AMD or anything like that.
RealTemp XS Score
This benchmark is only available when using a modern Intel CPU, so unlike the previous graph I do not have any AMD entries or Pentium 4 entries which is a shame. It is however another of those tests that can be run very quickly and doesn’t need to be installed onto the computer tested. I don’t however know much about what the test is actually doing to achieve it’s scores.
RealTemp XS Score (Greater is better)
Again, I hope this is useful. Let me know with a comment if it is. It’s a pity that it only supports recent Intel CPUs as this is a test that normalises against the result of a certain CPU, in this case the E8400 @ 3GHz obtains a score of 1000.
7-Zip Benchmark
The 7-Zip benchmark tests the PC’s ability to archive files into a compressed archive. It’s a new benchmark on mods-n-clocks so you won’t find it in any of the earlier reviews but I plan on using it in reviews going forward and also when displaying overclocked performance of CPUs. There are two charts for this test, the first is compression and the second is decompression.
This test takes advantage of the CPU’s multiple cores (if available) especially in the decompression test and as it stores the data to archive in system memory it is not disk access bound as you might expect from an archiving process. In reality compressing to archive and extracting out again will not achieve results mirroring these as data will need to be continually written and read from disk. However this test does give a more real-life, multi-core view of CPU performance than some synthetic benchmarks.
To obtain the results on this test I run the test once for 5 passes and then I restart and run for 5 more passes. This makes sure that the results are well averaged (5 passes) and that the data is fully loaded into memory (restart) so that disk access speeds cannot affect the performance. I use the default settings.
7-Zip data speed – Compression (Greater is better)
Decompression is a quicker process than compression so as we can see the KB/sec speeds are significantly higher across the board in the second test.
7-Zip data speed – Decompression (Greater is better)
This is a more useful benchmark as it shows what performance gains you might expect from one CPU to another in an application that you might actually run for real. Archiving data into a compressed archive remains a very useful process especially if you plan to send the data across the internet or have it available for download.
Conclusion
There are some interesting results here I think. Especially the battle between the higher clocked E7400 and the lower clocked but more parallel (more cores) Q9400. In tests that benefit from the additional cores like the 7-zip decompression we see the Q9400 miles ahead but in the entirely single threaded tests such as SuperPI and XS Score the E7400 pips it into second place.
Another interesting point to note is that the AMD 4850e lags behind the Intel T6670, T7200 and E2180 in the 7-zip compression test but it moves ahead of those three CPUs significantly in the decompression test. All these CPUs are dual core so it shouldn’t make sense on the face of it that in the most multithreaded test the AMD should do better. But we all know that AMD’s and Intel’s implementation of multi-core processing is quite different, i7 brings them a bit closer with Intel bringing the memory controller on board, but comparing the Core2 family to the X2 family there are definite differences.
Anyway, I hope you’ve enjoyed this article and found it useful – it should certainly help with reading some of the reviews that you find here on site and help you make comparisons between CPUs. As I have mentioned already this is an ongoing article and I’ll update it when I come across additional CPUs that I want to add.
Popularity: 20% [?]


Add A Comment