Jonathan Reichelt said:
...If you only understand overclocking from what you have read on the internet please do not bother replying to this post, I only want people who honestly know what they are talking about and have a lot of experience. Just because you made some minor changes in your Bios doesnt make you a pro at OCing, my problem cannot be fixed by changes in bios this is a problem with windows (unless someone who knows what they are talking about tells me otherwise).Thankyou for your help.
Its clear that windws is preventing the overclocking from being registered and factory settings are being reset when the OC is on, yet windows makes no changes when OC is not on and cpu is underclocked.
I'm sure more people would really like to help, JR, but perhaps they just don't like replying to questions from people who don't have "a lot of experience" to already know the answers.
Your conclusions as to what is "clearly" happening may be based on a false premise...but we can't know this for sure because you haven't told us whether your "7 benchmarking programs" tests consisted of
1) simply looking at reported frequencies or
2) actual benchmarking metrics. In other words...
1) Whether you are simply looking at the various acquired/reported CPU clocking parameters.
e.g. when CPU is NOT overclocked CPUz reports X.YZ GHz and when it IS overclocked CPUz reports the same
X.YZ GHz. By the way, CPUz and similar programs are NOT "benchmarking" programs. CPUz for example is a parameter reporting/analysis program that depends mostly on information provided to it directly by the system. Even a program that provides actual benchmarking tools, may
however in the clock/frequency reporting sub-section show the same data, even though actual benchmark results show OCing has in fact taken place. A number of factors can affect the data acquired and reported by programs like CPUz, including BIOS, Windows, power management, state of the CPU (e.g. whether frequencies are throttled-down and/or voltage-limited / protected by built-in CPU mechanisms or external programs), etc. Such factors would likely impact the parameter reports from similar Windows-based programs. However
, depending on the exact algorithms they use some might produce different numbers. And, if a particular program uses data the other programs ignore (perhaps by using system calls that are not considered the "Microsoft approved" method of obtaining the data, or
by executing some quick internal benchmark and then calculating
a clock frequency based on some assumptions
) it could possibly report a more real-world value...even if it did so by means intended to get around reporting limitations previously mentioned.
2) Whether you've actually performed benchmark tests and compared the execution times and determined there is no significant difference before and after OCing.
e.g. when NOT overclocked prime95*
takes X seconds to perform a suite of CPU and/or RAM-targeted tasks, but when it IS overclocked prime95 still
takes ~X seconds to perform the same suite of tasks. Presumably if your OC settings are being "reset (to default) by Windows" then before and after OCing performance under Windows would be virtually identical. If however they were substantially different then you know at least some, and perhaps all, of your OC settings were retained and not "reset by Windows". It would then be a parameter acquisition/reporting
issue...not a "Windows is resetting things" problem.
For example, if before you OC the CPU and/or RAM you run a Windows benchmark and it says it took 1000ms to execute that routine, and you then OC the CPU and/or RAM (by whatever means needed) and the benchmark program says it now takes only 700ms to execute the same
routine under the same
conditions, then clearly substantial OCing is taking place...regardless of what CPUz and other simple parametric reporting programs or sub-sections might be reporting to you
prime95 is used as just one example of a true Windows-based benchmarking program, and not a program that simply acquires and reports parametric data provided by the system.