AMD Ryzen 1600

Nothing at all wrong with using Prime95 to test stability and thermal limits. It's about HOW you do it that matters. Versions of Prime95 OTHER THAN version 26.6 are not suitable for this purpose unless you are planning to run applications that extensively use the AVX instruction set.

Prime95 v26.6 is THE primarily accepted way to do the majority of baseline stability and thermal limit testing running the Small FFT option.

Prime95 version 26.6: http://windows-downloads-center.blogspot.com/2011/04/prime95-266.html


Further, as explained to me and many others on this forum by Computronix, who has far more experience with CPU architectures and testing procedures than 95% of the people you will ever meet, speak to or read about. He is also the author of the Intel temperature guide, found here:

http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html



This pretty well sums things up and is equally relevant whether working with an Intel or an AMD system.

I can think of several reasons why x264 encoding or AVX / AVX2 / FMA3 apps won't work as a unilateral metric for thermal testing.

(1) A steady-state workload gives steady-state temperatures; encoding does not.

(2) Simplicity in methodology; most users would find encoding apps unfamiliar and cumbersome to accomplish a simple task.

(3) Most users such as gamers never run any apps which use AVX / FMA, so adaptive or manual voltage aside, it makes no sense to downgrade your overclock to accommodate those loads and temps.

(4) Standardization; Prime95 has been around since 1996; many users are familiar with it.

For the minority of users who routinely run AVX / FMA apps, then P95 v28.5 can be useful tweaking BIOS for thermal and stability testing.


Regardless of architecture P95 v26.6 works equally well across all platforms. Steady-state is the key. How can anyone extrapolate accurate Core temperatures from workloads that fluctuate like a bad day on the Stock Market?

I'm aware of 5 utilities with steady-state workloads. In order of load level they are:

(1) P95 v26.6 - Small FFT's
(2) HeavyLoad - Stress CPU
(3) FurMark - CPU Burner
(4) Intel Processor Diagnostic Tool - CPU Load
(5) AIDA64 - Tools - System Stability Test - Stress CPU

AIDA64's Stress CPU fails to load any overclocked / overvolted CPU to get anywhere TDP, and is therefore useless, except for giving naive users a sense of false security because their temps are so low.

HeavyLoad is the closest alternative. Temps and watts are within 3% of Small FFT's.


Computronix

Author - Intel temperature guide

Click to expand...

Click to expand...

The "Blend" test uses FFTs with sizes which range from very small to very large. The FFTs are not run in-place so the tests cycle through RAM. The blend test automatically picks a "Memory to use" value which uses up all the RAM in the computer. This is the version of the test which is most likely to fail if you have bad RAM. Note that on some computers running some versions of Prime95, the blend test allocates more memory than can fit into the physical RAM in the computer. When that happens, Windows has to use virtual memory and the test ends up accessing the hard disk a lot. That ruins the test because the CPU isn't doing anything while it is waiting for the hard disk to access the data whereas the test is really supposed to be keeping the CPU busy at all times in order to stress it. You can tell that the test isn't working properly if the hard disk activity light is on most of the time. When the blend test is running properly, the hard disk is only accessed occasionally. If you're running Windows 2000 or Windows XP then you can also check whether the CPU is fully utilized in the task manager. If the blend test has allocated too much memory then stop the test and bring up the "Run a Torture Test" window and make sure "Blend" is selected. Then select "Custom", reduce the value in "Memory to use (in MB)" to a somewhat smaller value, and then run the test again. Continue until you find a small enough value which works properly.

Click to expand...

Since everyone tests their rigs using X stress software at Y Ambient temperatures with Z measuring utilities resulting in CPU or Package or Core temperatures, it's impossible to compare apples to apples. This is why processor temperatures are so confusing.

There are only three relevant values; Ambient, steady-state 100% workload, and dead idle. Applications and games are partial workloads with fluctuating temperatures, which are unsuitable for thermal testing or accurate temperature comparisons.

Prime95 Small FFT's is the standard for CPU thermal testing, because it's a steady-state 100% workload. This is the test that Real Temp uses to test sensors. The link above is to version 26.6, which is well suited to all Core i and Core 2 variants.

Prime95's default test, Blend, is a cyclic workload for testing memory stability, and Large FFT's combines CPU and memory tests. As such, Blend and Large FFT's both have cyclic workloads which are unsuitable for CPU thermal testing.

Other stability tests such as Linpack and Intel Burn Test have cycles that load all registers with all one's, which is equivalent to a 110% workload, and are also unsuitable for CPU thermal testing. The software utility OCCT runs elements of Linpack and Prime95.

Click to expand...