cancel
Showing results for 
Search instead for 
Did you mean: 

Is your Win10 SkyLake system stable? Simple test.

MMikhail
Level 9
Are you sure that your system is stable under all conditions ?

Please try the following test:

1. Set one of default win10 screensavers (preferably animated, eg bubbles, 3D text) to activate in 2 minutes
2. Run RealBench stress test for 30 minutes
3. Relax a little

After 2 minutes ScreenSaver should start, but nothing should happen, or be very very slow.

Now watch, if the animation picks up speed, RealBench has aborted or finished.

Check the results.

I have tried it on two PC's, both fail within 10 minutes 😞
But can pass RealBench stress test for hours if SS is disabled.
4,331 Views
11 REPLIES 11

xeromist
Moderator
Do you succeed on stock clocks? Unless you can directly attribute this failure to an overclock it's not going to help you tune for stability. You may just be experiencing some other conflict.
A bus station is where a bus stops. A train station is where a train stops. On my desk, I have a work station…

No, it;s not an over clock, the behaviour is the same on OC, on Stock and on under clock.

xeromist
Moderator
Sounds like that might just be a software conflict due to some of the resources that Realbench uses. If those resources are tied up by the screensaver then it could certainly cause a halt. You don't really want your screensaver taking resources when benchmarking anyway so unless you are experiencing stability issues elsewhere I wouldn't worry about the stability of the two combined.
A bus station is where a bus stops. A train station is where a train stops. On my desk, I have a work station…

Praz
Level 13
Hello

This is a non-issue. With the subsystems that are tested with RealBench this is the outcome I would expect.

JustinThyme
Level 13
I have to agree with Praz on this.
Real bench will most likely abort when you consider moving the mouse aborts it. Its not at all usual to benchmark a PC with the screen saver enabled and I would expect the same to happen on any system and any benchmark. Your clue should be the fact you can run for hours on end without the SS.



“Two things are infinite: the universe and human stupidity, I'm not sure about the former” ~ Albert Einstein

JustinThyme wrote:
I have to agree with Praz on this.
Real bench will most likely abort when you consider moving the mouse aborts it. Its not at all usual to benchmark a PC with the screen saver enabled and I would expect the same to happen on any system and any benchmark. Your clue should be the fact you can run for hours on end without the SS.


Please do not confuse benchmark and stress test loop. Benchmark measure performance, and yes, it will abort if you move mouse or something odd happens since the results will be distorted. But the stress test loop is doing something different and what it checks for apart from some crc's is a mystery as well as it's reports that don't tell you any details as to why it decided there is a problem.
As for the Win10 ScreenSaver engine, there is something really odd going on there, since even selecting the simplest blank screen version apparently requires a lot of cpu power, you can actually see the mouse pointer going busy for a very long time during SS start, which in theory should use almost no cpu time.
My current guess is that Win10 SS engine does something with timers when it is activated which results is RB confusion and fail.

MMikhail wrote:
Please do not confuse benchmark and stress test loop. Benchmark measure performance, and yes, it will abort if you move mouse or something odd happens since the results will be distorted. But the stress test loop is doing something different and what it checks for apart from some crc's is a mystery as well as it's reports that don't tell you any details as to why it decided there is a problem.
As for the Win10 ScreenSaver engine, there is something really odd going on there, since even selecting the simplest blank screen version apparently requires a lot of cpu power, you can actually see the mouse pointer going busy for a very long time during SS start, which in theory should use almost no cpu time.
My current guess is that Win10 SS engine does something with timers when it is activated which results is RB confusion and fail.



Its not a mystery. Folks like me don't buy into the black box theory, or P.F.M. (pure effin magic)
Benchmarks and stress tests do pretty much the same operation, only the benchmark is taking a measurement of the performance while doing so.

What you have is simply a conflict of instructions.
SS are old and done unless you are one of the few that is still using a CRT. Some screen savers actually use more CPU power than average PC tasks, particularly the 3D animations.

If you want to call your system unstable because of software glitches that were never meant to run together, knock yourself out. The rest of us know that to get accurate stress test results and benchmarks you don't intentionally set it up to end your test with other software and disable or suspend other back ground operations. Ending a test is not unstable, post up screen shots of what message you get, maybe we can help you figure out exactly what is happening. I can say with a level of absolute certainty that is not an unstable system if it runs fine so long as you don't enable the SS.



“Two things are infinite: the universe and human stupidity, I'm not sure about the former” ~ Albert Einstein

I like SS to run when computer is not busy, it's more a tradition, beauty thing.
Unless it's something cpu / gpu intensive it should not interfere with background tasks anyway.
But apparently windows has it's own mind and does some house keeping and run services in background when in SS mode,
the opposite of upcoming 'gaming mode' when resources are freed, and normal operation.

Any ways, just completed my tests with RealBench 2.53b and everything is looking peaceful, nothing brakes and survived overnight torcher test.
So something has changed in the new RB version and it is not giving false instability warnings, or is missing something that 2.44 found.
Still, the update is nice and brings in some nice features like build in temperature monitoring.

chevell65
Level 12
Do you have a link for the 2.53b download?