Is this RealBench Strong Enough?

Frizzo

Distinguished
Sep 17, 2010
599
0
19,010
I've generally used Prime95 to stress test custom Bios configurations on main components -- seems to be too much for a 5820k, though. Question is, (which was probably answered 2 years ago, I'm sure): Is this RealBench actually strong enough to stress test stability? It doesn't seem to be pushing the system all that hard? yes/no ?
 
Solution
I couldn't have said it better myself.

Agreed and endorsed.

Respectfully, as a blanket statement, that doesn't fit well.

The comments, conflicting information and confusion on the Internet concerning Prime95 are due to misinformation regarding the "AVX" problem, which is strictly version dependent. When reading about Prime95, if the differences between versions AND torture tests are not clearly and specifically pointed out, then you're reading partial or...
Hi Frizzo :)

If you can pass Realbench then your system is stable. Realbench is as good as it gets for a realworld stress tester.
Prime95 is harsh on the CPU and will push your CPU to its limit with no indication why your workers have stopped if a core fails. AIDA64 is another good stress tester and also tests your sub systems.
 

Frizzo

Distinguished
Sep 17, 2010
599
0
19,010
How long would you recommend running it (RealBench) for ? If you want to be 100% certain of system stability ? It has 15 minutes up to 8 hours, and I'd imagine you want to select 100% of your system's RAM as well.
 


Yes run the test with all your RAM and a couple of passes is a good indicator that your 24/7 stable. I have never run it for 8hrs nor any stress tester. I don't recommend unattended stress testing especially with P95.
Very often if you fail Realbench, slight increases in Core voltage (.005 steps) is all you need for better stability however, keep an eye on socket and core temperatures when under load using HWINFO64.
 

Frizzo

Distinguished
Sep 17, 2010
599
0
19,010
Wow, must be decent amount of bad information out there. I've got this 5820k on air ($30 hs/f) and it is maxing at 53C stress tested on RealBench.... Thank you for the replies.
 

CompuTronix

Intel Master
Moderator
I couldn't have said it better myself.

Agreed and endorsed.

Respectfully, as a blanket statement, that doesn't fit well.

The comments, conflicting information and confusion on the Internet concerning Prime95 are due to misinformation regarding the "AVX" problem, which is strictly version dependent. When reading about Prime95, if the differences between versions AND torture tests are not clearly and specifically pointed out, then you're reading partial or incomplete information written by those who are poorly informed, which is always misleading.

Intel tests their processors on an open bench, without a case, under carefully controlled conditions at steady 100% TDP (Thermal Design Power) to develop and validate Thermal Specifications. Therefore, when performing your own thermal tests, the goal is to run a test that will get you as close as possible to a steady 100% TDP workload so you'll have valid results.

Prime95 version 26.6 Small FFT's is ideal for CPU thermal testing, because it's a steady 100% workload with steady Core temperatures that typically runs Core i variants with Hyperthreading and Core 2 processors within +/- a few % of TDP. No other utility so closely replicates Intel's proprietary test conditions. This is also the utility that Real Temp uses to test Core temperature sensors.

• Prime95 v26.6 - http://www.mersenneforum.org/showthread.php?t=15504

Here’s a sample of utilities grouped for thermal testing and stability testing, shown according to % of TDP at Default BIOS settings:

TDP … Thermal Test - Steady Workload

129% … Prime95 v27.7 through v29.4 - Small FFT’s (AVX, No Offset)
101% <-- Prime95 v26.6 - Small FFT’s
89% … HeavyLoad v3.4.0.234 - Stress CPU
87% … FurMark v1.19.1.0 - CPU Burner
78% … CPU-Z v1.82.0 - Bench - Stress CPU
66% … AIDA64 v5.95.4500 - System Stability Test - Stress CPU
55% … Intel Processor Diagnostic Tool v4.0 - CPU Load

TDP … Stability Test - Fluctuating Workload (Peak)

123% … OCCT v4.5.1 - CPU: OCCT (AVX, No Offset)
118% … LinX v0.6.5 - Default
116% … IntelBurn Test v2.54 - High
113% … OCCT v4.5.1 - CPU: Linpack (AVX, No Offset)
110% … AIDA64 v5.95.4500 - System Stability Test - Stress FPU
99% <-- Asus RealBench v2.56 - Stress Test (AVX, No Offset)
94% … Sandra 2017.09.24.41 - Burn in - Processor Tests
92% … CineBench v15.0 - CPU - Render Test
79% … Intel Extreme Tuning Utility v6.4.1.15 - CPU Stress Test

All versions of Prime95 later than 26.6 run a "brutal" workload, which is due to AVX. 100% CPU Utilization seldom equals 100% workload or TDP, yet all tests will show 100% CPU Utilization in Windows Task Manager, regardless of actual workload. Higher TDP tests produce higher Core temperatures. Results will vary according to Microarchitecture, Core count, Cache, Core speed, Turbo Boost, Core voltage, Hyperthreading, Instruction Sets, BIOS versions and CPU microcode.

2nd through 8th Generation i3, i5 and i7 CPU's have AVX (Advanced Vector Extension) Instruction Sets. Prime95 versions later than 26.6 run AVX code on the CPU's Floating Point Unit (FPU) which causes unrealistic temperatures up to 20°C higher. Certain other utilities may have similar results, as shown above.

AVX can be disabled in Prime95 versions later than 26.6 by inserting "CpuSupportsAVX=0" into the "local.txt" file in Prime95's folder. However, since Core temperatures will be the same as 26.6, it's easier to just use 26.6. AVX doesn't affect Core i 1st Generation, Core 2, Pentium or Celeron processors since they don't have AVX Instruction Sets.

If you’re overclocked and run AVX apps such as for rendering or transcoding, you may need to reduce Vcore and Core speed or upgrade your cooler and case fans so Core temperatures don’t reach 85°C. Many 6th, 7th and 8th Generation motherboards address the AVX problem by providing offset adjustments in BIOS. An offset of -2 or -3 (200 or 300 MHz) is usually sufficient.

Asus RealBench runs a realistic AVX workload typically within +/- a few % of TDP, and is an excellent utility for testing overall system stability, whether you're overclocked or not. RealBench is also the utility that Silicon Lottery (yes, that's actually a real company - https://siliconlottery.com ) uses for testing CPU stability.

• Asus RealBench - http://rog.asus.com/rog-pro/realbench-v2-leaderboard/

Prime95's default test, Blend, is a fluctuating workload for testing memory stability, and Large FFT's combines CPU and memory tests. As such, Blend and Large FFT's both have fluctuating workloads which aren’t well suited for CPU thermal testing.

Other stability tests such as OCCT have cycles that exceed 120% workload, which again isn’t well suited for CPU thermal testing. However, OCCT will by default, terminate the CPU tests at 85°C.

The "Charts" in SpeedFan span 13 minutes, and show how each test creates distinct thermal signatures.

gaaHaa3.jpg


Shown above from left to right: Small FFT's, Blend, Linpack and Intel Burn Test.

Note the steady thermal signature of Small FFT's, which allows accurate measurements of Core temperatures. A steady 100% workload is crucial for thermal testing, as the CPU, cooler, socket, motherboard, VRM's and power delivery components can then heat soak and stabilize.

tlNeU3E.jpg


Shown above from left to right: Small FFT's, Intel Extreme Tuning Utility CPU Test, and AIDA64 CPU Test.

Intel Extreme Tuning Utility is also a fluctuating workload.

While I agree that AIDA64's is useful for stability testing, although the CPU test is a steady workload, it's too far below TDP for thermal testing at just 66%, and the FPU test is too far above TDP at 110%. All other AIDA64 CPU test combinations are fluctuating workloads, which are suitable for stability testing, but they don't offer conclusive results for thermal testing.

If you'd like to get yourselves up to speed on this topic, then please read this Sticky: Intel Temperature Guide http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html

CT :sol:
 
Solution
Jul 15, 2018
5
0
20

Thanks for all this info; I saw your write-up about it, too. You are talking about using RealBench's stress test as opposed to it's benchmarking test, right? For how long do you recommend running Realbench?