Need BENCHMARK

[󠀡󠀡]

Which Is Best To Stress?

[󠀡󠀡]

https://store.steampowered.com/app/223850/3DMark/

Prob Go With This

[Bad 💀 Motha]

Nothing that runs through Steam. Here I will link a few...


https://benchmarks.ul.com/3dmark

https://benchmarks.ul.com/pcmark10

https://benchmark.unigine.com/superposition

https://www.maxon.net/en/downloads/cinebench-2024-downloads

https://www.aida64.com


Use the Free Edition or Demo first. Some have a limited time trial, such as AIDA64

[Tonepoet]

I honestly don't know what makes a good benchmark. However, as many of them are free you can just try them all.


I think furmark[geeks3d.com] is usually considered the gold standard of stress testing a G.P.U., although I've also seen stories of it damaging components. Pushes power draw higher than you'll realistically encounter while playing games. It's used as the basis of M.S.I. Kombustor[geeks3d.com], which is also a popular tool.

Uniengine has a few benchmarking tools, the latest of which is Superposition[benchmark.unigine.com].

Ezbench is another interesting test meant to stress test G.P.Us.:

https://store.steampowered.com/app/770170/EzBench_Benchmark/


Shadow of the Tombraider has a free trial version with access to the in-game benchmark:

https://store.steampowered.com/app/750920/Shadow_of_the_Tomb_Raider_Definitive_Edition/

It's useful for comparative analysis since it has proven a rather popular game to benchmark over the years.

However, the Black Myth Wukong Benchmark tool is probably more useful since that game was released just this year rather than five years ago.

https://store.steampowered.com/app/3132990/Black_Myth_Wukong_Benchmark_Tool/

[Illusion of Progress]

Benchmarks are benchmarks, not stress tests. While, yes, they can be fairly stressing, I'd caution against limiting stress testing efforts to just benchmarks.

There's different ways to stress different parts of the PC. At the end of the day, different applications and different use cases will load a system in different ways, and even good stress tests might not expose an instability where one exists. So it's best to use a range of stress tests (and even benchmarks), along with real world use.

There's a saying that "a system is only stable until it's not" and what that means is exactly what it says. You can run a system through 10 different stress tests and 10 different benchmarks and 50 different heavy games, but then if one game consistently crashes it, assuming it's not a software issue with the game itself, then you just found out you have an unstable system. Whether it's unstable enough to matter to you is a different question.

For the CPU (and memory subsystem to a lighter degree), Prime95 is a good one. It performs a lot of calculations, and checks the result against what it should be, so a mismatch will indicate an unstable CPU.

For the RAM, Memtest86 is one, but IIRC this one more so tests if the RAM is faulty or not faulty, and there's better ones to more specifically test if a given RAM configuration is entirely stable. This probably only matters more if you're tweaking RAM though (meaning, manually tuning frequencies, timings, voltages, etc.).

For the GPU, it's not as stressful as it once was, but I think Furmark is still pretty valid. I think some RMA/service centers still test with Prime95 plus Furmark (and some others).

OCCT is a good one because it's a suite of many different stress tests. The free version has a 5 or 10 second wait before each test, and a limit of 30 minutes per test (no cooldown so you can start it again).

CPU-Z and Aida 64 aren't all that heavy so I wouldn't put much worth in those (though I do like both anyway for other purposes, but their stress testing use is limited IMO). With Aida64 specifically, I think one or two of the tests (the FPU one is one of them I think) are pretty fair, but the rest are medium at best. CPU-Z's benchmark is ludicrously bad, and it's stress test isn't that good either, but it's fine for its hardware overview purposes.

There's also other ways to stress a system. Stressing stability and stressing cooling capability are two different things, for example, so some applications that are not all useful for one of those things, might be very useful for the other.

I read this a few years ago and it goes further into things.

https://www.tomshardware.com/reviews/stress-test-cpu-pc-guide,5461.html

TL;DR: Use as many as you want, because the more you try, the better. Some are time wasters more than others (since they poorly stress the system), but the less you limit your testing, the better your answer will generally be. There is no single inconclusive application that will answer the question of stability. "It's only stable until you've find something that demonstrates it is not." Therefore, try as much stress testing as you're willing to spend time on, and then just use it normally and let real world results speak.

[Bad 💀 Motha]

If you want a pure stress-test; use Prime95 (small FFTs loop test) + MSI Kombuster and run them at the same time for about 30 mins to ensure stability and to test your cooling.

Make sure you have a good app open and visible during testing so you can monitor CPU + GPU temps. If need be, once the Temp Monitoring software is running, set its Priority to Real-Time so it doesn't lose focus during stress-tests.

[DonMcK]

This: https://www.ocbase.com/occt/personal

[󠀡󠀡]

Thanks Guys!!