Is IBT recommended to stress test ryzen refresh?

[gasolin]

I have a ryzen 2600 and a Aorus x470 gaming 7 wifi mb, there was a new bios that i installed and ran prime 95 that was stable at 4ghz, ibt 2.54 was a whole different story.

It froze,graphic artifacts, vcore up and down,cpu and soc llc adjusted, ez oc of the ram, xmp profile ram manually set ram to 1.350volt 1.360 volt
after ibt x10 maximum with alot of graphic artifacts, black screen, freeze for many seconds you name it, i finally got it to finish ibt at maximum.

Is IBT recommed to stress test ryzen refresh?

What is recommeded top stresstest ryzen refresh with?

 

[phanbuey]

IBT is always good... sisoft sandra multimedia bench, Aida 64 stress test (although you have to leave that one for hours)...

Honestly what usually craps out with ryzen OCs (in my experience) is the memory - thats much harder to test, and some of my 'STABLE' OCs (after hours of stress tests) would crash after like day 20 of uptime on a development workstation.

So my way of testing stability is usually a quick 20 min stress test, about 10 cinebench runs while watching tv, and then a few hours of gaming. Then I just leave the rig on for a week or two, and if it stays up and keeps benching/gaming then it is stable.

 

[Caring1]

Intel Burn Test
There's the clue

 

[phanbuey]

Intel Burn Test
There's the clue

It's just linpack... it will load the FPU on a ryzen as well.

 

[TheoneandonlyMrK]

It's just linpack... it will load the FPU on a ryzen as well.

Well why not just run linpack then??

 

[hat]

I recommend a series of tests. Sometimes things will crash in test A but pass test B. I don't like that, I want my system to be rock solid. LinX and Prime95 are good, as well as memtest86+ for memory.

 

[gasolin]

I had my cpu at 4ghz 1.200 volt llc medium, could play all games no problems, 3200mhz cl 14 ram run at 3200mhz cl 14
It's basically stable.

Prime 95 100% stable, ibt i got alot of unstability,graphic artifacts freezing......

For some reason i think my cpu fans never really kicks in since it dead quiet at 100% and 75c, fan are running and set to cpu in the bios and should go above 50% at 65c 100% around 80c

Now at 3.9ghz it sort of newer really get's hot doing prime 95 small fft and passed ibt but not without probelms (not 100% stable but passed) it also passed aida64 system stability cpu and ram 30-50 min .

 

[MrGenius]

If it'll pass IBT on High it's pretty much guaranteed to be stable for any game or benchmark. And more stable than it needs to be for many games and benchmarks. IME. The term "stable" is relative though. Nothing is guaranteed 100% stable. If a random cosmic ray shoots through your system and flips the right/wrong bit...you're crashing whether you like it or not.

Well why not just run linpack then??

Because LINPACK isn't a GUI application that you can run with a couple clicks. Or let you select different levels of stress(test with various amounts of RAM usage). And there's the AVX thing. Which I think is what right clicking the start button on IBT does(Xtreme Stress Mode). Anyway, whatever it is, LINPACK probably doesn't do it. I dunno for sure though. Maybe I'm wrong. I've never run LINPACK on its own. Wouldn't even know how. Ain't got time for that.

 

[TheoneandonlyMrK]

If it'll pass IBT on High it's pretty much guaranteed to be stable for any game or benchmark. And more stable than it needs to be for many games and benchmarks. IME. The term "stable" is relative though. Nothing is guaranteed 100% stable. If a random cosmic ray shoots through your system and flips the right/wrong bit...you're crashing whether you like it or not.
Because LINPACK isn't a GUI application that you can run with a couple clicks. Or let you select different levels of stress(test with various amounts of RAM usage). And there's the AVX thing. Which I think is what right clicking the start button on IBT does(Xtreme Stress Mode). Anyway, whatever it is, LINPACK probably doesn't do it. I dunno for sure though. Maybe I'm wrong. I've never run LINPACK on its own. Wouldn't even know how. Ain't got time for that.

Strange so , it's both just linpack And it does extra shit especially for Intels AVX extensions hmm ,which is implemented by Amd differently,, and that is (Ibt)there fore part wrote for and definitely compiled for intel by intel, yeh i see no possible avenues for an issue their.

 

[phanbuey]

Strange so , it's both just linpack And it does extra shit especially for Intels AVX extensions hmm ,which is implemented by Amd differently,, and that is (Ibt)there fore part wrote for and definitely compiled for intel by intel, yeh i see no possible avenues for an issue their.

It was written by AgentGod, and it is a graphical gui on Linpack that:

You can take your intel tinfoil hat off now. Prime95 has Intel AVX as well, as does RealBench, and Sandra and Aida64. Hell - Timespy has an AVX feature...

 

[crazyeyesreaper]

Aida64 Stress CPU / FPU / Cache / Memory, Battlefield 4 multiplayer / Battlefield 1 multiplayer. Typically Aida64 if its stable for a few hours means it should be okay to test other applications. Typically if something is not quite right BF4 / BF1 will crash or freeze or stutter etc. With Various Memory OCs and testing those two games will typically show a problem pretty quickly.

 

[basco]

http://www.xtremesystems.org/forums/showthread.php?201670-LinX-A-simple-Linpack-interface

 

[Johan45]

I use Prime 95 for my testing and it hasn't failed me yet, Custom 128 FFT in-place to hammer the CPU and then custom blend with 75-80% ram usage selected, after a couple of hours start unigine heaven on a loop this really tests the whole system for stability and the ability to control temps.

 

[Vayra86]

Strange so , it's both just linpack And it does extra shit especially for Intels AVX extensions hmm ,which is implemented by Amd differently,, and that is (Ibt)there fore part wrote for and definitely compiled for intel by intel, yeh i see no possible avenues for an issue their.

You do realize that many applications we use every day use AVX yes? And if they don't because they can't they either don't run or lose a crapload of performance. Stop the guesswork that leads you to weird conclusions based on just a name...

Its one of the reasons AVX offsets don't really do much too. Effectively it just means a lower clock. I also love how people say 'hurr durr 5.2 Ghz and I don't stress test on AVX loads because never used anyway'... yeah. Dream on

 

[Athlonite]

There's also linpac binaries specifically made for AMD CPU's

 

[newtekie1]

This is why I use OCCT. It has Prime95 and Linpack bundled in one nice interface.

 

[Vayra86]

This is why I use OCCT. It has Prime95 and Linpack bundled in one nice interface.

Amen

 

[Vya Domus]

You do realize that many applications we use every day use AVX yes?

I am really curios as to how you figured that must be the case or is it just an assumption as I suspect ? Though I am unaware of any statistics with regards to this , I have many reasons to believe AVX is sparingly used at best in commercial software.

You are simply not going to find much of it outside synthetic test. The ways in which you include AVX support in your software are cumbersome and involve using certain functionalities which are compiler specific which are generally avoided. Hell ,there is still a fair amount of software out there which doesn't even use the previous SSE extension , these are instructions present in all CPUs for something like a decade , SSE2 is pretty mainstream and that's close to being two decades old. And of course there is also the fact that you simply can't always make extensive use of it , in fact more often than not that's the case.

Metro used to crash on CPUs that didn't have SSE 4.1 , you can bet developers would like to avoid cases like these at all cost. They would much rather skip AVX which is a high difficulty/low gains task and ensure their game runs on a toaster.

 

[gasolin]

Update: Cpu fan speed was related to bios im now on a earlier bios that works

 

[crazyeyesreaper]

I am really curios as to how figured that must be the case or is it just an assumption as I suspect ? Though I am unaware of any statistics with regards to this , I have many reasons to believe AVX is sparingly used at best in commercial software.

Its being used more and more. Hell even PUBG makes use of AVX and that games a mess.

 

[Vayra86]

I am really curios as to how you figured that must be the case or is it just an assumption as I suspect ? Though I am unaware of any statistics with regards to this , I have many reasons to believe AVX is sparingly used at best in commercial software.

You are simply not going to find much of it outside synthetic test. The ways in which you include AVX support in your software are cumbersome and involve using certain functionalities which are compiler specific which are generally avoided. Hell ,there is still a fair amount of software out there which doesn't even use the previous SSE extension , these are instructions present in all CPUs for something like a decade , SSE2 is pretty mainstream and that's close to being two decades old. And of course there is also the fact that you simply can't always make extensive use of it , in fact more often than not that's the case.

Metro used to crash on CPUs that didn't have SSE 4.1 , you can bet developers would like to avoid cases like these at all cost. They would much rather skip AVX which is a high difficulty/low gains task and ensure their game runs on a toaster.

https://forums.anandtech.com/threads/avx2-and-fma3-in-games.2460033/

I don't say these things because I'm assuming them, if that were so, my OC would have been done with AVX offsets, which is not the case

Not having AVX or AVX2 means the CPU 'drops' to an instruction set that is available. Either SSE (if the application supports it) or lower AVX. Been like this for years already...

This is an interesting post on that forum to make things more concrete - AVX is useful for physics calculations, for example, an area where the CPU is the weakest link;

As I mentioned in my OP, PhysX 3.xx uses AVX for cloth solving, so it appears that SIMD is seeing a lot more use in the various physics engines than in the main game engine...​

​

This comes directly from one of the programmers of PhysX, who was in fact responding to one of my own posts on Anandtech concerning CPU and GPU PhysX..​

​

If anything, CPU Physics is on the up and up, as NVidia has started to throw a lot of resources into developing their PhysX 3.xx SDK, and it runs primarily on the CPU and is extremely fast.​

​

Witcher 3 has some of the best cloth physics in a game, and it runs completely on the CPU. A lot of the effects seen in current PhysX 3.xx titles are even better than what was seen in GPU accelerated PhysX titles 5 years ago.​

​

#33Carfax83, Jan 9, 2016

 

[Vya Domus]

Not having AVX or AVX2 means the CPU 'drops' to an instruction set that is available. Either SSE or lower AVX. Been like this for years already...

Doesn't work like that , AVX is an extension like everything else added to the original x86 ISA. There is no interchangeability that takes place , code either use AVX registers/instructions or not. If something that needs AVX runs on a CPU lacking said extension it will not fallback to SSE. You need explicit code paths for that , which goes back to my explanation as to why AVX support is cumbersome and generally avoided.

This is an interesting post on that forum to make things more concrete - AVX is useful for physics calculations, for example, an area where the CPU is the weakest link;

As I mentioned in my OP, PhysX 3.xx uses AVX for cloth solving, so it appears that SIMD is seeing a lot more use in the various physics engines than in the main game engine...​

​

This comes directly from one of the programmers of PhysX, who was in fact responding to one of my own posts on Anandtech concerning CPU and GPU PhysX..​

​

If anything, CPU Physics is on the up and up, as NVidia has started to throw a lot of resources into developing their PhysX 3.xx SDK, and it runs primarily on the CPU and is extremely fast.​

​

Witcher 3 has some of the best cloth physics in a game, and it runs completely on the CPU. A lot of the effects seen in current PhysX 3.xx titles are even better than what was seen in GPU accelerated PhysX titles 5 years ago.​

​

#33Carfax83, Jan 9, 2016

Sure it's useful for some stuff , the emphasis being on some.

I don't say these things because I'm assuming them, if that were so, my OC would have been done with AVX offsets, which is not the case

I suspect some developers have the uninspired idea to let the compilers auto vectorize code rather that manually optimizing it. This explains why sometimes the AVX offset is triggered. If games really had proper support for AVX you would see quite a huge gap in performance that can't be explained by the drop in frequency alone when this occurs yet that doesn't seem be the case. Then again , these "AVX offset triggers" don't seem to always occur for everyone under the same situations , so it might just be a poor implementation on Intel's part.

 

[R-T-B]

Well why not just run linpack then??

Sure lemme just compile it... oh wait.

I mean yeah there are binaries, this is just one of the easier to use ones.

 

[eidairaman1]

Ryzen Blender, Unigen Heaven/Valley, 7Zip Max Thread Compression/Extraction

 

[TheoneandonlyMrK]

You do realize that many applications we use every day use AVX yes? And if they don't because they can't they either don't run or lose a crapload of performance. Stop the guesswork that leads you to weird conclusions based on just a name...

Its one of the reasons AVX offsets don't really do much too. Effectively it just means a lower clock. I also love how people say 'hurr durr 5.2 Ghz and I don't stress test on AVX loads because never used anyway'... yeah. Dream on

To you and others , i realise more than you realise i do.

I am of the opinion though ,that this old program adapted by(for) intel , on an intel compiler with a new Gui could contain glitchy code(ie the gui has code running on the same processor that's benching to display stuff ,with a guess i would say Java, it might not be the test causing issues directly, but issues do not imply reliability of the test or its minor or major fails more importantly, in this use case imho, im not a well experienced coder, more occasional but thinking logically about stuff this seam's a reasonable hypothesis) still.
I said that because the OP was using this software to test his CPU ,and was getting graphical artefacts?? Wtaf.

Now I am not a conspirational sort and I am not claiming to know anything for sure.....

But if i got a brand new cpu of any type or brand ,then ran code on it to test just it ,if i got gpu artefacts i would find something else to test with personally ,regardless of what some might say.

There are too many choices to use, to be holding Ibt precious, it isn't.

@phanbuey Agent is god eh , fine but he wrote it before Ryzen existed , and I get your point but i have experience of plenty of old software that doesn't work right after some time or has issues with a particular system/platform.
I implied it could be a slight issue with the software only ,graphical corruption is not common imho while benching CPUs.