Jump to content
HWBOT Community Forums

CPU benchmarks applicable for global points - 2019


Which CPU benchmarks should get global points?  

129 members have voted

  1. 1. Which CPU benchmarks should still get global points in 2019?


This poll is closed to new votes


Recommended Posts

To bring this back to benchmarks rather than ridiculous ad hominems;

For those who don't know, one of the ways chips fail is electromigration.  This is a phenomenon where the physical movement of electrons causes damage over time to the conductor.  Because this is a physical effect, it's entirely possible that a connection could be brought past the brink while cold but not actually physically disconnect until the slight expansion associated with warming back up happens.  This would appear to fit with what people have reported (I'd be really interested to hear the results of refreezing one of these 'walking dead' chips, in the event someone ends up with spare LN2).

Electromigration can be predicted according to an equation called Black's Equation.  Several of the factors are unknown to us mere mortals (though probably estimated/studied internally at Intel etc) but it does tell us that reducing the temperature gives a longer time to failure (but not indefinite extension with 'only' ln2), and increased current gives a shorter time to failure.

If the weak link that's failing is in the power delivery within the chip (or the package, I suppose), then the current draw by the entire core would be what matters.  That would mean something like whycruncher, XTU or x265 that pretty much maxes it out, hitting all the execution units most of the time (AVX2 is a good way to do this, I'd have to look up architectural info to know if it's possible without AVX2, it certainly could be), is going to be a worst case scenario for chip life because the current going into the cores is way higher for the same voltage and clocks.  It's also possible that x265 happens to hit different, weaker bits of Intel's chips than cinebench does.

However, you can just compensate for this with less voltage.  Technically semiconductors don't follow ohm's law, but in practice if you reduce voltage, current reduces roughly in proportion (buildzoid did a study demonstrating this).

The conclusion is that you absolutely can run x265 and other super-heavy benchmarks safely, just use less voltage to do so unless it's a suicide run.  There's no problem with the software, it's just a heavier load than people are used to and maybe requires that you back off on voltage more than people realise.

Regarding this original comment;

Quote

I'm sorry to say this because the developer works hard on it but x265 4k is a cpu killer and should be disabled for globals. I'm not talking randomly either, if you run it on ln2 with a modern cpu there is a greater than 50% chance it won't work the next day

You can absolutely run it on LN2, you just need to rethink the voltage.  Normal "safe" voltages may not apply.  Fundamentally, there is not some CPU instruction invoked by x265 that causes chip death only if you're under -100C.  It's just a heavier load, causing more stress than you're used to for the same voltage (probably very close to the maximum stress possible), whereas the "classic" benchmarks are clearly a long way from maximum stress.

An interesting example of this that's actually documented, by the way, is The Stilt's "strictly technical" analysis of PBO limits on Pinnacle Ridge.  AMD program a much higher boost voltage for single-core loads, not only way way above what's programmed for multi-core but also way above what's safe for 24/7 all-core use even well within safe temperatures (this isn't just a theory, people have tested safe voltages - 1.425V@60C degrades a 2700X noticeably over ~3 months).  It's not about keeping a lid on thermals, because the boost algorithm controls for them separately.  It'll be some part of the chip that can withstand the current from one core at the higher voltage, but not from multiple.

  • Like 4
Link to comment
Share on other sites

How somebody dont know that more voltage = more current on same benchmark? But with difference benchmark like with CPU-Z your current is lower than Prime95 at the same voltage. Voltage x Current = Power Draw (By simple, not including efficiency, power factor, etc.). Current kill cpu, thats why you dont use Prime95 under LN2.

  • Like 3
Link to comment
Share on other sites

  • Administrators

Finally something reasonable...

 

P.S You have to rethink the voltage - sure, you are expert so you should know that each chip has own vid, sacling and so on, one chip dies at 1,65, other live at 1,8v. And if you have no sign of this on LN2 ( on air you see heat, benchmark fail and so on) but the benchmark runs 24 hours (exaggereted) without signs of weakness, it is really hard to find volts that work and don´t kill aka find dead chip next day. In the end, this means you toss a coin or you don´t bench it on LN2 at all. This is the problem we talk about.

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

On Ryzen 7 1700X im not sure if it has some protection, but my chip can run cinebench scaling positively until 1.85-1.9v only, while Alva cpu can run 2v cinebench. If i put my cpu at 1.95v it will freeze while running it, or can run but with lower clock. But for X265, my chip scaling positive only until 1.68v, after that it freeze or still can run but with lower clock, probably AMD has some internal protection or logic like Maximum OCP to protect it.

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

3 hours ago, websmile said:

Finally something reasonable...

 

P.S You have to rethink the voltage - sure, you are expert so you should know that each chip has own vid, sacling and so on, one chip dies at 1,65, other live at 1,8v. And if you have no sign of this on LN2 ( on air you see heat, benchmark fail and so on) but the benchmark runs 24 hours (exaggereted) without signs of weakness, it is really hard to find volts that work and don´t kill aka find dead chip next day. In the end, this means you toss a coin or you don´t bench it on LN2 at all. This is the problem we talk about.

Yes you are just supposed to arbitrarily stop at a mythical voltage limit which is the square root of cotton candy divided by unicorns, then you will be safe. How do we even know vcore is the issue and not system agent or io voltage etc? We don't know and will never know. 

  • Like 2
  • Thanks 1
Link to comment
Share on other sites

I do wonder if Blender could be implemented as a Benchmark. It's agnostic to Intel/AMD. Would need a smaller test than Gooseberry. Do a HWBot logo or something along those lines.

I think there's GPU acceleration available. Might be a multifaceted Benchmark if it's secure from tampering.

  • Like 1
Link to comment
Share on other sites

20 hours ago, mickulty said:

To bring this back to benchmarks rather than ridiculous ad hominems;

For those who don't know, one of the ways chips fail is electromigration.  This is a phenomenon where the physical movement of electrons causes damage over time to the conductor.  Because this is a physical effect, it's entirely possible that a connection could be brought past the brink while cold but not actually physically disconnect until the slight expansion associated with warming back up happens.  This would appear to fit with what people have reported (I'd be really interested to hear the results of refreezing one of these 'walking dead' chips, in the event someone ends up with spare LN2).

 

Hmmm.........?

I too believe what's happening is related to the same basic effect of a chip that just dies one day after it's been ran hard awhile on LN2, but in this case the effect has been amplified to a ridiculous extent due to the thermal differences between what would be the hottest and coldest part(s) of the chip while running the bench on LN2.

Part(s) of the chip gets really hot during a run so that these part(s) of it expands thermally, the other part(s) remains very cool in comparision and has much less expansion creating severe thermal stress cracks in the die itself.

It's making me think the silicon die is just cracking up in the literal sense, staying together well enough while it's still being cooled to operate but once the entire chip warms back up and expands thermally, that's the point of death.

I'd also think how newer chips (Intel's) are being made today is a major factor with this - I mean we all know by now the PCB and other parts of these newer chips are thinner than older ones which makes them inherently weaker by comparison too.

Could be the effects of this are finally showing and the  X265 - 4K test ran under LN2 cooling is simply revealing it in spades vs older, thicker chips. I'm also not aware of AMD making their chips any thinner like Intel has been doing with the last few gens of chips made at least.

Maybe someone could actually refreeze one and see what happens. If it comes back to life frozen, then dies again once back at room temp that would more or less confirm it in the practical sense but in turn that would depend on how badly it cracked up in the first place to even have a chance of coming back, frozen or otherwise.

TBH I don't "Know" but it's a thought - Maybe someone could refreeze a chip and see guys.

Edited by Bones
  • Like 5
Link to comment
Share on other sites

8 hours ago, Splave said:

Yes you are just supposed to arbitrarily stop at a mythical voltage limit which is the square root of cotton candy divided by unicorns, then you will be safe. How do we even know vcore is the issue and not system agent or io voltage etc? We don't know and will never know. 

People run at hard memory settings at ambient and we don't hear of deaths from that, and LN2 should be safer at the same voltage, so there's no need for the mud of the memory controller to cloud the waters.

You should be able to set current limits in bios per intel datasheets - 193A for 8c coffee, 138A for 6-core (-K SKUs - some lower power bins have slightly less tolerance), 100A for 4-core.  These are specified on page 124 of the 8th gen core family datasheet, volume 1.  Above those limits, chips are going to be damaged - that's the nature of extreme overclocking.

As far as temperature scaling, it's a long time since I've done this kind of 'pure' maths but I think after filling two sides of A5 with rearrangements and simplifications of Black's equation the "safe" current (IE same MTTF as at the max specified by Intel)............................ does still ultimately depend on unknown factors.  I tried (even thought I'd nailed down a proportional relationship to e^(1/t) before I checked my work and found a mistake lol - it's actually proportional to (e^(1/T))^x where x is some unknown constant depending on the material properties, so fat lot of good that is).  But honestly, that's life.  It will be more but might be 0.00001A more, it all depends on the actual material properties of Intel's process, specifically of the aspect used for whatever the weak point is.

Of course, this stuff about setting current limits is all assuming that electromigration is THE explanation.  It could as bones said be related to mechanical stress from differences in thermals in different parts of the chip or substrate.  This would also have the same possibility of brief "reanimation" when refrozen.  It could even be a combination of both.

What we do know for sure is it's not x265's fault if Intel chips get really weak at true 100% load, and you no longer have roll-over there to hold your hand.

Link to comment
Share on other sites

I'm really starting to believe it's a side-effect of Intel making the chips thinner period.

Normally not a problem since 99.9999999% of them would never be ran frozen/Extreme conditions but for the select few that are, it's becoming a real issue.

  • Like 3
Link to comment
Share on other sites

9 hours ago, mickulty said:

Of course, this stuff about setting current limits is all assuming that electromigration is THE explanation.  It could as bones said be related to mechanical stress from differences in thermals in different parts of the chip or substrate.  This would also have the same possibility of brief "reanimation" when refrozen.  It could even be a combination of both.

What we do know for sure is it's not x265's fault if Intel chips get really weak at true 100% load, and you no longer have roll-over there to hold your hand.

^ This. + the now thinner processors as mentioned. 

__________

Electronic repairs in pcb's. The product turned on heated, turned off rapid cooled. This repeated action results in brittle solder joints, fractures.

 

Link to comment
Share on other sites

19 minutes ago, bigblock990 said:

LOL at everyone thinking thin cpu's is the problem. 9900k has thicker pcb and much thicker die than 7700k / 8700k. Unfortunately 9900k dies exactly same way as all the others on x265 4k.

When was the discussion re specific 9900K O.o Chat in here could be any modern processor from the last few years.

My experience on X265 4K ... is bench runs along time & requires additional Vcore to other benches I've done. Don't like it, whereas the 1080p completes fast.

Borrowed fasttrack writing below. Looks sound to me explanation

V=IR

Volts steady ( assumption ).
R falling as temps rise.
To keep a 3 parameter equation valid, with 1 steady parameter and 1 declining, means the 3 parameter ( in this case "I" = current ) has to rise
in order for the equation to be valid.
The problem with XTU and X265 4K is not the volts needed - neither the cooling method.
It is the brutal rise and fall of temps, which drives the "I" parameter nuts, because it has to follow these brutal changes up and down.
Silicon and millions of transistors ARE DESIGNED to keep up with these. But at stock settings or mild overclocks.
NOT record breaking runs.

_________

I do feel sorry for you LN2 guys losing nice processors, more lately. It's a big costing + the really great 7Ghz processors are gone forever :/

Give out less points for that bench, then you members are less attracted to run it

Link to comment
Share on other sites

1 hour ago, bigblock990 said:

LOL at everyone thinking thin cpu's is the problem. 9900k has thicker pcb and much thicker die than 7700k / 8700k. Unfortunately 9900k dies exactly same way as all the others on x265 4k.

Someone else said it's an issue of the smaller lithography not holding up as well, not the thinness of the chips and that could be it too.

As said earlier I don't really "Know", just trying to come up with a possible answer here that may help.  

 

Link to comment
Share on other sites

2 hours ago, avalanche said:

When was the discussion re specific 9900K O.o Chat in here could be any modern processor from the last few years.

My experience on X265 4K ... is bench runs along time & requires additional Vcore to other benches I've done. Don't like it, whereas the 1080p completes fast.

Borrowed fasttrack writing below. Looks sound to me explanation

V=IR

Volts steady ( assumption ).
R falling as temps rise.
To keep a 3 parameter equation valid, with 1 steady parameter and 1 declining, means the 3 parameter ( in this case "I" = current ) has to rise
in order for the equation to be valid.
The problem with XTU and X265 4K is not the volts needed - neither the cooling method.
It is the brutal rise and fall of temps, which drives the "I" parameter nuts, because it has to follow these brutal changes up and down.
Silicon and millions of transistors ARE DESIGNED to keep up with these. But at stock settings or mild overclocks.
NOT record breaking runs.

_________

I do feel sorry for you LN2 guys losing nice processors, more lately. It's a big costing + the really great 7Ghz processors are gone forever :/

Give out less points for that bench, then you members are less attracted to run it

 

55 minutes ago, Bones said:

Someone else said it's an issue of the smaller lithography not holding up as well, not the thinness of the chips and that could be it too.

As said earlier I don't really "Know", just trying to come up with a possible answer here that may help.  

 

Sorry guys, probably didn't need the LOL in my post. I read all the thin cpu comments as refering to thin pcb and thin/small die.

The lithography size theory could be plausible for sure. So much heat density. Too bad its impossible to test on older gens as they don't have avx.

For the record I don't care if x265 has points or not. I won't bench it either way. Its just not worth risking my hardware anymore. Too bad though, I actually really like the bench.

  • Like 2
  • Thanks 1
Link to comment
Share on other sites

23 hours ago, Splave said:

So the answer is to put amperage limits on my cpu while doing xoc....that sounds quite moronic no? 

It sounds counterintuitive that you can make your eyesight *better* by putting weird curvy bits of glass in front of your eyes as well, but you don't see me insisting to you that small books should just be got rid of and these "glasses" things are moronic.

Yes, there is an amount of stress that causes parts to fail.  A great way to stop them from failing is to limit the stress to below that amount.

  • Like 1
Link to comment
Share on other sites

1 hour ago, mickulty said:

It sounds counterintuitive that you can make your eyesight *better* by putting weird curvy bits of glass in front of your eyes as well, but you don't see me insisting to you that small books should just be got rid of and these "glasses" things are moronic.

Yes, there is an amount of stress that causes parts to fail.  A great way to stop them from failing is to limit the stress to below that amount.

LOL.....

So......did you know Splave is an Optometrist?  xD

  • Like 3
  • Thanks 1
Link to comment
Share on other sites

Folks,

Might sound irrelevant, not sure though if it is...

Did anyone else notice that only 71 members have voted ?

Is that a 0.00001% of active members or less ?

Does the community have no interest in poles ?

Is the community tired of multiple changes and changes and changes ?

Is the community not interested in Global points ?

Whatever the reason (s), the sample voting is so small, that I wonder if this pole can be seriously taken into account.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...