Linpack Xtreme Released

[Solid]

Sorry, forgot about ThrottleStop. Yes, just run the Windows, for a few hours. No need to fry the laptop for 12 hours.

thank you, I will try again from -130mV for 10GB of Ram and i will see

I will post some screenshots then, I'm repeating the test with -130mv and 35.000 equations (problem size at 10 GB), locating the laptop over a base with a fan. At the moment, no problems but it is still early.
The curious thing I'd like to ask to you @EarthDog and @Regeneration is: how is it possible that TS Bench built in Throttlestop gives tons of errors trying at 1024M with -130mV, but I'm going well (so far - 10 runs) with the equations of Linpack Xtreme??

This means that I shouldn't trust on TS Bench because it can give "false positives" or what??

PS: temperatures are around 75, 80°C but just because the system hits 97°C and start to throttle to reduce temps, with Thermals, VR Current , PL1, EDP OTHERS and PL2 alarms activated

 

[Regeneration]

thank you, I will try again from -130mV for 10GB of Ram and i will see

I will post some screenshots then, I'm repeating the test with -130mv and 35.000 equations (problem size at 10 GB), locating the laptop over a base with a fan. At the moment, no problems but it is still early.
The curious thing I'd like to ask to you @EarthDog and @Regeneration is: how is it possible that TS Bench built in Throttlestop gives tons of errors trying at 1024M with -130mV, but I'm going well (so far - 10 runs) with the equations of Linpack Xtreme??

This means that I shouldn't trust on TS Bench because it can give "false positives" or what??

PS: temperatures are around 75, 80°C but just because the system hits 97°C and start to throttle to reduce temps, with Thermals, VR Current , PL1, EDP OTHERS and PL2 alarms activated

I'm not familiar with TS Bench so I can't comment on it. I know the developer (@unclewebb) of ThrottleStop is also an active member on this forum.

It is possible the errors caused by memory or memory controller. But yet again, you should look at the time, rather then amount of runs.

 

[EarthDog]

Thermals, VR Current , PL1, EDP OTHERS and PL2 alarms activated

Even undervolted... wowzas. Hate to say I called that when using this app.

You shouldn't be throttling when running a stress test at stock speeds. That isn't testing much of anything except the thorttled clocks and voltage.
Your system simply can't handle what it is delivering. You are pinging hard against the wall on stock settings setting off multiple throttle reasons (thermals, VR current, Power limit 1/2, and EDP) even while undervolted. If that doesn't tell you to drop this program (and most likely a lot of laptop users)... not sure what does.

 

[Solid]

Even undervolted... wowzas. Hate to say I called that when using this app.

You shouldn't be throttling when running a stress test at stock speeds. That isn't testing much of anything except the thorttled clocks and voltage.
Your system simply can't handle what it is delivering. You are pinging hard against the wall on stock settings setting off multiple throttle reasons (thermals, VR current, Power limit 1/2, and EDP) even while undervolted. If that doesn't tell you to drop this program (and most likely a lot of laptop users)... not sure what does.

Yes, Dell XPS 9570, like predecessors, is a beautiful machine but it suffers high temperature like macbook pro, which hits also TJunctons (100°) while under stress. This is the reason why a lot of XPS users need to undervolt their pc.

 

[EarthDog]

Yes, Dell XPS 9570, like predecessors, is a beautiful machine but it suffers high temperature like macbook pro, which hits also TJunctons (100°) while under stress. This is the reason why a lot of XPS users need to undervolt their pc.

So why bother with a stress test that sets off all kinds of other throttling flags? That isn't right. Temps, fine... we know laptops generally run toasty... but its setting off a few other flags not related to temps and you are UNDERVOLTED!

Have you tried AIDA64 default stress test? Does it do the same thing (throw off several codes)? It may very well do that... but worth checking.

 

[Solid]

I'm not familiar with TS Bench so I can't comment on it. I know the developer (@unclewebb) of ThrottleStop is also an active member on this forum.

It is possible the errors caused by memory or memory controller. But yet again, you should look at the time, rather then amount of runs.

Yes I will leave it for 3 hours at least, i hope to talk with @unclewebb soon because it is an interesting question to understand at what TS Bench looks for... It sounds very odd, like the laptop answers 2+2=5 at TS Bench but it can solve thousands of equations on Linpack Xtreme.

So why bother with a stress test that sets off all kinds of other throttling flags? That isn't right. Temps, fine... we know laptops generally run toasty... but its setting off a few other flags not related to temps.

Have you tried AIDA64 default stress test? Does it do the same thing (throw off several codes)? It may very well do that... but worth checking.

i tried Aida and it stresses less the cpu. I'm trying with Linpack because of the incongruence with TS Bench.
I know that TS Bench isn't the best stress test available but how is it possible that at 1024M it gives a lot of errors at -130mV and under Linpack it can solve complex equations without errors (at least on calculation)?
In my case it's worth to undervolt the laptop because, doing it, my pc can gain around 15% of performance also it still throttles, like Notebookcheck said here: https://www.notebookcheck.net/Dell-XPS-15-9570-15-more-performance-by-undervolting.317738.0.html and how i can confirm using Cinebench R15

 

[EarthDog]

Ive been in computing and overclocking and benchmarking and stress testing for decades. Don't chase after tests like this unless the work you are doing on your laptop specifically requires such heavy heavy 'stability'.

Undervolting is worth it... nobody is denying that. But stress testing it across a few applications and comparing one to the other is living in the minutia. Yes, AIDA does stress it less in default form. If you only check of CPU and FPU, it will be in the ballpark (though still likely less than this). Or, Try Realbench stress test... another good one with real world applications.

That said, You'll get nowhere with Linpack since its hammering your system and bringing up multiple throttling flags. It is clear that for laptops, this seemingly can't be run.......at least on your laptop as due to it smashing temp limits and other throttling flags due to its inherent nature of 'overkill'.

 

[Solid]

but its setting off a few other flags not related to temps and you are UNDERVOLTED!

All of this flags are related to temps.. VR CURRENT i think relates to VMR Mosfets and they are voltage regulators which become too hot due the high temp, for example.

 

[EarthDog]

All of this flags are related to temps..

AFAIK, they are not related to temps outside of the thermal one.

VR Current is the CURRENT going to the VRMs, yes. They are likely hot when trying to run this test, but the flag is for CURRENT. There is a relationship between current and temps, I get that, but we are seeing it tripping for CURRENT not VR temps (which it may be doing? But not sure throttlestop does that? Not sure).

Power Limit 1/2 is the CPU crossing the default PL thresholds. Also not temp related.

EDP - I had to look that up and found Unclewebb himself explain what it is... "If at full load, ThrottleStop Limit Reasons shows EDP lighting up in red, that confirms that the CPU is throttling because it has reached the Current Limit." (LINK)

4/5 errors you listed have nothing (or little) to do with temperatures.

 

[Solid]

Ive been in computing and overclocking and benchmarking and stress testing for decades. Don't chase after tests like this unless the work you are doing on your laptop specifically requires such heavy heavy 'stability'.

Undervolting is worth it... nobody is denying that. But stress testing it across a few applications and comparing one to the other is living in the minutia. Yes, AIDA does stress it less in default form. If you only check of CPU and FPU, it will be in the ballpark (though still likely less than this). Or, Try Realbench stress test... another good one with real world applications.

That said, You'll get nowhere with Linpack since its hammering your system and bringing up multiple throttling flags. It is clear that for laptops, this seemingly can't be run.......at least on your laptop as due to it smashing temp limits and other throttling flags due to its inherent nature of 'overkill'.

Ok I understood what you are saying but it is really interesting the question I asked before: how can be possible that 2+2=5 in TS Bench 1024M with lots of errors at -130mV (and 0 errors with 0 offset) and passing at the same time a lot of equations on Linpack (at -130mV)... maybe only @unclewebb can answer to this.

 

[EarthDog]

Different tests testing different things/methods is my guess. One may test more or less RAM or test it in a different manner. This is why I suggest finding a more simple test (TS is fine, if it works for you) and sticking to it as it is quite common to be stable in one app, but not in another. Chasing ghosts this way.

The point of stress testing is to find one that works for you and your needs. And in the case of laptops, or at least your laptop, lol, Linpack is too stressful causing a whole bunch of flags to trip causing you to test lower than what would be tested in the real world.

 

[Solid]

AFAIK, they are not related to temps outside of the thermal one.

VR Current is the CURRENT going to the VRMs, yes. They are likely hot when trying to run this test, but the flag is for CURRENT. There is a relationship between current and temps, I get that, but we are seeing it tripping for CURRENT not VR temps (which it may be doing? But not sure throttlestop does that? Not sure).

Power Limit 1/2 is the CPU crossing the default PL thresholds. Also not temp related.

EDP - I had to look that up and found Unclewebb himself explain what it is... "If at full load, ThrottleStop Limit Reasons shows EDP lighting up in red, that confirms that the CPU is throttling because it has reached the Current Limit." (LINK)

4/5 errors you listed have nothing (or little) to do with temperatures.

So, I'm not sure about what VR current and PL1/2 mean if they start to blink when the laptop is under stress and the temperature hits above 90°. Anyway I started to search another stress test app because I saw so many errors with TS Bench also if i try to undervolt the laptop of -115mv and TS bench is considered not very stressful as stress test.
I would like to know if it can generate also "false positives" because otherwise i get confused.

 

[EarthDog]

So, I'm not sure about what VR current and PL1/2 mean if they start to blink when the laptop is under stress and the temperature hits above 90°.

It means the system is doing its job in regulating, power, temperature, and current trying to get below set thresholds. PL1/2 has nothing to do with temps... VR Current, nor apparently EDS either. You can get a PL/current limit flag with temps WELL under control.

I'll let you catch up to my posts... LOL.

 

[Solid]

I'm trying to understand as well as i can, because I'm not an english speaking user

Anyway, so far, no errors, no bsods, no freezes... with 10GB (35000 problems), if I had tried 30GB (62897 problem size) I would have got freezes or bsods for sure!! bah...

and it's been 2 and a half hours... I can say that my system is stable with Linpack Xtreme at 10GB (35000) with -130mV but NOT at 30 GB (62897) and not with TS Bench!

 

[EarthDog]

I can say that my system is stable

You can say its stable at the throttled values... that is all. This currently isn't testing the speeds and voltage the laptop will run outside of this stress test.

Edit: you should not be throttling when stress testing.

 

[Regeneration]

I'm trying to understand as well as i can, because I'm not an english speaking user

Anyway, so far, no errors, no bsods, no freezes... with 10GB (35000 problems), if I had tried 30GB (62897 problem size) I would have got freezes or bsods for sure!! bah...

and it's been 2 and a half hours... I can say that my system is stable with Linpack Xtreme at 10GB (35000) with -130mV but NOT at 30 GB (62897) and not with TS Bench!

Your processor is throttling and therefore the results are useless. You'll need to use something that produces less heat.

You're testing -130mV on 2893 MHz, you need to test -130mV on the highest clock.

Or try Linpack with the lowest amount of RAM to produce less heat. The 32-bit version allows 256MB and 512MB.

 

[Solid]

Sorry but of course the pc throttles.. because it has a i7 8750h inside a thin case.. anyway the pc wasn’t in throttling for the entire test, it alternated 2,893 ghz with the 3,9 ghz full speed... I don’t have any problems if pc isn’t under stress. When I tried with 30GB the pc crashed also if it was throttling.

Cinebench R15 produces less heat with around 80° C at 3,9 ghz and the pc never crashes.

 

[toyo]

The situation was never very good with laptops and cooling/OC. The 8700K which should be quite similar with your CPU is drawing 130W at stock (4.3GHz) under stress if you remove the power limit. Let's assume that the 8750H is much better binned and can run at way lower voltage, and you're still gonna look at a realistic 70W+ that need to be dissipated. The OEMs lock power limits for a reason, as they understand how flimsy their thermal solutions are.

I would just run it at stock, and yeah sure, undervolt it, but don't expect to be able to run the usual stress tests to guarantee stability. Can try running an older non AVX prime95 or AIDA64, but i fear you'll run into either a power or thermal limit anyway. Could also try to buy a laptop cooler for it.

 

[EarthDog]

anyway the pc wasn’t in throttling for the entire test, it alternated 2,893 ghz with the 3,9 ghz full speed...

this is expected... the PC throttles and bring it down, then loosens up once its cooler or doesnt hit PL/current limits. When it hits the throttling points again, it clocks down again. Normal. That and this isnt the same load throughout testing AFAIK.

As we said, you need to be running a stress test which doesn't throttle your PC. This isnt doing the job for you.

 

[Solid]

this is expected... the PC throttles and bring it down, then loosens up once its cooler or doesnt hit PL/current limits. When it hits the throttling points again, it clocks down again. Normal. That and this isnt the same load throughout testing AFAIK.

As we said, you need to be running a stress test which doesn't throttle your PC. This isnt doing the job for you.

I don’t think a stress test app which doesn’t throttle a Dell XPS exists... maybe you are used to look at desktop pc or laptop which are thick and heavy.. but it isn’t the case of Dell XPS.. look the reviews. Undervolting needs to be done to reduce the throttling time, not to eliminate it, anyway a difference must be if with 30GB of ram the pc crashes and with 10GB not. When I undervolt it I obtain 15% more in performance (cinebench R15). Anyway if I reduce the stress of course the pc doesn’t crash... my doubts regarded just the TS Bench and why I get errors

Have you ever had a MacBook Pro? What to say? This Dell is also a good step compared to it.. A MacBook Pro hits 100°C while it runs Cinebench R15 for some cores.

 

[mouacyk]

After glimpsing Linpack Extreme shoot my temps to 95C at 5GHz on all cores in the stress test, I've fallen back to more practical testing. That is -- actually looping encodes and compiles for several hours, which top out in the low 70's.

I do run some AVX workloads, but the GPU is enabled for those and can do it much quicker. If you must use AVX/AVX2 for quality encoding, then that's what you have to do and test it -- although, there's a reason for the AVX offset to maintain a balance between temperature and throughput.

 

[er557]

As you can see, your overclock is a bit excessive, and your cooling is insufficient, especially if as you say it took a short time to reach those temps.
But then again, everyone has it's own view of what is stable enough

 

[Solid]

I'm trying to undervolt a notebook, i78750H isn't overclockable.
I wanted to know from @unclewebb why Ts Bench gives me a lot of errors at 1024M till -113mV, why Linpack Xtreme freezes my laptop when I use 30 Gb of Ram if I undervolt it also of -105mV but I can go without problems with 10GB also undervolting of -130mV.
Just to understand how much I can undervolt this machine.

 

[freeagent]

Its a pretty heavy duty app, I cant say I am entirely surprised to see systems like Dells not being able to run it at full capacity. I was actually considering getting an XPS all done up it was such a good deal, especially with a 1070ti, even though you have to wait about 30 days for shipping vs 5 if you chose the 1060, 70, or 80. But I looked at the internal shots, and wasn't convinced.

 

[unclewebb]

why TS Bench gives me a lot of errors at 1024M

The TS Bench is a 32 bit benchmark which only uses x86 code so it does not use any AVX instructions. Basically it is testing a completely different part of the CPU compared to where the 64 bit AVX instructions are running when Linpack testing.

Interesting that the TS Bench needs slightly more voltage to run reliably compared to Linpack testing. These are two completely different tests so when you think about it, slightly different voltage requirements should not be too surprising. A couple of other users have reported that some quick TS Bench testing has helped them get their CPU voltage in the ball park a lot quicker compared to the more popular AVX based tests.

The TS Bench does millions of calculations, stops, and then does the exact same calculations again. The CPU cores are usually a little hotter second time around. If it finds two different answers to the same question, it reports this as an error. Any mathematical calculation should return the exact same answer no matter how many times you run the test, no matter what core the calculation is run on or where in memory the answer is saved. A 100% stable computer should be able to pass any of these tests.

You might also want to try doing some stability testing with turbo boost disabled. The 8750H runs at only 2.20 GHz when Intel Turbo Boost is disabled so its offset voltage requirement at this speed could be significantly different compared to when it is running fully loaded at 3.90 GHz. An offset voltage that passes full load, full speed testing, might be inadequate when running fully loaded at only 2.20 GHz. Some single thread stability testing is also important. This will allow the 8750H to hits its maximum speed of 4.10 GHz. Watch the reported MHz in ThrottleStop closely since it is one of the few monitoring apps that reports single thread MHz testing like this correctly.

I know that TS Bench isn't the best stress test available...

That is kind of an insult.
Your testing has shown that the TS Bench can detect errors quicker compared to Linpack testing. It also tests parts of your CPU that more popular testing software does not. That sounds like a pretty good test to me.