*Official* nVidia GTX 10xx Series notebook discussion thread

Then that's a systemic issue which Intel ought to have fixed. Going back to my jet engine analogy, it's like Boeing telling Delta/American/BA/Lufthansa/etc, saying 'Oh, we give you a throttle quadrant on your airliners, but please don't ever ramp them up to 100% as the engines will overheat and possibly catch fire'.

A CPU should absolutely be able to be cooled such that it doesn't thermal-throttle when under maximum load. No excuses.

 

The protection is called thermal/power throttling, there is a reason why its there. Please stop using Jet Engine as an analogy, its not accurate at all....
By all means, I think a CPU should be able to handle encoding/rendering etc. But P95 is starting to get unreasonable.
When do my overclocks, I dont expect my CPU to run P95 small FFT at all. When I said my CPU would have issues, I am talking about OCed... so I guess it doesnt apply here.

 

Is it normal to run into current limit throttling with small FFT? Ran prime95 for about 5 mins with a max temp of 78 using normal fan curve. My 6820hk is set to 4ghz but it only ran at 3.4ghz when using small FFT and drew around 74w package power for the entire duration.

 

what kinda voltage were u running at? have u ever seen higher voltages than during that run? have u seen higher wattages than 74W? could be a board limit that u hit when running p95

Sent from my Huawei Mate 8 NXT-AL10

 

Try raising your current limit to as high as it'll go. On haswell 256A+ is needed to kill this, but I don't know Skylake that intimately.

P95 is only THAT bad for Haswell and Broadwell though. Skylake doesn't generate NEARLY the heat nor consume as much power in the same situation. And as far as I know, neither does Sandy Bridge or Ivy Bridge.

 

Small FFT is basically a "power virus" like furmark. They stress the system in very specific ways that are designed to draw maximum power.

It's not really realistic, and I think the idea that Small FFT is the only way to test stability is now antiquated.

I definitely wouldn't run Small FFt's overnight anymore, and that used to be a recommendation for testing overclock stability way back when.

 

True, we all agreed that that's the best way to quickly kill a modern PC. We're talking about short runs of several minutes to test the cooling capability of systems here, not extended periods of torture to a brand new CPU.

 

Newer versions of Prime95 are basically power viruses in general - I only recommend version 26.6.

 

You guys are so cute, saying stuff - like it's woo-woo scary stuff - instead of as the factual stuff that everyone already knows it is, including the developers of prime95
https://en.m.wikipedia.org/wiki/Prime95

"Prime95 stresses a computer far more than the majority of software-based torture suites.

The nature of this is because the operating system usually shuts down the floating-point unit when unused by other programs, whereas Prime95 is well-optimized to continuously and effectively thread the FPU, causing it to be deeply pipelined, thereby generating significantly more heat because of elevated power consumption under the massive workload conditions.

In CPUs which are not adequately cooled, errors are likely to occur.

Prime95 also constantly accesses main memory at up to 60 MB per second. This constant activity will detect memory problems that other programs will not.

Lastly, power supply units of any machine running Prime95 are subject to the consistent ramifications of such harsh conditions.

Power must be maintained clean, while providing adequate voltage, particularly to the CPU, RAM, and chipsets (mainboard chipsets such as the Northbridge where the memory controller may or may not reside; see Athlon 64 or Intel Core i7 for on-die memory controllers) to provide peak performance while maintaining stability.

Cray Research used programs similar to Prime95 for over a decade for the purpose of stability testing."

I've been using it for decades, don't be afraid of it, prime95 is *not* a virus.

 

@hmscott

I don't really understand, since nothing I said was factually inaccurate. As a benchmark for stability, Small FFT is antiquated because it simulates unrealistic loads and draws unrealistic power through the CPU (much like furmark does through the GPU). Its operation is analogous to a power virus.

I was using Prime 95 more than a decade ago, too.

If you take umbrage with the way I said it "woo woo scary" [?], it should be noted that not only enthusiasts who know this stuff read this thread. I've been plugging NBR to casual laptop consumers many places for a long time. If someone stumbles across old information saying to run Small FFT for 12 hours on their new Aero 14 to test for "stability", then that's not going to be a good thing.

And Small FFT was absolutely *the* CPU stability test back in the day. Common overclocking enthusiast advice was to leave it running overnight.

 

Haha that would be the next Chernobyl right there right @D2 Ultima ?

Sent from my SM-G935F using Tapatalk

 

This is what a P95 small-FFTs run does to my CPU:

 

So that says you should turn on your fans 100% while running the test

Raise the rear exhaust area off the table, make sure there is a clear path for the hot exhaust to disperse without reflecting back off a wall or obstacle and back into your intakes.

Undervolt as well.

And, if you already did all that, then your laptop can't take the heat, likely a thinner laptop without heavy duty cooling.

That's why you run the test, to find out what your laptop / computer can do, and characterize it's capabilities and limitations.

Run other modes, 1/2/4/8 threads, downclock a bit to see if you can run 2/4/8 threads without thermal throttling - find that performance point for the CPU where the cooling can handle the heat output.

All fun stuff

 

What laptop? I can't read sigs now since I'm on Tapatalk.

Sent from my SM-G935F using Tapatalk

 

That's right, P95 is all about finding the limits of your power supply, CPU, cooling, etc, and then tuning things to match your laptops power and thermal capabilities.

No sense running an OC / or defaults that the laptop can't handle.

 

This is stuff said as woo-woo scary:

The word "virus" is a "scary" word for most people, computer virus or animal virus, it's not something they want. It's scares them off.

Prime95 is in no way a virus - you have a bad definition of the word "virus" in your head, go look it up.

Furmark has been "tamed", it's not woo-woo scary any more either, stop chasing people away from it.

I said you were presenting the information as if it was woo-woo scary instead of in a responsible factual frame.

I didn't say the information wasn't correct, but you presenting it as a "virus" or like "furmark" is totally wrong, and not factual.

That's it really, just saying please stop spreading FUD. Thanks.

 

Even if it's not a virus (adjectives can completely change the meaning of words, I think it is fair to call it a power virus), I still don't like running it. It literally meets the definition of a power virus: https://en.wikipedia.org/wiki/Power_virus

I don't run any Prime95 versions after 26.6 since they use AVX instructions which raises temps by literally 20C and I don't run Furmark. Prema himself suggests not to run Furmark and no offense but I trust him more than I trust you. I'm not going to recommend a stress tester that runs temps 20C higher than it needs to.

Prime95 26.6 + IntelBurnTest + Unigine Heaven is all I need for testing CPUs and GPUs. If I wanted to be extra careful I could replace Prime95 all together with AIDA64, but I haven't gone that far yet - I do still like Prime95.

 

Which working definition of "virus" are you thinking of when you want to call p95 a "virus"?

"a piece of code that is capable of copying itself and typically has a detrimental effect, such as corrupting the system or destroying data."

synonyms: worm, Trojan Horse => "a program designed to breach the security of a computer system while ostensibly performing some innocuous function." => "a computer virus"

"an infective agent that typically consists of a nucleic acid molecule in a protein coat, is too small to be seen by light microscopy, and is able to multiply only within the living cells of a host. "a virus infection""

I think of Prime95 as:

" Prime95 is the freeware application written by George Woltman that is used by GIMPS, a distributed computing project dedicated to finding new Mersenne prime numbers."

That's not a "virus"

And, since you are using prime95 constructively, not destructively, you think it's useful and not bad - or woo woo scary - so no need to overload it's real use with a "scary bad" word like "virus".

 

Did you really not read what I linked?

Like I said, adjectives can drastically change the meaning of words. For example, a good idea versus a bad idea has two completely opposite meanings.

Prime95, by definition, is a power virus: " A power virus is a computer program that executes specific machine code in order to reach the maximum CPU power dissipation (thermal energy output for the central processing units)."

Both Prime95 and Furmark are by definition, power viruses. There's not really much else I can say about that... It's important to understand what Prime95 and Furmark are, and what they do before considering or recommending their use. That is all I am saying here.

 

There is no such ill intent in prime95, it's not built or configured to act as malware.

That's the silliest mis-naming and mis-application of a mis-naming I think I have seen in quite a while.

You'll have to excuse my never supporting this phrase, as it's non-sensicle.

Just because it's in a Wiki doesn't mean it makes sense, and this one doesn't.

I'd suggest ignoring this name and move on, as it disrespects prime95's achievements in it's twisting of it's intent. Giving support and continuing usage of this phrase is an insult to the hard working people at http://www.mersenne.org/

https://en.wikipedia.org/wiki/Power_virus

In the source of the entry someone else with some perspective has also thoughtfully commented that needs cleaning up - I'd say it needs removal.

{{Cleanup|reason=largely unsourced, need cites, needs balance for benign use vs. malware|date=November 2013}}

At best, a program that "used" prime95 or a similar heavy power use tool to attempt to bring down a system and is an actual virus with mal-intent be called this, but not prime95 itself.

Edit: Some searching has brought up few hits, fortunately this mis-nomer hasn't caught on, first references are years ago now. Every time it's mentioned someone brings up the uncomfortable fact that a "virus" has a specific definition, and it doesn't fit in this usage.

I'd substitute "power hog" which makes more sense overall, but actually I've never found the need to add any such description for what prime95 does

 

Holy crap, gentlemen, it's a piece of software, there's no need to go to a war of words over semantics...

I started this P95 mess, so I'll end it: use it at your own risk. It's good for short testing of thermal cooling, but not recommended for extended periods of 'benching'. Full stop, thank you, good bye.

 

guys, thread topic

Sent from my Huawei Mate 8 NXT-AL10

 

On topic:

Furmark is the power virus, not P95.

 

Agreed

 

Back on the Power Virus wiki article:

"Stability Test applications are similar programs which have the same effect as power viruses (high CPU usage) but stay under the user's control."

Soo the only power virus is my brother's Pentium D 940

Anyway, what do you think about the new chips made by Samsung? Will it be worth waiting for them before buying a 1070 lappy? What are the theoretical benefits?

 

IMHO, P95 is OK for a 15 minute thermal paste test, not for much more than that. For simulating high GPU use, I prefer to loop Fire Strike.

Sent from my SM-G935F using Tapatalk

 

For gpus I find playing a game a better test. Often it is possible to pass looped firestrike but run a demanding fps for 90 minutes or more at same clocks/voltage you often end up needing to step down a bit.

Sent from my SM-A500FU using Tapatalk

 

Fair point. Something like Crysis 3/Witcher 3 for GPU testing?

Sent from my SM-G935F using Tapatalk

 

Yup. I can imagine any modern fps will do the trick. Usually a plethora of settings you can toggle and you can even run a higher non native resolution on nvidia cards to push further if needed.

Sent from my SM-A500FU using Tapatalk

 

BF1 is a pretty good benchmark lol.. It really pushes your CPU + GPU to the limit and is much more realistic then Prime95..

 

So, any word on when the Nvidia 1040 and below are coming out, or their specs?

 

They will possibly be rebranded Maxwell 1 GPUs—I somehow can't see nVidia asking TSMC to waste their latest 16 nm fabs on low-end chips that will be outdone by Intel iGPUs and AMD APUs anyway. So take the GTX 860M, 950M and 960M, and OC a little bit, and you get the GeForce 1040. Why do you ask, though, seeing that you've already got your notebook?

 

Curiosity really, as there's been little discussion of anything lower than the 1050. It'll be interesting to see if nVidia decides for once to make the entire lineup all current gen tech and not borrow GPUs from previous generations, but you're probably right that they won't.

And correct me if I'm wrong, but I assume that nVidia will likely release these low-end GPUs before they roll out Volta. With rumors that Volta might be released as early as mid-2017, a timeline as to when the low-end Pascal GPUs will be released might shed some light on the veracity of those rumors (which I personally doubt are true).

 

For good reason. If one is on a budget, or has no need for a high-end GPU—meaning he/she just needs a display device—then buying a low-end GPU is, frankly speaking, a waste of money. There's no need to cough up an extra hundred bucks for a XX30/XX40 series GPU just to get monitor output when the Intel CPU already has a fairly powerful on-die graphics adapter, that supports a fairly wide variety of outputs.

In a nutshell, anyone who wants a discrete GPU immediately goes to the mid-range segment at least, and anyone who wants a low-end GPU doesn't buy a dGPU at all, because CPUs have the GPUs for them. So the low-end discrete GPU market is severely neglected.

 

Interesting, and good to know - if I can't afford to buy a GTX xx50+ laptop in the future, I'll avoid buying a laptop with a xx40, and just get a laptop with a high-end iGPU instead. So who do you think will be in the market for 1040s and below, then? People who are fooled into thinking they're getting a better deal than they are because of nVidia's brand?

 

Very likely. Take a look: the highest end Iris Pro 850 offers more performance than the GTX 660M, and more performance than the 940MX, too. There's no reason at all why one would want the lowest-end of dedicated GPUs any more (except the brand and advertising) since iGPUs have caught up massively.

 

So high-end iGPUs can compete with the 945M - impressive. Makes me wonder how much further iGPUs will close the gap with dGPUs in the future.

Thanks for sharing your insight, I now will never be suckered into buying a dGPU-equipped laptop that has anything less than a GTX xx50 inside

 

igpus are such a waste of cpu die area, if you ask me...they take up more than HALF of the die nowadays, imagine what kinda cpu power we could have had already, if they had just for ONCE decided to let the igpu stay on the same performance level or even leave it out altogether...

Sent from my Huawei Mate 8 NXT-AL10

 

Fair enough, but you see, from Intel's perspective, they win: they have the biggest GPU market share of the three manufacturers. And as I mentioned, they effectively killed off nVidia's and AMD's low-end GPU sector.

Not to mention that having the iGPU on-die actually is fairly convenient, for 90% of PC users who never, ever will run a 3D program in their lives. Why waste board area, circuitry and power on a discrete GPU, or worse, an off-die iGPU when you can just chuck it on the CPU die?

 

Might still be worth buying a machine that has a 1040 inside because iris gpu laptops are generally NOT cheap. And the gap between a 940mx and 960m is marginal. The 1040 is going to be roughly on par with a 960m and probably similarly priced as iris igpu laptops. I've only seen iris laptops in higher end quality build laptops at more or equal cost as average gaming laptops


Sent from my iPhone using Tapatalk

 

i totally and completely understand and this argument makes sense. but, since those people never ever run a 3D application, why does intel have to increase igpu performance by like 1000% while cpu performance only gets a measly +50%? why not just integrate a BASIC igpu that takes up 5-10% of the die area and is sufficient for desktop stuff and call it a day?

but no, instead they dedicate more and more ressources, time, money and die area to stupid igpus that are slow no matter what! *ugh*

Sent from my Huawei Mate 8 NXT-AL10

 

Yep, and without competition we get what we're given.

 

As an engineer, all this talk of P95's FFT calculations being a power virus is very funny. Are the users saying this ok with the blend configuration?

 

Yup, it's like aboriginal natives trying to describe "traffic"

 

"This one crazy app uses instructions your CPU was never meant to attempt!"
"10 ways vector instructions will DESTROY your PC!"

 

And, it randomly runs the "Halt and Catch Fire" instruction, beware!!

 

Your PC? The whole chassis?

 

P95 runs exceeding 15 minutes have been known to destroy entire neighborhoods, don't you know!

 

I ran 8 hours of Prime last night to verify my overclock on my newly built desktop, now I'm homeless & still don't know if my overclock was stable or not!

But, joking aside, it's one of the most intensive tasks a CPU can do, so I can understand loose dramatic descriptions of 'power virus' - especially on some laptops that can't handle the heat - it feels like destructive CPU abuse!

 

Prime95 will give false negatives in failures though. You wouldn't want to think your girlfriend isn't pregnant but really is would you now?