Dell XPS 15 9570 benchmarks + temps

Hehe, the irony is that this is exactly what Dell appears to be selling, considering that they declare thermal limitations nowhere in the specs. The i7 9550/9560 could at least be modded to run unthrottled under full load. I haven't seen this demonstrated with the i7 9570 yet, not to mention the i9. I don't think this is a good direction, and I don't think it is impossible to make better cooling within the same dimensions. Imagine a car factory would start selling a new model with 30% more powerful engine but the same radiator and the engine would go into limp mode after a minute of sporty drive? I couldn't appreciate a review that wouldn't bring that up.

 

Don't you get all logical on us, you'll be a part of the cricket club before you know.

 

So, what is the experience of increasing the powerlimit for the i7 or i9? Any differences? After repaste, PL throttling seems to be more frequent than Thermal throttling.

 

Temps are not good enough to increase powerlimits sadly.

 

Someone needs to buy a $20 used case bottom, drill some "large" holes in it, run a big fan under the opened laptop.

I did that with an old MacBook Pro and even with protective screen material it made a massive difference in temps.

 

When I have some time, I'll try to work on the long term CPU power limit. There is a bit wiggle room with the temps if the GPU is not simultaneously stressed.

I don't plan on doing that. Alternatively, there's this cooling pad where you can place the fans in the right spots to hit the ventilation grills.

https://www.amazon.com/dp/B00ED3WMT...olid=3FLN2KC9J6NXH&psc=0&ref_=lv_ov_lig_dp_it

 

I like that design although the fans are not so powerful. Someone (I think @Mr. Fox ) purchased that cooler and replaced the anemic fans for a monster gaming laptop a few years ago.

You will see on the 9550 and 9560 threads that cooling pads have been disappointing with these XPS units, so this might not be such a good investment.

 

i9 is nowhere able to function fully in xps 15.

 

So is there any other laptop that comes close in the size/design/price of the XPS 15?

I know there are some limitations to the cooling solution but I'm 100% OK with the manual work just to get that design and build quality, unless I find a better alternative.

 

@improwise

I think several people have different interpretations of what "unthrottled" and "under full load" mean.
After you repaste and do some modding, you end-up hitting the 45W power limit for CPU bound stress tests.
But that is not a fallacy of Dell (or any other manufacturers), It's the "specification" of the chip.

Should we consider reducing the frequency/voltage to stay at the 45W limit a form of "unwanted throttling", a manufacturer fraud, a misrepresentation of the performance?
Even with perfect cooling, you just can't run prime95 at 3.9Mhz while consuming 45W on these chips. It downclocks to 3.0Ghz to consume 45W in the long-term? That's part of the specification.

@_sem_ On the 9560 you had 45W to feed 4 cores. Here you have to divide 45W for 6 cores and the manufacturing process is the same 14nm.

 

Not sure that I have ever said something else than what you say, so not sure why you quote me, but never mind, as I agree with you. So lets agree to agree

My point was that after a repaste I mostly see power throttling instead of thermal throttling so it seems obvious that in order to get some more performance, the power limit should be focused on, if possible. But even if we could raise that, I assume that we would instead hit thermal throttling again and so the circle continues. I'm actually quite surprised all in all how well the XPS15 seem to hold up, at least once you replace that abysmal Dell paste job. And this is without the fans being to loud either. But then I mostly use this for "short bursts", like starting debuggers in Visual studio etc, not any long running CPU heavy processes like video editing. And I never use the nVidia GPU.

 

@improwise I cited you because I wrote here a long answer to what you linked in the other thread.

 

I must admit I was a bit confused, even thought this was some thermal throttling initially. Intel not listing a table of turbo clocks vs busy cores doesn't help either. Seen i5 run faster which does seem precisely due to less cores. It still isn't totally clear to me why certain other i7 laptops run faster long-term than the i9 9570, likely due to PL1 allowed higher than 45W or more undervolt?

 

Is anyone able to run all cores @3.9? Not wanting to change the topic, but I'm deciding between 9570 and gs65. I've read people being able to run gs65 3.9 on all cores. But the 9570 will be worth more in 2-3 years when I'm going to buy a new one.. Life is tough haha

 

Depends on what you are running. You're not going to be able to run something that is really CPU intensive like prime95 or realbench at 3.9, but something less intense may run there.

 

Someone stated that he's " getting 39x multiplier on all cores when running Cinebench, occasionally dropping to 38x on some for a second." (on gs65)

 

Running a round of Cinebench isn't really a difficult thing. I am pretty sure even stock it will run 3.8-3.9ghz. If you could loop it then you'd find issues.

 

I actually noticed that HWInfo reported all my cores (i9) as being above 4 GHz in idle, which I thought was not even possible.

Sent from my ASUS_A001 using Tapatalk

 

What type of battery life are people generally seeing? Any thoughts on the 9750 i7/1050ti vs the Aero 15 w/ 1060?

 

~8 hours.

 

7-8 hours

Sent from my ONEPLUS A5010 using Tapatalk

 

It seems like the i7 choice is useable without throttling by re-pasting + undervolting.

How about the i9? Any benchmarks on that one available?

 

Huh? The i7 still throttles. The i9 will also throttle.

 

Oh crap.. guess it will be the Thinkpad P52 then.

 

Hi, Great thread. I read through a lot of it but need some help. I ordered the i9 before i even knew about the throttling issues and it arrives tomorrow. I will be posting my temps after i try to reduce throttling as much as possible with suggestions from this thread. I got some questions though:

1. Could somebody show me with a picture where the VRM's are? ( i think theyre different locations than the 9560 model) Im going to try stacking it with 6kw/h thermal pads to backplate and see if theres a difference.
2. When padding the ram: Am i just padding the black chips on the RAM in OP, or padding the whole of it? Is that with 6kw/h? Maybe show in picture?
3. When padding, would it be best to cut it to the exact size of each individual VRM/chip to allow as much airflow as possible?
4. What settings will be best when trying to game? I will turn off turboboost. Should laptop be on high performance? Should i be trying to cap the CPU?

thankyou for all the help

 

Responded above in red.

 

Cant believe i didnt see the aluminium heatsinks! woops

I regret getting the i9 but cant return as dell take forever to ship and im leaving country in a week.
thanks for the reply


what mm padding do i need for the DIMMS?

 

I believe it was 2mm for the DIMMS. Dell gives you 30 days to return the system (at least in the US), so I say use it and if you can't get temps under control or you feel like you would be better off with the i7 then return it later.

 

It's important to separate burst mode vs long running CPU tasks. As a software developer, I got the i9 mainly for the former. If I had demands for video editing, massive builds, heavy number crunching etc, the XPS15 would not have been my choice. Neither of serious gaming was a priority.

Sent from my SM-N950F using Tapatalk

 

So true and so understated!
Most people focus on looong CPU intensive tasks when they talk about performance, whereas hardly anyone talks about the short bursts (even a bit longer ones!).
I think the short CPU bursts actually might be the most common usage pattern for most people, and the i9 definitely wins there. Unless you're encoding videos literally non stop...

 

i9 isnt supported by throttlestop


edit: older version. ill edit this post later with i9 results

 

I dont know if ive done a poor job here but the results compared to OP are not great.
My biggest worry is that i didnt repaste properly. is it normal to have 6-8 difference in CPU temps when running prime?



Prime 10 mins running
Stock :
idle temp : 52
max temp: 98
avg temp: 83
AVG clock speed: 2600

Repaste + undervolting + VRM/ram padding:
idle temp: 48
max temp: 94
avg temp: 83
avg clock speed: 3050


It never once went over 2900 on stock. It flucuates a little bit between 2800-3300 after changing it all.

I tried realbench but i lost the results but it looked something like

Stock:
avg temp 83
clock avg:1900
Dimm temp: 60


After:
avg temp: 91
clock avg: 2850
dimm temp: 60



Im currently at -127 for undervolting and i need to test for stability and if i can still go lower. but it doesnt look like its worth the money at all. unless i just done a really terrible paste job.


i assume theres not way to reduce power limit throttling?

I dont know if im going crazy but i swear my mouse stops for about 1/4 of a second every 3 mins. drives me insane

 

Power limit throttling is the main reason I am thinking if repasting is really necessary for the 9570. Sure, it runs a little cooler, but what goes does that to IRL if actual performance is based on the the PL throttling kicks in. It should be said that I rarely tax the GPU (at least not in this machine), if you do, things might change. Also, fan noise could of course also decrease with better cooling, but then it would need to be more than a few degrees

I actually compared a stock 9570 with a repaste one (exact same specs as it was a replacement) and real life performance was actually almost the same...

Sent from my SM-N950F using Tapatalk

 

Lower temps is good for hardware longevity.

 

my temps are dangerously hitting 99degrees when running 4 memu android emulators. CPU package hit 101degrees

i think im going to repaste but i dont think it will change anything and the i9 will just always get this hot. disabling turbo obviously reduces this but would be nice to have to reduce turbo on every thing i do

Package/Ring Thermal Throttling

i keep hitting this. is there a way to reduce this?

 

I guess this is not normal and a repaste should help, nothing uncommon unfortunately. Your GPU is probably idle then? I understand the i9 shouldn't be getting so hot because of the power limit (45W after short turbo time expires, a bit more initially) - but it probably isn't any faster than the i7 with longer workloads, just in short bursts.

 

You really need to undervolt and repaste these XPS units else thermal performance is terrible.

Those two easy steps make thermals much better. You may still face significant throttling but performance could be improved. . . YMMV

 

Well, yes, at least in theory. But how often do we actually see hardware failures due to overheating during the time people in here consider the same hardware "relevant"? If my 9570 breaks down in year five, it will be long gone anyway. Also, CPUs etc AFAIK are designed to survive temps up to 100 degrees. Or am I wrong?

Sent from my SM-N960F using Tapatalk

 

Sadly it's not just a theory, heat kills electronics on an exponential scale in terms of failure rates. While the time-frame may not be within the "usable" period, I would prefer to keep things cool and not take that chance. CPUs are designed to throttle at ~100C but that's to prevent a much quicker failure.

 

And that is why you should always bring an umbrella even if it is sunny outside, because there is always a risk it might start to rain. My point is about managing risks, not to argue if they exist.

For me personally, a decision to repaste and similar would be mostly based on possible performance gain. If no such were to be expected, I probably would not bother to do it. There are also risks involved with pulling your laptop etc appart, which might also affect the longlivety of the unit/components. Others might have a different opinion and I respect that.

That said, I would probably still repaste just because I can

Sent from my SM-N960F using Tapatalk

Edit:
Just to avoid any unnecessary confusion, custom90gt is of course correct about high temps over time being a problem. Just that with the 9570 and the Power Limit throttling that seems to be the main problem, repasting might not give as much as it would have done if thermal throttling was the main problem. Since I would never game on this anyway, active usage of the dGPU might change things.

 

Hey All-

I have a i9 9570 - I disable the 1050ti completely and use a eGPU with a 1080ti in it.
-140mv undervolt, "max performance" plan in dell power manager, c1e disabled, speedstep disabled, speedshift on 0

I currently have repasted (well, technically I am using a Innovation Cooling Graphite Thermal Pad ) and padded all the VRMs.

What I am thinking about is if I should unpad the VRMs associated with the GPU (since its disabled) and just leave the (3?) rightmost VRMs padded.

I have not padded the memory, is that worth doing, by connecting it to the backplate? (This thing lives 100% of its life on a notebook cooler, so it wont get saturated)

Additionally: I was thinking that maybe I should pad directly above the CPU so as to apply more 'pressure' from the back of the case when its closed to get a tighter seal. Thoughts on that?

Can anyone tell me where the thunderbolt chip is? Since I run all my graphics over a egpu, I think it might make sense to pad that chip.

Also, not sure it was discussed here much but whats the verdict on padding the PCH?

Also- I can run a CPU stress test all day long at 4.1ghz, as I saw someone asking about that. Temps will stay at around 88 during that process.

 

Guessing that you are not using something like Prime95 for that stress test as no amount of thermal paste and pads should be able to solve the power throttling?

How do you like the Graphite thermal pad BTW? Seems like an elegant solution.

 

Put my responses in red above.

 

Dear All,

I am very new to this forum and I recently purchased my first XPS. I have the i5-8300h model with the FHD display and the 1050.

I am having a few issues and thinking of returning it but dont want to return it if I can get the issues solved.

1. When on battery , it seems like the 1050 is severely downclocked especially when running benchmarks. It doesnt seem like this is happening when the AC adapter is plugged in. I have set the slider to best performance as well as im using the ultra performance option in the dell power software but this doesnt help when on battery. I have GeForce experience installed and am thinking on uninstalling this and letting the dell updater keep the driver updated.
2. When on power and I run a CPU/GPU combined benchmark it seems like my CPU sometimes downclocks to 700Mhz for a few seconds then will bounce back up above 3Ghz. I think based on this forum it is VRM related but I am not sure as when I run a CPU benchmark without GPU, this does not happen and i get excellent results with the CPU consistently staying between 3.6 to 3.9Ghz.

Hope someone can help.

 

All laptops downclock their dGPU when on battery. You can try setting the nvidia power settings in the control panel to see if it helps, but most of the time it's because the battery simply can't handle the load of both the CPU and GPU. Sadly I don't know of any laptop that runs full speed when on battery.

The CPU downclocking could be due to a few things. The best way to check would be to use HWinfo and save a log file so you can see what the reason is. It may be the DIMM sensor getting too hot. If you want to save the log file and post it here, there are a few members that wouldn't mind helping you figure out why you're throttling.

*on edit*
It looks like you posted the same message in the owners thread as well and got some good info there. In the future I would stick to one thread so you don't have multiple people giving you the same advice.

 

I recall someone measured the 9550 actually drawing 130W from the battery under Linux. Myself I've seen that the CPU also on battery actually slowly ramps up the clock all the way to the max under long-term loads, which in addition to the 30FPS-capped GPU must have drawn about 100W from the smaller battery (got the HDD version with upgraded SSD). It didn't crash during the couple of minutes of testing, but it surely isn't healthy long-term. Too many swollen battery reports.... I use it to overcome the restricted power mode with the 45W charger for travel though, then the battery just covers the excess during CPU workload spikes (GPU still throttled heavily), recharges during idling.

 

I would be surprised to see the 9550 draw 130W from anything. I've measured the draw from the AC adapter on my 9560 and 9570 and even under prime95 and furmark neither has gone over 105w from the wall. AC adapters aren't super efficient, I'd be shocked if it was 90% efficient.

 

Weird, I've measured the 9550 under Prime95 + Heaven and it did pull just above 130W, both with the default 130W AC charger and the 180W AC charger of the WD15. Dunno how much of that came into the 9550, and how accurate the "Watts up pro" is.

My old post: http://forum.notebookreview.com/thr...0-owners-lounge.783377/page-349#post-10499329

 

Odd that your 9550 would draw more than my 9570 or 9560. I wonder how efficient the dell ac adapter is? Maybe 80% So around 104W of power going into the system.

 

The chargers also have different efficiencies, depending on 110V or 240V.