14" UltraNote IV Low performance

ilidim

Member
Hello, I am writing this here because I was thinking that if there is a solution, it would be nice to exist in the forums for future users.
If no one responds here, I will open a support request and paste the solution here.

I bought the laptop of the title and also provided a review on the link below.
https://www.pcspecialist.co.uk/foru...500-16GB-RAM-M2-SSD-256GB&highlight=UltraNote

In short, my main specs are:
i7-7500U, M2.SSD 256GB by Samsung, 26GB Ram.
I am using ubuntu linux, but also tried another distribution as well. Haven't tried winodws because there is no need.

Here is my problem:
I am on hold of two other laptops.
The one laptop is a Lenovo Y50-70 which is quite strong, 8GB RAM, 128SSD drive, 6th generation i7 and a NVidia 940M.
The second in the office though, an HP Pro, is an old laptop that should not even come close to my UltraNote. It comes with a simple i5-32xx and again with an SSD of 256GB and 8GB of RAM.

I checked benchmarks online, and the i5 of my HP pro is times slower than the i7-7500U. And of course the M2 drive is much faster than the SSD (like 3 to 4 times). The RAM is faster as well.
So the UltraNote should behave close, and maybe even much better, with the Lenovo laptop which was a high performance when it was released, for any non Graphics demanding job.

So I am a developer and I do websites.
We have a bash script that installs automatically the website and it is running php.
Before we go into solutions of 'did I turn it off and on again', I should mention I tried 2 different installations while formatting 4 times and testing with the bare minimum packages, no debuggers running, only the absolutely necessary packages in Linux.

The results are really bad. While the browser, the IDE, and everything that need RAM and the drive, work much better than both of the above laptops, scripts and other processes that require the CPU are failing misserably.
I am working on Drupal (which is irrelevant) and a clean bare site installation, for both the above old laptops takes ~1 minute. The UltraNote IV takes ~2 minutes and 15". What's even worse, is that my current project, which is larger than normal, in the first 2 laptops takes between ~2-3 minutes and in the new laptop it takes an enormous amount of ~6-7. I am calling it enormous because I would expect to be even faster than the other two.

I tried really everything. I tried disabling the powersaving mode and selecting performance as a governor for my CPU and I only managed to lower it by about a minute, so still > 5'.

This is really bad for a laptop of that magnitude.
I read that there could be a thing with the CPU throttling but this is insane. It's a laptop that can reach high performance in simple tasks and scripts processing and it is actually behaving worse than old competitors.

Please advice... or if anyone knows something and has solved that already....
 

Oussebon

Multiverse Poster
The -U CPUs are dual cores focusing on low power rather than performance. If your Lenovo had a quad core i7 then it will be better despite being a generation older. And even if it's an i7 6500U rather than a quad core in the Lenovo, that's barely any slower than the 7500U. The CPU in the HP won't be all that far behind the i7 7500U either, given the relatively small IPC gains in each generation.

None of that necessarily explains why the new laptop takes 6-7 minutes for a project vs 2-3 minutes for the other two older ones - I'm just saying that when you fix what's making it so slow, it might not be much faster than the others.

Have you monitored temperatures? You're quite right that the laptop should be able to handle them fine, but that doesn't preclude there being a fault with the cooling. What temps are you getting during the task that takes a long time? And what frequency is the CPU running at? Something like Realtemp would let you monitor those.
 
Last edited:

ilidim

Member
If your Lenovo had a quad core i7 then it will be better despite being a generation older. And even if it's an i7 6500U rather than a quad core in the Lenovo, that's barely any slower than the 7500U.
True. I know that generations of these versions don't make much of a difference, and I know that still, my lenovo might have a slight advantage, though the fact that my HP Pro has a inferior processor and still makes it to the same time means that my new PC should more or less manage this or a better one.
Here is a comparison with my ProBook's 4740s i5-3210M: http://cpu.userbenchmark.com/Compare/Intel-Core-i7-7500U-vs-Intel-Core-i5-3210M/m171274vs2719

I will publish today reports from the temperature and frequency status of the process.
 

ubuysa

The BSOD Doctor
Are you certain there are no background tasks running on the Ultranote and consuming CPU cycles? When comparing two computers like this you need to be sure they are running exactly the same workload, and that means the same background tasks as well of course.....
 

ilidim

Member
Ok, first of all I just noticed that this is not in the technical support group, rather than the sales something. Can a moderator move it there?

Are you certain there are no background tasks running on the Ultranote and consuming CPU cycles? When comparing two computers like this you need to be sure they are running exactly the same workload, and that means the same background tasks as well of course.....
I am actually running the process in a lower load as in the other pc's I usually have an IDE (PHPStorm) open, browsers with multiple tabs etc.
Plus, as I said before, I even tried it twice in a bare minimum installation of ubuntu just to make sure..

Below there is an image of the process. The window is broken into 4 terminal tabs.
Top left is the process running. The longer process is the one using the database but it is through php and needs processing power I guess.
Bottom left is the cpu temperature. The average temperature was around 52 degrees with some peaks reaching at ~63 max which dropped fast when the fan started working harder.
On the top right there are the processes and the CPU load. The average CPU load was < 20% and only in certain cases it went above 40-50%.
On the bottom right I have results from powertop which shows processes and consumption.

View attachment 10822

I have really tried everything I could. Different versions of database, different versions of php, tried to install modules that cache commands, that make the whole procedure supposedly faster, all I managed is to take it to 6 minutes from 7 and now with the performance settings it dropped down to 5... Still not an acceptable result in my opinion...
Ok, first of all I just noticed that this is not in the technical support group, rather than the sales something. Can a moderator move it there?
 

Attachments

  • idimopoulos@pcsp: ~_001.png
    idimopoulos@pcsp: ~_001.png
    604.4 KB · Views: 230

ubuysa

The BSOD Doctor
I'm no Linux expert and those four reporting screens are new to me, so apologies if I'm way off-base....

The figures at the top of the top right screen (htop?) appear to show CPU, RAM and Swap Space total usage? None of them appear to be in any way stressed, though I guess these are averages over some time interval (or made up of an average of averages over some time interval) and it's not impossible (though unlikely I'll grant) that there are some instantaneous peaks at 100% that we don't see here that are slowing processes down...

The only thing that stood out to me was in the bottom right screen (powertop?) where there are two audio codes that seem to be 100% utilised? It is possible that they are flooding the bus somehow and restricting the internal flow of data for your your wanted processes? If you don't need audio for your tests can you not stop those two and see whether that makes any difference?
 

ilidim

Member
Well, audio codecs should not make that much of a difference anyway.. The only difference with my work PC that I can identify is a separate graphics card, which plays absolutely no role in that script (I guess), and that my work PC needs a 65W charger, which means more power.

I really cannot think of anything else. I will open a request ticket and let this forum know..
 

Tony1044

Prolific Poster
Drivers? Have you put the relevant chipset support on? That could potentially create a bottleneck.

Appreciate you've said you have tried a different distro but was it a completely different rather than a different variant of, say, Ubuntu?
 

ilidim

Member
No it indeed was Ubuntu variant! It is actually Elementary OS which is based on Ubuntu. But I do not think that this makes a big difference.
And update, today at the office I talked with a friend that also uses an i7-3xxxU processor. The other guy at the office uses Arch Linux and its a custom lightweight installation. His timings are more or less like mine. Slow. My best guess here is that the power limit also limits the processor. The 45W is messing with the performance limits. I opened a request already.
I haven't tried a different distro because that would take me quite a bit of time and I cannot re format it now as I have already set it up. I might try a live CD though. I don't think it will make much of a difference.

As for the drivers, I will check if there is something specific because I did not need to install any. I am using the latest LTS release of Ubuntu and the current Kernel already supported all the drivers. I will google some tutorials today on drivers for this processor and let you know.
 

Tony1044

Prolific Poster
No it indeed was Ubuntu variant! It is actually Elementary OS which is based on Ubuntu. But I do not think that this makes a big difference.
And update, today at the office I talked with a friend that also uses an i7-3xxxU processor. The other guy at the office uses Arch Linux and its a custom lightweight installation. His timings are more or less like mine. Slow. My best guess here is that the power limit also limits the processor. The 45W is messing with the performance limits. I opened a request already.
I haven't tried a different distro because that would take me quite a bit of time and I cannot re format it now as I have already set it up. I might try a live CD though. I don't think it will make much of a difference.

As for the drivers, I will check if there is something specific because I did not need to install any. I am using the latest LTS release of Ubuntu and the current Kernel already supported all the drivers. I will google some tutorials today on drivers for this processor and let you know.

Yeah if you have a similar spec (your friends' in the office) and it's similar performance it's probably just how it's going to be by the sounds of things.

Might still be worthwhile trying a live CD of a different distro but I'm not hopeful from what you've described.
 

Oussebon

Multiverse Poster
I have to say I'd be quite surprised if the laptop was sold in a state where it couldn't feed the processor enough power to run at its full potential, if that's what's being said here?
 

Tony1044

Prolific Poster
Nah my take is that it's running as well as it will for the tasks being performed. That the OP mentioned a colleague has a similar spec with similar results tends to back that up some, too.
 

ilidim

Member
Nah my take is that it's running as well as it will for the tasks being performed. That the OP mentioned a colleague has a similar spec with similar results tends to back that up some, too.
I don't know. If you checked my previous comment, normally the processor has more capabilities than that. Check one of my previous comment where I provide a comparison with my office laptop which is HP Pro 4740s.
My laptop should beat it in every aspect. My home laptop even runs on less software because I wanted to keep it light.
If it was just a comparison to my Lenovo, which was a strong laptop, I wouldn't be pushing it. But I find this a bit dissapointing..

PCSpecialist laptop - HP ProBook 4740s
i7-7500U - i5-3210M (Comparison provided in my previous comments through external link)
Samsung M2-SSD - Kingston SSD (M2 reports 3-4 times faster)
GPU Intel 620 - Radeon HD 7550M
16GB Ram DDR4 - 8GB Ram DDR3

Drupal installation srcipt (Same LAMP stack and other services settings)
5 minutes, 14 seconds after setting everything in performance mode - 2minutes 30 seconds for HP


I bought a mobility laptop so I already did not expect to compete with high end laptops but this is quite disappointing. Am I seeing it wrongly?
 

Oussebon

Multiverse Poster
I don't know. If you checked my previous comment, normally the processor has more capabilities than that. Check one of my previous comment where I provide a comparison with my office laptop which is HP Pro 4740s.
My laptop should beat it in every aspect. My home laptop even runs on less software because I wanted to keep it light.
If it was just a comparison to my Lenovo, which was a strong laptop, I wouldn't be pushing it. But I find this a bit dissapointing..
I have checked your posts and read them mate, before I posted mine.

The i7 7500U has a 15W TDP and seemed to use 20-21W under torture tests: https://www.notebookcheck.net/Kaby-Lake-Core-i7-7500U-Review-Skylake-on-Steroids.172692.0.html

I'm not suggesting you should be satisfied with the performance especially compared to the older 2C/4T HP laptop you mentioned in one of your first posts. I'm just saying that I would find it odd if PCS were selling a Clevo chassis inherently unable to supply enough power to the CPU, or if 45W wasn't enough for the system.

Since we're focusing on hardware rather than software at the moment - out of interest is the HP 4740s running memory in single or dual channel? I would be really surprised if the Ultranotes's single channel memory accounted for all of the difference even if the probook is running dual channel. But if we were being exhaustive I'd certainly look at that. If the probook is currently using RAM in dual channel maybe see how it does with 1 x 8gb rather than 2 x 4gb, if you have a suitable 8gb stick lying around.
 
Last edited:

ubuysa

The BSOD Doctor
Well, audio codecs should not make that much of a difference anyway.. The only difference with my work PC that I can identify is a separate graphics card, which plays absolutely no role in that script (I guess), and that my work PC needs a 65W charger, which means more power.

I really cannot think of anything else. I will open a request ticket and let this forum know..

They could if they are reading/writing to the system bus....
 

Tony1044

Prolific Poster
...today at the office I talked with a friend that also uses an i7-3xxxU processor. The other guy at the office uses Arch Linux and its a custom lightweight installation. His timings are more or less like mine. Slow...

I read your posts and this is what I personally was referring you back to.

If you have the same CPU but in different distro's and on different vendors hardware but with the same timings then I'm not sure there's much anyone else can do or advise?

Unless I'm missing something which is always a possibility.
 

ilidim

Member
Guys thanks all for the effort you put into this. I did not manage to do anything so instead I tried to find the most logical explanation.
I think that Oussebon had the most related argument indeed.

The UltraNote 14 comes with a i7-7500U which has a TDP [1] of 15W. TDP is in plain words and by leaving out the details (as far as I understood) the heating produced by a processor that the cooling system must handle.
It seems that even if the HP Probook 4740s's i5-3210M is slower than the current one, the corresponding TDP is 35W [2]. The U processors are meant for low consumption. They have a configurable TDP which balances between 7.5 and 22 Watt [1] but still is lower than the i5-3210M's 35W.

I can imagine that since also the Ultranote 14 is meant for mobility, the processor throttles after a while in order to maintain the temperature and the maximum Watt.

I cannot think of anything else. I might try overclocking in the future or try to find another workaround but I think that's it. A mobility laptop is going to restrict resources for energy.

I might be wrong but that's as far as I can go. Generally very good laptop but yeah..!

Anyway, thanks guys.

P.S., The support request answered me.. That they do not provide support for linux.. I could really argue that it is understandable that even if I am writing about linux, this is not meant to be a linux issue.

P.P.S.
Since we're focusing on hardware rather than software at the moment - out of interest is the HP 4740s running memory in single or dual channel?
They both run on single channel :/


[1] https://en.wikipedia.org/wiki/Thermal_design_power, https://en.wikipedia.org/wiki/Kaby_Lake#Mobile_processors
[2] http://www.notebooktest.pl/recenzja...-173-intel-core-i5-3230m-amd-radeon-hd-7650m/
 
Last edited:

Oussebon

Multiverse Poster
I can imagine that since also the Ultranote 14 is meant for mobility, the processor throttles after a while in order to maintain the temperature and the maximum Watt.
My point was the complete opposite though. :) The CPU only uses a small amount of power, such that 45W should be enough for it, and everything else in the chassis. Kaby Lake ULV processors were more efficient than the Skylake ones (more performance for less W) which was one of the relatively few absolute good things about Kaby Lake.

If the CPU has to throttle performance because it's not being fed enough power, the chassis, or at least the unit you have, isn't fit for purpose and should be returned.

But as I say I would be extremely surprised if it was power throttling.
 

sverx

Active member
TL;DR ... did you check if the CPU cores are working at 100%?
If it isn't so, it means I/O is slowing your processes (is disk cache enabled? I'm saying that because I found out it's disabled by default in Linux Mint... and maybe in some other Linux distros...)
 

ilidim

Member
Hey @scerx, sorry for the late reply. I have already given up on this matter. Yes, the cache is enabled btw.I can just assume that it's 'not made for these things'. That's all.
 
Last edited:
Top