Benchmarking/Performance Troubleshoot

[MPeckover]

Thanks in advance for your thoughts!

I'm pretty new to Benchmarking and LTT in general, but lately its been feeling like my laptop has been chugging a lot more than usual. Especially while running Adobe CC programs or multi-display setups with a few programs open.
I don't really game on my PC and use it primarily for the Adobe CC stuff and other workstation-related things.

Specs:
2016 Dell XPS 15 9560

Windows 10 Home x64
Intel Core i7-7700HQ 2.8ghz

NVIDIA GeForce GTX 1050
16gb Kingston DDR4 2.4ghz
(Sorry if I missed anything)

I ran a test using the free https://www.userbenchmark.com and got pretty abysmal results... 7th percentile among equally spec'd PCs. I attached a screenshot of the test results in case that is of use to anyone.

Is there any software or benchmarking tools I can use to try and figure out how well my PC should be performing and why it's suffering so dramatically?
Any help is hugely appreciated!

Thanks everyone.
Mike

[minibois]

Hey there, welcome to the forum!

You could try to benchmark the components separately from each other using specific tests.

For example, I am personally a fan of Cinebench to test a CPU; it's a free benchmark that shows CPU multicore and singlecore performance, but also has a decent built in GPU test. The CPU test is more interesting  

Quickly consulting the Cinbench R15 score spreadsheet (found here: https://linustechtips.com/main/topic/62476-post-your-cinebench-r15r1152003-scores-over-1000-submissions/) you should be able to get around 730-740 CB (their measurement of performance) in multicore.

 

For a GPU benchmark, you can consult 3DMark. It has a plethora of gaming benchmarks which are quite reliable and measure GPU score (Graphics score) and CPU score (Physic score) and roll it into an average number. It's pretty easy to, from there, compare your scores to what others get (just be sure to compare the right benchmarks, you for example have a couple versions of Fire Strike; some more intensive on your system than others).

 

If the scores of all these check out; you could also run a memory test just to know for sure that is also okay.. But it's also somewhat possible that the people with similar hardware just have better overclocks/better temps, etc. I  just find these percentage score (in Gaming, Desktop and workstation) hard to believe; those look like scores I got on my, at this point, 8 year old PC  

 

EDIT: oh yeah, I totally forgot about thermals and maintaining clockspeeds. As said below, a poor thermal paste application may also induce higher temperatures, which in turn makes it so your components don't speed up as much (or even slow down, to cool a bit down) and thus will lower your scores.

Just keep an eye on those temperatures and if you see it go up a ton and suddenly stay stable at a high temp (say, 90-100 C) thermal throttling might be an issue.

[suchamoneypit]

Maybe its having thermal issues and throttling itself. Have you tried running a program like MSI afterburner and seeing what kind of temps the CPU and GPU are reaching when doing intensive tasks? Might need to change out the thermal paste if thats the case. Since Its a laptop I think this is the most likely cause. Any throttling during benchmarks will severely affect your score.

[Sorenson]

It looks like you either have a software or hardware issue. I would back up your data and do a fresh OS install. 

[NunoLava1998]

Are you using the iGPU or GPU? 3% is a very low score for a 1050 and just right for the iGPU

[MPeckover]

13 minutes ago, Minibois said:

Hey there, welcome to the forum!

You could try to benchmark the components separately from each other using specific tests.

For example, I am personally a fan of Cinebench to test a CPU; it's a free benchmark that shows CPU multicore and singlecore performance, but also has a decent built in GPU test. The CPU test is more interesting  

Quickly consulting the Cinbench R15 score spreadsheet (found here: https://linustechtips.com/main/topic/62476-post-your-cinebench-r15r1152003-scores-over-1000-submissions/) you should be able to get around 730-740 CB (their measurement of performance) in multicore.

 

For a GPU benchmark, you can consult 3DMark. It has a plethora of gaming benchmarks which are quite reliable and measure GPU score (Graphics score) and CPU score (Physic score) and roll it into an average number. It's pretty easy to, from there, compare your scores to what others get (just be sure to compare the right benchmarks, you for example have a couple versions of Fire Strike; some more intensive on your system than others).

 

If the scores of all these check out; you could also run a memory test just to know for sure that is also okay.. But it's also somewhat possible that the people with similar hardware just have better overclocks/better temps, etc. I  just find these percentage score (in Gaming, Desktop and workstation) hard to believe; those look like scores I got on my, at this point, 8 year old PC  

I'll get on these tests and get back to you! Thank you!

5 minutes ago, NunoLava1998 said:

Are you using the iGPU or GPU? 3% is a very low score for a 1050 and just right for the iGPU

I have to admit I'm not sure what this means... can you clarify for me?

[NunoLava1998]

4 minutes ago, MPeckover said:

I'll get on these tests and get back to you! Thank you!

I have to admit I'm not sure what this means... can you clarify for me?

Are you using the CPU's integrated GPU (which is very weak), or are you using the GTX 1050?

[MPeckover]

4 minutes ago, NunoLava1998 said:

Are you using the CPU's integrated GPU (which is very weak), or are you using the GTX 1050?

That is a good question. I sort of assumed the two GPUs shared the workload simultaneously... Is that not the case? How do I switch if I find that I am using the iGPU, and how can I tell which is in use?
Thank you!

[NunoLava1998]

18 minutes ago, MPeckover said:

That is a good question. I sort of assumed the two GPUs shared the workload simultaneously... Is that not the case? How do I switch if I find that I am using the iGPU, and how can I tell which is in use?
Thank you!

That isn't the case. You can download GPU-Z and see what GPU you're currently using. If it's Intel Graphics, you're using the iGPU. If it's the GTX 1050, you're using the GPU and there's something wrong with that.

If you're using the iGPU, you can typically switch in BIOS or somewhere else (no idea how in laptops, in desktops you can typically go to the BIOS or plug your video cable into the motherboard instead of the GPU)

[MPeckover]

12 minutes ago, NunoLava1998 said:

That isn't the case. You can download GPU-Z and see what GPU you're currently using. If it's Intel Graphics, you're using the iGPU. If it's the GTX 1050, you're using the GPU and there's something wrong with that.

If you're using the iGPU, you can typically switch in BIOS or somewhere else (no idea how in laptops, in desktops you can typically go to the BIOS or plug your video cable into the motherboard instead of the GPU)

I figured this out. There's a setting in the NVIDIA control panel to default to using the nvidia GPU or "auto-select". I switched it to the nvidia gpu and my benchmark results improved somewhat. I'll till run the other benchmarks to see what happens!
Apparently its a known issue on Dell XPS models that when on "auto-select" gpu mode they rarely (if ever) use the nvidia gpu. weird.

[MPeckover]

7 hours ago, Minibois said:

You could try to benchmark the components separately from each other using specific tests.

For example, I am personally a fan of Cinebench to test a CPU; it's a free benchmark that shows CPU multicore and singlecore performance, but also has a decent built in GPU test. The CPU test is more interesting  

Quickly consulting the Cinbench R15 score spreadsheet (found here: https://linustechtips.com/main/topic/62476-post-your-cinebench-r15r1152003-scores-over-1000-submissions/) you should be able to get around 730-740 CB (their measurement of performance) in multicore.

 

For a GPU benchmark, you can consult 3DMark. It has a plethora of gaming benchmarks which are quite reliable and measure GPU score (Graphics score) and CPU score (Physic score) and roll it into an average number. It's pretty easy to, from there, compare your scores to what others get (just be sure to compare the right benchmarks, you for example have a couple versions of Fire Strike; some more intensive on your system than others).

So I ran both the cinebench and 3D Mark benchmarks.
The cinebench seemed extremely low, especially compared to the other 730-740 benchmarks (see screenshot) and the 3D Mark tests seemed passable? I'm not too sure. I ran both NightRaid and FireStrike (basic version).
Let me know what you think. Again, I"m new to all this software so I may not be using it correctly, I'm not too sure lol

On the same note, I used MSI Afterburner to monitor the CPU temps and it seemed it didn't reach any unusually high temps.

Any help is appreciated again and thanks!

[minibois]

7 hours ago, MPeckover said:

So I ran both the cinebench and 3D Mark benchmarks.
The cinebench seemed extremely low, especially compared to the other 730-740 benchmarks (see screenshot) and the 3D Mark tests seemed passable? I'm not too sure. I ran both NightRaid and FireStrike (basic version).
Let me know what you think. Again, I"m new to all this software so I may not be using it correctly, I'm not too sure lol

On the same note, I used MSI Afterburner to monitor the CPU temps and it seemed it didn't reach any unusually high temps.

Any help is appreciated again and thanks!

 

 

 

Thanks for running the benchmarks and providing the scores, really helps with seeing what is going on!

 

Looking at some comparable benchmarks, the scores do seem really low.

Just to compare your FireStrike score, I found a benchmark for an i5 7300HQ (basically the same chip you have, but with less threads) and it gets much higher scores.

I'll list some scores, in the Graphics (GPU), Physics (CPU) and combined catogories, in that order.

You get 5135, 2291, 1365.

While a 7300HQ, GTX 1050 gets: 7090, 6545, 2630. Definitely some higher scores, especially in the CPU department. 

This was the test I was talking about: https://www.3dmark.com/fs/12026191

 

Just a sec ago I realized I have a pretty similar specced laptop (an Asus one) so decided to also run these benchmarks. It's also an i7 7700HQ, GTX 1050 (non Ti), only difference being it's an 8GB model. I also ran test on and off battery. Often on battery the components in your laptop can go faster.

I also set the battery mode to 'best performance', you might want to do that for the tests too so you know how your laptop performs at best. I can imagine at the lower modes, it can be significantly slower.

 

Soo... I just ran FireStrike on my laptop without the power adapter and afterwards with it plugged in and the scores are as follows:

Without adapter: 3554, 5315, 1286

With adapter: 6067, 9643, 2295.

That's going from 3154 average to 5472 average, by just plugging it into power.

 

I did a similar test with Cinebench, here are those numbers:

Without adapter: 309

With adapter: 722 cb

 

I later just opened HWMonitor while benchmarking and saw that my CPU was dropping its clockspeed from 3.7Ghz to 0.8Ghz (likely to conserve power), which causes these terrible scores.

Your Cinebench score is still quite low, compared to what I was able to get.. But that can be due to many other things too; for example, I had minimal things open on my laptop, put the brightness all the way down and was not running HWMonitor/Task Manager the time I got that score. Usually I run Cinebench a couple times and only like twice with monitoring open to see what the temps and power and such is all about.

 

Long story short:

on battery, it's likely dropping the CPU clockspeeds (probably even when you have it set to high performance) causing the lowering of the performance.

You might be able to tweak some stuff in Intel XTU so it doesn't drop the clockspeeds on battery.. but I am not familiar with that program

[MPeckover]

4 hours ago, Minibois said:

Thanks for running the benchmarks and providing the scores, really helps with seeing what is going on!

 

Long story short:

on battery, it's likely dropping the CPU clockspeeds (probably even when you have it set to high performance) causing the lowering of the performance.

You might be able to tweak some stuff in Intel XTU so it doesn't drop the clockspeeds on battery.. but I am not familiar with that program

Thanks for the breakdown!

I ran a few more tests and found that my CPU is also capping at 0.8ghz but even when on AC power... I'll do some digging and see if I can see what is causing this, including BIOS updates, Dell Hardware diagnostic tests, etc. Hopefully I'll be back with good news!
I'll do some good old fashioned Google work too and see what I can find. Thanks again!

[MPeckover]

@Minibois @NunoLava1998 @suchamoneypit

All figured out!
All known issues on Dell XPS models. First, Nvidia was defaulting to using the iGPU and never the Nvidia GPU, which I fixed.
Second, core speed was limited to 0.8ghz on both AC and battery mode. This is apparently caused by the motherboard not detecting battery power correctly and switching to a "safe mode" to save power. Disconnecting the battery inside the laptop for a minute and reconnecting resolves the issue.

GPU now functioning properly, and CPU running at full capacity. Cinebench and 3DMark scores reflect this as well.

Thanks everyone for the help! This was a great introduction to the forum
Mike