Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Thanks. In Canadian website, it states up to 1TB so I want to check if 2TBx1 and 2TBx2 would work in practice.
 
On my Yoga C930, it could handle Office work well but when I tried to test its performance by flipping the image upside-down (a simple task?) using Photoshop, I saw lag in performance. In other tasks, I could see the program processing blocks of images on the screen before the tasks were completed. I don't recall seeing lag on the X1E.

On my L390Y I just went into Photoshop and I opened a 57.7M RAW photo (5184x3888 pixels) and I flipped and rotated it in various ways - it was absolutely instant. No lag whatsoever. I then upscaled and doubled the size of the picture to 10,000 pixels across and repeated the process - still instant. I then did some liquify filter work on the same picture and it was very fast. I then made 7 adjustment layers and worked on those, and flipped it again - virtually instant.

On my X1E there was a bit of lag doing anything graphical in 4K resolution, but never 'very slow' or blocky. When I switched it down to 1080p it was faster. Maybe you should try that on your C930 to see if it makes a difference, but if you're seeing blocks on the screen then that sounds very odd.
 
  • Like
Reactions: hajime
On my L390Y I just went into Photoshop and I opened a 57.7M RAW photo (5184x3888 pixels) and I flipped and rotated it in various ways - it was absolutely instant. No lag whatsoever. I then upscaled and doubled the size of the picture to 10,000 pixels across and repeated the process - still instant. I then did some liquify filter work on the same picture and it was very fast. I then made 7 adjustment layers and worked on those, and flipped it again - virtually instant.

On my X1E there was a bit of lag doing anything graphical in 4K resolution, but never 'very slow' or blocky. When I switched it down to 1080p it was faster. Maybe you should try that on your C930 to see if it makes a difference, but if you're seeing blocks on the screen then that sounds very odd.

So your L390Y is faster than the X1E?
 
I've explained the situation with my X1E in this thread - there may have been something wrong with it which is why I am going to re-order. Or the slowness could simply be down to having the 4k version. I'll soon find out.
 
I've explained the situation with my X1E in this thread - there may have been something wrong with it which is why I am going to re-order. Or the slowness could simply be down to having the 4k version. I'll soon find out.

When will you know?
 
I haven't had time to order it, when I do it will take up to a month to arrive from Lenovo, then I'll need time to test it.

But as has been said several times now - it makes no difference what anyone else's experience is because each user case is different and each machine appears to be somewhat different. And each owner is different in their needs and preferences. What might be 'slightly laggy' to me might be acceptably fast to somebody else, what may be 'fairly quiet' to somebody else may be too noisy for the next person.

If this is the only laptop you're still considering then I think the only solution is for you to re-order it and see how it goes.
 
I have decided to give the X1E Gen 1 one more chance. If it does not work out this time, I wait for Gen 2. After receiving the machine, I should update all the drivers and make sure that the BIOS is updated to 1.19. Then, what shall I do next? Just try it for a few days before deciding if I should undervolt and/or do a thermal re-paste or jump directly to do the undervolting? Am I correct that -150mV is the recommended value for undervolting the i7-8850H? With the 1.19, do I still need to install the dual-fan version of TPFanControl? https://github.com/cjbdev/TPFanControl

Is there a guideline to follow in regard to changing various settings on new machine to make it quiet, have good performance and battery life?
 
Last edited:
It's all in this thread, these questions have been answered several times. Plenty of tips and experiences have been shared so if you re-read everything you can make a list of things you can try if you need to.

There is no 'recommended value' for undervolting. As has been said already - all chips are cut differently to they will respond a little differently. Personally I would start at -0.110v and then go up on 0.005v increments till you hit your limit, then drop down by 0.005 and retest, until stable. There are lots of guides online for both XTU and Throttlestop, whichever you choose. You can undervolt the GPU as well.
 
It's all in this thread, these questions have been answered several times. Plenty of tips and experiences have been shared so if you re-read everything you can make a list of things you can try if you need to.

There is no 'recommended value' for undervolting. As has been said already - all chips are cut differently to they will respond a little differently. Personally I would start at -0.110v and then go up on 0.005v increments till you hit your limit, then drop down by 0.005 and retest, until stable. There are lots of guides online for both XTU and Throttlestop, whichever you choose. You can undervolt the GPU as well.
What do u use to benchmark and test how stable your underclock is?
 
  • Like
Reactions: hajime
What do u use to benchmark and test how stable your underclock is?
I just used cinebench and used the computer. If the computer crashes, then the underclock setting is over zealous. I found that a setting in the -1.6 range didn't crash the computer immediately, it booted up, but after a few hours of usage, or even the next day it would crash
 
I just used cinebench and used the computer. If the computer crashes, then the underclock setting is over zealous. I found that a setting in the -1.6 range didn't crash the computer immediately, it booted up, but after a few hours of usage, or even the next day it would crash

From Reddit, the consensus seems to be -150mV. They have taken a tolerance of 10% into consideration.
 
From Reddit, the consensus seems to be -150mV. They have taken a tolerance of 10% into consideration.
I just went on my own experience and confirming my settings, I have it set to -1.40. Mostly, because I don't want to deal with any instabilities and the computer is fast.
 
ThrottleStop has it's own self-test routine built into it, TS Bench any errors the Offset Voltage needs to be raised as the CPU is unstable.

As for a default undervolt value none exists as each CPU is unique. Computer needs to be monitored for stability on and off load, and battery etc. My own notebook with 8750H can go as deep as -170mV, equally to ensure stability across the full operating range including TDP Down I typically run it at -140mV.

Q-6
 
What do u use to benchmark and test how stable your underclock is?

I use the built in test in XTU. I'm not a gamer so I don't need to do a long thrash test (which is unnecessary for most users). If at any point the computer becomes unstable I'll know why.

My X1E undervolt was, if I remember correctly, -0.125v as it wasn't stable at -0.145. Bearing in mind I had the i5 processor.
 
ThrottleStop has it's own self-test routine built into it, TS Bench any errors the Offset Voltage needs to be raised as the CPU is unstable.

As for a default undervolt value none exists as each CPU is unique. Computer needs to be monitored for stability on and off load, and battery etc. My own notebook with 8750H can go as deep as -170mV, equally to ensure stability across the full operating range including TDP Down I typically run it at -140mV.

Q-6
Ok I'm using handbreak instead to bench by encoding a mkv movie 1080p to mp4 on my main pc which I'm overclocking now
 
Handbreak wont show if the CPU is stable, TS Bench will as that's it's purpose. Once your certain stability is solid, I'd then just get on with using the system.

Q-6
Quite the opposite. Handbreak is more valid then ts bench.

Ts bench always passes yet when I use lr or play games it will crash randomly.

I then started using handbreak to test and since its passed that my system has been stable for everything
 
Quite the opposite. Handbreak is more valid then ts bench.

Ts bench always passes yet when I use lr or play games it will crash randomly.

I then started using handbreak to test and since its passed that my system has been stable for everything

Mine are fine, no lock ups. Like I said your only achieving stability under full load there's a lot more variables with modern CPU's.

Q-6
 
Last edited:
Min are fine, no lock ups. Like I said your only achieving stability under full load there's a lot more variables with modern CPU's.

Q-6
So how come it crashes on full load and not in ts benchmark? I don't just browser the Internet on my computers. I make them work!
 
So how come it crashes on full load and not in ts benchmark? I don't just browser the Internet on my computers. I make them work!

Lots of differing power variables now, best to run TS Bench 1024M test repeatedly on lowest power settings. At higher power settings CPU voltage is likely already boosted by W10 or OEM setting, you can see an indication of this in FIVR with EPP

There's no right or wrong, equally nobody likes surprises, nor would I rule out other issues. As for working:
3100CB (No Taskbar).jpg

Nor does it crash on or off load.

Q-6
 
So how come it crashes on full load and not in ts benchmark? I don't just browser the Internet on my computers. I make them work!
If I change my undervolt to -1.6, it will pass under benchmarks, but will eventually crash. As noted, the benchmarks cannot fully exercise all aspects of the computer and the overly aggressive undervolt will hit a situation that crashes it. I've seen some people offer a prime95 test for 24 hours, but that seems a little extreme. For the X1E, I have mine set to -1.40 and I'm happy. Its a nice balance of power and temperature. I've said this before, I'm not looking to squeeze out every ounce of performance. I've largely stopped monkeying with my system at this point, its running exactly how I need it to run.
 
If I change my undervolt to -1.6, it will pass under benchmarks, but will eventually crash. As noted, the benchmarks cannot fully exercise all aspects of the computer and the overly aggressive undervolt will hit a situation that crashes it. I've seen some people offer a prime95 test for 24 hours, but that seems a little extreme. For the X1E, I have mine set to -1.40 and I'm happy. Its a nice balance of power and temperature. I've said this before, I'm not looking to squeeze out every ounce of performance. I've largely stopped monkeying with my system at this point, its running exactly how I need it to run.

Same once I've set it up I rarely change settings. I do however take advantage of ThrottleStop's profiles and automation, leaving W10 & OEM power settings at maximum.

Q-6
 
  • Like
Reactions: maflynn
If I change my undervolt to -1.6, it will pass under benchmarks, but will eventually crash. As noted, the benchmarks cannot fully exercise all aspects of the computer and the overly aggressive undervolt will hit a situation that crashes it. I've seen some people offer a prime95 test for 24 hours, but that seems a little extreme. For the X1E, I have mine set to -1.40 and I'm happy. Its a nice balance of power and temperature. I've said this before, I'm not looking to squeeze out every ounce of performance. I've largely stopped monkeying with my system at this point, its running exactly how I need it to run.

Agree with maflynn here.

Finding the optimized voltage is a process to be done over a couple of weeks. Start out with the benchmarks as a testing tool at the voltage you think you are good at, and then use your notebook regularly and increase & lower it as the stability becomes proven or disproven.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.