Even we at Notebookcheck are sometimes surprised by test samples. The entry-level CPU of the ThinkPad X1 Tablet is about 25% faster than the more expensive Core i7 under sustained workloads. How is that possible?
https://www.notebookcheck.net/Unpleasant-surprise-Core-i5-Lenovo-ThinkPad-X1-Tablet-25-faster-than-i7-SKU.394986.0.html
What a mess. I've seen this on other brands too (where i5-U is faster than i7-U). Don't brands test their equipment to optimize their performance before selling them? It just looks like they just slap parts together and hope for the best, which doesn't seem like top engineering to me. Hopefully Lenovo (and other brands) learn from this and start to take things seriously.
This is not Lenovo's fault(at least not entirely), it's intel's.
The Core i7 model is supposed to be the better silicon, running the same speed with lower heat or or the same heat with more speed. Intel clearly is just "overclocking" the Core i5(a midtier CPU) and naming it as a Core i7 thus putting a higher price tag on the same CPU and selling it to the OEMs.
@Kian
Better silicon means more core speed, not better temperatures and more speed means it gets hotter. The device wasn't properly tested OR the design was to sustain max boost over short periods of time.
Inadequate cooling and too much throttling. Why anyone surprised?
I have seen this in other brands one was my Sony Vaio it was the last update for my model I upgraded to SSD and went to crucial memory test to see if it would accept more or faster memory to my surprise it would accept 1600 according to the test but Sony said no 1333 was all at the Take well I took crucial at their word and they were correct it was one extra digit at the end of my board number slash b and apparently there was an update that made it possible to run it faster quite a bit faster it was like a whole new computer and I only had it a year and when I received it because already outdated it was the last of the model run got a surprising gift that's the way I took it I think it's better to be surprised to the good side and not look for the bad
It is obvious a I7 needs more cooling than a I5 (ie : higher frequency = more heat load)
In fact, this is a good question : how to archived "good" thermal design in products, when Intel issues "skewed" marketing product specs ?
An exemple ?
I5 8250u - base freq.= 1.6 Ghz / max turbo freq. = 3.4 Ghz
I7 8550u - base freq.= 1.8 Ghz / max turbo freq. = 4.0 Ghz
I7 8650u - base freq.= 1.9 Ghz / max turbo freq. = 4.2 Ghz
But Intel said TDP (thermal design power) is the same for all (15 W) !
They even said configurable TDP-up is the same (25 W) !
How is this possible ? the answer lies with Intel's TDP definition : "average power, in watts, the processor dissipates when operating at Base Frequency with all cores active under an Intel-defined, high-complexity workload"
Then, you can be sure the "high-complexity workload" is not the same for I5 and I7, or there's a change with measurement.
As the T- junction (Junction Temperature is the maximum temperature allowed at the processor die) is 100°C for I5 and I7, you can see the consequences in Andreas' report...
Additionaly, bezels become more thinner every year, brands have to design thermal solutions that perfectly fits for all Intel processor : not easy !
@ Andrea : please, how is this taken into account in your reviews ? This is an issue because when you test and rate a I5 8250U config, people may prefer to order I7 8550/8650, and unfortunately sometime they will not experience the same story...
Thank you
I purchased this Tablet from Lenovo about 2 months ago as a replacement for a Surface Pro. I wanted a larger display. I got the i7 sku and 16gb of ram. After two days I had to return it. The machine overheated significantly, making it nearly unusable. I was worried about the longevity of the machine because I've never used a laptop that got that hot under my typical workload. I assumed it was defective unit. It appears it is a heat management design issue, or an Intel problem. I purchased a Surface Laptop 2 as a replacement. Microsoft's heat management for their i7 sku appears to be much better. The machine runs cool and performs well for me under load. I also own a Huawei Matebook X Pro i7 for home use. It's heat management for the i7 sku is somewhere between Lenovo and Microsoft. It gets much warmer under old and doesn't appear to be as fast. I believe this website indicates the Huawei Matebook X Pro i7 is no faster than the i5 and I believe that based on my experience.
@Dantoine:
Our performance rating is automatically calculated by the benchmark results. However, we also have a look at the sustained performance and deduct points in the magnitude of the deficit. For example: The i5 performance drops by ~15%, so we deduct 15%. This is to ensure that an i7 with a higher initial result (but lower score under sustained workload) does not get a better rating than an i5 with a higher performance under stress, like in this case.
Yes, this is obviously an Intel issue, but Lenovo should have noticed it, too, and they should have set a fixed limit to avoid this behavior. Other manufacturers and other devices from Lenovo do this as well.
Nah this is entirely Lenovo's fault, unless you mean to say it is acceptable to just shove a bunch of parts into a chassis and sell it without doing proper stress and performance testing.
Apple had the same problems with the recently refreshed 6-core MacBook Pros, but they were able to quickly issue a firmware fix. Lenovo should do the same.
Andreas said :
QuoteLenovo should have noticed it, too, and they should have set a fixed limit to avoid this behavior
This is unacceptable and a risk for brands : Lenovo sells expensive X1, and Dell sells XPS as well because people think they get value for the money and they think an Intel I7 is worth its price.
Before testing a new device, I think Notebookcheck could ask manufacturers if the same thermal design is being used for all Intel processor. In case of positive answer, I suggest you focus 1st on high-end I7 configs.
Can you please test the Core i7 underclocked to i5 speeds? (is it possible to underclock?) And maybe another test for the i7 underclocked to i5 speeds and underclocked as well. Just do this as a control, so we know that the results are not some kind of QC/manufacturing problem with Lenovo, but rather a design issue.
I literally unboxed my i7 X1 tablet less than 48 hours ago. In that time, without putting it through its paces, I noticed that it was hotter than I expected. I suspect that I have exactly the problem outlined here. Hoping for a firmware fix.
To think...at least the i7 should get a better bin!
Do you think this problem is just your model or the entire i7 allotement.
I believe you are experiencing a problem in the settings.
Goto Lenovo Vantage. Hardware Settings. Smart Settings.
Turn off Intelligent Cooling and select Performance. Now re-test.
My Desktop benchmark on "userbenchmark dot com" went from 69% to 88% what that setting change.
Rule of thumb is to always go for lowest i5 U series and i7 H series.
Last year those were i5-8250U and i7-8750H.
You could expect better performance on i7 U or better efficiency on i5 H but the value trade off is usually not worth it. This probably applies to the 98% of people and Intel processors since 4th gen Haswell.
Quote from: james A Tomlinson on January 19, 2019, 01:09:47
My Desktop benchmark on "userbenchmark dot com" went from 69% to 88% with that setting change.
FYI, my laptop has the i7-8550U
Is anyone having issues with screen coating scratching and scuffing only after a few weeks of use? Check the lenovo forums
who will use the laptop doing Cinbench over and over? Is it a realistic scenario?
Quote from: heffeque on January 16, 2019, 18:44:44
What a mess. I've seen this on other brands too (where i5-U is faster than i7-U). Don't brands test their equipment to optimize their performance before selling them? It just looks like they just slap parts together and hope for the best, which doesn't seem like top engineering to me. Hopefully Lenovo (and other brands) learn from this and start to take things seriously.
Hi, I'm wondering if the manufactures' tests are somehow different and probably more like real world use?
Can some one compare different benchmark methods/ real world office workload/ video editing photo editing tasks?
I feel there are quite some surprises, too.
Quote from: DavidEngineer on January 17, 2019, 13:36:02
I purchased this Tablet from Lenovo about 2 months ago as a replacement for a Surface Pro. I wanted a larger display. I got the i7 sku and 16gb of ram. After two days I had to return it. The machine overheated significantly, making it nearly unusable. I was worried about the longevity of the machine because I've never used a laptop that got that hot under my typical workload. I assumed it was defective unit. It appears it is a heat management design issue, or an Intel problem. I purchased a Surface Laptop 2 as a replacement. Microsoft's heat management for their i7 sku appears to be much better. The machine runs cool and performs well for me under load. I also own a Huawei Matebook X Pro i7 for home use. It's heat management for the i7 sku is somewhere between Lenovo and Microsoft. It gets much warmer under old and doesn't appear to be as fast. I believe this website indicates the Huawei Matebook X Pro i7 is no faster than the i5 and I believe that based on my experience.
Hi, David,
I'm more curious about your typical work load. why you could make it warm up so quickly. Maybe that could help me to decide if I could use that. Right now probably I'm thinking to get a gaming model for office/photo work because of the cooling solutions.
Quote from: Andreas Osthoff on January 17, 2019, 14:48:46
@Dantoine:
Our performance rating is automatically calculated by the benchmark results. However, we also have a look at the sustained performance and deduct points in the magnitude of the deficit. For example: The i5 performance drops by ~15%, so we deduct 15%. This is to ensure that an i7 with a higher initial result (but lower score under sustained workload) does not get a better rating than an i5 with a higher performance under stress, like in this case.
Yes, this is obviously an Intel issue, but Lenovo should have noticed it, too, and they should have set a fixed limit to avoid this behavior. Other manufacturers and other devices from Lenovo do this as well.
Andreas, would you fill me in on the benchmark methods? what really does Cinbench do during that time and how does it compare to real world usage? And usually what people would choose an i7 Thinkpad? Video editor? Photographer? or Powerpoint/PDF content editors? programmer?
Maybe most of us normal people just use the i5?
Could you clarify if the keyboard was on or off for these tests? I have found that the TDP management is different with the keyboard off. My 8650u x1 is able to sustain scores ~580 when the keyboard is attached and package wattage hovers around 14w. With the keyboard detached the score goes to ~420 and the wattage ~8w.