News:

Willkommen im Notebookcheck.com Forum! Hier können sie über alle unsere Artikel und allgemein über Notebook relevante Dinge disuktieren. Viel Spass!

Main Menu

Post reply

The message has the following error or errors that must be corrected before continuing:
Warning: this topic has not been posted in for at least 120 days.
Unless you're sure you want to reply, please consider starting a new topic.
Other options
Verification:
Please leave this box empty:

Shortcuts: ALT+S post or ALT+P preview

Topic summary

Posted by Niels
 - May 29, 2018, 22:48:29
Certainly, throttling will be an issue to users wanting to edit audio, video or 3D drawings.

Anybody who plans to play an occasional game will also suffer.

Pure office users may not run into throttling, but for them, there's still a major issue with Microsoft's product:  Its glued together and can't be upgraded or repaired by yourself or the shop at the corner.

Not to mention that Microsoft thinks your computer is really theirs and you just get to use it in the way they deem fit, as demonstrated by Microsoft's admission, that Windows updates can and will remove software from your system unasked, without prompt.  We're not talking pirated stuff here either, we're talking anything the logic of the present windows update doesn't like... 

After its finished, you want to continue your work, but your software, its gone!  This has included Office Pro 2016, Capture One Pro, for example.

Please send Microsoft the bill for missed deadlines  :)
Posted by Star1
 - February 10, 2018, 22:22:51
Owned a Surface Pro 2017 i7, 1TB, 16GB model for 7 months now.

Of the 3 power modes when plugged in (4 in battery mode), amazingly the lowest setting of the three actually has the "(Recommended)" in brackets. I find this is only justifiable in the most lowest of mundane tasks that require hardly any performance, otherwise apps will stutter, some video playback/streaming will stutter, etc. Medium level "Better Performance" is probably the true "(Recommend)" setting for when plugged in. The highest "Best Performance" can actually serve to lower CPU Utilization percentage.

I've found that the troublesome "Power Limit Throttle" will repeatedly alternate on any of the three settings, just at different frequencies. Sadly after some time of use, the Power TDP will be throttled substantially and I believe it is primarily from when the enclosure becomes warm. Personally despite liking the Surface Pro 2017 i7, 1TB, 16GB very much, the throttling prevents the unit from being so much more.
Posted by Howard Cadwell
 - July 23, 2017, 19:29:09
The throttling and performance review didn't specify whether max performance or max battery life option was selected.

Other reviewers have found both no throttling and good thermal control when max performance is selected.  Other modes, such as max battery life, enable throttling by design.

Unless this is taken into consideration, I wonder if the results here are relevant.
Posted by Fuzz E. Loujek
 - July 17, 2017, 16:05:33
POW, Mac man

https://www.reddit.com/r/gadgets/comments/6no840/microsoft_surface_pro_series_facing_heavy/dkb8e3p/
Posted by Obviously
 - July 17, 2017, 15:18:06
garbagedisposal, take a time out and let the adults talk; name-calling and posting drivel is not the way to get people to listen to your argument, especially when it's clueless.

Little surprised this article is coming out of notebookcheck as they usually have some pretty solid articles. Obviously the passive cooling solution on the i5 7300U gets overwhelmed under extended load, but it's odd that the article authors call this surprising when it's not. It's helpful to know, but as already mentioned no one buying a Surface Pro is using it to render for an extended period of time; any designer with half a brain is going to export renders to a much more powerful CPU anyway.
Posted by garbagedisposal
 - July 17, 2017, 13:26:53
Quote from: Carl on July 09, 2017, 22:13:16
Using looped synthetic benchmarks as a guideline for user experience, grossly miss represents the device.

The vast majority of surface users don't run workloads that continually stain thr CPU and will experience throttling far, far less frequently, if at all.

No it doesn't.  They tell you straight up that they loaded all cores and watched what happened to performance, it tanked.

Misrepresentation? Are you joking?
All these surface feet kissing tards in this thread complaining. Fact is it's an expensive machine with a misleading spec sheet.  I feel sad for you clueless people that would pay good money for an inferior machine, 'but it's OK because it wont throttle when I check my emails' lmao
Posted by garbagedisposal
 - July 17, 2017, 13:23:34
Quote from: Charles on July 17, 2017, 07:17:38
Testing the Surface Pro with Cinebench and calling it disappointing proves only that those performing the test failed to understand what they were testing.  These conclusions are about as valid as a 1917 drivers permit.  The best analogy I can come up with is saying the Prius is a failed automobile because it fails to cross the Moab desert, a task best left for  a Jeep. The surface wasn't designed to be rendering platform.  It's  meant for casual to mobile convenience  professionals (photography a la lightroom and photoshop on the go i.e. for those in the field edits)  Surface Pro and its predecessors aren't meant to render 3D scenes hence the use of cinebench is not the best task to throw at it.

It's a computer and that's what it was advertised as - not a faebook-only machine.
If all cores are loaded and the clocks scale back, that means it throttles.

You don't get a cop out by saying 'that's not what the machine was intended for'.
Your car analogy is so dull it could have only been dreamed up by a SIDS baby.

It throttles and the cooling solution is not good enough. End of story, deal with it.
Posted by garbagedisposal
 - July 17, 2017, 13:20:19
Quote from: GreyFox on July 10, 2017, 13:13:21
There are a number of problems with the methodology on this that make any comparison invalid.   

I get the whole benchmark it until it breaks mantra and in that respect they achieved their goal however, the variable settings for power, thermal, and clock manipulation doesn't lend itself to an objective conclusion. To say the Max Turbo Clock rate was not maintained is throttling is a misrepresentation.  Unless you can pin the configuration to the same settings you cannot make a comparison from one to another.  In a multidimensional system you cannot pick one parameter and draw a conclusion based on that alone.             

Turbo Boost is variable and not continuous, as evidenced by the XTU setting for microseconds of Turbo Boost.  At the same time clock rates are variable, voltage and current are variable, boost duration is variable, thermal parameters are variable.  Each of these contributes or detracts from the total work accomplished over time which is the ultimate desired result. The instantaneous parameter values may not be a 100% accurate predictor under dynamically variable conditions.   

If as shown on a youtube video, the sustained performance of a system is maintained at 16W and the rated power of that system is 15W how do you conclude it throttled at all even though under temporary boost conditions it may have peaked at 30W?  The message being sent is misleading and incorrect.

You're an idiot.
There's no problem with the methodology at all - load all the cores and see if it throttles. It does throttle, and in barely any time at all.

If max turbo is not achieved, that is the very definition of throttling.
Whether intel calls the max attainable frequency 'super turbo' or 'beyond max turbo+' - if whatever max frequency can only be reached transiently before thermal/power limits are hit - then it's called throttling.

The misleading nonsense BS in your post about variable clocks/voltages are completely meaningless. You're not educating anyone. Please refrain from commenting, it's pretty clear you have no experience with hardware.
Posted by Charles
 - July 17, 2017, 07:17:38
Testing the Surface Pro with Cinebench and calling it disappointing proves only that those performing the test failed to understand what they were testing.  These conclusions are about as valid as a 1917 drivers permit.  The best analogy I can come up with is saying the Prius is a failed automobile because it fails to cross the Moab desert, a task best left for  a Jeep. The surface wasn't designed to be rendering platform.  It's  meant for casual to mobile convenience  professionals (photography a la lightroom and photoshop on the go i.e. for those in the field edits)  Surface Pro and its predecessors aren't meant to render 3D scenes hence the use of cinebench is not the best task to throw at it.
Posted by Charles
 - July 17, 2017, 07:16:17
Testing the Surface Pro with Cinebench and calling it disappointing proves only that those performing the test failed to understand what they were testing.  These conclusions are about as valid as a 1917 drivers permit.  The best analogy I can come up with is saying the Prius is a failed automobile because it fails to cross the Moab desert, a task best left for  a Jeep. The surface wasn't designed to be rendering platform.  It's  meant for casual to mobile convenience  professionals (photography a la lightroom and photoshop on the go i.e. for those in the field edits)  Surface Pro and its predecessors aren't meant to render 3D scenes hence the use of cinebench is not the best task to throw at it.
Posted by nonua
 - July 17, 2017, 01:11:04
Why do you keep saying "SKU" (stock keeping unit) rather than the more accepted term "model"
Posted by Klaus Hinum
 - July 14, 2017, 18:49:43
We set the Turbo Boost Power Max via XTU to 35 W to test it and still the Cinebench loop did throttle to similar levels, so I guess its just a cooling issue and not a TDP limit.
Posted by Lorenzo
 - July 13, 2017, 21:17:15
you can unlock power limit from intel xtu but un that case you need a fan. https://www.youtube.com/watch?v=suFQIbt3Dxc.
Posted by GreyFox
 - July 10, 2017, 13:13:21
There are a number of problems with the methodology on this that make any comparison invalid.   

I get the whole benchmark it until it breaks mantra and in that respect they achieved their goal however, the variable settings for power, thermal, and clock manipulation doesn't lend itself to an objective conclusion. To say the Max Turbo Clock rate was not maintained is throttling is a misrepresentation.  Unless you can pin the configuration to the same settings you cannot make a comparison from one to another.  In a multidimensional system you cannot pick one parameter and draw a conclusion based on that alone.             

Turbo Boost is variable and not continuous, as evidenced by the XTU setting for microseconds of Turbo Boost.  At the same time clock rates are variable, voltage and current are variable, boost duration is variable, thermal parameters are variable.  Each of these contributes or detracts from the total work accomplished over time which is the ultimate desired result. The instantaneous parameter values may not be a 100% accurate predictor under dynamically variable conditions.   

If as shown on a youtube video, the sustained performance of a system is maintained at 16W and the rated power of that system is 15W how do you conclude it throttled at all even though under temporary boost conditions it may have peaked at 30W?  The message being sent is misleading and incorrect.

Posted by Carl
 - July 09, 2017, 22:13:16
While I agree, tge results aren't great and quite disappointing.

Using looped synthetic benchmarks as a guideline for user experience, grossly miss represents the device.

The vast majority of surface users don't run workloads that continually stain thr CPU and will experience throttling far, far less frequently, if at all.