Loading user information from Channel 9

Something went wrong getting user information from Channel 9

Latest Achievement:

Loading user information from MSDN

Something went wrong getting user information from MSDN

Visual Studio Achievements

Latest Achievement:

Loading Visual Studio Achievements

Something went wrong getting the Visual Studio Achievements


androidi androidi
  • Surface Book battery life test

    I think there should be some tool that checks the configuration and what's running and then runs a standardized test. eg. for video playback the specs say 12 hours with i5 model w/o adaptive brightness & wi-fi connected (?).

    Seen a few user reviews and threads suggesting that some aren't getting that but without some battery perf validation tool it would be hard to say if the machine is in same state as that 12 hour test was done.

    For future it might be nice also if the products battery perf was characterized such that the validation tool could take into account how much use the batteries has seen. Gets a bit complicated to track that though but since MS now controls the firmware, it's certainly possible.



  • Hiring: Looking for a GREAT manual TESTER

    Idea for hard to fix intermittent bug repro contest (those seem to pile up on these latest gen devices)..

    But now that I thought about this, there is a problem. If the awards for producing fix/repro for hard to repro bugs was really worthwhile (for talented people) then there could be incentive to cheat. However I don't think this is a problem if the contest is limited to only short period after product launch. (so that no component vendor gets to create driver/fw with more bugs added in order to cheat in the contest)


    1. Make internal list of reported intermittent/hard to repro issues

    - If there is a report of such issue internally (or say 5+ external reports) but is not reproducible:

    2. Make external list where anyone that can repro the intermittent issue can win say next iteration of the product and whatever.. also have a way of submitting say some proof of how much work was it to repro, so if someone can externally prove they did put in some serious effort, they should have right to name their hourly price if they had to go around with kernel debugger or logic analyzer etc to track stuff down.



  • Hiring: Looking for a GREAT manual TESTER

    Not really related to this but some comments on the SB/SP4 issues which there seem to be a bit too many.

    I tried to think some way to preserve a release schedule while still allowing for larger internal QA.

    Back in early days of C9 recall "dogfooding" being talked about (prior to automated testing hype). When the decision was made to put brand new custom firmware to Surface, I think the launch should have gone 1. Launch 2. public reserve to assess demand (don't bill yet) & keep dogfooding (say produce some 1000 units and have them assigned to biggest internal complainers & no-men - assuming they weren't all fired lol) 3. don't bill yet (or just take email for availability reminder) and extend the shipping date until most frequent complaints are addressed. 4. if those 1000 units didn't have serious hardware issue then allow the testers to keep them or something as payment for the testing .. Put some kinda prizes that go down over time, so the testers that most quickly find serious issues get paid, so you get a bit of a window to fix them before starting to mfg the units going to public.




    note "been mentioned more than once" ... so it's already filtered list. Such issues should be caught in QA.

    So bleeding edge hw and firmware has some issues. There's probably an expectations management mismatch here - public is likely expecting 4th gen Surface Pro 4, when infact if you put in some brand new custom firmware made from scratch, it's really a 1st gen device, there's no going around that. Analog: Same hardware SSD but different vendor firmware - close study of performance finds the performance may have large anomalies despite same hw.



    I doubt consumers have much recourse beyond returning the product but I could see a scenario where there is unexpected amount of returns and poor tracking of what was the fault in each return. If there's a bunch of intermittent faults that are hard to repro and aren't recorded during the return process, you could end up with similar reputation as the rest of the PC oems.

    While all the comparable competition is still stuck in 16:9 displays, MS does have time to fix this but damage (lost Mac converts) probably piles up quickly given how impressive the launch hype was yet the fw/sw side is sounding like not quite meeting the expectations (reality or atleast perception of certain "polishedness" that the Apple devices tend to give, reality may not quite be that but competition with Apple is a lot about this perception is reality stuff).

    I read some of the updates that fix some big issues are not actually critical so the user has to do something to patch those. Probably the right decision in long term, but given the half-baked release state, might have been better to do some one-off first use mega patch that patched the FW etc as critical update.


  • October 6th ​Announcemen​ts

    I wish next year we'd see a Surface Pro with Book size display (maybe 14" 3:2) as then it's more of a portable desktop display+computer instead of a tablet or laptop.  (the lapability is probably a lot worse but what I really want is a decent size large & stable touch display/computer when sat on train/airplane plastic tray tables - then you can use bluetooth keyboard on your lap)

    Apple needs to do a bluetooth keyboard for that Surface Pro 14" with full in-home-pgup, del-end-pgdn, del-end-pgdn, arrow keys block.

    Now you might say, why just not buy a 15" MBP. Well maybe if Apple makes 14" 3:2 - 4:3 MBP and somehow manages to get full size navigation key block there ... No such laptop has ever been made afaik.


    I think the width of the laptop could simply be determined by what is the smallest size where you can cram standard size keys with navigation block. You could also do navigation keys inside numpad if you had 2 modes, where the numpad 0/ins was split into left,down and numpad del was right. (don't want to use standard numpad as nav keys as the layout sucks and remapping doesn't do any good as numpad-0/INS is wide key)

  • October 6th ​Announcemen​ts

    Found this video linked from reddit. Interesting demo of a tablet support, similar to the Apple smart keyboard what I was thinking might solve the SBook stability.


    Doesn't really bode well for solving the stability issue. Still if you had something like that with very high friction material at the bottom, you could atleast solve the problem of the device sliding on some airplane/train plastic tables (though the wobble could still present a big issue in train).

    The video actually goes to suggest that SP4 is likely more stable in lap (when used in certain manner) than the Book. I don't really think one should use these things on lap for extended times due to ergonomics but when you call it a "laptop" that just kinda warrants testing the lap usability.


  • Does the Surface Book have Apple quaking in their boots already?

    @spivonious: Except these devices were pitched for creatives, not average consumer - or perhaps I didn't get the joke you were making. I'm half thinking perhaps there was desire to put in the SM951 model but for whatever reason that didn't happen. 

    After looking at some more video reviews, the wobble some complained about appears to not be due to hinge but simply because the overall weight distribution causes the front of the base to lift off from lap/table easily. Lightest solution is completely ridiculous - some sort of way to make the bottom/base secure itself to whatever it is on and detect when you try to actually lift it off so it unsecures itself. In biology there are many kind of insects with methods to secure themselves to glass and clothes equally well.

    Given there's also that scratch issue, it seems the best 'after market' solution to fix the scratch and the wobble (and lack of kickstand) is "surface book cover" that is pretty much a copy of Apple Smart keyboard - so it goes to the bottom and extends the surface area of the base from behind which in turn prevents the front from lifting off. Maybe Apple should do a big smart keyboard that secures to the Surface, you can then use it either without the base (prop up the display lacking kickstand) or without...




  • Does the Surface Book have Apple quaking in their boots already?

    Looking at the reddit comments*, it would appear that the 128 GB models are PM951 and some say 256 GB of the reviewers are Toshiba.

    Kinda ugly situation when you don't know really what you're getting. (or atleast the rough seq/iops perf if no exact model)

    I looked at some Apple benchs** and it seems their write speeds are ~400 MB/s for the 13" MBP 128 GB model.


    Case could be made that if average consumer read a review of 256 GB model and then bought the 128 GB model, if the specs didn't mention anything about the write speed being less than half in the smaller model, the average consumer may have a case for returning the product for some period after they notice the issue. Atleast using a liberal interpretation of the EU consumer protection stuff... There's some pros to not committing to (too) exacting component specs but if you end up having large perf differentials in higher price device that aren't obvious prior to purchase I would be more concerned about getting bad reputation! Last time I bought a PSU I had to return one after I found the capacitors weren't all well known japanese brands as that particular detail is generally seen as big deal in PSU reliability after the bad cap fiasco.


    * https://www.reddit.com/r/Surface/comments/3r3dbi/the_128gb_surface_pro_4_models_use_perhaps_the/


    ** http://blog.macsales.com/25770-owc-unboxes-tests-ssd-speeds-of-new-13-and-15-retina-macbook-pros

  • Does the Surface Book have Apple quaking in their boots already?


    Apologies - you were right entirely now that I rechecked the model numbers. I didn't even know Samsung was doing a low performance OEM version of the NVMe nevermind it being found in this price class device!

    PM951 is indeed just "up to 150 MB/s" and SM951 128 GB model is "up to 650 MB/s".

    That's certainly below expectations - though given the thermal limitations the best you can get may just be some sort of "allow a quick burst if cool enough". Which would be fine if all you need the burst for is for say suspending or resuming a VM. That's really what I want the seq.speed for, so you can have a bunch of VM's suspended and quickly switch between them, take snapshots etc. I reckon some 10 second burst @ ~2000 MB/s would be enough for pretty much any snapshot/app load needs if it could be done say once per 5 minutes. I can accept that tablet device can't have the full speed going all the time.

    Now whether there's any power consumption issues in regards to having the 2000 MB/s (bursted if cool enough) read+write instead of 150 MB/s:

    Initially I thought there is but after some thinking I figured there actually is not. So it certainly does look like a cost cutting measure. The reason is that even if the data is concurrently interleaved to 8 nand's during write, it's not like you have to power up all the 8 nands simultaneously while reading the data. So after 10 seconds of concurrent bursty reading/writing just read/write the interleave sequentially powering the chips one at a time.



  • Does the Surface Book have Apple quaking in their boots already?

    Here's a note how the 950 nvme drives throttle if written to and gets too hot.


    the PM951 is supposedly OEM version of this drive. So in theory it should be capable of that 1700 MB/s but it's likely that to stop it from heating things up too much, the write speed might be capped.


    Despite the throttling and the TBW based warranty, given my current SSD is getting flaky and there's not much choices yet I'm seriously looking at the 950 pro because Intel has not promised OPAL for the 750 that I was looking to get. Kind of important if you want to run Bitlocker otherwise you lose the bleeding edge perf you're paying for when looking at nvme drives.


  • Does the Surface Book have Apple quaking in their boots already?


    Well looking at reviews (of large capacity models) the samsung has generally been near the top. So if I was buying a large capacity laptop, I certainly wouldn't mind having one of the components that have received good reviews. (Of course those reviews are often so cursory that they may ignore things like throttling due to overheating which has been said to be a problem with atleast the previous iteration of samsung pcie ssd's)

    Those models you suggested look rather low end if I as a buyer was expecting nvme pcie ssd like Apple has. (they too have the pm951 I believe)

    If there is a "dumb mistake" here it IMO is all about failing to notice that the lowest capacity model of pretty much every (premium) ssd model I've looked at past couple years is always much worse than the rest. To me the problem is not the choice of SSD model but starting the Book lineup at the lowest end model of a premium line up with high capacity - because they have optimized the configuration for the large capacity models. So for small size you'd have to go for different model. And then you'd surely have to mention in the Book specs that there's different brand components depending on the configuration. Right now I'm not sure they are mentioning the brand at all so MS is free to switch it (though if they switch all the capacities to one of the models you suggest and then the people buying the 256+ GB model will be undoubtedly questioning whether the performance is lower now than in the current PM951).

    I would say that if someone proved the pm951 still has throttling due to overheating that would not happen on some other brand ssd, then there's a good case for switching the brand. Right now the case I see is to remove the 128 GB model or atleast add a note in the specs that different SSD capacities have different write performance so buyers know to avoid the 128 GB model for reasons other than size.

    Though you do have a point in that at this price range, even if it added a bit into costs, sourcing the 128 GB model from another vendor (all else equal) seems reasonable. But you'd probably have to "pitch" why you did it as the "all else equal" may not hold - the write speed might be faster but those other models might be lower in 4K read for example, which is typically where Samsung has benched well. (this would show up in  a small file copy test when comparing equal size large capacity models, so it's likely all the drives you suggested are slower than pm951 in 512+ capacities when copying)

    Perhaps the most relevant bench (lacking in many reviews) I look is simple real world small file copy inside same ssd. Though in case of some of the "premium" models they may have some 3 GB of fast cache in front of the slower memory, or drivers (samsung) that may allow skewing results of this bench if the total amount of small files doesn't far outweigh the cache size.