wacom's newest tabletPC driver issues!

Discussion in 'Fujitsu' started by stoneseeker, Oct 12, 2012.

Thread Status:
Not open for further replies.
  1. Plognark

    Plognark Scribbler - Standard Member

    Messages:
    176
    Likes Received:
    25
    Trophy Points:
    41
    Update: I'm reaching out to a bios mod forum to see if we can get the bios unlocked for the T902. I'm giving myself a crash course in system throttling on laptops :p

    ETA: One of the things I've noticed during testing based on some of the latency monitoring software is that the USB has really bad latency; I don't know if this is a result of all the tweaking and tuning I've been doing or not.

    I've played around with some of the available settings I can get to through throttle stop, but all I was able to do was make the glitch worse.
     
    Last edited: Jul 19, 2013
  2. stoneseeker

    stoneseeker Animator and Art Director Senior Member

    Messages:
    3,604
    Likes Received:
    1,074
    Trophy Points:
    231
    Sadly, throttlestop is near useless on our systems with the BIOS locked down so hard. I believe the maker of our BIOS/motherboard is Phoenix..?
     
  3. Plognark

    Plognark Scribbler - Standard Member

    Messages:
    176
    Likes Received:
    25
    Trophy Points:
    41
    Yeah, it's Phoenix. A securecore Tiano, to be more specific.

    I'm still researching when I get the chance, although work and general life obligations are making it tough.

    I think what's happening, after trying everything else on the planet, is that we're getting a USB buffer underrun. You usually see these with audio, from what I've read, but the repeating brush stroke seems to match up with the behavior of an audio underrun where a ring buffer will repeat whatever it has stored again if the data being fed to it is too slow. In this case I think it's looping over the pen data, rather than an audio feed.

    I don't think there's anything wrong with the buffer, I just think that the power management lockdown in BIOS is preventing an adequate data rate getting to the USB buffer. I think whenever it kicks in and throttles down the CPU/GPU there's an underrun that bogs everything down.

    I've had a little bit of luck reducing the glitch with throttlestop by lowering my CPU multiplier down to 20-ish, but it still happens enough to be annoying. The i7 is such a beast that cutting it back by a third still leaves a ton of surplus computing power.

    I don't know if the throttling is based on temperature or wattage; maybe it's both. There are a lot of throttling schemes out there; it sucks that this is common practice with laptops, because buyers really aren't getting the advertised hardware.

    Anyway, I haven't gotten a reply yet on unlocking the bios from the forum (there's a huge backlog of requests), so I guess I'm going to have to learn how to do it myself. In the meantime, I may try to contact Fujitsu and beg for them to release an unlocked bios. I doubt I'll get anywhere, but it can't hurt to ask, right? :)
     
  4. stoneseeker

    stoneseeker Animator and Art Director Senior Member

    Messages:
    3,604
    Likes Received:
    1,074
    Trophy Points:
    231
    Your summary is ringing VERY true. It seems that the 2 issues I have noticed with this machine may very well be related to an extent then. Aggressive power throttling when a certain limit is approached that was set by Fujitsu to cover their asses causing a buffer underrun makes perfect sense and seems highly likely.

    If they are related, then I believe thermals are very much the reason it throttles. If you use Realtemp to monitor the CPU heat, it is often quite high while doing medium - intense tasks. When gaming or stressing the GPU especially, the CPU and GPU die get easily up to the low 90's before throttling power kicks in.

    One time while trying to figure out if it was indeed the thermals or something else, on a particular COLD day (our heater was broken in the winter) I tried playing Skyrim while monitoring the GPU/CPU temps. The temps for the first time ever were able to stay under 80 for a long time, and as a result, the system DIDN'T throttle once for the entire duration of play, which had never and has never happened since.

    My thought is, what if we can solve this issue in a round-about way by improving thermals on the T902. If one of us under arranty with in-home service can place a free appointment to have the thermal paste reapplied with better (arctic silver 5 or something) thermals paste, maybe our issue will be quelled to some degree. Perhaps some of our systems are running hotter than others on average due to temperature and pasting variables, which is why some are experiencing it worse than others and it can be terribly inconcsistent. Thermal throttling would also explain why it gets worse after extended use of the system or when it sunder heavy load.

    If one of us is bold enough to ope it up and replace the thermal paste and have a look a tthe heat sink, perhaps we can come up with a way to limit or prevent the throttling that may be causing our glitch.
     
  5. jan_c

    jan_c Pen Pal - Newbie

    Messages:
    48
    Likes Received:
    4
    Trophy Points:
    16
    That correlation makes total sense. I've been a bit too busy to deal with this lately, but I'm getting started with some USB packet capture. I hope to log some interesting results soon.
    [​IMG]
    If anyone wants to give this a try, get USBPcap to do the capturing and Wireshark to analyze the logfile
     
  6. jan_c

    jan_c Pen Pal - Newbie

    Messages:
    48
    Likes Received:
    4
    Trophy Points:
    16
    The high frequency at which the glitch happens to me supports your claim. I assume your temps are in Fahrenheit? I work in a pretty warm environment.. about 31°C (88°F), and my i7 idles at 50°C (122°F). Loads typically hit ~70°C (~158°F)...
     
  7. stoneseeker

    stoneseeker Animator and Art Director Senior Member

    Messages:
    3,604
    Likes Received:
    1,074
    Trophy Points:
    231
    No, I am talking Celsius. When I just have some tabs open, and things are at typical light use, I am still hitting 50-56°C.
    When I am working, things easily spike to the 70's and up and thats when I start seeing the issue... when gaming, I've seen it hit all the way to 94°C!! The TDP max is 105°C.

    This thing is packing a FULL voltage CPU into a fairly tight space for one. Still, something feels like its mishandling that heat. I knwo the T901 had a very well-known heat issue as well. It often overheated and throttled. In my monitoring in the past though I haven't seen the speed of the CPU throttle like the GPU does ...
     
  8. Plognark

    Plognark Scribbler - Standard Member

    Messages:
    176
    Likes Received:
    25
    Trophy Points:
    41
    Whoa, that's pretty hot. Mine idles at around 50 C, runs about 57-60 C while drawing. I've noticed that those repeated hatching strokes that set this issue off will push the temps to around 65-70 C if I'm drawing quickly and not taking a break.

    This also fits with my experiences with the T901. It had the same sort of glitch, and I imagine it was also a thermal issue.

    Stoneseeker, I've come across tons of past threads of yours trying to sort out the GPU issues with the T902.
    They actually helped me dig into the possibility that this is a thermal throttling issue quite a bit. Your discussions with the throttlestop guy were incredibly helpful :)

    I wonder what would happen if I set this thing on top of an ice pack during testing...
     
  9. darkphantom

    darkphantom Scribbler - Standard Member

    Messages:
    139
    Likes Received:
    0
    Trophy Points:
    30
    Since I have plenty of TIM lying about (thermal paste for those not in the know :p ) I will go ahead and monitor the temps today and log it. I will then, when I get a chance, replace the stock TIM with something like MX-5 or this AS Nano diamond compound I have sitting here.
     
  10. Plognark

    Plognark Scribbler - Standard Member

    Messages:
    176
    Likes Received:
    25
    Trophy Points:
    41
    Cool, I'll try to give it a shot.

    I've noticed with latencymon that the T902 has periodic USB latency spikes that max out the meter. The warnings from that software were the first time I've ever heard the term underrun :D

    Thermal issues also fits with some of the past testing we've done, where it will seem like the problem is gone, but will come back after a period of use. How many times have we seen claims that something or other fixed the glitch, only to get a report half an hour later that the damn thing is back?
     
Thread Status:
Not open for further replies.

Share This Page