Is This Computer Reporting The Wrong Temperature?


Recommended Posts

  On 11/01/2014 at 22:54, Boo Berry said:

Didn't older AMD CPUs (e.g. Phenom II's) report the wrong temperture? I remember hearing something like that, not sure.

Possibly on actual board probes, but doubtful on the CPU register. What I'd imagine happened is that software misinterpreted the CPU temperature register or other ACPI probes. I've definitely seen the latter (my Dell Core 2 laptop had this issue).

 

If there was a known CPU issue, we should be able to find it in AMD's errata list for their processors.

 

EDIT: Found a reference to K10s: http://en.wikipedia.org/wiki/Talk%3AAMD_K10#Unreliable_thermal_sensors.3Fhttp://support.amd.com/TechDocs/41322.PDF

 

EDIT2: I took a look and you are absolutely correct. This was for specific Phenoms, Phenom IIs, and Athlons.

  On 11/01/2014 at 22:53, snaphat (Myles Landwehr) said:

I see, well do report back here if you find out which components are replaced.  (Y)

Aye Aye Captain!

  On 11/01/2014 at 23:10, Boo Berry said:

I found this regarding AMD CPU temperatures (6 + 7).

 

EDIT: Apparently AMD's FM2 CPUs don't give accurate readings either.

 

Well there is a difference between the former and the latter. The temperature sensor is known to be completely unreliable in the K10s and acknowledged in the official errata (based on what I found in my edited post above). The temperature on the FM2 is just supposedly not as accurate as per core temperatures, which is what I've been saying. That doesn't mean that is completely wrong like the former. In a second, I'm going to try to check the FM2 errata and see if they list anything.

 

EDIT: Nope, I can't find any errata listed, so the FM2s temperature sensor is reliable if not particular accurate at temperature estimations.

  On 11/01/2014 at 23:20, Boo Berry said:

Which honestly makes me glad I don't have an AMD CPU anymore. Good luck to the OP, hopefully the replacement will address your issue! :)

Thanks for the input :) This is actually for a friend, I'm buying the parts for my new rig on Friday and I'll be getting a Core i-5 3350p :P Thank you though!

  On 11/01/2014 at 23:24, The Dingus Diddler said:

Thanks for the input :) This is actually for a friend, I'm buying the parts for my new rig on Friday and I'll be getting a Core i-5 3350p :p Thank you though!

Oh! Well in that case good luck for your friend and excellent choice for you. ;)

  On 11/01/2014 at 23:24, The Dingus Diddler said:

Thanks for the input :) This is actually for a friend, I'm buying the parts for my new rig on Friday and I'll be getting a Core i-5 3350p :p Thank you though!

 

I think we can guarantee your core temps will say what they really are. But there is an important clarification here. In the case AMD K10s, this wasn't a random on/off manufacturing defect, this was a design issue that affected entire processor families until they fixed it. Such things are completely normal. Every processor design has errata because they are too complex to know exact operation even by the engineers who design them. The K10 hardware bug is an example of something that probably should have been found before manufacturing though. Seems to me that it would be obvious under stress testing. 

  On 11/01/2014 at 20:24, snaphat (Myles Landwehr) said:
sniped

 

i know that the northbridge is part of the CPU die; it used to be a separate processor (combo with the southbridge) but not anymore, sorry if i didn't explain well. The fact here is that it's going to be hard to find the real cause of this problem. And the fact that the OP RMA'ed the whole computer would make very little possible to find if the CPU or the mainboard was the culprit.

After thinking about it, i believe i can hazard a guess as to why the k10 temperature hardware bug wasn't found. The engineers probably use undocumented per core sensors during testing and didn't test the aggregate sensor on high loads. undocumented features wouldn't exactly be abnormal. One of the tricks hardware engineers do is implement processor features a few generations prior to an actual documented unveiling of said features. This way they can silently internally do testing on the features and fix any hardware bugs in them without the public ever knowing they existed. For example, I would be willing to bet that a number of "32-bit" x86 processors are actually 64-bit processors (with 64 bit bus lines, 64-bit decoders, etc.) with the 64-bit feature-set completely disabled. You wouldn't ever know because the microcode flashing of the processors that went for sale would disable all undocumented features. 

  On 12/01/2014 at 01:16, snaphat (Myles Landwehr) said:

For example, I would be willing to bet that a number of "32-bit" x86 processors are actually 64-bit processors (with 64 bit bus lines, 64-bit decoders, etc.) with the 64-bit feature-set completely disabled. You wouldn't ever know because the microcode flashing of the processors that went for sale would disable all undocumented features. 

 

so very true.

  • Like 1
This topic is now closed to further replies.
  • Posts

    • I wish one of the windows updates hadn't broken glass. Is there a workaround for that I'm not aware of?
    • OpenAI announces o3 Pro, its most intelligent reasoning model by Pradeep Viswanathan OpenAI today announced o3-pro, its flagship reasoning model that uses more compute to "think harder" and provide consistently better answers. This new model will be replacing o1-pro in ChatGPT since it consistently performs better in math, science, and coding. To help everyone make the most out of the model, o3-pro supports tool calling. So, based on the user prompt, the model can do a web search, analyze files, reason about visual inputs, use Python, personalize responses using memory, and more. This is a big improvement over o1-pro, which lacked access to tools. But the downside is that o3-pro will take more time to respond. OpenAI recommends users use o3-pro in cases where reliability matters more than speed. According to OpenAI's evaluations, users consistently preferred o3-pro's responses over o3 in key domains like science, education, programming, business, and writing help. o3-pro was also rated consistently higher for clarity, comprehensiveness, instruction-following, and accuracy. The new o3-pro model is available in ChatGPT's model picker for Pro and Team. ChatGPT's Enterprise and Edu users will get access to this latest model next week. For developers, the o3-pro model comes with a 200,000 context window and is priced at $20 per million input tokens and $80 per million output tokens. OpenAI recommends developers use background mode with o3-pro to prevent timeouts. And the model has a May 31, 2024 knowledge cutoff. The OpenAI team also noted the following as the current limitations of the o3-pro model: At the moment, temporary chats are disabled for o3-pro as we resolve a technical issue. Image generation is not supported within o3-pro—please use GPT-4o, OpenAI o3, or OpenAI o4-mini to generate images. Canvas is also currently not supported within o3-pro. As OpenAI continues to refine its models, the balance between speed and reliability will likely remain a key consideration for users choosing the right tool for their needs.
    • What annoys me is updates take forever to install on my Pixel 9 Pro XL, it was never so slow on my Samsung devices.
  • Recent Achievements

    • Reacting Well
      rshit earned a badge
      Reacting Well
    • Reacting Well
      Alan- earned a badge
      Reacting Well
    • Week One Done
      IAMFLUXX earned a badge
      Week One Done
    • One Month Later
      Æhund earned a badge
      One Month Later
    • One Month Later
      CoolRaoul earned a badge
      One Month Later
  • Popular Contributors

    1. 1
      +primortal
      542
    2. 2
      ATLien_0
      269
    3. 3
      +FloatingFatMan
      209
    4. 4
      +Edouard
      203
    5. 5
      snowy owl
      140
  • Tell a friend

    Love Neowin? Tell a friend!