Observations with an FX-8350

Discussion in 'CPUs and Overclocking' started by Idontcare, Dec 15, 2012.

  1. Idontcare

    Idontcare Elite Member

    Joined:
    Oct 10, 1999
    Messages:
    21,130
    Likes Received:
    0
    Figured I'd make a thread regarding the trials and tribulations with my FX-8350. This is a new area for me, so expect some noobness to my efforts, feedback and pointers will be much appreciated as I attempt to scale the new learning curve :thumbsup:

    A trip down memory lane:

    The last time I had the opportunity to really dive into AMD hardware it was circa 1998 or 1999 and I had just acquired 24 800MHz K7 systems with a government grant in the name of assembling what was called, at the time, a "beowulf class supercomputer". It was basically a cluster of desktop PC's, connected by a high-speed interface (100Mbit ethernet at the time) that ran multi-threaded applications much as we all do today with our multi-core processors.

    [​IMG]

    ^ 12-node cluster of 800MHz K7's.

    [​IMG]

    ^ 24-node cluster of 800MHz K7's, built by me and my lab mate Jason.

    The primary function of this cluster was to run a parallelized version of a computation chemistry application called Gaussian98. (quantum chemistry modeling) We were designing and building molecules which would be used to harvest/absorb sunlight and split water into hydrogen and oxygen gas. The modeling helped us decide which molecules would be worth attempting to engineer in the lab, which we also did right next door.

    At any rate this cluster had long been my benchmark for "performance in a single CPU" because I longed for the day when I would have the same performance as this cluster but in a single cpu that could be purchased at commodity prices.

    Enter the FX-8350! 4GHz of 8 core goodness (or 4 cores, depending on how you count ALU vs FPU).

    From a straight dumb-flops computation (thanks to guskline for motivating me to think about this) the old cluster was capable of 24*800MHz = 19,200 MHz of K7 IPC. The FX8350 would be capable of 8*4000MHz = 32,000MHz of K11 IPC. (yes I know piledriver is not a K11, but there is no harm in making the analogy IMO)

    Surely K7 IPC < K11 IPC, so from a brute flops standpoint the FX-8350 probably has at least 2x the capability of my previous benchmark, the 24-node beowulf cluster.

    Not to mention the cost and power-consumption reduction that comes with an FX-8350 over the old cluster, and just from a span of ~12yrs.

    All right, enough nostalgia, lets talk about the FX-8350 itself.

    First thing is the testbed, I bought the Crosshair V Formula-Z for one specific reason - the ProbeIt belt.

    I learned with my MIVE-Z in testing the 2600k and 3770k that relying on CPUz, or any other software voltage reporting tool for that matter, to document the applied CPU voltage was a pursuit of one blind man by another. If you truly want to know what voltage your CPU is experiencing you simply have no choice but to get a physical voltmeter connected to the circuit.

    [​IMG]

    Now one obvious issue with the probe-it belt is the matter of relative dimensions. Those probe pabs are silly tiny and close together. I also did not like my prospects of being able to maintain electrical isolation between the probes themselves (the alligator clips in my case).

    So I realized first thing I needed to do here was to electrically isolate the pads, and the clips, lest I make some sparks by accident and let out the blue smoke :eek:

    I turned to the oldest trick in the book, black electrical tape and masking off the work area:

    [​IMG]

    ^ here I've isolated the two probe pads of interest.

    [​IMG]

    ^ then I masked off the alligator clips themselves, and using a razor blade I opened just one metal tooth at the very front of the clip, but taking care to keep the sides of the tooth still covered with the electrical tape. (I don't want the alligator clips to short out each other)

    And the final assembly:

    [​IMG]

    OK, all is well and good, no risk of shorting now.

    And the result?

    [​IMG]

    For my mobo, which is updated with the latest BIOS, CPUz under-reports the idle voltage by a hefty 0.108V! That is a rather large miss.

    And at full load (LinX, 8 threads):

    [​IMG]

    The gap between actual and CPUz reported has markedly narrowed but it is still the case that CPUz is under-reporting the actual Vcore. In this case it is off by ~0.010V.

    I don't have the screen grab comparison to show you here, but in my various runs of LinX so far the delta between CPUz and actual Vcore tends to run about 2x this - i.e. ~0.020V error on behalf of CPUz at load volts.

    Is this gap specific to my mobo? Or is it common to everyone's Crosshair V Formula-Z? I can't say, but it just goes to show that relying on CPUz values is pretty much a crapshoot.

    Another anecdotal observation, not captured in these static images but notable by the eye in real-time, is that CPUz likes to shake around the reported Vcore like a cat playing with a ball of string; whereas the measured values are much more stable and consistent.
     
  2. OVerLoRDI

    OVerLoRDI Diamond Member

    Joined:
    Jan 22, 2006
    Messages:
    5,470
    Likes Received:
    0
    Looking forward to your testing and I'm sure the comparison between your current data sets with the 2600k and the 3770k!
     
  3. thilanliyan

    thilanliyan Diamond Member

    Joined:
    Jun 21, 2005
    Messages:
    8,884
    Likes Received:
    2
    Thanks for the detailed info on this...I may pick up a 8350/8320 if the price is right on boxing day.

    If you are able to provide some power consumption numbers while overclocked that would be great...I'm trying to figure out if my 2x120mm + 1x120mm rads can handle my 7950 and a 8350. Currently my CPU at 4.2GHz maxes out at about 68C running P95 and running bitcoin mining as well (100% CPU and GPU load).
     
  4. guskline

    guskline Diamond Member

    Joined:
    Apr 17, 2006
    Messages:
    4,627
    Likes Received:
    11
    Great job! I'm fascinated by the depth of testing.
     
  5. Idontcare

    Idontcare Elite Member

    Joined:
    Oct 10, 1999
    Messages:
    21,130
    Likes Received:
    0
    This kind of surprised me, but I'm sure people who use these FX chips won't be surprised, but the following is a power-consumption measurement at stock:

    Idle:

    [​IMG]

    Loaded:

    [​IMG]

    TDP for this chip is 125W. But the loaded CPU, at stock mind you, is pulling nearly 200W D:

    I'm use to seeing that kind of power-draw for an OC'ed Intel setup, but the stock power consumption is always well below the TDP rating (even with LinX).

    But this is quite an eye-opener. Why bother rating the CPU as 125W if it is going to consume 200W at stock? Why not just rate it at what it is - a 200W TDP SKU? There is no shame in being true to what you are, but why mislead? I just don't get it.

    For my i7-2600k, a 95W TDP SKU, at 3.8GHz LinX loaded the chip uses 85W.

    For my i7-3770k, a 77W TDP SKU, at 3.9GHz LinX loaded the chip uses 66W.

    For my FX-8350, a 125W TDP SKU, at 4GHz LinX loaded the chip uses ~195W.

    I know everyone says AMD TDP != Intel TDP, but they aren't even close in this regard.

    I know folks will be saying "but the AMD chip is at 4GHz, what if you OC the Intel chips to 4GHz too?"

    i7-2600k at 4.0GHz LinX loaded uses 95W.

    i7-3770k at 4.0GHz LinX loaded uses 70W.

    FX-8350 at 4.0GHz LinX loaded uses 195W (and probably more than that as I haven't fully accounted for the leakage contribution yet).

    I also notice that the reported idle temps are impossible, some 10C below ambient. Now this can be due to a poorly calibrated thermal probe or the TJmax used by CoreTemp is wrong.

    Does anyone know what TJmax is for the FX-8350. What I'm most interested in is an actual spec value from AMD, not just opinion or expectation per se.

    Is 90°C the correct TJmax?
     
  6. Ferzerp

    Ferzerp Diamond Member

    Joined:
    Oct 12, 1999
    Messages:
    6,107
    Likes Received:
    1
    Intel ones do this too. I often see idle temps that are at or below ambient. From my understanding, the farther you get from Tjmax, the less accurate they become. At idle temps, they are so inaccurate that they're nearly useless.
     
  7. AtenRa

    AtenRa Lifer

    Joined:
    Feb 2, 2009
    Messages:
    11,270
    Likes Received:
    51
    IDC,

    281W is for the entire system not the CPU alone. Dont forget you have a northbridge (990FX) that you luck in Intel's 1155 socket because it is integrated in to the CPU die.

    Edit: I have found that HWiNFO64 provides more accurate Idle temperature readings.
     
  8. Subyman

    Subyman Moderator <br> VC&G Forum
    Super Moderator

    Joined:
    Mar 18, 2005
    Messages:
    7,825
    Likes Received:
    2
    I'm wondering if there could be any backlash from the unrealistic TDP given by AMD. A lot of people use these numbers to build servers, HTPCs, or workstations. These can be power sensitive environments, especially if many units are purchased. Could there be a class action against AMD for falsely advertising their chips to be more power efficient than they really are? I would understand if there was a gray area, such as +/-20W or so, but almost doubling the TDP seems like blatant misrepresentation of their product.

    Just curious if something like this has ever come up.
     
  9. ShadowVVL

    ShadowVVL Senior member

    Joined:
    May 1, 2010
    Messages:
    754
    Likes Received:
    0
    Can you test the wattage from the cpu and ground points?

    Not sure if the make multi meters that test wattage.cant seem to find anything for that except rc car testers.

    http://www.amazon.com/Watts-Meter-An...a-watt+monitor

    I never power tested a cpu before but maybe you can multiply the amps and volts to find the watts unless this wont work for cpu.
     
    #9 ShadowVVL, Dec 15, 2012
    Last edited: Dec 15, 2012
  10. thilanliyan

    thilanliyan Diamond Member

    Joined:
    Jun 21, 2005
    Messages:
    8,884
    Likes Received:
    2
    Thanks for the info...lol, looks like my setup won't be enough.

    For temps, I went here:
    http://products.amd.com/pages/DesktopCPUSideBySide.aspx?id=772&id=811&id=809&id=810&id=770

    FOr the 95w chips it seems to be 70C, and for 125w it is 61C...although I have no idea whether that is accurate as you can probably hit over 61C on the stock cooler at stock.
     
  11. Idontcare

    Idontcare Elite Member

    Joined:
    Oct 10, 1999
    Messages:
    21,130
    Likes Received:
    0
    It is for the entire system, but so too is the 87W at idle.

    The only difference between the 281W at load and the 87W at idle is the CPU which is loaded at 281W, versus the CPU being idle at 87W.

    So to be sure a portion of the 87W at idle is attributable to the CPU, but even if we (for now) make the assumption the attribution is zero, the difference between idle and load is 194W which has no place to come from but the CPU.

    I usually extract the CPU's power consumption by means of running a series of tests which are designed to isolate the power usage of the platform from that of the CPU.

    For working examples, see here and here. I will do the same with my FX-8350, but for now I just have a handful of data to speak to in terms of lower-limits.
     
  12. khurios2000

    khurios2000 Diamond Member

    Joined:
    Dec 11, 2003
    Messages:
    7,530
    Likes Received:
    1
    Love that 9'c idle temp, while 19'c ambient.
     
  13. inf64

    inf64 Platinum Member

    Joined:
    Mar 11, 2011
    Messages:
    2,250
    Likes Received:
    7
    Something else on the motherboard is pulling the additional Watts besides the CPU,maximum rated power for FX is 125W. I will try and see if I can find a review that measured the CPU power directly at the socket level.

    Another note: are you sure the "auto" voltage values in bios are correct for your CPU? Sometimes auto is just plain wrong(read high).

    edit: got it I think.
    http://www.hardware.fr/articles/880-4/consommation-efficacite-energetique.html

    Click "ATX12V" under the first graph. Maximum they get for 8350 is 116W. This is within the rated TDP but still lot more than intel chips (or even X6). On the system level FX8350 is pulling 189W in their case. That's around 86W more than 3770K system so efficiency of other components come to play too.
     
    #13 inf64, Dec 15, 2012
    Last edited: Dec 15, 2012
  14. AtenRa

    AtenRa Lifer

    Joined:
    Feb 2, 2009
    Messages:
    11,270
    Likes Received:
    51
    Question, have you enabled C6 mode in the bios ??? if yes then out of the 87W measured in Idle mode the CPU only uses a fraction of that. The rest is from the system.

    I believe here is the error, you have no idea how much power each component of the system uses when you run LinX. You assume that every other component (NorthBridge, Memory, SouthBridge etc) operates at the same capacity and voltages and current you have measured when the system was at Idle. That is INCORRECT.

    Can you measure the Current (I) for the CPU at Idle and Full load ??? if yes then you can have a closer estimate of the CPU power consumption since you know the Voltage.

    Because it is very difficult to measure the Current(I), i would suggest to disable any C states from the bios. By doing this you will make the system to use the highest power in Idle mode. Now, measure the Idle power and then measure the Full load power. You will see that the difference will not be 200W ;)
     
  15. ShintaiDK

    ShintaiDK Lifer

    Joined:
    Apr 22, 2012
    Messages:
    20,091
    Likes Received:
    15
    Maybe its 125W ACP ;)
     
    #15 ShintaiDK, Dec 15, 2012
    Last edited: Dec 15, 2012
  16. sequoia464

    sequoia464 Senior member

    Joined:
    Feb 12, 2003
    Messages:
    870
    Likes Received:
    0
    This 'feature' is fairly well known by the people that are AMD users. Apparently the algorithm used to calc temps is accurate under load but sucks at idle.

    Don't really know anything technical about it myself other than that what you are seeing is echoed fairly often.

    Look at posts 2668-2669 here .. http://www.overclock.net/t/1318995/official-fx-8320-fx-8350-vishera-owners-club/2660#post_18551388 - a search through that thread will bring up other similar posts on the idle temp. As always, not certain of the voracity of these posts, just sharing what I have seen.
     
    #16 sequoia464, Dec 15, 2012
    Last edited: Dec 15, 2012
  17. SlowSpyder

    SlowSpyder Lifer

    Joined:
    Jan 12, 2005
    Messages:
    10,133
    Likes Received:
    4
    Wow, that's a lot of power consumption. I'm sure the memory contributes somewhat to that, but that is still a low of power being pulled. If it makes you feel any better, I believe I can get my PC up to ~480 watts (if my memory is correct) when I load up my Thuban. :)



    *edit - Thanks for sharing your trip down memory lane with us. You have done some pretty cool stuff... you make me regret not continuing with school! ;)
     
    #17 SlowSpyder, Dec 15, 2012
    Last edited: Dec 15, 2012
  18. GammaLaser

    GammaLaser Member

    Joined:
    May 31, 2011
    Messages:
    173
    Likes Received:
    0
    Some percentage of that system power consumption could also be attributed to power delivery losses in the VR. It certainly wouldn't be enough to bring the CPU-only power to be below its rated TDP though.
     
  19. VirtualLarry

    VirtualLarry Lifer

    Joined:
    Aug 25, 2001
    Messages:
    33,374
    Likes Received:
    34
    I've got two rigs, with Thuban 1045T chips running at stock (2.7Ghz), both with two 9600GSO cards. (One of them was an SC model, and one was a 768MB model.)

    One of the rigs, with an Antec 650W Earthwatts PSU, pulls 306W at the wall (KAW), the other with a Xion PowerReal 800W (less efficient, I believe, no active PFC), pulls ~360W.

    Load was measured running BIONC 64-bit on Win 7 64-bit, running WCG WUs on the CPU cores, and PrimeGrid on both GPUs.
     
  20. BD231

    BD231 Diamond Member

    Joined:
    Feb 26, 2001
    Messages:
    9,754
    Likes Received:
    1
    Its like bulldozer all over over again.
     
  21. bononos

    bononos Diamond Member

    Joined:
    Aug 21, 2011
    Messages:
    3,198
    Likes Received:
    1
    I thought AMD temps (for load and idle) reported from internal thermistor were not real temps. And this has been true since the beginning when AMD reported core temps I believe. Thats why some utilities report -ve temps at idle and cpus throttle or freeze in the high-60s.
     
  22. Greenlepricon

    Greenlepricon Senior member

    Joined:
    Aug 1, 2012
    Messages:
    468
    Likes Received:
    0
    That power draw is one reason I'm afraid to overclock both piledriver and my gpu at the same time on my 600 watt psu. Still that seems a little high. Maybe it's time for me to buy a kill o watt so I can test my own power usage. I'm interested to see if it's worth the extra power to get the performance bump when you overclock. So far amazing job testing. IDC you never fail to impress.
     
  23. Ferzerp

    Ferzerp Diamond Member

    Joined:
    Oct 12, 1999
    Messages:
    6,107
    Likes Received:
    1
    What happens to power usage if you use a smaller ram size for linpack (or undervolting the ram if it's still stable, or even overvolting it to get data)? I'm sure you see where I am going with this. There is another potentially isolatable source of significant power draw here.
     
    #23 Ferzerp, Dec 15, 2012
    Last edited: Dec 15, 2012
  24. Torn Mind

    Torn Mind Platinum Member

    Joined:
    Nov 25, 2012
    Messages:
    2,723
    Likes Received:
    1
    Wow, that is truly a guzzler we have here. Even if it isn't all the CPU and you can adjust settings, for the poor fellows who buy this, a 500W+ PSU and aftermarket cooling is practically a necessity.
     
  25. Markfw900

    Markfw900 Super Moderator <br>Elite Member<br> CPUs & Overcl
    Super Moderator

    Joined:
    May 16, 2002
    Messages:
    13,447
    Likes Received:
    5
    Well, only somewhat related, but I have 2 opteron 6234's, 12 cores each, and 2 GTX 460's and with both cards @ 100% and all 24 cores @ 100%, they take 470 watt's total. It's the same cores, just 12 instead of 8 and 2 chips, except they are at 2.4 ghz. But there are 3 times as many cores !!! And with the 2 460's taking at least 200 watts. not bad total.