![]() |
1700 + Taichi insane power use |
Post Reply ![]() |
Author | |
Atan87 ![]() Newbie ![]() Joined: 01 May 2017 Status: Offline Points: 25 |
![]() ![]() ![]() ![]() ![]() Posted: 01 May 2017 at 5:10pm |
Hello. According to HWInfo im using ~140A and almost 200W of power as cpu core power. Can this be correct? I thought AM4 socket has limit of 128W. Also it seems that other users having even higher clocks and volts are getting less power usage than me, are my settings wrong or is taichi really pushing this much current to the core? Second question, which is the correct reading for core voltage? As im seeing two different values for core voltage, other one reported from cpu and other from MB.
HWInfo screenshot provided for analysis http://imgur.com/a/LjIat |
|
![]() |
|
parsec ![]() Moderator Group ![]() ![]() Joined: 04 May 2015 Location: USA Status: Offline Points: 4996 |
![]() ![]() ![]() ![]() ![]() |
While I normally trust HWiNFO more than any other monitoring program, it might not be correct on all of those power readings. The programmer does a fantastic job getting things right, but most of the time he does not get cooperation from mother board manufactures about the details of the monitoring chip's outputs, etc. Also, I'm not to sure that Ryzen is as power efficient as it is made out to be. But that must be taken in context of comparing apples to apples. Yes Ryzen may be more power efficient than Intel's 140W TDP HEDT processors, but comparing Ryzen to four core processors of all types is not a fair comparison. What CPU are you using? First, what are you doing on that PC to cause that much power usage? Running CPU stress tests? I see 100% CPU usage in your screenshot, but we don't know what exactly you were doing. One thing I don't see in your screenshot is something I see in HWiNFO with my ASRock X370 Killer SLI/ac board. That is the CPU and SOC VRM readings, that appear below the readings on the lower right in your screenshot. Mine for example have the title, "ASRock X370 Killer SLI/ac (IRF IR35204)", which are the type of VRM chips used. I think the VR VROUT voltage and power readings are the most accurate, and closely matches the VCore reading. For example: ![]() Your VCore seems higher than necessary, at least compared to my 1700X. That may increase your power usage. I don't know what you are comparing your power usage to, but unless they are identical hardware wise and doing the same things like CPU stress testing, the comparison may not be identical. |
|
![]() |
|
Atan87 ![]() Newbie ![]() Joined: 01 May 2017 Status: Offline Points: 25 |
![]() ![]() ![]() ![]() ![]() |
Those readings are from single cinebench run. That vcore is minimum i get my cpu stable at 3825mhz.
Also im doing some testing atm, sofar by disabling SMT my power usage has gone down like 20% while being stable at 3900mhz with same voltages. Can smt really cause so much more power being used and instability in overclocks? Or is my instability caused by high power usage overall? |
|
![]() |
|
Atan87 ![]() Newbie ![]() Joined: 01 May 2017 Status: Offline Points: 25 |
![]() ![]() ![]() ![]() ![]() |
For the first time ever i managed to get 4ghz cinebench stable by upping voltage a little and disabling smt. As you can see, im using much less power for this. Amps are down by 28A and 40~ watts less.
Also cinebench score is down more than i anticipated. I thought 1 smt thread is ~0.25 real cores in optimal conditions. So by disabling smt i go from 10 effective cores to 8 but my CB score is down to 1250@4ghz from 1701@3.85ghz. Bug in the microcode or bios perhaps? Screenshot for analysis http://imgur.com/a/VbGsu Edited by Atan87 - 02 May 2017 at 1:03am |
|
![]() |
|
wardog ![]() Moderator Group ![]() Joined: 15 Jul 2015 Status: Offline Points: 6447 |
![]() ![]() ![]() ![]() ![]() |
You want, actually need to, Disable SMT when OC'ing for stability. It's just the way SMT is and what it accomplishes in NON-OC'd useage.
HPET should always be Enabled. OC'ing or not. Or so I've discovered in OC'ing my 1500X. Change either from the above and instability rears its ugly head. |
|
![]() |
|
Atan87 ![]() Newbie ![]() Joined: 01 May 2017 Status: Offline Points: 25 |
![]() ![]() ![]() ![]() ![]() |
![]() |
|
wardog ![]() Moderator Group ![]() Joined: 15 Jul 2015 Status: Offline Points: 6447 |
![]() ![]() ![]() ![]() ![]() |
Leakage would show in higher temps
|
|
![]() |
|
parsec ![]() Moderator Group ![]() ![]() Joined: 04 May 2015 Location: USA Status: Offline Points: 4996 |
![]() ![]() ![]() ![]() ![]() |
It looks like we are seeing the differences between Ryzen processor models here. It looks like there are some significant differences in the capabilities of the different models, although the sample size here is tiny.
One thing first, your board is not pushing current to the CPU, that is simply what the CPU is using. You changed the amount of power being used by disabling SMT, and the board supplied what was needed. Here's my Cinebench R15 run with my 1700X at 3.925GHz, 3.917GHz due to BCLK droop, VCore set to 1.300V, SMT enabled: ![]() Why Cinebench R15 sees Windows 10 as Windows 8 is weird. My "OC" is hardly an OC, more like the Multi-core Enhancement feature seen on ASRock Intel chipset boards. My CPU Core Current is lower than yours, but my CPU Package Power is higher, since I'm using SMT. Your SoC Current is about three times higher than mine, which adds to the higher CPU + SoC Power total. It looks like, at least on your CPU, the OC takes more power and VCore for stability. That's normal for any processor, the steps in power usage vs 100MHz steps in OC are not linear, and it gets worse as the OC increases. Are you using the AMD/Ryzen Power Plan for Windows? Or the latest chipset drivers, 17.10? |
|
![]() |
|
Atan87 ![]() Newbie ![]() Joined: 01 May 2017 Status: Offline Points: 25 |
![]() ![]() ![]() ![]() ![]() |
![]() |
Post Reply ![]() |
|
Tweet
|
Forum Jump | Forum Permissions ![]() You cannot post new topics in this forum You cannot reply to topics in this forum You cannot delete your posts in this forum You cannot edit your posts in this forum You cannot create polls in this forum You cannot vote in polls in this forum |