The AMD FX-9590 "FX Centurion" Preview

[bhtan83]

And that's actually quite different from what I had posted above.

" accurate only from around 40C onwards" ie. inaccurate below ~40C, more accurate above ~40C

Most DTS on CPUs are calibrated to get more accurate the closer it gets to the throttling temperature (the primary reason for having the DTS ie. prevent thermal damage) and it's not just AMD. Intel CPUs have inaccurate idle DTS readings too and some have even stuck idle DTS as well ie. the DTS does not read below a certain value even if you cool it down further.

All in, the DTS purposes to serve more as a guide to trigger thermal throttling in the CPU and not so much for reading idle temps.

HTH

i see. thanks for sharing man.

 

[hlots123]

jus curious, does "CPU" circled in blue refer to temp at the motherboard cpu socket? if yes, then shudn't the cpu itself also be in that range too (actually slightly higher)...? eg. maybe abt 70-75C instead of 49.4C.

 

[Kyo_kusanagi86]

suspicious kicks in on TS cpu cooling temps. Can any fellow hwz bro with haswell i5 processor do a comparison with TS ma?

 

[MacClipper]

jus curious, does "CPU" circled in blue refer to temp at the motherboard cpu socket? if yes, then shudn't the cpu itself also be in that range too (actually slightly higher)...? eg. maybe abt 70-75C instead of 49.4C.

Till date, think no one has specified where the Nuvotron CPU sensor on the mobo being reported in hwinfo is situated- I suspect it may be actually reporting the CPU VRM temps cos running a FX-8350@4.5GHz brings it > 80C In my testing, I have ever brought it past 90C without triggering any CPU thermal throttling.

Screenshots are from my older 990FX Extreme9 posts starting from here
http://forums.hardwarezone.com.sg/73608645-post180.html

My PB accounts are overloaded to the max so remove the space after http and copy & paste in your browser window to view directly

http ://i755.photobucket.com/albums/xx191/MacClipper2/Asrock/990FX%20Extreme9/P95Blend45GHz_15mins_OKCopy_zps7a3812b1.png

http ://i755.photobucket.com/albums/xx191/MacClipper2/Asrock/990FX%20Extreme9/ripbot26445_OKCopy_zps252ce14c.png

suspicious kicks in on TS cpu cooling temps. Can any fellow hwz bro with haswell i5 processor do a comparison with TS ma?

Maybe Try reading my HW thread
http://forums.hardwarezone.com.sg/h...survey-h-o-t-s-4215572.html?highlight=lga1150

 

[hlots123]

oic, but hwinfo author's guess is a location close to cpu instead of vrm...

I'm not exactly sure what CPU temperature is measured by the Nuvoton chip, I would need to see the mainboard schematics how the chip is connected (and this is not possible). Usually the CPU temperature reported by a sensor chip like Nuvoton is measured from a diode that is placed close to the CPU (i.e. beneath it). So that explains why the internal CPU temperatures are always higher than those measured by a diode close to the CPU.

HHWiNFO64 is Excellent!

 

[MacClipper]

and this is not possible... the internal CPU temperatures are always higher than those measured by a diode close to the CPU

2nd part of statement does not correlate with above screenshots unless it is measuring something else. Which is why I suspect it is measuring the VRM instead cos nothing else gets hotter under load.

In the end, I guess that... we are all just guessing then.

 

[haylui]

hm..Mac if you have thermal couple measurement device maybe can clear this doubt.

it seems like a lot of people could not believe that AMD's chip could run cool....

 

[MacClipper]

imo probably pointless to try, by now if they don't believe then most likely they won't believe.

If you scrutinise the earlier screenshots with the FX-8350, you may have noticed that the FX-8350 does not run so cool either at lower clocks/volts so this FX-9590 sample is really a super duper fantastically binned Vishera, the crème de la crème of AMD consumer processors.

 

[hlots123]

no lah not i dun want to believe, i jus curious only - dun misunderstand!

asrock info page seems to also state it has 2 sensors, but the phrasing doesn't seem to be vrm. does its uefi screen also phrase it as cpu or vrm?
Hardware Monitor:
- CPU Temperature Sensing
- Chassis Temperature Sensing

ASRock > 990FX Extreme9

but ur test results & ur reasoning does seem that it may be vrm... why both amd & asrock don't make it clear for all of us...?

 

[Wisperknot]

Just curious at this price range, performance, and power consumption, what demographic is AMD targeting with this chip?

 

[MacClipper]

no lah not i dun want to believe, i jus curious only - dun misunderstand!

No worry, just being realistic cos you can't win them all.

Maybe when I switch over to the Asus Crosshair V mobos then we can get another viewpoint on the FX-9590 temps cos hwinfo has specific readouts for the Asus. Here's a screenshot from my older FX-8350 post.

http://forums.hardwarezone.com.sg/75474610-post335.html

My PB accounts are overloaded to the max so remove the space after http and copy & paste in your browser window to view directly
http ://i1236.photobucket.com/albums/ff450/MacClipper3/AMD/Vishera/P95Blend45GHz_20min_OKCopy_zps7ff50bd6.png

Note that the ITE CPU sensor also reads higher than the DTS and the ROG CPU temps so again one is left to wonder where the ITE CPU sensor actually is.

 

[hlots123]

Note that the ITE CPU sensor also reads higher than the DTS and the ROG CPU temps so again one is left to wonder where the ITE CPU sensor actually is.

confused, ITE & ROG CPU sensors are different sensors on the motherboard altogether?

 

[haylui]

imo probably pointless to try, by now if they don't believe then most likely they won't believe.

If you scrutinise the earlier screenshots with the FX-8350, you may have noticed that the FX-8350 does not run so cool either at lower clocks/volts so this FX-9590 sample is really a super duper fantastically binned Vishera, the crème de la crème of AMD consumer processors.

clear doubt as in to identify which sensor/whereabouts are those sensors.
I know most people thought AMD is actually slowly the world down....

 

[MacClipper]

You can google for more info on the onboard CHV sensors and hwinfo64 readings.

 

[ArcticCheetah]

hm..Mac if you have thermal couple measurement device maybe can clear this doubt.

it seems like a lot of people could not believe that AMD's chip could run cool....

Actually for me, IMO AMD well is heading down the wrong road.
5GHZ is actually the wrong move.
A good processor should work towards better IPC (Instruction Per Clock) and instruction set optimization.
It is easy to just up the frequency, people have been doing that all along.
Nowadays it is about efficiency getting more performance per watt.
The days of playing the Clockspeed game is over...
I stopped reading at 200Watt TDP regardless of cooling.

 

[MacClipper]

Actually for me, IMO AMD well is heading down the wrong road...

1st post

★ limited edition ★

So they are not heading down any road that you fear, imo it's more a case of making a limited edition of the best that they have at the present moment ie. milking it for all that it is worth.

btw, a minor storm is raging over at OCN forums regarding Intel's compiler and its crippling performance bias. Do read this blog too for a refresher and more background info
Agner`s CPU blog - CPUID manipulation program for AMD

 

[bhtan83]

1st post


So they are not heading down any road that you fear, imo it's more a case of making a limited edition of the best that they have at the present moment ie. milking it for all that it is worth.

btw, a minor storm is raging over at OCN forums regarding Intel's compiler and its crippling performance bias. Do read this blog too for a refresher and more background info
Agner`s CPU blog - CPUID manipulation program for AMD

lol good read. thanks for sharing man. this problem stated in the blog has been there for a very long period of time.

 

[MacClipper]

Apparently, may happen in some games too like what this guy posted
AMD CPU ID vs Intel - Forums World of Warcraft

Instead of being outright trashed, the AMD fps jumped to a much closer fight suddenly...

Frames per sec,

No CPU ID change:
AMD --> 76min/102max
Intel -->148min/212max

CPU ID spoofed:
AMD --> 127min/162max
Intel -->88min/122max

 

[MacClipper]

I stopped reading at 200Watt TDP regardless of cooling.

Power Consumption
Note that this is the whole test rig power consumption off the wall excluding the LCD display.

Tested with my current undervolted settings of Idle@0.875V, 4.7GHz@1.4375V, Turbo 5GHz@1.488V


Component List
In addition to those listed in the screenshot, factor in a Swiftech 655 pump (set at speed 3), 4 x 120mm 1.5K rpm fans (cooling rad & mobo mosfets), 1 x 80mm 2K rpm fan (cooling my HDs).

Idle At Desktop

WinRAR Load
Chose WinRAR as loading example since it has a small window suitable for camera capture

Erm, official 220W rating for just the FX-9590? Looks like quite a bit of overkill to me...

 

[ArcticCheetah]

Can you run programs like IntelburnTest on all cores?
Doesn't make sense to me why they stated 220W when it isn't necessary.
I am sure there would be a test case for 220W.