Solved FX 8350 - Inaccurate Core Temperature

[Mellon Head]

Question what is your HT speed with that chip 2600 or 2400 ?

It's at its default of 2200. I never change it until I've set my main clock speed/multiplier. There's a sweet spot/ratio that you have to follow, according to some guides that I read.

I knew it that isn't the default HT for that chip it's 2600 the mother board is the issue Proper Default NB and HT link speed
That motherboard has lowered the real value of that chip and may be overvolting it as well i only stuck with the boards that were made with the chips the other one's don't control them well

Thanks for the link. Maybe I will have to get another mobo. I'll try setting this one up "Non-auto" and see what happens. It's a bit sluggish and needs a tuning anyway. lol

 

[Solarstarshines]

I think it's the mobo because the 970 and 980 are two generations back and i notice when people use a older board for a FX chip the voltages are just way higher

 

[Mellon Head]

I should try dialing the voltage back and seeing what happens at stock.

 

[Mellon Head]

Here is the official answer from AMD on reading Core temps and what their values should be

Official FX-8350 temps - AMD - CPUs

I'm marking this one closed.

 

[Solarstarshines]

Well reading it you should be well in spec
but still way off

 

[Mellon Head]

If I enable Load Line Calibration then it reads like it should. The CPU gets real hot though - into the 60's and the socket temp is in the 70's. I shut it down real quick.

Looks like I'd better invest in some water cooling if I want to OC this thing.

 

[Solarstarshines]

If I enable Load Line Calibration then it reads like it should. The CPU gets real hot though - into the 60's and the socket temp is in the 70's. I shut it down real quick.

Looks like I'd better invest in some water cooling if I want to OC this thing.

Yes i highly recommend water i got my 8350 stable at 4.8 Ghz on water full load never went above 55c
Look at Thermaltake water 2.0 professional or antec 920 khuler very good even the 620 khuler is awesome

 

[Mellon Head]

The Corsair Hydro H80i is on sale at Newegg right now. $30 off. Know if it's any good?

 

[madcratebuilder]

Check the CM Glacer 240L. What ever you get I would highly recommend it be of all brass and copper construction. A lot of the aio's have aluminum rads. Before buying anything do some reading here. This site is 100% BS free.

 

[Mellon Head]

Thanks for the link. Good review of the H100i.

Edit:

I can't use the 240L. I can't use a dual fan setup in my case as I don't have a large enough exhaust hole. The best I can do is a 120mm or 140mm push-pull setup. That's why I was considering the H80i. I'm not really considering a new case, because it was a bear to get all of this gear that I have into this one and I don't really want to repeat the experience. lol

 

[madcratebuilder]

I've run the 80i with a [email protected], normal everyday work fans run under 1200 rpm, when you stress test they go up to max and are loud but only when pushing the system to the max.

I tried a few different rad arrangements on the back of the case. Rad mounted inside in push/pull, rad mounted outside in push/pull, temps within margin of error. I settled for rad and push fan inside case and the pull fan on the outside. That gave me a little more room inside for air circulation around my xfired gpu's.

I hear you about stuffing gear inside the case, I'm about half done with the wiring.


It's a little cleaner on the other side.

 

[Indianatone]

I've run the 80i with a [email protected], normal everyday work fans run under 1200 rpm, when you stress test they go up to max and are loud but only when pushing the system to the max.

I tried a few different rad arrangements on the back of the case. Rad mounted inside in push/pull, rad mounted outside in push/pull, temps within margin of error. I settled for rad and push fan inside case and the pull fan on the outside. That gave me a little more room inside for air circulation around my xfired gpu's.

I hear you about stuffing gear inside the case, I'm about half done with the wiring.
View attachment 304690

It's a little cleaner on the other side.
View attachment 304691

Very nice rig , stupid question of the day, what is the purpose of the 2 small fans to the rear of the CPU socket?

 

[ganjiry]

Nice looking rig Madcratebuilder, good job mate

@Indianatone, The Crosshair v Formula can get a pretty toastie Northbridge when overclocking so I'd imagine those fan's are too help cool that. I know of people that hit cpu oc's of 7Ghz + on that board with North bridge temps of 100c without additional cooling for the north bridge without issue though.

 

[Mellon Head]

Nice rig, madcratebuilder.

 

[madcratebuilder]

I've run the 80i with a [email protected], normal everyday work fans run under 1200 rpm, when you stress test they go up to max and are loud but only when pushing the system to the max.

I tried a few different rad arrangements on the back of the case. Rad mounted inside in push/pull, rad mounted outside in push/pull, temps within margin of error. I settled for rad and push fan inside case and the pull fan on the outside. That gave me a little more room inside for air circulation around my xfired gpu's.

I hear you about stuffing gear inside the case, I'm about half done with the wiring.
View attachment 304690

It's a little cleaner on the other side.
View attachment 304691

Very nice rig , stupid question of the day, what is the purpose of the 2 small fans to the rear of the CPU socket?

Like ganjiry said, added air cooling. When you have the cpu on water you no longer have a fan moving any air around the cpu socket or the vrm/nb array. Those two 50mm fans cool the socket (lower left) and the upper right fan cooling a row of chips on the back side of the vrm's.

There's been some discussions on the OC boards if this is worthwhile or not. On some boards it is more effective and it's not a lot of temp difference, 2-6C is what most report. A pair of $5 silent fans is a minor investment, adding more wires to the rats nest. I'll use a Y connecter and run off the board for BIOS control of these fans initially. This board has a two pin temp probe connection, I think I can feed that to my cooling control center (Aquacomputer AT6) and ramp the fan rpm based on mother board temps.

 

[MagusMagnus]

I think it's the mobo because the 970 and 980 are two generations back and i notice when people use a older board for a FX chip the voltages are just way higher

Bingo.

The main problem with the M5A97 R2.0 paired with the FX-8320/50 rests with the fact that that board only has a 4+2 VRM (voltage regulator module) power phase whereas the M5A97 Evo R2.0 has a VRM 6+2 power phase and also the M5A99FX Evo/Pro R2.0 has a 6+2+2 VRM power phase.

The VRM is responsible for dampening down the 12V supplied by the PSU and is therefore in turn actually feeding the CPU its-converted-voltage directly... it literally keeps the CPU from frying. Essentially the rule of thumb is the more powerful the CPU the more VRM power phases are necessary to adequately feed the CPU the proper voltage, being neither too much nor too little. Ideally, these chips wouldn’t be paired with any board with less than an 8 VRM power phase (which would include the three aforementioned because technically, they are at 6+2 and 6+2+2 respectively).

In other words, that board (actually all 4+2 VRM power phase boards in general) simply was not designed with the FX 8-core chips in mind, the 4+2 VRM power phase boards really shouldn’t be paired with anything more than an FX 6-core and that in fact is the FX series available at the time of their design and release in the first place.

Asus should never have revised this to be “compatible” with the FX 8-cores because it really isn’t, not in its actual, physical design. The only reason it ‘works’ is the technicality of fact that it is still AM3+ socket but that is truly where the “compatibility” ends here. That boards VRM is just not good enough for that chip at stock, let alone compounding the problem further with overclocking. Yikes.

 

[Mellon Head]

Bingo.

The main problem with the M5A97 R2.0 paired with the FX-8320/50 rests with the fact that that board only has a 4+2 VRM (voltage regulator module) power phase whereas the M5A97 Evo R2.0 has a VRM 6+2 power phase and also the M5A99FX Evo/Pro R2.0 has a 6+2+2 VRM power phase.

The VRM is responsible for dampening down the 12V supplied by the PSU and is therefore in turn actually feeding the CPU its-converted-voltage directly... it literally keeps the CPU from frying. Essentially the rule of thumb is the more powerful the CPU the more VRM power phases are necessary to adequately feed the CPU the proper voltage, being neither too much nor too little. Ideally, these chips wouldn’t be paired with any board with less than an 8 VRM power phase (which would include the three aforementioned because technically, they are at 6+2 and 6+2+2 respectively).

In other words, that board (actually all 4+2 VRM power phase boards in general) simply was not designed with the FX 8-core chips in mind, the 4+2 VRM power phase boards really shouldn’t be paired with anything more than an FX 6-core and that in fact is the FX series available at the time of their design and release in the first place.

Asus should never have revised this to be “compatible” with the FX 8-cores because it really isn’t, not in its actual, physical design. The only reason it ‘works’ is the technicality of fact that it is still AM3+ socket but that is truly where the “compatibility” ends here. That boards VRM is just not good enough for that chip at stock, let alone compounding the problem further with overclocking. Yikes.

Thanks for the good information. I came to the conclusion that my motherboard wasn't up to it and I ended my quest at overclocking a while ago. At least until I can get a better motherboard. Something like a Sabertooth if it's in my wife's budget. lol.

 

[Smartcom5]

VRM-Design …

I know necrophilism has become a hot topic of debate these days (pun intended), but if i may just add some of my findings in here - as i was searching for a friend of mine as he got also heat-issues using the FX-8320 …

I think it's the mobo because the 970 and 980 are two generations back and i notice when people use a older board for a FX chip the voltages are just way higher

[…]
The main problem with the M5A97 R2.0 paired with the FX-8320/50 rests with the fact that that board only has a 4+2 VRM (voltage regulator module) power phase whereas the M5A97 Evo R2.0 has a VRM 6+2 power phase and also the M5A99FX Evo/Pro R2.0 has a 6+2+2 VRM power phase.
[…]
Ideally, these chips wouldn’t be paired with any board with less than an 8 VRM power phase […].

In other words, that board (actually all 4+2 VRM power phase boards in general) simply was not designed with the FX 8-core chips in mind, the 4+2 VRM power phase boards really shouldn’t be paired with anything more than an FX 6-core and that in fact is the FX series available at the time of their design and release in the first place.

Asus should never have revised this to be “compatible” with the FX 8-cores because it really isn’t, not in its actual, physical design. The only reason it ‘works’ is the technicality of fact that it is still AM3+ socket but that is truly where the “compatibility” ends here. That boards VRM is just not good enough for that chip at stock, let alone compounding the problem further with overclocking. Yikes.

I just wanna drop in that a FX-8350 without any difficulty can be used on boards which may only featuring a 4+2 VRM power phase design. This isn't to proof anyone wrong but to confirm that the board‘s manufacturer ain't that wrong as they releases or flag given boards to be compatible with the highest Bulldozer‘s FX-8xxx series.

As far as i'm concerned and where the message comes in here … I bought a rig a few years ago, custom build which featured the following parts:

• CPU: AMD FX-8350
• CPU-Cooler: be quiet! Dark Rock Advanced C1
• MB: AsRock 970 Extreme3 R2.0
• Mem: Crucial Ballistix Tactical, 16 GByte, 1866 MHz, CL9
• GPU: AMD Radeon HD 7850 Black Edition, 2 GByte GDDR5
• PSU: be quiet! Straight Power E9 580W-CM
• Case: CoolerMaster HAF 912+

The 970 Extreme3 R2.0 has only a 4+1 VRM design. But despite lacking any good VRM design (read: ±4+2 phases) i never ever ran into any trouble at all. No throttling, no drop-out or something similar.

Remarkably though, the whole system stayed and stays pretty cool - barely reaching any temperatures above 55°C (package or core) even under heavy load (Prime 24hrs). The GPU idling down at 23°-25 or 27°-29°C (depending on ambient temperature) reaching a maximum at ~45°C under heavy load with games/applications and ~50°C on FurMark (24hrs).

I was quite surprised and set my bet on false read-outs as i even bought some infrared thermometers (Voltcraft IR-2200-50D & FLUKE 62 MAX+, Infrared Pyrometers) and tried every monitoring software around - knew about AMD‘s changed measuring scheme though - to just be eventually proven wrong.

The system in fact is just amazing cool!
It turned out that the VRM reaching ~25°-30° C when idling and come close to 45° C under load.
I guess it's just a nearly perfect mix of the heavy weight CPU-cooler, the really nice case (it isn't branded as HAF; → reads „High Air Flow“ for no reason …) and the quality PSU which ensures some reliability.
It's even capable to boost given cores via Turbo-Core™

For storage my own rig got some 500GB, 1.5TB & 2TB HDD and a 256GB SSD i used before.
The whole system taxed 750,50 € back then including the 2TB HDD and a Cherry™ CyMotion Expert Combo beyboard.

As a matter of fact i bought this identical setup seven times to date, to configure and assemble it for other guys around - as some came by and kept wondering how surprisingly cool and freaking quite it run. No-one ever complained about anything and all of them are still in service - some where upgraded with better GPUs over time but didn't struggle to maintain silence and temperature after all. Mine got a R 290X a while ago too. Just one of them had serious heat-issues and therefore throttling as he tenaciously insisted to use his own case and CPU-fan he salvaged from an older rig.

So, a minor amount of phases on a VRM design doesn't necessarily mean it couldn't fire a high-performance setup …


In this sense

Smartcom