CPU stable at 4.7ghz @1.452V

[KemoKa]

crosshair v formula -z

 

cpu just passes 1024k on prime 95

go out to the back of your house and watch your power meter slowly go up...

[Bajantechnician]

go out to the back of your house and watch your power meter slowly go up...

already used to it with a titan z.

 

anyways, just passes 1152k in prime 95. seems stable

[AlwaysFSX]

You mean the video with the 970FX Krait? That thing had a 4+1 VRM, of course it was going to cook.

 

Anything lower than a 6+2 would probably die eventually, too. 220W going through a CPU... ridiculous.

I haven't seen that one but I've seen a few others that have caught on fire.

[harrynowl]

9590s are validated up to 1.55v..

 

as long as you cool it it's fine

 

this is 32nm soi not intel 22nm tri-gate guys.

[Benji_w]

Update, Prime 95 says self test 720k passes on all 8 cores

 

24hours? because prime needs 24 hours to test stability... IBT AVX will take an hour or so on maximum.

Download here and run.

 

My next question is, is that a constant 4.7ghz or is your speed dropping to 1.4ghz due to thermal throttling?

[Bajantechnician]

24hours? because prime needs 24 hours to test stability... IBT AVX will take an hour or so on maximum.

Download here and run.

 

My next question is, is that a constant 4.7ghz or is your speed dropping to 1.4ghz due to thermal throttling?

dropping

 

set it on 4 ghz base 4.5 turbo. no more throttling

[Benji_w]

dropping

 

set it on 4 ghz base 4.5 turbo. no more throttling

You still need to test for stability, keep upping the mhz while youre testing, if it fails you need more volts.

[Bajantechnician]

You still need to test for stability, keep upping the mhz while youre testing, if it fails you need more volts.

but it throttlees at 1.45v 4.7ghz.

 

running 4.5 at 1.42 is not throttling

[vagabond139]

9590s are validated up to 1.55v..

 

as long as you cool it it's fine

 

this is 32nm soi not intel 22nm tri-gate guys.

This. Too many people on here seem to think AMD cpu's will melt or die or something at 1.5V.

[harrynowl]

This. Too many people on here seem to think AMD cpu's will melt or die or something at 1.5V.

Too many people here know nothing about anything that isn't Haswell and assume all platforms are the same

[vagabond139]

Too many people here know nothing about anything that isn't Haswell and assume all platforms are the same

I know. I have had to clear quite a few misconceptions about AMD cpu's since my short time here, I even had tell a mod that you have to use AMD Overdrive to get a accurate temperature of the cpu and not HWmonitor. I really should make a post about it clearing all of the misconceptions but I am too lazy to do so.

[arniks8]

I'm not sure on AMD CPUs, so that voltage may be fine. If it was Intell..ohh boy.

If it was Intel's CPU you just would need good cooling.

[Samfisher]

If it was Intel's CPU you just would need good cooling.

It will also degrade much faster, 1.47v is quite a bit higher than the recommended max of 1.35v

[arniks8]

It will also degrade much faster, 1.47v is quite a bit higher than the recommended max of 1.35v

If temperature is low (like 60^o c and lower), then 1.45 wouldn't degrade the CPU much faster. 1.47 is still somewhat near 1.45, but serious cooling is needed for that voltage.

[Benji_w]

I know. I have had to clear quite a few misconceptions about AMD cpu's since my short time here, I even had tell a mod that you have to use AMD Overdrive to get a accurate temperature of the cpu and not HWmonitor. I really should make a post about it clearing all of the misconceptions but I am too lazy to do so.

 

Overdrive tells you how much thermal headroom you have, HWmonitor (I suggest using HWinfo, it has graphs so you can follow your temps) tells you the recorded temps as they are.

 

If it was Intel's CPU you just would need good cooling.

 

Jury is still out on this, there is always a big discussion about what makes things degrade quicker, voltage or heat or even both? Who knows, why bother anyway just overclock and enjoy your fast PC.

[vagabond139]

Overdrive tells you how much thermal headroom you have, HWmonitor (I suggest using HWinfo, it has graphs so you can follow your temps) tells you the recorded temps as they are.

 

 

Jury is still out on this, there is always a big discussion about what makes things degrade quicker, voltage or heat or even both? Who knows, why bother anyway just overclock and enjoy your fast PC.

HWmonitor isn't accurate at all...It says my cpu is running at 58C at nearly idle but my thermal margin is about 65, that is a huge difference. With my thermal margin at 65C I have 65C of thermal headroom before I started throttling and with it at 58C I would have probably about 30C before I start to throttle assuming it has a physical max operating temperature of 90C which is reasonable to guess.

[Suika]

Im running a fx 9590,

Another person on toms hardware, reply 1, says he runs 1.56 v on his 9370

http://www.tomshardware.com/answers/id-2650784/amd-fx9590-voltages.html

I had 1.49v on my 8350 with an H100i and that was around 70C, 1.56v seems exaggerated without some high-end cooling.

[Bajantechnician]

I had 1.49v on my 8350 with an H100i and that was around 70C, 1.56v seems exaggerated without some high-end cooling.

 

 

HWmonitor isn't accurate at all...It says my cpu is running at 58C at nearly idle but my thermal margin is about 65, that is a huge difference. With my thermal margin at 65C I have 65C of thermal headroom before I started throttling and with it at 58C I would have probably about 30C before I start to throttle assuming it has a physical max operating temperature of 90C which is reasonable to guess.

ordered x61

[Benji_w]

HWmonitor isn't accurate at all...It says my cpu is running at 58C at nearly idle but my thermal margin is about 65, that is a huge difference. With my thermal margin at 65C I have 65C of thermal headroom before I started throttling and with it at 58C I would have probably about 30C before I start to throttle assuming it has a physical max operating temperature of 90C which is reasonable to guess.

 

HWmonitor is off a few degrees, Download HWinfo64 and use the graphs. Also 58c idle? are you reading socket temp there and it that isn't idle...

 

[Benji_w]

I had 1.49v on my 8350 with an H100i and that was around 70C, 1.56v seems exaggerated without some high-end cooling.

 

70C must have been your socket temp, or your fans aren't fast enough on the H100i, you can use 1.5v on an air cooler like the NH-D14/15 or a phanteks PH-C14PE

[Suika]

70C must have been your socket temp, or your fans aren't fast enough on the H100i, you can use 1.5v on an air cooler like the NH-D14/15 or a phanteks PH-C14PE

Socket temp was 60C, given the 8350's inability to give an accurate temperature I take off of socket + 10C. I haven't had the 8350 for over a year now so yea I can't really reconfirm.

 

Also the NH-D14/15 are high-end air coolers that cool just as well as the H100i so I don't know what you mean by "you can use an air-cooler instead"? Air coolers aren't far behind AIO's, the NHD-15 beats a lot of them.

[Benji_w]

Socket temp was 60C, given the 8350's inability to give an accurate temperature I take off of socket + 10C. I haven't had the 8350 for over a year now so yea I can't really reconfirm.

 

Also the NH-D14/15 are high-end air coolers that cool just as well as the H100i so I don't know what you mean by "you can use an air-cooler instead"? Air coolers aren't far behind AIO's, the NHD-15 beats a lot of them.

 

They can give accurate readings just not at idle, at idle you use socket temp, no plus or minus needed. Under load the temp given is accurate. An aircooler like the NH D14/15 will cool the VRM while also cooling the CPU so is better than an AIO given VRM cooling is highly recommended.

 

This is from people who know more about FX chips than me and you combined:-

 

The CPU temp (aka "socket"), as Stilt recently stated, is actually on die temperature (so let's say die surface). That's accurate. Then there is the "core" (or "package") temp. This one is inaccurate up to 40C, because it's not a Celcius temperature. It has its own scale. The value we get from software, is calculated through an algorithm that becomes accurate over 40C.

[vagabond139]

HWmonitor is off a few degrees, Download HWinfo64 and use the graphs. Also 58c idle? are you reading socket temp there and it that isn't idle...

 

That is more than a few degrees off, probably around 30 degrees off. Yup 58C idle according to HWmonitor. I am reading the cpu temperature on HWMonitor at idle. HWinfo is saying 53C, very close to HWMointor. Under full load my thermal margin is 25Cish and according to HWInfo and HWmonitor my cpu temperature is 95C. I can send you some screenshots if you don't believe me.

 

[Benji_w]

That is more than a few degrees off, probably around 30 degrees off. Yup 58C idle according to HWmonitor. I am reading the cpu temperature on HWMonitor at idle. HWinfo is saying 53C, very close to HWMointor. Under full load my thermal margin is 25Cish and according to HWInfo and HWmonitor my cpu temperature is 95C. I can send you some screenshots if you don't believe me.

 

 

Either your software of your motherboard is borked.

[Benji_w]

That is more than a few degrees off, probably around 30 degrees off. Yup 58C idle according to HWmonitor. I am reading the cpu temperature on HWMonitor at idle. HWinfo is saying 53C, very close to HWMointor. Under full load my thermal margin is 25Cish and according to HWInfo and HWmonitor my cpu temperature is 95C. I can send you some screenshots if you don't believe me.

 

 

Or is this an APU? If it's an APU then ignore HWinfo and HWmonitor and use just overdrive, the thermal margin is the only way to measure APU temps IIRC.