AMD Ryzen 7 1800X (14nm Zen)

[watercooled]

Yeah my idle clocks are also fairly high, around 2.11 GHz. Do you know any more about this/resolutions?

I still don't think that explains the temperatures though, I just don't think they're accurate. Perhaps they need correcting with an offset.

[CAT-THE-FIFTH]

Yeah my idle clocks are also fairly high, around 2.11 GHz. Do you know any more about this/resolutions?

I still don't think that explains the temperatures though, I just don't think they're accurate. Perhaps they need correcting with an offset.

It will contribute to it though which is not helpful,although you are not alone:

https://forums.overclockers.co.uk/th...temp.18771900/

Somebody else on OcUK forums had a similar problem too.

[watercooled]

Do you know what the normal idle clocks are supposed to be? I just assumed it might have high-ish idle clocks due to clock gating.

[CAT-THE-FIFTH]

Do you know what the normal idle clocks are supposed to be? I just assumed it might have high-ish idle clocks due to clock gating.

Not sure ATM - but its all a bit hit and miss.

TBH,as long as the CPU is not exceeding its safe temperature I probably would not worry too much.

I still remember a mate who had their Athlon II X3 unlocked to a Phenom II X4 with 6MB of L3 cache,didn't bother to use a better heatsink than the Athlon II stock cooler for some reason,and it was running at a stupidly high temperature for years!!

[watercooled]

Ughh. Early adopter issues I guess...

Well hopefully it gets resolved sooner than later! I'll keep you posted.

[watercooled]

Just found this screenshot of AMD's own OC tool: http://core0.staticworld.net/images/...12478-orig.jpg

The temps and idle clock speeds are about the same there, too. I might drop AMD an email too, see what they have to say about it.

[DanceswithUnix]

The rather old 65W quad core APU in my home server idles at 1.7GHz, so a better architecture on a more modern process 2.1GHz would be less than a 25% increase and sounds reasonable.

[watercooled]

Yeah that part didn't really concern me, what with power gating etc. I just wonder if it's normal for that processor.

The temperature readings are so far off I wonder why no reviewers sanity-checked them? Some setups seem to show sensible readings (like Hexus) but how can an experienced reviewers see 60C at idle for a soldered die and think everything is correct, then complain about it being 'hot-running' due to it increasing by 20C under load. TBH I suspect even the lower values are off, and AMD's reported CPU values are inaccurate yet again.

Another thing, which TBH isn't than unusual, is the vcore reported by HWmonitor is insanely high, like over 2v (again this is not logical as wall power consumption is fine). And fan speed doesn't match what is reported in BIOS.

Come on Gigabyte - waiting on you now!

[CAT-THE-FIFTH]

The rather old 65W quad core APU in my home server idles at 1.7GHz, so a better architecture on a more modern process 2.1GHz would be less than a 25% increase and sounds reasonable.

The R7 1700 idles at 1.5GHZ:

http://www.legitreviews.com/amd-ryze...ssor_192191/10




You need to consider the R7 1800X is consuming more power at idle than a Core i7 6900K alone since it is an SOC(the Core i7 6900k platform power is also being shared with a larger chip for the motherboard functionality) and also the chip is nearly 15% smaller too,so the cooler has to dissipate more heat anyway over a smaller area.

The FX CPUs are over 50% larger than a Ryzen chip and also have a more traditional platform layout.

[watercooled]

You're not talking anywhere close to the sort of power consumption/density that would cause a CPU to reach 40C over ambient though - the readings are just plain wrong, unless the IHS TIM is faulty/cracked (which I doubt because the CPU survives without throttling under heavy load).

My 1055T, also with a soldered die, 95W TDP, *no* clock gating, much higher idle power consumption at the 8pin, in the same case, with the same cooler, didn't even hit 60C *under load*. Even at idle I could feel the 1055T was lukewarm, the 1700X is just cold.

[CAT-THE-FIFTH]

You're not talking anywhere close to the sort of power consumption/density that would cause a CPU to reach 40C over ambient though - the readings are just plain wrong, unless the IHS TIM is faulty/cracked (which I doubt because the CPU survives without throttling under heavy load).

My 1055T, also with a soldered die, 95W TDP, *no* clock gating, much higher idle power consumption at the 8pin, in the same case, with the same cooler, didn't even hit 60C *under load*. Even at idle I could feel the 1055T was lukewarm, the 1700X is just cold.

The Phenom II X6 die was like 300MM2 too,so again its much smaller on Ryzen and you have static chipset functions too,onboard the Ryzen CPU. Remember the Phenom II X6 figures are shared with a much more functional chipset which consumes a larger percentage of the idle power.

Ryzen is like 190ish MM2 in size,so even if the motherboard is not reporting the things right,Ryzen is still going to probably run hotter with similar cooling and the Ryzen stock coolers seem to have a larger fan than the older AMD ones.

You also need to remember,that link I posted yesterday - someone with the R7 1700X with a better than stock cooler had the same issues as you.

They didn't with the R7 1700.

Their R7 1700X was significantly hotter running than their R7 1700 sample too.

It was on an MSI motherboard.

So,my viewpoint is not to worry as I said before,unless the CPU is going past its safe temperature.

[DanceswithUnix]

You need to consider the R7 1800X is consuming more power at idle than a Core i7 6900K alone since it is an SOC(the Core i7 6900k platform power is also being shared with a larger chip for the motherboard functionality) and also the chip is nearly 15% smaller too,so the cooler has to dissipate more heat anyway over a smaller area.

Those power figures must be from the wall system power, so SOC issues wash out. You are thinking too hard on this one.

Look at the AM4 motherboard used in that review, just the LEDs on it are probably using 5W. Each chip on that board requires power, and £260 buys you a lot of chips.

I would love to see what the idle power is like on an A300 board, and I would also like to see AMD avoid shooting themselves in the foot with their motherboard choice as AIUI they are the ones sending boards out for review and the most sparkly one is not necessarily the best, unless you are 12 yrs old or in marketing

[CAT-THE-FIFTH]

Those power figures must be from the wall system power, so SOC issues wash out. You are thinking too hard on this one.

Look at the AM4 motherboard used in that review, just the LEDs on it are probably using 5W. Each chip on that board requires power, and £260 buys you a lot of chips.

I would love to see what the idle power is like on an A300 board, and I would also like to see AMD avoid shooting themselves in the foot with their motherboard choice as AIUI they are the ones sending boards out for review and the most sparkly one is not necessarily the best, unless you are 12 yrs old or in marketing

Why would it wash out - the chip in an I7 6900K is 230ish MM2 - the R7 series is around 190mm2,is around 20% larger and the X99 motherboards have a much larger chipset too. You know very well that the AMD AM4 platform chipset really does not do that much. So when you have the Intel CPU being 20% larger and when 99% of the chipset functionality is under the same cooler,unlike with X99 OFC it is going to run hotter.

A lower percentage of the idle and load power is dissipated from the CPU in the X99 platform.

What is the likelihood if Intel push Broadwell D onto the desktop and push up the clockspeeds it would start to show the same issues?

I am not sure why you are trying to hide this??

The R7 1700 also seems to be idling at closer to 1.5GHZ instead of 2.2GHZ too.

Plus you also need to get that GF 14NM is way past its optimal performance/watt once you go past 3GHZ - look on OcUK. Somebody with both an R7 1700 and R7 1700X in the same MSI motherboard with both the R7 1700 stock cooler and an H110i,noted that their R7 1700X sample was significantly hotter at idle.

Even if the motherboards are over-reporting stuff,that is two different motherboards with the R7 1700X showing the same thing now.

The R7 1700X and R7 1800X are closer to the edge of 4.0GHZ/4.1GHZ than the R7 1700,so I am not surprised that it will be worse than a R7 1700.

Edit!!

The Phenom II X6 had a 346MM2 die size and the FX8350 a 315MM2 die size - compare that to a 190ish MM2 die size for a Ryzen R7,and that is with 90% of the chipset functionality onboard.

Plus people get too worried about temperatures - my IB Core i7 with a low profile cooler used to run very hot too,and even with an AIO water cooler, idle is not very low - it can be easily in the early 40 degree range,and that is a 69W TDP chip with an OTT cooler and the IGP disabled at the factory.

My Q6600 overclocked in a Shuttle was not that great either and that was on a 975X chipset and those ran hot.

[watercooled]

So,my viewpoint is not to worry as I said before,unless the CPU is going past its safe temperature.

I wouldn't, if it wasn't messing with the fan speeds. The 212's fan is *loud* at 100%.

I know what you're saying about possibly marginally higher idle consumption of the 1700X vs 1700 - it's most definitely not enough to cause it to idle so hot, and just to re-iterate, I can *literally touch* the heatspreader with the cooler attached - it's *cold*. Bear in mind that 40C would be quite warm on skin, 50C is uncomfortably hot, 60C enough for reflex to pull hand away. The IHS is as cold as any other metal object in the room, and not more than about 30C. Even Intel's problematic Haswell IHS didn't have an idle temperature delta of 40C - remember Ryzen is soldered.

Either it's coincidental, or the temperature bug is somehow more pronounced on certain models. Just to add to that, I can unplug the CPU fan, even after sitting for a while it doesn't get warm and the reported value doesn't move - that just doesn't add up if a CPU is somehow burning through enough power at idle to cause a 40C temperature delta (it's not).

A faulty thermal path seems very unlikely, but is still something that concerns me, so I hope a BIOS update can reassure me.

[CAT-THE-FIFTH]

I wouldn't, if it wasn't messing with the fan speeds. The 212's fan is *loud* at 100%.

I know what you're saying about possibly marginally higher idle consumption of the 1700X vs 1700 - it's most definitely not enough to cause it to idle so hot, and just to re-iterate, I can *literally touch* the heatspreader with the cooler attached - it's *cold*. Bear in mind that 40C would be quite warm on skin, 50C is uncomfortably hot, 60C enough for reflex to pull hand away. The IHS is as cold as any other metal object in the room, and not more than about 30C. Even Intel's problematic Haswell IHS didn't have an idle temperature delta of 40C - remember Ryzen is soldered.

Either it's coincidental, or the temperature bug is somehow more pronounced on certain models. Just to add to that, I can unplug the CPU fan, even after sitting for a while it doesn't get warm and the reported value doesn't move - that just doesn't add up if a CPU is somehow burning through enough power at idle to cause a 40C temperature delta (it's not).

A faulty thermal path seems very unlikely, but is still something that concerns me, so I hope a BIOS update can reassure me.

Like I said on OcUK the R7 1700X seemed to be much worse than the R7 1700 at idle. They had an H110i which is a 280mm rad AIO water cooler. Their R7 1700X sample was hitting over 50C at idle and their R7 1700 sample was just over 30C.

In fact I read another reply in that thread:

I think its normal with the X? I have 110I but changed the fans to Noctua NF-A14 3000. If you have on auto it will go up and down all the time from 3.4 - 3.8GHz -+ Volts. Mine 1800X have same temp idle 53-60, clocks goes from 3.6-4.1 on 2 cores. When I do nothing. Have ca 60-65C when I game. But have seen someone says the new beta bios for CH6 shows 15+ degrees too much. There is no warm flow from the radiator soo something is up And 1700 have - 15-20C just beacuse its a none X.

It seems to be the X versions have much higher idle temperatures. He is talking about an Asus motherboard.

So its being seen with three different motherboard companies and the R7 1700 seems to be fine.

[Xlucine]

Idling at 1.5GHz is quite high - my 4690K drops to 750MHz under low load. With AMD's fancy auto-overclocking gubbins I'm surprised it doesn't drop the clockspeed lower at idle