Speculation: Ryzen 3000 series

[Abwx]

It is not



This looks like a flat object with no heat dissipation properties at all to you?

Your argument would look better if it didnt require to "flat" deny thermophysics for it to work.

Just to be clear, Intel states 4.1w tdp for the Z97 chipset. Even with the red gigabyte logo cover, does it seem like a small heatsink for dissipating that heat output? Now imagine 2.75× that. We even dont know if the 11w tdp figure is just tdp or absolute peak power consumption of x570 under heavy pci-e 4 load

11W TDP is forcibly at full power, it is defined in AMD s methodology as the max power under full usage for worst silicon at 10% higher voltage than what is actually implemented.

Beside the cooler in your pic is the one for a Baytrail that was known to exceed 10W, in this case you can see that the board must be implemented such that the cooler is in vertical position in the direction of its legnth for the air flow to be efficent, but apparently they anticipated that it could be hoizontaly implemented, hence the relatively big size.

Now you can make a test with a device that dissipate 11W and different coolers and aluminium plates...

 

[PPB]

You are quite wrong, the stock heatsink for this J5005 is this one on the right and is an active heatsink design that is made with the closed enclosure of the nuc in mind and keeps said nuc in the 70-80c at load.

Have in mind the difference of design between both, the one of the left (which is the one i said hovers 90+c for the j5005 at 13.5w sustained load). Fin stacks are made with forced air renovation in mind, if not they lose a lot of value. Also notice how much smaller is the stock j5005 heatsink is. In this same logic, a tiny fan enables aibs to not require a massive heatsink that would otherwise get in the way of the sheer ammount of pcb components high end mobos have today.

11W TDP is forcibly at full power, it is defined in AMD s methodology as the max power under full usage for worst silicon at 10% higher voltage than what is actually implemented.

Beside the cooler in your pic is the one for a Baytrail that was known to exceed 10W, in this case you can see that the board must be implemented such that the cooler is in vertical position in the direction of its legnth for the air flow to be efficent, but apparently they anticipated that it could be hoizontaly implemented, hence the relatively big size.

Now you can make a test with a device that dissipate 11W and different coolers and aluminium plates...

 

[Abwx]

At first glance 100cm2 (10 x 10 cm) verticaly positionned can exhaust 10W with the device (and the plate) temperature increasing 60°C over ambiant, so 200cm2 should be enough for 40°C at most higher temp than ambiant.

As for fanlsess J5005 see by yourself :

 

[PPB]

At first glance 100cm2 (10 x 10 cm) verticaly positionned can exhaust 10W with the device (and the plate) temperature increasing 60°C over ambiant, so 200cm2 should be enough for 40°C at most higher temp than ambiant.

As for fanlsess J5005 see by yourself :

But i never talked about that product lmao. Also that heatsink is pretty bigger than the one I'm using, and the one I'm using is one of the bigger Z97 chipset ones. You are totally making a non-point here.

Besides that, by height that heatsink you show is at least 3 centimeters tall, how do you plan to put a GPU if the chipset heatsink right beside the PCI-E port obstruct the area?

 

[Abwx]

i was talking of the total surface, not the shape, total area is small, below 200cm2, the aluminium cover used by Gigabyte is more than this cooler and way enough, it s overdimensioned given the usage that will be made with the MB...

 

[PPB]

The shape impacts the heat transfer behaviour heavily, there is a reason they chose a tall radial fin shape for that heatsink. On height restricted heatsinks even if you match surface area like that heatsink you wont be able to match it's thermal performance because heat wont spread the same way across the heatsink's mass. This is why also the p5q-e vrm+chipset heatsink i showed doesnt perform as well as it's looks would suggest.

Also if you cant match its height, you need to have more mass, to have more mass on heigh restricted heatsinks you need to go really wide. The wider you go in x/y directions the more surface not able to renovate air you have (the one facing the chipset/pcb), so you need more material to achieve the same thermal performance.

If you need more x/y surface then you need to watch out for height clearance of the surrounding pcb components, that means you need to make it even more flat with a big baseplate to make up that height clearance required (a wide T shape like the z97 heatsink i showed). If you do that, you restrict more the thermal performance.

To say shape doesnt matter is asinine. Seriously. And you take the 11w figure for granted whereas that is still a rumor laying around the halls of computex. Until we see actual power figures i wouldnt even attempt to trust 11w is absolute max peak. The heatsinks we have been shown in the motherboards showcased would suggest otherwise.

i was talking of the total surface, not the shape, total area is small, below 200cm2, the aluminium cover used by Gigabyte is more than this cooler and way enough, it s overdimensioned given the usage that will be made with the MB...

Also, so we put this failed comparison of yours to rest. That heatsink on the J5005-ITX measures roughly 56mmx92mm. The Z97X gaming 7 PCH heatsink measures 60x80mm. If my math isnt wrong, that is 5152mm2 of base area for the J5005-ITX vs 4800mm2 for the Z97x-gaming 7 pch heatsink, and that is without taking in account the difference in height between the 2 heatsinks. Also, a little point you excluded when mentioning that heatsink, per https://androidpctv.com/review-asrock-j5005-itx-j4105-itx/ it reaches 87c of peak temperatures when stress tested, 18° shy of J5005's TJMAX. FOr sure everybody here would want x570 temperatures be a little lower than that, am i right?

Please, just stop. Or at least try to elevate the argument a little bit.

 

[DrMrLordX]

If you ask me, i would pay a 200 bucks x570 board only if it had vrm + chipset heatsinks like these, but then again, they cant guarantee you will properly feed this kind of design with forced ventilation. A AIO would be the bane of this heatsink design for example and the fin stack would lose purpose if air is stagnant and doesn't

I am somewhat inclined to agree on your financial assessment (again, I can tolerate a little fan if it doesn't break). Personally I rely on my case for airflow and ventilation - not my HSF. I've been on L-shaped towers long enough that my heatsinks never help the board components stay cool anyway. I used to use a desk fan to just blow air through the open case (which worked well). Now I use the considerable array of case fans on my Thor v2 to do essentially the same thing. Even if you have "fake" copper heatsinks that are actually aluminum, if they're of the size that I showed (or of the size of anything you showed - love that HR-09u by the way) then you'll do "just fine" in a windy case like mine, or even an older windtunnel-style case like the Antec 900.

You do bring up a good point about PCIe card obstruction of the chipset area. I can't see any low-profile chipset coolers doing very well if they spend their entire lives obstructed by a hot video card. Vapor chambers could be implemented here to good effect. Vapor chambers can be pretty expensive, though.

 

[extra]

It seems like maybe this time around we'll have ball bearing fans with better tolerances on the chipset? I'm not sure that the chipset fans are a big deal if that's true. Let's wait and see the motherboard reviews before freaking out about 'em. I have bad memories from 15 years ago (maybe more) with tiny fans. But that was long ago--I have bad memories from 90mm fans from that era, too! The rumors are that the max power consumption really only happens under raid with PCIe 4 SSDs as well. Who runs raid NVMe? So let's wait and see. It'll be something for reviewers to focus on--chipset fan noise, if they are ball bearing, if they are easy to replace, etc.

 

[AMDrulZ]

Hi everyone i mostly just read the forums. But i find all of this discussion about the x570 chipset fans hilarious. I solved the chipset fan issue 15 years ago. I had an old pci evga nvidia tnt2 graphics card that ran hot so i grabbed an old 486 cpu cooler fan and installed it on my graphics card and it never over heated and it was still working 15 years later. I also started installing those fans on every chipset that required a fan and never had any failures. People are definitely over thinking the x570 chipset fans. The old 486 and pentium cpu fans were pretty small fans and i believe they were ball bearing fans all the way back then...lmao

 

[Kedas]

We know that the heat is mainly due to PCIe4.0 vs PCIe3.0 (double freq)
We know that your main SSD will be directly connected to the Ryzen I/O so doesn't influence chipset heat.

Conclusion: Most users will never need this chipset fan, but test will tell.

 

[AMDrulZ]

I would love to see this kind of chip-set cooling on an x570 motherboard though...lol

 

[Kedas]

Maybe more interesting do we know how much the I/O die uses at full load?
(PCIe4 M.2, 2 GPU's, Fast DDR4, 12/16 core)

 

[Abwx]

And you take the 11w figure for granted whereas that is still a rumor laying around the halls of computex. Until we see actual power figures i wouldnt even attempt to trust 11w is absolute max peak. The heatsinks we have been shown in the motherboards showcased would suggest otherwise.


Please, just stop. Or at least try to elevate the argument a little bit.

Is what i bolded an elevated argument..?

11W TDP was stated by AMD at Computex and reported by Computerbase.de but also AT in Cutress s article, so what rumour are you talking about..?


Beside i posted AMD s own tests conditions for chipsets, it will never see said 11W, FTR they were published for the 990FX chipset wich has a 19.6W TDP, as said they test with the lowest grade silicon at 10% excess voltage and of course full flow of the I/Os, so it s overstated by at least 21%.

 

[PPB]

I am somewhat inclined to agree on your financial assessment (again, I can tolerate a little fan if it doesn't break). Personally I rely on my case for airflow and ventilation - not my HSF. I've been on L-shaped towers long enough that my heatsinks never help the board components stay cool anyway. I used to use a desk fan to just blow air through the open case (which worked well). Now I use the considerable array of case fans on my Thor v2 to do essentially the same thing. Even if you have "fake" copper heatsinks that are actually aluminum, if they're of the size that I showed (or of the size of anything you showed - love that HR-09u by the way) then you'll do "just fine" in a windy case like mine, or even an older windtunnel-style case like the Antec 900.

You do bring up a good point about PCIe card obstruction of the chipset area. I can't see any low-profile chipset coolers doing very well if they spend their entire lives obstructed by a hot video card. Vapor chambers could be implemented here to good effect. Vapor chambers can be pretty expensive, though.

A "hot" video card that is not a blower design and has the fin stack perpendicular to the motherboard will actually help moving air through. Hot moving air is still better than stagnant but still above ambient air.

The problem is, when you design something you have to think of all the use cases possible.

So for a heat perspective, you design that pch heatsink with the worse case scenario in mind, that is little air renovation/airflow obstruction (blower gpu not moving air at all at that area of the motherboard). From the dimensions perspective, is a balance of z height clearance of pcb components and maximum z height limited by add in card interference.

Then you consider where to put the pch. Its not random that is mostly on the same area except on itx cases. And this is because on itx you don't have more pci e slots that are feeded by said pch that noe because it is pci e 4 lanes, they need the less routing possible to keep signal integrity. On itx then you have another problem, amd heatsink clearance which is a lot higher in the Y axis of the motherboard (50mm give or take from the center of the socket both upwards and downwards). This is why asrock made the good choice of going with 115x mounting holes, because it frees you up some space in the middle of the itx mobo area to put stuff that otherwise would be a hard fit, specially now that the pch needs at least 40x40 space to at least house a fan.

On atx forget doing a fin stack with heatpipes connecting vrm and pch, there is just too much distance. On itx its totally possible as per z390 fatality itx 's layout. But still, a fin stack needs to be fed. A big chunk of aluminum will keep a pch at a better temperature than a fin stack if there is no airflow involved. And on enthusiasts that happens more often than not because of tall tower coolers and aios.

A fan+ alu heatsink is just simpler and cheaper. There is no way around it.

 

[AMDrulZ]

I agree 100% that an aluminum heat sink and fan is the simpler and cheaper solution. But a well designed heat pipe solution can allow more flexibility for fan placement esp in area's that have little vertical space. That's basically why all laptop's use heat pipe cooling.

 

[Abwx]

Also, so we put this failed comparison of yours to rest. That heatsink on the J5005-ITX measures roughly 56mmx92mm. The Z97X gaming 7 PCH heatsink measures 60x80mm. If my math isnt wrong, that is 5152mm2 of base area for the J5005-ITX vs 4800mm2 for the Z97x-gaming 7 pch heatsink, and that is without taking in account the difference in height between the 2 heatsinks. Also, a little point you excluded when mentioning that heatsink, per https://androidpctv.com/review-asrock-j5005-itx-j4105-itx/ it reaches 87c of peak temperatures when stress tested, 18° shy of J5005's TJMAX. FOr sure everybody here would want x570 temperatures be a little lower than that, am i right?

On second thoughts you gave those numbers after i referenced a 200cm2 aluminium plate as being enough, do you realize that 5152mm2 is 51.25cm2 , 4800mm2 (60x80mm) is no more than 48cm2...?.

And they are enough..?..for how much power.?.
Granted there are fins that add considerably to this surface, but still, does it reach 200cm2 , that is 14.14 x 14.14 cm, wich is 20 000 mm2..?..

 

[Ajay]

Well, I guess we’ll be stuck with whatever the board manufacturers give us. I don’t think many ppl with pass on a top rated x570 board just because of a chipset fan. It looks like we’ll see some boards that offer heat sink only solutions. I wish I was looking to build a new system this summer, but I’m just doing b450/2400G build for my mom.

 

[Ajay]

On second thoughts you gave those numbers after i referenced a 200cm2 aluminium plate as being enough, do you realize that 5152mm2 is 51.25cm2 , 4800mm2 (60x80mm) is no more than 48cm2...?.

And they are enough..?..for how much power.?.
Granted there are fins that add considerably to this surface, but still, does it reach 200cm2 , that is 14.14 x 14.14 cm, wich is 20 000 mm2..?..

Really depends on case temperature and air flow. Plus, afaik, this is just an ASIC (pretty much switches and amps), which may well have higher temperatures limits than a CPU.

 

[amrnuke]

As long as they don't fail, I don't particularly mind. It just seems like they're being NASA using special pens designed to write in space when they could just use grease pencils instead (like the Soviets).

1) NASA didn't pay to develop the "space pen". It was developed independently (by Fisher) then NASA started using it, paying $6 per pen x 400 pens initially, where Fisher spent over $1 million independently developing it.
2) Grease pencils used by the Soviets were less durable, the writing was smudgy, they created waste, and they were flammable.
3) The Soviets eventually adopted the same Fisher pen because it was a much better and safer product.

Just so there's no misinformation about this.

 

[PPB]

On second thoughts you gave those numbers after i referenced a 200cm2 aluminium plate as being enough, do you realize that 5152mm2 is 51.25cm2 , 4800mm2 (60x80mm) is no more than 48cm2...?.

And they are enough..?..for how much power.?.
Granted there are fins that add considerably to this surface, but still, does it reach 200cm2 , that is 14.14 x 14.14 cm, wich is 20 000 mm2..?..

That 200cm2 that you pulled out of your ass? Sorry, I have enough debunking the other garbage you spewed in this thread already. The area claim by me is to debunk that the heatsink I used for my nuc is bigger in area than the one you linked in the J5005-ITX. Not only it's false, but by per the review I linked that J5005-ITX is barely adequate for very high loads. And that are 13.5W loads.

So you need a heatsink at least that size to keep the """"11w"""" (yet to be confirmed) PCH at not toasty temperatures. With the caveat you cant fit that heatsink in the space dedicated for the PCH, both in ITX and mATX-ATX configurations. Do you understand now or I need to bring the crayons? You are just embarrasing yourself at this point.

 

[extra]

1) NASA didn't pay to develop the "space pen". It was developed independently (by Fisher) then NASA started using it, paying $6 per pen x 400 pens initially, where Fisher spent over $1 million independently developing it.
2) Grease pencils used by the Soviets were less durable, the writing was smudgy, they created waste, and they were flammable.
3) The Soviets eventually adopted the same Fisher pen because it was a much better and safer product.

Just so there's no misinformation about this.

To expound on this, the flammable part is key. Pieces breaking off and getting into things in space is bad news. Both NASA and the Soviets made questionable safety choices in that period to meet deadlines and political goals but they both did some amazing things. Those soviet venus missions are fascinating (like the soviet scientist's being shocked by the pressures when they first got probes there, and having to really redesign the next series of probes, culminating on finally actually landing on the surface!) The US had great success with Mars missions by contrast. Good stuff.

Anyway, off topic.

AMDRulz: i think I have an old 486 or pentium overdrive or something fan in my basement. I don't recall them being very reliable back then though! I remember them getting clogged with dust and making awful noises. And I also remember at my old work having so many of those geforce gtx210s (I think that's what they were) with those tiny fans die on our clients and replacing them constantly! Maybe it's just bias and I'm mis-remembering. Still, I'm going to wait and see, I think things will have improved.

 

[Ajay]

You haven’t lived till you’ve had two high rpm 60mm Delta fans running @ 100% in two PCs with all copper heatsinks from Thermalright. I’m surprised my wife didn’t banish me to the garage.

 

[Abwx]

That 200cm2 that you pulled out of your ass? Sorry, I have enough debunking the other garbage you spewed in this thread already. The area claim by me is to debunk that the heatsink I used for my nuc is bigger in area than the one you linked in the J5005-ITX. Not only it's false, but by per the review I linked that J5005-ITX is barely adequate for very high loads. And that are 13.5W loads.

.

I said that 200cm2 is enough for the chipset while all you did was to display a cooler that has barely this surface yet are claiming that it s enough for 13.5W albeit at 87°C, that s all you stated so far, i see no formal evaluation of yours, hence your tendency to use a derogative language as a pathetic mean to compensate for this lack.

Now getting back to the matter and say 200cm2 sized plate, wich is not as much as what Gygabyte used, and check how much it can dissipate at an ambiant of 300°K (27°C).

The plate has two sides, so its total radiative surface is 400cm2 and with the condition i stated this allow to dissipate 11-12W just with the radiation even before counting convection.

Q = (5.87 x 10^-8 x T^4)/m2

That make about 31W at 340°K but at the same time there s 20W returned by the ambiant that is at 300°K

If you have noticed the shape of your cooler you can see that fins are faced, this way the energy radiated electromagneticaly (by the fins) is returned to the fins that are facing the ones the other, for the same reason usual CPU coolers are very bad when fanless because most radiated energy is returned to the cooling device, i wont even talk of the air cooled part of a laptop cooler, because basicaly it can use only forced convection as a mean to evacuate heat....

Really depends on case temperature and air flow. Plus, afaik, this is just an ASIC (pretty much switches and amps), which may well have higher temperatures limits than a CPU.

Dunno the exact node but i dont think that they use lower than 28nm, wich is surely the reason of this relatively high TDP, although the late 990FX was 19.6W (at 40nm node) and i dont remember that those AM3+ boards where unreliable chipset wise..

Anyway we ll soon see the MB production tested, methink that this chipset TDP frenzy will end as it begun, abruptly...

 

[DrMrLordX]

Just so there's no misinformation about this.

Yeah I figured the grease pencils could amount to a problem if they fragmented. They still saved a lot of money just using those upfront. I'm surprised they actually got ahold of the Fisher pens eventually.

You haven’t lived till you’ve had two high rpm 60mm Delta fans running @ 100% in two PCs with all copper heatsinks from Thermalright. I’m surprised my wife didn’t banish me to the garage.

Let's hope X570 boards have nothing quite that loud. Granted, I'm running two Noctua industrialppc fans (3000rpm!) full bore 100% of the time on my NH-d15 so I'm not going to be the one bothered by such things.

 

[dnavas]

Granted, I'm running two Noctua industrialppc fans (3000rpm!) full bore 100% of the time on my NH-d15 so I'm not going to be the one bothered by such things.

Hah, I've got some of those on my enermax. At full speed, they are definitely NOT quiet!