Avalon ASIC users thread

[runeks]

That entire heatsink on each card is saturated at 50C!  Holy hell on a crutch! 

Are those IR pics accurate?  That seems excessive... if it actually is 40 - 50C, what's going to happen to them in hot weather? 

Maybe Jeff can heat up a room to 85F or so and let it run to see what happens.

Is 50C really a problem? I had the impression that even 80C isn't a problem? My GPU is at 62C right now not doing anything. (FOSS Linux Radeon driver has no power management )

Are there components on the board that might not be able to withstand >50C temperatures?

Remember that the hotter the heatsinks get the more heat they dissipate. Both by radiation but also by warming up the exhaust air more. If Jeff's room is 20C now and he heated it up to 35C I don't think we would see an increase to more than 60C, with the fans running at the same speed.

[1714060437]

1714060437

[1714060437]

1714060437

[1714060437]

1714060437

[1714060437]

1714060437

[MonsterZero]

That entire heatsink on each card is saturated at 50C!  Holy hell on a crutch! 

Are those IR pics accurate?  That seems excessive... if it actually is 40 - 50C, what's going to happen to them in hot weather? 

Maybe Jeff can heat up a room to 85F or so and let it run to see what happens.

Is 50C really a problem? I had the impression that even 80C isn't a problem? My GPU is at 62C right now not doing anything. (FOSS Linux Radeon driver has no power management )

Are there components on the board that might not be able to withstand >50C temperatures?

Remember that the hotter the heatsinks get the more heat they dissipate. Both by radiation but also by warming up the exhaust air more. If Jeff's room is 20C now and he heated it up to 35C I don't think we would see an increase to more than 60C, with the fans running at the same speed.

I know some computer components can run extremely hot.. what I'd be more concerned with high temperatures over long time periods is the lifespan of the device.  Just how long can these keep running at those temps... 3 months... 6 months.. 3 years?... Hopefully we won't find this out anytime soon

[runeks]

That entire heatsink on each card is saturated at 50C!  Holy hell on a crutch!  

Are those IR pics accurate?  That seems excessive... if it actually is 40 - 50C, what's going to happen to them in hot weather?  

Maybe Jeff can heat up a room to 85F or so and let it run to see what happens.

Is 50C really a problem? I had the impression that even 80C isn't a problem? My GPU is at 62C right now not doing anything. (FOSS Linux Radeon driver has no power management )

Are there components on the board that might not be able to withstand >50C temperatures?

Remember that the hotter the heatsinks get the more heat they dissipate. Both by radiation but also by warming up the exhaust air more. If Jeff's room is 20C now and he heated it up to 35C I don't think we would see an increase to more than 60C, with the fans running at the same speed.

I know some computer components can run extremely hot.. what I'd be more concerned with high temperatures over long time periods is the lifespan of the device.  Just how long can these keep running at those temps... 3 months... 6 months.. 3 years?... Hopefully we won't find this out anytime soon

Why would these temperatures break down the device though, as you seems to suggest is possible? Which components are slowly degraded because of 80C temperatures? The PCB itself? Surely the silicon itself can stand temperatures much much higher than that. But of course there are capacitors, voltage converters and all sorts of components I don't know what are, and it's possible they somehow break down at higher temperatures.

I'm genuinely interested in an answer, because we all hear all the time that, for example, running one's CPU/GPU at high temperatures causes its lifespan to decrease, but I haven't actually seen any data to support this. I mean, it obviously gets too hot at some point, where some of the components will not be able to handle this, but whether this is 60C, 80C, 120C or 150C I don't know.

[Photon939]

Usually high temperatures affect the capacitors first, electrolytic capacitors constantly baked in high temperatures will fail rapidly. Also the design of the semiconductor structure in the chip itself can have an impact on the maximum junction temperature. My old laptop's Athlon64 has a max junction temperature of 70C while my new laptop's i7 won't shut down until nearly 100C.

High temperatures will also cause devices to age more rapidly in terms of electromigration and other nasty stuff that some users have observed here when their cards can no longer run stable at high clocks after long periods of operation.

[kaerf]

Bitsyncom, here's a request for information to include in the Avalon newsletter:


1)
What is the status of "shipped" batch #1 orders? (Are they all stuck in customs? Any updates on ETA? How many have shipped?)

2)
What is the status of batch #2 orders?
-From my understanding only a fraction of the 600 units were processed by WalletBit.
-What's the status of confirmation emails for paid orders? (How is order # being determined?)
-Is there going to be another official re-re-opening of remaining batch #2 orders or will they just show up as "in stock" in the store (http://store.avalon-asics.com/ -- which is in a bad state right now)?

[Unacceptable]

Bitsyncom, here's a request for information to include in the Avalon newsletter:


1)
What is the status of "shipped" batch #1 orders? (Are they all stuck in customs? Any updates on ETA? How many have shipped?)

2)
What is the status of batch #2 orders?
-From my understanding only a fraction of the 600 units were processed by WalletBit.
-What's the status of confirmation emails for paid orders? (How is order # being determined?)
-Is there going to be another official re-re-opening of remaining batch #2 orders or will they just show up as "in stock" in the store (http://store.avalon-asics.com/ -- which is in a bad state right now)?

1) Secret..............

2) Secret..............

 

[Phinnaeus Gage]

Come on Bitcoin Foundation, you gotta give up more details than that

I honestly don't know the answer...Yifu will be here today, I dont have the logins.

Can you kindly do us a favor and ask him what the shipping status of batch #1 is and what's happening with batch #2 ?

There should be excuses for TBF to not ask serious questions of Yifu.

[ProfMac]

That entire heatsink on each card is saturated at 50C!  Holy hell on a crutch! 

Are those IR pics accurate?  That seems excessive... if it actually is 40 - 50C, what's going to happen to them in hot weather? 

Maybe Jeff can heat up a room to 85F or so and let it run to see what happens.

Why would he want to do that?   I think we can guess what happens. Either they slow down or it hits a temp cutoff that shuts down mining.  FYI, high temperatures will have adverse effects on mining operations.  What happens to BFL singles when they get too hot, they down-clock until they are within the optimal temperature range. 

I don't know if you noticed, but many parts of the world get to be 85F or more during the summer months.  Personally, I would want to know if my expensive mining hardware was going to shutdown when the temp rose as it would dictate how much power I had to consume to cool the thing with an AC unit.  Maybe you don't care, but I sure as hell would.

In many places where I have lived it is 85F when the air conditioner is running.

[Inspector 2211]

That entire heatsink on each card is saturated at 50C!  Holy hell on a crutch! 

Are those IR pics accurate?  That seems excessive... if it actually is 40 - 50C, what's going to happen to them in hot weather? 

Maybe Jeff can heat up a room to 85F or so and let it run to see what happens.

Why would he want to do that?   I think we can guess what happens. Either they slow down or it hits a temp cutoff that shuts down mining.  FYI, high temperatures will have adverse effects on mining operations.  What happens to BFL singles when they get too hot, they down-clock until they are within the optimal temperature range. 

I don't know if you noticed, but many parts of the world get to be 85F or more during the summer months.  Personally, I would want to know if my expensive mining hardware was going to shutdown when the temp rose as it would dictate how much power I had to consume to cool the thing with an AC unit.  Maybe you don't care, but I sure as hell would.

In many places where I have lived it is 85F when the air conditioner is running.

And if Al Gore is right, more and more places will be like that.
Rumor has it, he and his cronies are buying up huge swathes of land in Alaska and northern Canada for future pineapple and sugarcane plantations...

[Dalkore]

That entire heatsink on each card is saturated at 50C!  Holy hell on a crutch! 

Are those IR pics accurate?  That seems excessive... if it actually is 40 - 50C, what's going to happen to them in hot weather? 

Maybe Jeff can heat up a room to 85F or so and let it run to see what happens.

Why would he want to do that?   I think we can guess what happens. Either they slow down or it hits a temp cutoff that shuts down mining.  FYI, high temperatures will have adverse effects on mining operations.  What happens to BFL singles when they get too hot, they down-clock until they are within the optimal temperature range. 

I don't know if you noticed, but many parts of the world get to be 85F or more during the summer months.  Personally, I would want to know if my expensive mining hardware was going to shutdown when the temp rose as it would dictate how much power I had to consume to cool the thing with an AC unit.  Maybe you don't care, but I sure as hell would.

In many places where I have lived it is 85F when the air conditioner is running.

Guess what, those places might not be suitable for bitcoin mining.  Power cost and climate are things to take into consideration.

[Inaba]

That entire heatsink on each card is saturated at 50C!  Holy hell on a crutch! 

Are those IR pics accurate?  That seems excessive... if it actually is 40 - 50C, what's going to happen to them in hot weather? 

Maybe Jeff can heat up a room to 85F or so and let it run to see what happens.

Why would he want to do that?   I think we can guess what happens. Either they slow down or it hits a temp cutoff that shuts down mining.  FYI, high temperatures will have adverse effects on mining operations.  What happens to BFL singles when they get too hot, they down-clock until they are within the optimal temperature range. 

I don't know if you noticed, but many parts of the world get to be 85F or more during the summer months.  Personally, I would want to know if my expensive mining hardware was going to shutdown when the temp rose as it would dictate how much power I had to consume to cool the thing with an AC unit.  Maybe you don't care, but I sure as hell would.

In many places where I have lived it is 85F when the air conditioner is running.

Guess what, those places might not be suitable for bitcoin mining.  Power cost and climate are things to take into consideration.

Yeah, you are probably right... so it's wise to investigate whether or not the hardware you are buying will fail miserably in the climate you live in, don't you think?

[ab8989]

I see these temperatures as something not to worry about much. The heatsink seems to be about 50C and the board where the chips sit on is about 55C. It is quite significant issue to notice that the power consumption per chip is ridiculously low ( about 1.7W ) and so the thermal resistance between the silicon and the surrounding ambient temperature around the chips can be really really poor and still because of the thermal power that needs to be transferred is so low it simply cannot heat the silicon to a temperature worth worrying about.

It is a totally different situation when there is only one chip that is consuming huge wattage. In that case the cooling of the one chip becomes critical and a problem. The amount of chips or total power consumption does not matter in here because that has already been taken into consideration as that is the power that heats up the heatsink and the board and they have been measured in here. It is only question how much the 1.7 watts can heat the silicon of one chip above its surroundings that matters and it cannot be that much.

I see the text from Inaba not show much understanding of basic fundamentals of thermal engineering.

400W/(3*8*10)=1.66W

[kano]

So, looking from outside the box, by someone who doesn't know much about thermal dynamics ...

Why does the heat sync get to 50C?
If the chips aren't very hot, how can that whole big heat sync get hot?

i.e. if it's not the chips, then something else must be applying the heat to the heat sync to heat it up ...
What is this something else?

[Phinnaeus Gage]

So, looking from outside the box, by someone who doesn't know much about thermal dynamics ...

Why does the heat sync get to 50C?
If the chips aren't very hot, how can that whole big heat sync get hot?

i.e. if it's not the chips, then something else must be applying the heat to the heat sync to heat it up ...
What is this something else?

It's either that little red transistor in far corner or the power supply unit. My money's on the latter.

[ab8989]

So, looking from outside the box, by someone who doesn't know much about thermal dynamics ...

Why does the heat sync get to 50C?
If the chips aren't very hot, how can that whole big heat sync get hot?

i.e. if it's not the chips, then something else must be applying the heat to the heat sync to heat it up ...
What is this something else?

What heats up the heatsink is not any one of the chips, it is all of them and there sure is lots of them in that box.
All together there is 620W being converted into heat inside that box. That is quite enough to make things 50C.

What I was talking about that once you measure what is the effect of 620Watts in a box, then addtionally on top of that there is much less going on temperaturevice than on normal board and therefore the fact that the heatsink gets 50C is not an indication of a problem at all. Usually there are additionally on top of that individual components that are running much hotter than the general environment, but not so in here. BFL quite probably is having some problems with individual chips and there they would be worried about hot general environment to start with.

Perhaps the hottest component in Avalon is found from power supply units, but I assume that is also under control. They would not publish the pictures othervice. That picture was meant to calm things down to show something that proves that they have no problems and it sure looks like that way.

[kano]

Yeah but for the heat sinks themselves, correct me if I'm wrong, but isn't there only 3 ways to get to 50C
1) The ambient temperature is at or above 50C (the heat sink on this side of the box away from the power supply?)
2) The devices touching it are at or above 50C
3) Electricity is flowing into the heat sinks and heating it that way (clearly not)
What have I missed?

[gyverlb]

Yeah but for the heat sinks themselves, correct me if I'm wrong, but isn't there only 3 ways to get to 50C
1) The ambient temperature is at or above 50C (the heat sink on this side of the box away from the power supply?)
2) The devices touching it are at or above 50C
3) Electricity is flowing into the heat sinks and heating it that way (clearly not)
What have I missed?

I don't understand why 50°C on heatsinks is a problem. The most likely explanation is that the chips run above 50°C but most of the chip technologies are able to withstand 70°C and for many of them well beyond that (I know some can reach 100°C: the oldest Pentium IV did).
Avalon stated that the fans are regulated and ramp up when the temperature does. If it's indeed the case (jgarzik should be able to answer that), unless they are already running at full speed, I don't see what the problem is.

I would be more worried about their inability to ship other units from China: I've a package in transit picked up by DHL February the 7th in China. If I'm not mistaken other units should have been ready on that date so they clearly should have been able to ship them. So either the units weren't ready or they choose not to send them (maybe DHL didn't fit in their margins but that wasn't a reason given by Avalon).

[ab8989]

Yeah but for the heat sinks themselves, correct me if I'm wrong, but isn't there only 3 ways to get to 50C
1) The ambient temperature is at or above 50C (the heat sink on this side of the box away from the power supply?)
2) The devices touching it are at or above 50C
3) Electricity is flowing into the heat sinks and heating it that way (clearly not)
What have I missed?

It is the number 2. All of the electrical components are hotter than 50C and they are generating the heat and from them the heat is flowing to the board and heatsink. But how MUCH hotter they are is the question and it looks like they are not so MUCH hotter than 50C that it makes a problem as their temperature still looks to stay at acceptable levels ( or at least there is no indication of that not being the situation.)

[punin]

So, looking from outside the box, by someone who doesn't know much about thermal dynamics ...

Why does the heat sync get to 50C?
If the chips aren't very hot, how can that whole big heat sync get hot?

i.e. if it's not the chips, then something else must be applying the heat to the heat sync to heat it up ...
What is this something else?

Ofc it's mainly the chips that heat up the heat sink. Then there's the power supply circuitry which is attached to the side of the heat sink. All-in-all there's about 500W of heat that needs to be dissipated by those heat sinks. Thermal resistance from chip to heat sink should be around 15C°/W if copper filled vias are used. This would give Tj of max 70°C which is quite acceptable for an ASIC.

[kano]

Yeah but for the heat sinks themselves, correct me if I'm wrong, but isn't there only 3 ways to get to 50C
1) The ambient temperature is at or above 50C (the heat sink on this side of the box away from the power supply?)
2) The devices touching it are at or above 50C
3) Electricity is flowing into the heat sinks and heating it that way (clearly not)
What have I missed?

It is the number 2. All of the electrical components are hotter than 50C and they are generating the heat and from them the heat is flowing to the board and heatsink. But how MUCH hotter they are is the question and it looks like they are not so MUCH hotter than 50C that it makes a problem as their temperature still looks to stay at acceptable levels ( or at least there is no indication of that not being the situation.)

Thanks - yep - sounds good to me.
Yes I do understand that even at 70C the chips may not be shortening their lifetime, but I was just getting the impression that it wasn't the chips or somehow they weren't even 50C.
Hopefully that clears it up for anyone else other than me too