Topic: Avalon ASIC users thread - page 208. (Read 438340 times)
Won't be long until it's a month since they shipped! Someone start a countdown!
I love there still aren't any avalon users in this thread... =P
Well there is one don't know if I have ever seen the foundation post them being in possession of the only other machine that seems to have made it into the wild in all these weeks...
I love there still aren't any avalon users in this thread... =P
LOL so you're speaking on Jeffs behalf. He cowardly denied such implication.
Its a rhetorical question, btw.
Its a rhetorical question, btw.
Jeff isn't in a position to say what he's actually thinking
Then why posting this here?
Avalon users thread?
Avalon users thread?
Because the Avalon works best with BTCGuild?
And its highly likely this is one user who just switched on their Avalon ASICS?
Because this user is now up to 940 GH/s? Which is 14 avalon machines.
The real question is why does it bother you so much?
LOL so you're speaking on Jeffs behalf. He cowardly denied such implication.
Its a rhetorical question, btw.
800gh?!?!? Holy Retired Pope! Is this the Friedcat / D3 project or some unknown entity???
No clue, but I just ran the numbers with BFL-like power efficiency and these guys are banking 99,000 a month.
So Jeff, out of curiosity, just how many BTC have you collected so far since having this beast?
[...] More heat on the cables means more losses.
While in theory you are correct (resistivity of metals raises with aprox. 4% per each 10C), in practice, because the power loss in cables is very low compared with other losses, the effect can be neglected.
Quote
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 Smiley
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 Smiley
The problem with 50°C is not related to the ASICs. I guess they gonna work correctly even with 70°C.
But 50°C enviromental temperature and the efficiency of the PSU is where im pointing at. More heat on the cables means more losses.
You might be right about efficiency though but it's difficult to know how much losses there are unless someone mods an Avalon to bring cool air directly to the PSU. The losses probably aren't big enough to matter: there's ~33% of power lost in the system bringing power to the chip and controlling them: even if you are 10% more efficient doing so (generous estimation), that's less than a 4% overall gain, when it will start to matter it will be time to buy the next generation of ASICs...
Could anyone come up with an explanation for the lack of Avalons reaching consumers although I could make a package come from Beijing February the 7th?
Quote
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 Smiley
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 Smiley
The problem with 50°C is not related to the ASICs. I guess they gonna work correctly even with 70°C.
But 50°C enviromental temperature and the efficiency of the PSU is where im pointing at. More heat on the cables means more losses.
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?
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.)
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
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?
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.
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?
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.)
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?
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).
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?
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?
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?
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.
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?
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.
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?
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?
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
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
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.
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?
Jump to: