[Work in progess] Burnins Avalon Chip to mining board service

[CumpsD]

Fyi, the K16's of BkkCoins are finished. Looking eagerly to see Bitburners get into production

[SebastianJu]

Fyi, the K16's of BkkCoins are finished. Looking eagerly to see Bitburners get into production

I hope some overclocking tests are done too. And i would like if burnin could make a overclocking test with watercooling so one could check if its worth.

[RoadStress]

Fyi, the K16's of BkkCoins are finished. Looking eagerly to see Bitburners get into production

I hope some overclocking tests are done too. And i would like if burnin could make a overclocking test with watercooling so one could check if its worth.

Actually i think that's exactly what he is doing now. I feel/hope that an update will come very soon.

[nemercry]

Fyi, the K16's of BkkCoins are finished. Looking eagerly to see Bitburners get into production

Finished ? Did you read the Thread entirely ?
The only thing i read there is that they are currently not able to get higher than 250 - 300 mhz without having a lots of Hardware errors.
Nothing is finished there in my point of view.

[eros]

Eros:
If you want to 11 000 Gh then use BitFury device, electricity company will thank you.
Also i am sure that if you will order 40000 avalon chips and pay to burnin $200000 plus delivery then you will get on your plate everything you want.

Yes, but I think order only ~10pcs bare pcb. And like to get ethernet version, but I have also junk computers with usb if ethernet are not availabe.
 10000pcs avalons need ~30kw. Over 50A three phase 400volt mains are needed. 40000pcs avalos needs so big mains that only rare places have it. And I don't have 40000pcs avalons and no ideas how to get them in my hand
 btw 200000$ is not enough for 40000pcs if follow burning list price but maybe he can give some bulk price..

bitfury standard size is 400Ghash. I'll like to get them say few containers but lack some 10⁶ euros..

[burnin]

Fyi, the K16's of BkkCoins are finished. Looking eagerly to see Bitburners get into production

I hope some overclocking tests are done too. And i would like if burnin could make a overclocking test with watercooling so one could check if its worth.

Done some measurements as requested.
But again didn't dare the 1.3V. (maybe tomorrow)
But i am pretty sure that with 1.3V the 450Mhz could be doable.
The stabilizing effect of voltage adjustment is obvious with this dataset.
Core Voltage can be adjusted in fine steps just like your GPU, not just 1.1/1.2/1.3 like planned before.
The Water cooler did not have the impact i had hoped for.
But the undervolting tests were encouraging.
See for yourself:

ambient: 25°C, water temperature 26°C
Chip Frequency - Chip surface temp, cgminer board-temp, Core voltage, power consumption, comments

Air cooling:
282 - 45.8, 40.5, 1.2, 46W, stable
340 - 50, 43, 1.2, 56W, stable
372 - 51.5, 45, 1.2, 61W, stable
390 - 52.5, 46, 1.2, 2% hw errors
390 - 53.3, 47.5, 1.25, 70W, stable
409 - 55, 48.5, 1.25, 2% errors
409 - 56.5, 51, 1.28, 77W, stable
434 - 57, 52, 1.28, 82W, 10% hw errors

Water cooling by Anfi-Tec:
282 - 40, 30, 1.016V, 33W, stable 170Mhash/watt undervolting
282 - 42, 30, 1.085V, 37W, stable
282 - 45, 30, 1.20V, 46W, stable stock 122Mhash/Watt
340 - 48, 30, 1.20V, 56W, stable
372 - 50, 30, 1.20V, 61W, stable
390 - 53, 30, 1.25V, 70W, stable
409 - 54, 30, 1.28V, 77W, stable
434 - 57, 31, 1.28V, 82W, 4% hw errors

Project status:
First batch of boards is in production so I can react immediately when the chips are shipped.
Website - almost ready to launch.

[ecliptic]

That's damn impressive, 409 MH/sec @ 56.5C, 1.28V, 3A dead on the money if you assume 100% efficiency (77W / 20 / 1.28V ) = 3.0078A.  2.7A with 90% efficiency, so probably around 2.6-2.7A in reality

Too bad about water cooling, but it looks like the limiting factor here isn't cooling, since the air cooling did almost equally as well.  The main thermal resistance is probably in the PCB.  Was that with forced airflow on the heatsink?

[madmax_ger]

@burnin, these are promising measurements. did you try undervolting with aircooling too or are there specific reasons you didn't? Am I right guessing you just didn't wanted to rollback your setup to aircooling?

and..

Project status:
First batch of boards is in production so I can react immediately when the chips are shipped.
Website - almost ready to launch.

great news, that's fantastic!!

[eraziel]

I am going on vacation Friday to Monday and with my luck I can bet that's when the site will be opened for orders... :DDD

Now the question is whether Avalon will deliver the chip orders on time or not

[tarmi]

Now the question is whether Avalon will deliver the chip orders on time or not

they are already late.

[ujka]

Now the question is whether Avalon will deliver the chip orders on time or not

they are already late.

For orders made in April, yes, a week or two behind '10 week delivery'. But I think Avalon did not place each  batch order immediately at the fab - 10000 chips can be produced on 3 wafers. They probably waited and made an order for bigger quantities. In latest 'newsletter update' they said chips are done, and shipping will start mid-July. So let's hope first batches will be here soon.

[jlsminingcorp]

Done some measurements as requested.
But again didn't dare the 1.3V. (maybe tomorrow)
But i am pretty sure that with 1.3V the 450Mhz could be doable.
The stabilizing effect of voltage adjustment is obvious with this dataset.
Core Voltage can be adjusted in fine steps just like your GPU, not just 1.1/1.2/1.3 like planned before.
The Water cooler did not have the impact i had hoped for.
But the undervolting tests were encouraging.
See for yourself:

ambient: 25°C, water temperature 26°C
Chip Frequency - Chip surface temp, cgminer board-temp, Core voltage, power consumption, comments

Air cooling:
282 - 45.8, 40.5, 1.2, 46W, stable
340 - 50, 43, 1.2, 56W, stable
372 - 51.5, 45, 1.2, 61W, stable
390 - 52.5, 46, 1.2, 2% hw errors
390 - 53.3, 47.5, 1.25, 70W, stable
409 - 55, 48.5, 1.25, 2% errors
409 - 56.5, 51, 1.28, 77W, stable
434 - 57, 52, 1.28, 82W, 10% hw errors

Water cooling by Anfi-Tec:
282 - 40, 30, 1.016V, 33W, stable 170Mhash/watt undervolting
282 - 42, 30, 1.085V, 37W, stable
282 - 45, 30, 1.20V, 46W, stable stock 122Mhash/Watt
340 - 48, 30, 1.20V, 56W, stable
372 - 50, 30, 1.20V, 61W, stable
390 - 53, 30, 1.25V, 70W, stable
409 - 54, 30, 1.28V, 77W, stable
434 - 57, 31, 1.28V, 82W, 4% hw errors

Project status:
First batch of boards is in production so I can react immediately when the chips are shipped.
Website - almost ready to launch.

Wow, those are some seriously impressive numbers! Congratulations burnin, these are going to be some really impressive devices . Hurry up Avalon!!!!!

Does anybody have a feeling for how overclocking might affect the longevity of the Avalon chips/board? I'd be really happy if my GPUs ran at 55 oC and would expect them to last for ages at that temp. However, I guess that they are engineered to withstand higher temps. Is there any reason why the Avalons would degrade quickly at these (relatively low) temps (I'm afraid that I'm no EE)?

[f.baasch75]

really impressive, really brilliant!

[Schrankwand]

Thanks Burnin.


I will in fact not go for water cooling thanks to your tests, the additional cooling and mh/s ratio is not worth it for me. Also for the newer values calculated.

[ummas]

I`m in with WC.
AC will be noissy. With WC i can put hest whre ever i need/wont.
We know nothing about lifetime of the chip. We know it works for 6 mounths with no OC. But what about after OC?
AVALONS can be oc`ed for haw long, month or so?
I`ll better pay little bit more and stay safe.
Pmps, radioators and all the stff i have alredy,

[nemercry]

I`m in with WC.
AC will be noissy. With WC i can put hest whre ever i need/wont.
We know nothing about lifetime of the chip. We know it works for 6 mounths with no OC. But what about after OC?
AVALONS can be oc`ed for haw long, month or so?
I`ll better pay little bit more and stay safe.
Pmps, radioators and all the stff i have alredy,

Does not sound bad at all. @Burnin: Do you have tried to cool the other components too with some air cooling ? Maybe you could reach 450MHZ with extra air cooling on the left components while using watercooling on the chips.

[ecliptic]

I`m in with WC.
AC will be noissy. With WC i can put hest whre ever i need/wont.
We know nothing about lifetime of the chip. We know it works for 6 mounths with no OC. But what about after OC?
AVALONS can be oc`ed for haw long, month or so?
I`ll better pay little bit more and stay safe.
Pmps, radioators and all the stff i have alredy,

Huh?  Based on burnin's numbers, 372 MHz is the limit for Vcore = 1.2V, regardless of cooling method.  Water cooling is only slightly cooler (50C vs 54C).

Watercooling's only advantage is it would be quieter


and actually, cgminer's temps are basically useless on water..  they're all 30, even when the chip itself is getting hotter

[madmax_ger]

I`m in with WC.
AC will be noissy. With WC i can put hest whre ever i need/wont.
We know nothing about lifetime of the chip. We know it works for 6 mounths with no OC. But what about after OC?
AVALONS can be oc`ed for haw long, month or so?
I`ll better pay little bit more and stay safe.
Pmps, radioators and all the stff i have alredy,

Huh?  Based on burnin's numbers, 372 MHz is the limit for Vcore = 1.2V, regardless of cooling method.  Water cooling is only slightly cooler (50C vs 54C).

Watercooling's only advantage is it would be quieter


and actually, cgminer's temps are basically useless on water..  they're all 30, even when the chip itself is getting hotter

perhaps the WC guys should start thinking about downcooling water
what about putting the radiator / mora in a fridge?

[nemercry]

I`m in with WC.
AC will be noissy. With WC i can put hest whre ever i need/wont.
We know nothing about lifetime of the chip. We know it works for 6 mounths with no OC. But what about after
 OC?
AVALONS can be oc`ed for haw long, month or so?
I`ll better pay little bit more and stay safe.
Pmps, radioators and all the stff i have alredy,

Huh?  Based on burnin's numbers, 372 MHz is the limit for Vcore = 1.2V, regardless of cooling method.  Water cooling is only slightly cooler (50C vs 54C).

Watercooling's only advantage is it would be quieter


and actually, cgminer's temps are basically useless on water..  they're all 30, even when the chip itself is getting hotter

What do you mean with useless on water? I dont get your explanation could you try this in more detail ?

Still does the board with the cooler temperature has slighty less errors. I think errors does mean, that the board answer back with wrong or non nonces, what should mean that the calculated hashes are wrong or not existing. So i  would assume that
434 - 57, 31, 1.28V, 82W, 4% hw errors means 434 Mhz - 17.36Mhz(434*0.04Hwerrors) = 416.64 Mhz (actual usable speed with WC)
434 - 57, 52, 1.28, 82W, 10% hw errors means 434 - 43(434*0.1Hwerrors) = 390.6 Mhz(actual usable speed with AC)
[im not sure if this behaves linear but i take it as assumption.]

If WC would have zero effect on the boards, there should be no drop at the hw rates.
But there are drops in errors, even they are low. But they could be low because of the fact, that the reason of the hw errors does not effectively lies in the chips but instead in the heat of the sourrounding components of the board.
 

[Schrankwand]

434 - 57, 31, 1.28V, 82W, 4% hw errors means 434 Mhz - 17.36Mhz(434*0.04Hwerrors) = 416.64 Mhz (actual usable speed with WC)
434 - 57, 52, 1.28, 82W, 10% hw errors means 434 - 43(434*0.1Hwerrors) = 390.6 Mhz(actual usable speed with AC)

And that means, the choice here is between:

434 - 57, 31, 1.28V, 82W, 4% hw errors means 434 Mhz - 17.36Mhz(434*0.04Hwerrors) = 416.64 Mhz

and 409 - 56.5, 51, 1.28, 77W, stable


If you already have everything in place, I doubt you need the water blocks. But if you ask me, that means that I am going to buy a 7mh/s increase for 15€ per board with additional costs of about 200€ per 10-16 boards.

Im not sure that is worth it.