Topic: Cairnsmore5 Pre-Order (Read 2034 times)
Amoveo algorithm is now available by downloading our latest microSD image at https://www.enterpoint.co.uk/products/cairnsmore-5/.
We have found an Ethernet bug that appears with some but not all gigabit hubs/switches. This is purely at startup and there are a number of ways to get past this initial startup issue. We are posting one of these on the main product webpage https://www.enterpoint.co.uk/products/cairnsmore-5/ . Once the gigabit link is up there are no further issues. There are no issues with 10/100 hubs and switches.
A work around is available now on the product webpage. We also noted that we don't tend to have the issue with D-Link gigabit switches but do see it with Netgear gigabit Switches. Others we don't know yet.
A work around is available now on the product webpage. We also noted that we don't tend to have the issue with D-Link gigabit switches but do see it with Netgear gigabit Switches. Others we don't know yet.
We are now testing radiator solutions for the first rack solution which will be between 4 and 6 boards in a 6U 19" rack together with radiator, pump and power supply.
We have a choice of connections to the coldplate. The standard is 6mm stems shown below.
Seems like you have quite the power expectations from how the board looks but how much power are you expecting to go through these cards with only dual pcie? I don't doubt your cards living, especially with the liquid cooling as standard but I think there will definitely be foolish people melting some power cables.
Amoveo and Denarius bitstreams are running roughly at about 200W in the current versions. I think with add-on modules kept reasonable I think a max power of maybe 300W is to be expected over all bitstreams. We have run 200W off a single PCIe cable (good quality PSU) but I would not recommend it. Using twin PCIe keeps the cables in the comfortable zone and that doesn't meant using splitters. Use proper off the PSU cables. We have seen splitters melt in the past and nearly go on fire.
Thermally it is probably just about possible to run with air cooling at 200W on CM5 but there are noise and efficiency aspects to consider. Like many performance FPGAs and similar the Arria10 we use has a thermal runaway aspect which means the hotter it gets the more power it takes. Doesn't necessarily mean yet get more hashing but it certainly drops efficiency. With the current liquid cooling coldplate typically we get the Arria10 (2off -200W running) sitting 35-45degC above the ambient temperature that the radiator sits in. So with radiator in a 20degC room the Arria10 die might be around 60-65degC. That is comfortable and good for Arria10 lifetime and reliability. I'd expect an air cooled solution to be more like 90-95degC. Apart from efficiency running at 90-95degC that could be a lifetime limiting factor. At 60-65degC we expect a very long lifetime on Arria10.
Noise on a liquid cooled setup is good. I can stand next to a test rig with 10 boards hashing and talk in a normal voice. That's with radiators and pumps next to boards and not semi-remote outside which can be even more thermally efficient if it is colder outside. With a air cooled setup of the same numbers of boards I would be wearing ear defenders and shouting at people to communicate.
Is dedmaroz a part of your Bitstream development team or is he a third party and is he supporting this specific board?
Yes dedmaroz is working with us. He has said in public that he is only workng on 2 boards one of those being Cairnsmore5. We also have 3 or 4 other developers working with us as well and once our team has reduced load we will be doing something in that area as well.
Are you accepting pre orders or is there a potential for a group buy?
There is a new pricing structure now in place that has a small discount for numbers it's slightly more expensive(5-10%) in low numbers that the early adopter pricing mainly to reflect addition costs we added to the design at various points in the design process.
Cairnsmore5 ex-works pricing (excluding VAT, sales tax, shipping, insurance etc.)
Purchase Quantity USD $ Price GBP £ Price Euro € Price (monthly changed exchange rate)
1-9 $1,750.00 £1,350.00 €1,620.00
10-24 $1,730.00 £1,330.00 €1,600.00
25-49 $1,700.00 £1,300.00 €1,560.00
50+ $1,680.00 £1,290.00 €1,550.00
We are just starting to ship the early adopter preorders and generally production is picking up now. If you want to order email [email protected] and we will send you the formal offer document.
Nobody as yet has talked about a group offer and previously flat and now near flat pricing structure don't make much sense as a group buy.
Update on bitstreams - Amoveo and Denarius(Tribus) are now available. More coming soon.
Bitstream page created at https://www.enterpoint.co.uk/products/cairnsmore-5/cairnsmore-5-bitstreams/ . More coming on this shortly.
Updates for the Controller Linux subsystem and various other tools now linked off the main product webpage https://www.enterpoint.co.uk/products/cairnsmore-5/ .
Are you accepting pre orders or is there a potential for a group buy?
Sure, they will copy aria10 and mass scale produce it - where?
can this shipping to China,
This card have any bitstream now?
This card have any bitstream now?
Yes can ship to China.
Bitstream announcements coming shortly.
Don't send it to China. They will copy it and sell it cheep.
can this shipping to China,
This card have any bitstream now?
This card have any bitstream now?
Yes can ship to China.
Bitstream announcements coming shortly.
can this shipping to China,
This card have any bitstream now?
This card have any bitstream now?
Some pictures of the Simple Interface Board which the first of the accessory items for this product. It's aimed at users with small numbers of boards. There will be backplane and rack offerings as well.
Board fitted with simple interface.
Board fitted with simple interface.
legendary
Activity: 1848
Merit: 1166
My AR-15 ID's itself as a toaster. Want breakfast?
*snip of reply above*
I appreciate the detailed answer; It should help people make a decision for sure.
im going to be the skeptic; don't take it the wrong way.
Why arent you doing this? :
Things such as currently working bitstreams, guarantee of algos that will be available by what date (specifically) as well as hash to power figures....
As of this point; in this day and age; you should have at least a physical demo and review of the hardware verified by a heavily trusted member before taking in any funding.... as in the past; taking in funds before us seeing some actual results on this website from a trusted reviewer, has almost always resulted in disaster of many different levels and proportions.
At first glance, I see a big up front hardware cost; with no promise of return other than a stated interest in algos (no word on active development or working hashing bitstreams for instance...); especially since you mention algos that are constantly changing to be as asic/FPGA resistant as possible... we aren't not even any sort of efficiency datum or anything; just a spec sheet for a publicly sourceable fpga.
The hardware looks like it should be very versatile; and it looks like the product as shown is a sound concept/design.. but; you are asking a bunch of people to jump off a cliff right now without knowing if there's a parachute attached to their back while boarding the plane.
I hope you understand my concern, and its meant in the best of light to help you succeed. These are the main red flags I have to share and ask about.
Why arent you doing this? :
The expected production process is as follows:
(2) Gather enough pre-orders to be viable - aim over next 1-2 weeks.
(1) Give out full details of product and offer.
(3) Payment for pre-order/ hard commit by customers.
(4) First production boards get delivered approximately 8-10 weeks after stage (2)/(3). We will try to better this but this number determined by our parts suppliers.
(2) Gather enough pre-orders to be viable - aim over next 1-2 weeks.
(1) Give out full details of product and offer.
(3) Payment for pre-order/ hard commit by customers.
(4) First production boards get delivered approximately 8-10 weeks after stage (2)/(3). We will try to better this but this number determined by our parts suppliers.
Things such as currently working bitstreams, guarantee of algos that will be available by what date (specifically) as well as hash to power figures....
As of this point; in this day and age; you should have at least a physical demo and review of the hardware verified by a heavily trusted member before taking in any funding.... as in the past; taking in funds before us seeing some actual results on this website from a trusted reviewer, has almost always resulted in disaster of many different levels and proportions.
At first glance, I see a big up front hardware cost; with no promise of return other than a stated interest in algos (no word on active development or working hashing bitstreams for instance...); especially since you mention algos that are constantly changing to be as asic/FPGA resistant as possible... we aren't not even any sort of efficiency datum or anything; just a spec sheet for a publicly sourceable fpga.
The hardware looks like it should be very versatile; and it looks like the product as shown is a sound concept/design.. but; you are asking a bunch of people to jump off a cliff right now without knowing if there's a parachute attached to their back while boarding the plane.
I hope you understand my concern, and its meant in the best of light to help you succeed. These are the main red flags I have to share and ask about.
Fair points and stages 1,2,3 have stretched out for a combination of reasons. The first part of this was it took us longer to complete the initial design than anticipated and that was mainly due to us having to redesign the controller and power sections a few times to shrink circuit size and to allow running at higher power levels as requested by our potential bitstream providers. The next part of the delay was the crypto wobble we saw in late 2018 and we ourselves took a few weeks to evaluate the situation and basically delayed building dev boards and going forward to making the offer to customers. The same wobble also meant that it has taken longer to get to the critical mass of orders. We also slipped in a large unrelated rush customer job to improve our funding and give us more options on how we did the build.
The next part of what is happening is that we are mainly reliant on outside developers for bitstreams. We do have a very junior member of our team doing an initial first algorithm backup it needs one of our senior people to be free to help get that running. This option was only ever meant to be a last resort and our senior team focus is still on the evaluation of the Issue1, getting Issue 1.1 to developers, and the now running customer build. External devs are delayed on lack of hardware and basically overloading and of course this is a concern. However in the worst scenario we can simply turn most our very experienced FPGA team to algorithm development and that will happen about late February/early March should no other bitstreams be available. We do have have public statements by Whitefire/Zetheron on Discord stating they will be providing bitstreams and they are just one of several devs working with us.
The main FPGAs are not a secret on the design and there is more on this in Discord but they can actually be read off the block diagram available on the product webpage. They are Intel Arria10 GX1150 and there are two of those on board. Arria10 selection table giving resource outline here - https://www.intel.com/content/dam/www/programmable/us/en/pdfs/literature/pt/arria-10-product-table.pdf . The board has a USB hub inbuilt and off that an inbuilt JTAG programmer that works directly with Intel Quartus tools. That's the crude programming mechanism. The on board Cyclone5 SoC will run Linux and offer a pile of other bitstream loading options.
We did consider forms of escrow with senior member community but we have difficulty in justifying the costs given out low margin and also that we are a reputable business trading for 30 years next month without any black marks on our record. That's public record if anyone cares to look it up and a longer public record than pretty much anyone else in crypto community. The crypto community also has a lot of experience in our previous product Cairnsmore1 from 2012/2013 where we used a very similar product launch mechanism and dare I say lots of people made good money out that product. Nobody lost out with us on that development. On the CM1 development/product launch we took peoples funds, we kept to timeline pretty much, and external developers delivered bitstreams in time for the first boards shipping. That was before external developers even had a monetary reason to do the work which is one major change today with dev fees being common practise.
We have recognised early customers risk with an extended purchase discount scheme going forward and there will be a more formal structure coming for that in the next few weeks.
Hope that explains where we are on this project to everyone here. I'll be happy to answer questions as soon as I see them although there may be more delay on here versus the cairnsmore5 email address or on Discord as I have to reconfigure my email alerts here. Just one of many jobs on my task list currently.
legendary
Activity: 1848
Merit: 1166
My AR-15 ID's itself as a toaster. Want breakfast?
im going to be the skeptic; don't take it the wrong way.
Why arent you doing this? :
Things such as currently working bitstreams, guarantee of algos that will be available by what date (specifically) as well as hash to power figures....
As of this point; in this day and age; you should have at least a physical demo and review of the hardware verified by a heavily trusted member before taking in any funding.... as in the past; taking in funds before us seeing some actual results on this website from a trusted reviewer, has almost always resulted in disaster of many different levels and proportions.
At first glance, I see a big up front hardware cost; with no promise of return other than a stated interest in algos (no word on active development or working hashing bitstreams for instance...); especially since you mention algos that are constantly changing to be as asic/FPGA resistant as possible... we aren't not even any sort of efficiency datum or anything; just a spec sheet for a publicly sourceable fpga.
The hardware looks like it should be very versatile; and it looks like the product as shown is a sound concept/design.. but; you are asking a bunch of people to jump off a cliff right now without knowing if there's a parachute attached to their back while boarding the plane.
I hope you understand my concern, and its meant in the best of light to help you succeed. These are the main red flags I have to share and ask about.
Why arent you doing this? :
The expected production process is as follows:
(2) Gather enough pre-orders to be viable - aim over next 1-2 weeks.
(1) Give out full details of product and offer.
(3) Payment for pre-order/ hard commit by customers.
(4) First production boards get delivered approximately 8-10 weeks after stage (2)/(3). We will try to better this but this number determined by our parts suppliers.
(2) Gather enough pre-orders to be viable - aim over next 1-2 weeks.
(1) Give out full details of product and offer.
(3) Payment for pre-order/ hard commit by customers.
(4) First production boards get delivered approximately 8-10 weeks after stage (2)/(3). We will try to better this but this number determined by our parts suppliers.
Things such as currently working bitstreams, guarantee of algos that will be available by what date (specifically) as well as hash to power figures....
As of this point; in this day and age; you should have at least a physical demo and review of the hardware verified by a heavily trusted member before taking in any funding.... as in the past; taking in funds before us seeing some actual results on this website from a trusted reviewer, has almost always resulted in disaster of many different levels and proportions.
At first glance, I see a big up front hardware cost; with no promise of return other than a stated interest in algos (no word on active development or working hashing bitstreams for instance...); especially since you mention algos that are constantly changing to be as asic/FPGA resistant as possible... we aren't not even any sort of efficiency datum or anything; just a spec sheet for a publicly sourceable fpga.
The hardware looks like it should be very versatile; and it looks like the product as shown is a sound concept/design.. but; you are asking a bunch of people to jump off a cliff right now without knowing if there's a parachute attached to their back while boarding the plane.
I hope you understand my concern, and its meant in the best of light to help you succeed. These are the main red flags I have to share and ask about.
Just wanted to say the Cairnsmore 1 was my first ever FPGA and an awesome piece of hardware in its time. Wishing you guys luck with Cairnsmore 5
Maybe a little off topic but Amoveo miner implementation now available for Cairnsmore1...
...
We are hoping this will get rebuilt and appear on the Cairnsmore5 platform before too long.
...
We are hoping this will get rebuilt and appear on the Cairnsmore5 platform before too long.
It's great that you got the attention of dedmaroz! Using CM1 devfees to fund CM5 development is the circle of life!
I would be very interested as I like to be the first one in on new tech like this, but with the recent price crash I gotta sit this out till things recover. At least a little. If this had come out a few months ago...
I should have sold a few coins back when they were at $8k a few months ago. If I had I would probably buy quite a few.
I should have sold a few coins back when they were at $8k a few months ago. If I had I would probably buy quite a few.
I think most of us wish that. We took a short pause on this to improve a few things in the CM5 design and evaluate the situation. We think it is still very much viable given how much better ROI we think there is in the product versus others. In context Bitcoin particularly has crashed many times before and always come back stronger. I think that will be true for all the stringer coins. It might be the more obscure coins might die snd leave a clearer more concise set of coins for people to work with. That's my guess.
Pictures of the Issue1 are now on our webpage https://www.enterpoint.co.uk/products/cairnsmore-5/.
Offical offer released with first batch scheduled to ship March 2019.
Maybe a little off topic but Amoveo miner implementation now available for Cairnsmore1. Get it here https://github.com/dedmarozz/Amoveo-Cairnsmore1-FPGA-Miner/releases .
We are hoping this will get rebuilt and appear on the Cairnsmore5 platform before too long.
I would be very interested as I like to be the first one in on new tech like this, but with the recent price crash I gotta sit this out till things recover. At least a little. If this had come out a few months ago...
I should have sold a few coins back when they were at $8k a few months ago. If I had I would probably buy quite a few.
I should have sold a few coins back when they were at $8k a few months ago. If I had I would probably buy quite a few.
interested...
but i would like know better
but i would like know better
Jump to: