The cooling stuff was not very expensive. There definitely was some secret sauce involved, though. The electrical work was where most of our money went.

Correct. Yellow labels, I think.

For our phase 3, we're thinking of trying to get around 5 MW of machines running in about 5,000 sq ft. That should be fun.

We're having no trouble with cooling. We should be able to fill that rack without a hitch. If there were something 2x as power dense than an SP30, we should be able to cool that too. I'm a little disappointed with how easy it's been.

I was doing some work on our router/server today, and I found the most comfortable place to be (as long as you have earplugs on) is inside the hot aisle, with about half your body in front of the SP30s and half in front of our half-full medium-density rack. That way, you can lean forward to get in the cool airstream if you get too hot, or lean back to take on the SP30 exhaust if you get too cold. Sitting outside the hot aisle in our DC is just too windy and cold to be comfortable.

From one of the machines inside that rack:

Mining Rate: 4528.75Ghs

Temp Front / Back T,B
21 °C / 63,66 °C

Socket Voltage
251/250 volt

Fan Speed
90

Max Watts
1340 / 1335

Excuse me for intercepting. Here's a photo of some of our customers' machines. There's 48 kW running in that rack so far, and we only stopped filling it because we ran out of SP30s.

http://toom.im/images/almost_full_rack.jpg


Yes, I know we need to clean up those cables. We're still changing the port connectivity around; we'll make it neat when we're done rearranging.

1200W will make your PSUs go into thermal overload. When I was using 120V, the best I could get in a cold room was around 1170. I wasn't able to maintain that on both PSUs, though, and not for more than a few hours. The best I could get in a warm room sustained was around 1120.


http://toom.im/faq#orders

Toom.im requires customers to pay for hosting in advance. If they don't pay on time (and we don't grant them an exception), then the machines stop hashing, and we sell the capacity to someone else. If they want their machines back, then the customer pays return shipping costs to the shipping company. No handling charges. Customers never owe us a debt, and therefore can't default. It's simpler this way.

I'm curious about this 300 SP10 default. Link?

$52 UPS Ground, $132 3-Day Select, etc. 18 kg. Contact orders@toom.im if you want a shipping address.

The PSUs are also much louder on 120V than on 200+V. Lower efficiency means more heat, which means faster fans. You could also reduce the noise by putting them on a 240V line.

Mining margins are going to be getting much thinner as time wears on. Only those with the cheapest (and hopefully also cleanest) electricity running the most efficient hardware will be able to make a profit. That's when people will stop adding to the network hashrate, and the so-far inexorable push of the difficulty will finally meet its end. There are places in the world that pay 1/7th of what you do for electricity. I'm in one of them.

If you insist on mining at home, then you have three options, as far as I can see it. you can either

1) Move somewhere with cheaper electricity,
2) Mine at a loss, or
3) Stop mining.

I don't understand this question.


We at Toom.im host SP10s for $390.60 for 3 months.

By the way, an SP10 uses about 1390W. The 1050W figure is the DC rating of the power supply. SP10s are typically run at about 1250W, or about 20% above their rating. (Emerson, the PSU manufacturer, approved this.) When supplying 1250W of DC, the PSU uses about 1390W of AC, for an AC2DC conversion efficiency of about 90%.


3989 GH/s sounds about right for 120V.

I was able to get better hashrate on 120V when using 2.2.28. I got the performance of my SP30 up to 4150 GH/s on that version by tweaking the PSU limits manually. Zvisha hard-limited the PSU performance to something like 1100W in later versions of the FW, so even though you see 1300W on the settings page, you're not getting it.

On the next difficulty (27B), the extra 500 GH/s you'd get by running on 200+V would earn you about $137/month. This is about half of our hosting fee at Toom.im, where we use 254V. In order for hosting on our 254V to be a worse deal than running it at home, you'd have to get your electricity for less than $0.07/kWh, and you'd have to have free cooling and not have the noise be a problem at all.

http://blog.comparemysolar.co.uk/electricity-price-per-kwh-comparison-of-big-six-energy-companies/

Electricity alone in London is about $0.25/kWh, or roughly twice Toom.im's all-in rates.

We at http://Toom.im have been extremely busy for the last week and change, so I apologize for not being more communicative and for getting behind on responding to emails.

By the way, it's Toomim, not Toomin. It's important to spell it correctly, or else you go to a jewelry website in India. We have a tricky name. So sorry.

DevonMiner's machines arrived the day before yesterday and are hashing at 4455 and 4558 GH/s.

We had 6 other SP30s from the group buy arrive today.

Our 37 units from the group buy have also shipped, as of yesterday. We expect them early next week, along with a few other GB SP30s.

We had an issue with an SP10 power supply failing yesterday morning. Failing power supplies is to be expected on occasion, so no surprise. We didn't have any replacement power supplies from Spondoolies in stock yet, but we did have a few DS1050-3 and DS1200-3 PSUs from Emerson (that's the model number used in the SP10 and SP30, respectively) that I bought on Ebay. Oddly, those ebay power supplies didn't work in the SP10. It appears that the 5V connection isn't being made (no CPU, serial communication, LEDs on ethernet), but the 12V connection is working (fans run). However, the power supplies do work in my SP30 just fine, probably since the SP30 shares the 5V connection between the two PSUs, and one PSU's 5V rail is plenty. My SP30's PSU works fine in the SP10, so what I did was I moved my SP30s PSU to the troubled SP10, and put one of the ebay DS1200-3 PSUs in my SP30. We should have some real Spondoolies-approved spare PSUs next week to make this kind of thing unnecessary.

WARNING: There are some serious issues with the I2C interface in most Emerson PSUs. I happened to get a couple off of ebay that did not have these issues. There is a way to get them to work, but it is in no way sanctioned by Spondoolies Tech, and Zvi will probably become moderately irate with me for mentioning publicly that it exists. As a bare minimum, make sure to limit your PSUs to something absurdly low (like 1000W) before attempting any PSU shenanigans. You should also probably just not try it at all unless you need some more bricks.

In other news, we also received our first X-3 today. The customer paid over $5k for this one in November, expecting a February 24th shipment date. One of the hashing units is only running at about 700 GH/s instead of 1.1 TH/s, probably due to a heatsink dislodged during shipping--I'll try to fix this tonight. Each hashing unit has a touchscreen LCD on the front running Android. You can literally browse the web with your mining rig. It's mildly amusing. Each hashing unit hashes at 1/4 the rate of an SP30 in 50% more space while consuming about 25% more power per GH/s. But hey, at least they shipped.

The orange blinking LEDs will be common at first. It indicates a PSU failure of some sort or another. The SP30 is designed to push the PSUs to the point of thermal failure, then remember where that was, and push it slightly less hard next time.

With the Emerson PSUs (everything that's been shipped so far, AFAIK), the PSUs will usually automatically recover after about 20 seconds. Occasionally they don't. I've seen one fail to automatically recover once. In that case, a hard power cycle (unplugging the affected PSU, waiting 30 seconds for it to cool down, then plugging it back in) fixed it. That PSU has had no problems since.

More information on our former business relationship can be found in the Hosting Directory and Reputation thread.

https://bitcointalk.org/index.php?topic=622998.msg8565456#msg8565456


I built a datacenter in Grant County, WA. It is online at 500 kW capacity, soon to be expanded to about 800 kW capacity. We are currently accepting customers. This datacenter is operated by me and my cohort of minions at http://toom.im. We are no longer affiliated with ASICSPACE in any way, nor do they have a stake in our datacenter. Initially, we were building our datacenter with the intent of transferring it to ASICSPACE after it was finished for them to administer and maintain, but during our collaboration I got the impression that they were not competent or reliable enough, so I severed our ties with them and are now administering the facility ourselves. So far, our customers have been giving us good reviews on our service.

ASICSPACE appears to also be attempting to build a datacenter in the same area. I don't know much more about it than that. They were privy to many of the details of my build plan, so they will probably be building something that looks a lot like mine. Now that I've released a video of my facility, they will probably study that for clues about the things I never got around to teaching them. As a result, they may be able to successfully build something. Eventually. If they can find and keep the funding.

Edit: In case anybody read the above message without context, it is a retaliatory negative feedback for the negative feedback I gave them immediately prior.


Edit 2, Jan 5th, 2015: ASICSPACE's message above now says this:

The photo posted by ASICSPACE above (http://imgur.com/WqLy4n9) was uploaded December 16th, 2014 and taken not long before that. Their DC was not quite finished when the photo was taken. The claim of a preorder scheme was made in late August. Their DC was not almost done in August, if it had even been started. They ninja-edited a photo they took in December into a post they made in August, in order to claim that they weren't asking for preorders in August. I think this is misleading at best. I have found this type of behavior to be typical of Damir Kalinkin, ASICSPACE's CEO. Be cautious.


I thought this would get ugly. I just find it difficult to watch those guys cruising for easy marks here to pump and dump. QuiveringGibbage created this thread for people to describe their experiences with other hosting providers, and I felt that I owed it to everyone else to state my experience with ASICSPACE as a customer, even though I knew it would result in retaliatory mudslinging.

I don't want to get drawn into a long flamewar here, because I have way too much to do as it is, so I'll stop replying to this topic soon. This is probably my last post on the topic of ASICSPACE [Edit: except for the edits above], unless someone other than ASICSPACE asks me a question about my relationship with them.

As some serious accusations have been leveled against me as a hosting provider, I do feel that I should explain a little more about our relationship, and make some comments about my experience with ASICSPACE as a former business partner. A few points:


I "posed" as an investor? That's an interesting take. Not accurate, though. I approached them as a customer. Once it became apparent to me that they were not capable of competently hosting my machines, I decided to do it myself.

I did not initally intend to start a hosting company. I had another successful business making these:

http://brain-trainer.com/product/heg-starter-package/
https://www.upgradedself.com/products/upgraded-focus-brain-trainer

I was in China at the time, working on an ultrasonic echolocation project (https://www.facebook.com/video.php?v=10103162599232593 -- please do not friend me, as I use facebook for actual friends). I was also mining bitcoin on the side, and was looking for a hosting provider for my mining operation. ASICSPACE (then MyRigSpace) seemed to be it. I approached them as a customer. They said they didn't have enough capacity yet for the machines I had coming (7 Black Arrow X-3s, a Cointerra, an SP10, and a July SP30) in their Hillsboro warehouse, but that they were getting a site in Moses Lake. This was April. They thought they would have it in about a week. Their prices seemed good, and Robert seemed pretty smart, so I thought I would go with it. Later, my brother and I doubled down, and we decided to buy 37 SP30s from the August group buy. Later still, we ordered another 30 SP30s from the September batch.

Robert and I had some nice conversations via Skype. We discussed different ways of building a datacenter. I had a lot of ideas about ways in which they could save money, and I shared them with Robert freely. Eventually, Robert dropped this on me:


[5/5/14, 1:53:31 PM] Robert Van Kirk: this is what i want to offer you:
[5/5/14, 1:54:16 PM] Robert Van Kirk: i will do my best to learn as much about electrial work, but what would really calm my nerves is this:
[5/5/14, 1:54:42 PM] Robert Van Kirk: first i want to offer you:
[5/5/14, 1:54:52 PM] Robert Van Kirk: full ownership over your racks for their lifetime
[5/5/14, 1:55:08 PM] Robert Van Kirk: and no fees
[5/5/14, 1:55:18 PM] Robert Van Kirk: we'll just pass our basic costs onto you
[5/5/14, 1:55:37 PM] Robert Van Kirk: so those racks will be useful for a long time after your equipment is no longer profitable
[5/5/14, 1:55:51 PM] Robert Van Kirk: they may be worth up to 100k a year
[5/5/14, 1:56:07 PM] Robert Van Kirk: maybe 60k
[5/5/14, 1:56:24 PM] Robert Van Kirk: we can also offer you a seat on our board of directors
[5/5/14, 1:56:52 PM] Robert Van Kirk: in return, we want you to 1) finance the cost of constructing your racks
[5/5/14, 1:56:57 PM] Robert Van Kirk: including any overruns
[5/5/14, 1:57:32 PM] Robert Van Kirk: and especially 2) i want you to work with us, to teach us how to install the electrical equipement.
[5/5/14, 1:57:40 PM] Robert Van Kirk: if we can learn how to do that ourselves, that's incredibly valuble to us
[5/5/14, 1:57:53 PM] Robert Van Kirk: as we can then start building on a mass scale
[5/5/14, 1:58:24 PM] Jonathan Toomim: full ownership of our racks, but what about the electrical and cooling equipment?
[5/5/14, 1:58:44 PM] Robert Van Kirk: yes, everything that is required to create your racks
[5/5/14, 1:59:58 PM] Jonathan Toomim: man, i need a clone
[5/5/14, 2:00:10 PM] Robert Van Kirk: also, we may want a small stipend for our salaries, so we can afford pizza until we've ramped up production
[5/5/14, 2:01:49 PM] Jonathan Toomim: i think we'd also want partial ownership or a share of future profits, especially if we're paying you guys a stipend


We eventually settled on 30% ownership of ASICSPACE for helping them build an initial 400 kW, of which we would have exclusive at-cost access to 200 kW. I would fly in from China and help them with the build, and we'd also pay for setting up the 200 kW (since we would have to do that anyway if Mike and I wanted to make use of our SP30s). Later, they decided they didn't want the stipends, and wanted to trade those for 5%. I thought that wasn't a fair exchange, but agreed nonetheless.

The essence of the agreement was that they would provide premises on which to build the DC and 3 people for the summer. My brother and I would provide capital and guidance. The goal was to do most of the construction during June, so that we would have 1 month of buffer before the SP30s arrived. They thought they would have the lease signed by the middle of May at the latest. I was supposed to fly over to the USA from Shenzhen around May 15th, and come up to WA around June 1st and help them. I got to work on designing the datacenter, sourcing components, and ordering the ones with long lead times.

When late May came around, they still didn't have a space. I got a bit worried, so I flew up to join them on a scouting expedition. We checked out the places that they had been looking at so far, and each one turned out to be unsuitable for some reason. I soon had to return to California to take care of some business with my other company, expecting them to keep looking. They took a week off from the search. I kept working on the datacenter design and sourcing throughout. When they got back and started informing me of their search, I got the impression that they were again looking at the wrong stuff, and negotiating poorly with the landlords. I flew back up to help again.

We ran into several problems on that trip. The hood of Damir's car was broken, and on his way up, it flipped up in the wind and smashed his windshield. He continued to drive it for several days until the police noticed. They detained him for a day for driving such a dangerous vehicle. While he was detained, my miners (SP10 and CT) in their Hillsboro datacenter stopped hashing. Upon inspection, it became apparent that the issue with the Hillsboro warehouse was that Damir had neglected to pay the internet bill. Also, they couldn't pay the bill any longer because they were out of money in their joint account. When I heard this, I handed Robert $500 cash to put in their account so they could pay the bill. However, they weren't able to pay the bill because Damir was in police custody at the time. It wasn't until several days later that they were able to get my (and their other customers') miners running again.

By this time it was mid- to late-June, near the end of the period I had originally scheduled for building. I had already ordered tens of thousands of dollars worth of equipment for our build (all my money, none of theirs), but I had to tell our vendors to hold onto it all because we had nowhere to ship our equipment to. Also, one of the three people they had promised would help us for the summer only showed up for three days.

While I was a customer of MyRigSpace/ASICSPACE, my machines experienced roughly 85% uptime. When an issue cropped up needing a miner reboot, or clearing a circuit breaker that tripped due to an overloaded circuit, or resetting a switch that overheated in the direct sunlight, it usually took several days for them to get in and fix the issue.

After many episodes like these, it became clear to me that not only could I not rely on them to actually find the location as per our agreement, I also didn't trust them to take care of our miners once they arrived.

I reexamined the deal that I had made with Robert and Damir. Basically, what I was doing was helping them find a location (which they had promised to find entirely on their own), providing labor (I was going to have to hire people to replace the worker they had lost, plus many others to be able to have the DC ready by the time our X-3s and SP30s arrived), providing 100% of the capital, and providing nearly all of the buildout planning. In exchange, my brother and I would get 25% ownership of their company, plus 50% of the profits for any capacity above 400 kW that we built.

On Thursday, the 26th of June, I told Robert that I was thinking of "jumping ship." I told him that I no longer had faith that they could deliver the premises that they had planned, and I told him that the informal agreement we had made had turned out be extremely unbalanced and untenable. I told him that even though we had never actually signed anything, I felt bound to the agreement; however, as they had failed to deliver on essentially all of their obligations under our agreement in a timely manner, I told him that I considered that agreement to be null and void. I told him that I might be willing to continue collaboration if we could renegotiate the terms of the deal to be less ludicrous. Over the next day, we came close to an agreement between Robert and I. We hadn't discussed it with Damir yet, because he was still AWOL after having been detained by police.

On Friday, the 27th of June, I drove Robert with me to check out another warehouse that we had seen listed for lease earlier. It turned out this warehouse was entirely suitable. We talked over a price and terms with the landlords, and we came to an agreement. Everything was looking up. On the ride back, Robert and I had essentially reconciled. When we got back to the house, Damir was there. We first told him that we had finally found our warehouse. I then told Damir that over the last few days, I had been thinking of jumping ship, and that I might be willing to collaborate with them if we renegotiated the deal to be at least 50/50%, if not 75%/25% in my favor.

Damir's response was that if I thought I would be better off without them, that I should leave, that there was nothing in our agreement that obligated me to keep the deal (Damir, why did you even say that?), and with the general sentiment of anger that I would think that my contribution was worth 3x their contribution. After listening to Damir yell at me for about 3 minutes, and hearing him repeat the words "you should leave" several times in several different grammatical constructions, I decided to follow his advice. A person whose idea of hardball negotiations is to tell his business partner to leave so that he can get his former business partner's share wasn't the kind of person I wanted to work with anyway.

Damir and Robert then called the landlord, and made an appointment to meet with them at 7:00am on Saturday morning. After talking with them, they had the impression that they had gotten the building. In order to do so, they offered the landlord the price we had agreed upon the night before plus $5 per month per kW that they used, as a balance against the "amortized value of the switchgear and transformer" in the building, or something like that. The landlords said that this seemed like a really sweet deal, and could be a lot of money.

I met with the landlords at 4 pm the same day. I explained that I was no longer working with Robert and Damir of ASICSPACE. I walked away with the lease. I offered them less money than Robert and Damir did. I didn't even offer them any more money than we had agreed upon the night before. The reason that they chose to go with me instead of them was pretty simple: I convinced the landlords that I knew enough to successfully build a datacenter, and that I would actually be able to pay rent.

Let me show you guys the math. The lease on the section of the building we both wanted to use was $2700 per month. Robert and Damir only had about 20 kW worth of customers at the time, and the building only had about 30 kW of immediately usable capacity. If they sold all of that at $100/kW, and if electricity and everything else were free, that would be at most $3000 of revenue, leaving the two of them with $300 per month for food. Their biggest investor other than my brother and I had given them about $5000. They were down to a few hundred dollars in their bank accounts. Damir has wealthy relatives, but they had viewed Damir and Robert's inability to secure a lease as a sign that they weren't going to succeed with their venture, and had refused to invest. Essentially, the gambit that Robert and Damir were making was to sign a lease that they would default on immediately in the hopes that the fact that they had a building would either allow them to get money from other investors (Ponzi scheme anyone?) or convince me to return to their team.

Anyway, I got the lease, and they didn't, and then they defaulted on the $950/month rent at the warehouse that they already had, etc. etc.


Robert and Damir refused the stipend we offered them at the beginning in order to retain another 5% ownership. After we separated, they started demanding some sort of payment for their time. Sorry guys, that's not how it works. You don't get to negotiate something away at the beginning and then claim you still deserve it at the end.


I didn't have any interest in it when we started. I was left with little choice but to build the DC myself, since I had 230 kW worth of miners already ordered and arriving in around 1-2 months.


[5/15/14, 9:40:21 AM] Robert Van Kirk: one way you could help us
[5/15/14, 9:40:36 AM] Robert Van Kirk: is we have $2k tied up in the security deposit on the location.
[5/15/14, 9:41:05 AM] Robert Van Kirk: we really could use that cash to help us ship some of the gpu miners back to customers.
[5/15/14, 9:41:32 AM] Robert Van Kirk: or even just part of it.

We gave them that money. We got 1/4 of it back.

Anyway, I'm curious to hear whether ASICSPACE will deny that they defaulted on their lease, or that they allowed their internet bill to go unpaid for several months, or that they took hashrate from their customers for themselves, or that they were unable to return property to their customers in a timely fashion. Those claims are verifiable or falsifiable. They're factual.

I used to be a business partner of ASICSPACE. I no longer am. I also used to be a customer of ASICSPACE. I no longer am. My comments below are primarily as a former customer, not as a former partner or current competitor.

Less than one month ago, ASICSPACE was in default on their warehouse in Hillsboro. They had two of my miners (an SP10 and a Cointerra) in their warehouse. (Oh, and they also owed me and my brother $2000 that we loaned them in May against their security deposit, plus another $300 I loaned them in June when they were unable to pay their internet bill.) I had asked them to return their miners to me on August 1st, but they couldn't do this. They hadn't been paying their bills, and their landlord had locked them out of their warehouse as a result. This meant that my equipment was locked in their warehouse, and I could not access it or get it sent back to me. I also could not get my SP10 rebooted (as it had been misconfigured by one of their agents the last time they had had access to the warehouse, but that's a different story), and they were unable to reboot it for about a week, if memory serves.

How did they deal with this situation? They did two things: They reconfigured my miners' pool settings without my permission or knowledge to mine into their ghash.io account. They then asked me for $999.05 beyond what I was contractually obligated to pay them for the hosting service, so they could pay their bills. The implication was that I would not be able to regain possession of my miners unless I did. I refused to do pay them this. Their next move was to tell me that they couldn't ship me my miners on August 1st as previously agreed, and charged me $465 for rent for the first half of August. They also dropped some of the other charges they were asking for, and in total asked me for $1178.51. I eventually acquiesced when their asking price dropped to around $849, or rent and electricity plus $300. (The $300 was to compensate for two errors they had been making for several months when estimating their electricity costs.) We agreed that they would debit this from the debt they owed me. They still haven't paid me the remaining $1451.

If you want to hear their landlord's take on this story, you can email rich at techspeed.com. Ask about Robert and Damir.

When it comes to this preorder scheme of theirs, my suggestion is this: Please take them up on their offer so that they can pay me back the money they owe me.

Edit 4/22/2015: Added two sentences about their bills.

No. For one thing, our cooling system would probably blow them away. Literally.


Probably, as long as it meets our technical requirements and you pay our rates. What kind of franchise are you thinking of? MegaBigPower? I don't think their hardware meets our technical requirements.

Electric hand drills with screwdriver bits are awesome.

If the area costs are twice as much for both 14 and 16 nm vs 28 nm, and the transistor area is just the feature size squared, then the transistor costs are:

2 * (16^2 / 28^2) = 65% for 16 nm vs 28 nm
2 * (16^2 / 28^2) = 50% for 14 nm vs 28 nm

This is an oversimplification in many ways, but maybe it will get you an idea.

The use of FinFETs in the smaller process is another improvement in performance which isn't captured by the simple XX nm metric.

Hosting your machine with me would be cheaper in capital costs and also electricity/recurring costs if your rates are above about $0.10.

A 20 amp 220V circuit would be fine. A 15 amp 240V circuit is right on the edge of possible, and not recommended.

Reserved.