[ANN] Spondoolies-Tech - carrier grade, data center ready mining rigs

[crazyates]

I'm not digging through dozens of pages of political bickering. I'm looking for specs.

I know it will be ~4.5TH/s @ ~3000W in a 2U form factor. How loud will the SP30 be?

[1714044470]

1714044470

[1714044470]

1714044470

[1714044470]

1714044470

[Grix]

I'm not digging through dozens of pages of political bickering. I'm looking for specs.

I know it will be ~4.5TH/s @ ~3000W in a 2U form factor. How loud will the SP30 be?

Like 5 dB or so quieter than SP10 IIRC

[Biodom]

I'm not digging through dozens of pages of political bickering. I'm looking for specs.

I know it will be ~4.5TH/s @ ~3000W in a 2U form factor. How loud will the SP30 be?

Like 5 dB or so quieter than SP10 IIRC

The most relevant point (to choose home mining vs datacenter)-does it screech (a la Sp-10) or whoosh?

[Bicknellski]

Please remember we got the ASICs just about 10 days ago and working around the clock to make the FW as good and stable as possible.

Regards`

Thanks for all the hard work. Can't wait to see what magic you can shake out of the box with the FW.

[Guy Corem]

Have an august order with hosting... Write few mails about it 1-2 days ago still nothing from them...

Hi Whisper, We are doing our best to answer all requests in a timely manner, we will get to your email as well.

As Guy mentioned we are working to increase our customer service department but this process takes time.

Thank you for your patience,

Gadi

[bobsag3]

Please remember we got the ASICs just about 10 days ago and working around the clock to make the FW as good and stable as possible.

Regards`

Thanks for all the hard work. Can't wait to see what magic you can shake out of the box with the FW.

+1

[Marvell1]

Please remember we got the ASICs just about 10 days ago and working around the clock to make the FW as good and stable as possible.

Regards`

Thanks for all the hard work. Can't wait to see what magic you can shake out of the box with the FW.

+1

well at least we know what is going on now, hopefully the tuning progresses well

[zvisha]

Ok, Guy taught me how to block users and after blocking RawDog, s1gs3gv and Zig I am back at the forum
Please share your performance statistics etc`, and if your miner provides less then 4.4TH let me know so I can check what's wrong.

I will say again that the miner will restart mining several times because it checks for PSU limit over the first day. We have 2 PSU so when 1 of them fails, the system remembers it's power limitation and restarts it. We start trying to pull ~1360 watt, but usually in the end of day it will settle down on 1320 watt. Note that the miner will change it's own "max PSU" settings.

Just thinking about it:
- the miner works few watt away from PSU failure
- it works 110% PSU capacity 100% of the time
- it works 5 mhz away from ASIC over-clocking failure
- 70% of normal ASIC voltage
- 24 hours a day 7 days a week
- at 99.9% ASIC load all the time (0.01% idle time when ASIC changing frequency and when new block arrives)
- at temperature of ~75c

This is the craziest business...

[jtoomim]

Thanks.
I wanted to read the experiences of  people in the morning with the new SP30s, and I read so much bullshit to get to 1 or 2 interesting  posts.

Anyway, there was a bug with 110V that I fixed in 2.2.23 - upgrade if you have 110V. It should provide almost 4.00Th now in 110V, I will try to increase it a bit next week.

I wrote long time ago ago when someone asked that 110V will take 7% to 20% of your hashing. This is what we see - ~12%.
https://bitcointalk.org/index.php?topic=525469.msg7218217#msg7218217

Zvi, I know exactly what you're talking about regarding the bullshit. It sucks. It's mostly only been a problem over the last week, basically since the announcement of underperformance. I think it will die down in a week or two, once people get used to the new ratings. In the mean time, we might want to create a separate thread for technival discussions only, and moderate it heavily to keep people on topic. Oh, and don't feed the trolls, people.

About the firmware stuff: I can confirm this. Zvi logged in to my SP30 last night via a port forward I set up for him, diagnosed the issue, and compiled and installed a new version of the FW before I woke up. I've done some temp measurements on the 122V cables, and they're only 6°C above ambient. The new firmware improves performance by fixing a bug in the DC2DC voltage selection, and fixes an issue with the PSU power limit. It hardcodes the PSU power limit pretty low, though (1080 W), which is why the performance is only about 3.98 TH/s with my machine. I am optimistic that we'll be able to bring performance up higher on 120V, possibly as much as 4.3 TH/s. I think I saw 4.3 TH/s briefly on the 2.18 firmware without the 1080 W limit installed before I ran into the bug, so I think it possible we will be able to get it again.

This particular SP30 shows in its stats log that it was hashing at a solid 4.5 TH/s for 24 hours on the 29th of July (burn in testing in Israel), so this is just a firmware+120V issue. Once I get this on 200V+, I expect I'll be able to get it back up to 4.5 TH/s or maybe even higher with some tweaking (and 254V?).

Power consumption at the wall is temporarily unavailable in the stats, but Zvi said it would return in one of the next firmware revisions.

Thanks again to Zvi and Harel and Guy and the rest of Spondoolies for their hard work. Bugs aside, this is a gorgeous piece of engineering, and I'm excited to soon have a room full of them running at their full potential.

Code:

uptime:8535
-----TOP-BOARD-----
PSU-TOP[EMERSON1200]: 0->1084w[1084 1084 1084] (lim=1080) 0c cooling:0/0x0
LOOP[0] ON
 0: DC2DC/1/:[vlt1:675 vlt2:677(DCl:794 Tl:794 Ul:727) 58W  86A  56c] ASIC:[ 85c  675hz(BL: 675)  327 (E:193) F:0]
 1: DC2DC/1/:[vlt1:671 vlt2:677(DCl:794 Tl:794 Ul:727) 56W  82A  54c] ASIC:[ 85c  660hz(BL: 660)  302 (E:193) F:0]
 2: DC2DC/1/:[vlt1:673 vlt2:677(DCl:794 Tl:794 Ul:727) 59W  88A  50c] ASIC:[ 85c  675hz(BL: 675)  296 (E:193) F:0]
LOOP[1] ON
 3: DC2DC/1/:[vlt1:673 vlt2:680(DCl:794 Tl:794 Ul:727) 60W  89A  57c] ASIC:[ 85c  680hz(BL: 680)  305 (E:193) F:0]
 4: DC2DC/1/:[vlt1:675 vlt2:680(DCl:794 Tl:794 Ul:727) 55W  81A  60c] ASIC:[ 85c  660hz(BL: 660)  293 (E:193) F:0]
 5: DC2DC/1/:[vlt1:671 vlt2:677(DCl:794 Tl:794 Ul:727) 59W  87A  63c] ASIC:[ 85c  690hz(BL: 690)  306 (E:193) F:0]
LOOP[2] ON
 6: DC2DC/1/:[vlt1:671 vlt2:677(DCl:794 Tl:794 Ul:727) 56W  83A  64c] ASIC:[ 85c  675hz(BL: 675)  309 (E:193) F:0]
 7: DC2DC/1/:[vlt1:671 vlt2:677(DCl:794 Tl:794 Ul:727) 53W  79A  63c] ASIC:[ 85c  620hz(BL: 620)  279 (E:193) F:0]
 8: DC2DC/1/:[vlt1:671 vlt2:677(DCl:794 Tl:794 Ul:727) 51W  76A  52c] ASIC:[ 85c  590hz(BL: 590)  250 (E:180) F:0]
LOOP[3] ON
 9: DC2DC/1/:[vlt1:675 vlt2:680(DCl:794 Tl:794 Ul:727) 66W  97A  67c] ASIC:[ 85c  725hz(BL: 725)  316 (E:193) F:0]
10: DC2DC/1/:[vlt1:671 vlt2:680(DCl:794 Tl:794 Ul:727) 66W  97A  70c] ASIC:[ 85c  730hz(BL: 730)  315 (E:193) F:0]
11: DC2DC/1/:[vlt1:673 vlt2:680(DCl:794 Tl:794 Ul:727) 62W  92A  70c] ASIC:[ 85c  735hz(BL: 735)  328 (E:193) F:0]
LOOP[4] ON
12: DC2DC/1/:[vlt1:673 vlt2:680(DCl:794 Tl:794 Ul:727) 61W  90A  73c] ASIC:[ 85c  715hz(BL: 715)  309 (E:193) F:0]
13: DC2DC/1/:[vlt1:675 vlt2:680(DCl:794 Tl:794 Ul:727) 64W  94A  74c] ASIC:[ 85c  740hz(BL: 740)  344 (E:193) F:0]
14: DC2DC/1/:[vlt1:677 vlt2:680(DCl:794 Tl:794 Ul:727) 66W  97A  62c] ASIC:[ 85c  720hz(BL: 720)  336 (E:193) F:0]
----BOTTOM BOARD------
PSU-BOT[EMERSON1200]: 0->1080w[1076 1080 1076] (lim=1080) 0c cooling:0/0x0
LOOP[5] ON
15: DC2DC/1/:[vlt1:666 vlt2:672(DCl:794 Tl:794 Ul:727) 51W  76A  51c] ASIC:[ 85c  635hz(BL: 635)  277 (E:193) F:0]
16: DC2DC/1/:[vlt1:667 vlt2:672(DCl:794 Tl:794 Ul:727) 50W  74A  50c] ASIC:[ 85c  625hz(BL: 625)  278 (E:193) F:0]
17: DC2DC/1/:[vlt1:667 vlt2:672(DCl:794 Tl:794 Ul:727) 58W  86A  49c] ASIC:[ 85c  675hz(BL: 675)  290 (E:193) F:0]
LOOP[6] ON
18: DC2DC/1/:[vlt1:666 vlt2:672(DCl:794 Tl:794 Ul:727) 56W  84A  57c] ASIC:[ 85c  675hz(BL: 675)  311 (E:193) F:0]
19: DC2DC/1/:[vlt1:666 vlt2:672(DCl:794 Tl:794 Ul:727) 57W  85A  59c] ASIC:[ 85c  690hz(BL: 690)  338 (E:193) F:0]
20: DC2DC/1/:[vlt1:669 vlt2:672(DCl:794 Tl:794 Ul:727) 52W  78A  58c] ASIC:[ 85c  640hz(BL: 640)  293 (E:193) F:0]
LOOP[7] ON
21: DC2DC/1/:[vlt1:667 vlt2:672(DCl:794 Tl:794 Ul:727) 53W  78A  65c] ASIC:[ 85c  660hz(BL: 660)  317 (E:193) F:0]
22: DC2DC/1/:[vlt1:667 vlt2:672(DCl:794 Tl:794 Ul:727) 57W  85A  67c] ASIC:[ 85c  695hz(BL: 695)  298 (E:193) F:0]
23: DC2DC/1/:[vlt1:666 vlt2:672(DCl:794 Tl:794 Ul:727) 56W  84A  56c] ASIC:[ 85c  685hz(BL: 685)  314 (E:192) F:0]
LOOP[8] ON
24: DC2DC/1/:[vlt1:667 vlt2:672(DCl:794 Tl:794 Ul:727) 44W  66A  57c] ASIC:[ 85c  545hz(BL: 545)  230 (E:191) F:0]
25: DC2DC/1/:[vlt1:669 vlt2:674(DCl:794 Tl:794 Ul:727) 62W  92A  70c] ASIC:[ 85c  730hz(BL: 730)  329 (E:193) F:0]
26: DC2DC/1/:[vlt1:671 vlt2:674(DCl:794 Tl:794 Ul:727) 61W  91A  72c] ASIC:[ 85c  740hz(BL: 740)  342 (E:193) F:0]
LOOP[9] ON
27: DC2DC/1/:[vlt1:669 vlt2:674(DCl:794 Tl:794 Ul:727) 62W  92A  77c] ASIC:[ 85c  750hz(BL: 750)  382 (E:193) F:0]
28: DC2DC/1/:[vlt1:671 vlt2:674(DCl:794 Tl:794 Ul:727) 60W  90A  78c] ASIC:[ 85c  740hz(BL: 740)  340 (E:193) F:0]
29: DC2DC/1/:[vlt1:671 vlt2:674(DCl:794 Tl:794 Ul:727) 62W  92A  64c] ASIC:[ 85c  750hz(BL: 750)  343 (E:193) F:0]

[H:HW:3961Gh[1985+1975],W:1746,L:0,A:30,MMtmp:0 TMP:(30)=>=>=>(67,67)]
Pushed 30 jobs , in HW queue 4 jobs (sw:3, hw:3)!
min:55 wins:9297[this/last min:38/64] bist-fail:1060, hw-err:0
leading-zeroes:42 idle promils[s/m]:0/0, rate:4473gh/s
Fan:90, conseq:200
I2C BUSY with BUG: 1
R/NR: 7855/0
Adapter queues: rsp=2, req=20

[RawDog]

Zvi, I know exactly what you're talking about regarding the bullshit. It sucks. It's mostly only been a problem over the last week, basically since the announcement of underperformance.

Yeah, people get pissed when you try to fuck them in the ass after they send you money.  Imagine that.

You can't tell me SPD didn't expect a serious backlash in view of all they said about certaintey and how serious the shortcoming was.  I am just trying to help give them the grief they know they deserve.  It will motivate them to speak more humbly in the future.  And possibly to avoid major fuck ups which cost their customers thousands of euros.

I am pretty fucking sure Guy knows this anti-PR is well deserved.  As soon as he sends me a personal note to that effect, I'll stop. 

[jtoomim]

You must be planning to install a wind tunnel, lol!
And that 200,000 btu's would be the heat that the units are generating - that many units would create enough heat to get a very large room above 40C in only a few minutes.  I question if you would even be able to put 21 of them into a 42U rack without literally cooking the units (unless you're literally in a 100% hot aisle/ cold aisle contained facility that has a huge pressure differential)
What kind of cooling system are you putting in place to dissipate that kind of volume of heat?

A picture of me with some of my biggest fans:



We bought 10 of these. Humans in this picture may be larger than they appear.

[1369]

So let me understand, when will the August Batch start delivering?

By the looks of it the pictures (at least the last one) were taken 2014-07-29 around 4:20PM with a Nikon D3000 camera  (EXIF says).
Since most of the units there waiting for QA (by the paper tags) have rack ears, I'm guessing that what we see in the pictures are the units going to the data centers for hosting or were installed in the local datacenter as the July batch.

RoadStress said earlier that the group buy units are expected during the second part of August. Knowing DHL/UPS guys .. I'm expecting something in the week starting Aug 25 .. Thu or Fri delivery.

Your question still stands though.

just wanna make clear that anyone can download that photo, right click on the file, choose properties, then details & make the "date taken" whatever they want.

[Biodom]

This particular SP30 shows in its stats log that it was hashing at a solid 4.5 TH/s for 24 hours on the 29th of July (burn in testing in Israel), so this is just a firmware+120V issue. Once I get this on 200V+, I expect I'll be able to get it back up to 4.5 TH/s or maybe even higher with some tweaking (and 254V?).

Does it make a loud screeching sound as Sp-10 does or is it making a whooshing sound? I understand that it is ~65-67dB at 1m, correct?

[jtoomim]

Does it make a loud screeching sound as Sp-10 does or is it making a whooshing sound? I understand that it is ~65-67dB at 1m, correct?

I don't have a dB meter handy. As best as I can tell (which isn't very good), it's about the same as an SP10 in both quality and quantity. I don't have my SP10 here to compare against, though. If you want to hear it, you could Skype me (jtoomim) or Facetime me (jtoomim@jtoomim.org).

[Biodom]

Does it make a loud screeching sound as Sp-10 does or is it making a whooshing sound? I understand that it is ~65-67dB at 1m, correct?

I don't have a dB meter handy.

thanks for invite
i got this
https://itunes.apple.com/us/app/decibel-meter-pro/id382776256?mt=8
but here is the android equivalent
https://play.google.com/store/apps/details?id=kr.sira.sound&hl=en

[TracerX]

thanks for invite
i got this
https://itunes.apple.com/us/app/decibel-meter-pro/id382776256?mt=8
but here is android equivalent
https://play.google.com/store/apps/details?id=kr.sira.sound&hl=en

Cheers for the link, I had not considered using my iPhone as a meter.

[s1gs3gv]

Ok, Guy taught me how to block users and after blocking RawDog, s1gs3gv and Zig I am back at the forum

So you can't stand somebody telling you the Sept. SP30 machines need to be 2.5 BTC to compete with the S3s or
you don't like somebody pointing out that you advertised these widely as 6 - 6.5 TH machines without telling anybody 'but not on 110' ?
Or maybe you just don't like people who think you are playing down all the issues with SP30 performance so that you can avoid making adequate and timely compensation to your customers ? Or perhaps you just don't like people who don't constantly stroke you.

I have a suggestion, why don't you fellas just  ban anybody who asks hard questions or makes statements that make you uncomfortable just like they do on the Alpha-T forum.

I expected better from ST … but that was when Guy was telling everybody 'don't worry, be happy, we'll meet our targets'.

Perceptions have changed Z. Its up to you to restore them. Ignoring people won't get you there.

[s1gs3gv]

They created the best miner on the planet,

There is nothing about the SP30 that makes it 'the best miner on the planet'. Who are you kidding ? The best miner on the planet is the one that produces the best ROI, not the one with the nicest case or heat sinks.

I am so glad I passed on that Farmer's SP30 deal back in April.

[Biodom]

They created the best miner on the planet,

There is nothing about the SP30 that makes it 'the best miner on the planet'. Who are you kidding ? The best miner on the planet is the one that produces the best ROI, not the one with the nicest case or heat sinks.

I am so glad I passed on that Farmer's SP30 deal back in April.

i don't know why you keep pressing your case so hard, dude.
it is what it is, we are all (or most) adults here, we deal with the current situation.
was it a bit overmarketed-perhaps, yes, but i don't blame anyone for trying and not achieving the exact number, especially if an adequate compensation is promised.
It is not the first or the last miner, it is a flow...stay with the flow.

[s1gs3gv]

They created the best miner on the planet,

There is nothing about the SP30 that makes it 'the best miner on the planet'. Who are you kidding ? The best miner on the planet is the one that produces the best ROI, not the one with the nicest case or heat sinks.

I am so glad I passed on that Farmer's SP30 deal back in April.

i don't know why you keep pressing your case so hard, dude.
it is what it is, we are all (or most) adults here, we deal with the current situation.
was it a bit overmarketed-perhaps, yes, but i don't blame anyone for trying and not achieving the exact number, especially if an adequate compensation is promised.
It is not the first or the last miner, it is a flow...stay with the flow.

Well, I'm not sure what case you think I'm pressing so hard here. Perhaps it is this:

I had some very early exchanges with Guy regarding pre-order nature of the SP30 and the  absence of a refund policy. Guy repeatedly and forcefully asserted that no refund policy was necessary because there was nothing that could go wrong and they would deliver on time and to spec.

And they didn't. And they didn't in a way that seriously affects customers expectations of ROI, which were marginal to begin with. And their compensation plan doesn't seem to address the costs associated with increased power usage as well as decreased hashing rate, and it seems like it is deferred until October in any case. A compensation plan based on a decrease in performance from 5.4 TH when Guy advertised 6 - 6.5TH is disingenuous. A compensation plan that doesn't consider customer losses due to decreased power efficiency is disingenuous. And statements that assert that substandard performance at 110 volts is not a problem because customers can always host in a datacenter instead of running their machines in a home or office environment are disingenuous.

And finally, the pricing on the Sept. SP30 is not competitive with the alternative, which is S3 delivered in a week or so.

So, I'm not sure which case I am pressing so hard here … take your pick.

ST need to directly and openly acknowledge these issues and address them fairly and promptly if they want to regain the trust of this community.