600watt
Legendary
 Offline
Activity: 2338
Merit: 2106
|
 |
June 08, 2013, 08:44:31 PM |
|
So he's really pro KNCMINER. Why the bashing? **** knows?! Bizarre. easy: when invested into knc it is bad for your roi when too many people also invest in those machines. it is not only "homo homini lupus est" but also jupiter jupitii lupus estyou bitcoinorama actually hurt my roi much more than the knc-haters.  How's that?! i was kidding. when knc gets just enough orders to get going, those who will buy those will have a decent roi if difficulty isn´t at 500 mio by september. but if knc gets run over with orders and they decide to push all those orders through 500 mio difficulty will look small. the more jupiters in the universe, the lesser the individual jupiter shines. (psst: secret plan... after you attend the open day post a big SCAM into the threads. that will rock... & shake off the weak hands...) |
|
|
|
|
yeemartin
Member
Offline
Activity: 95
Merit: 10
|
|
June 08, 2013, 09:05:00 PM
Last edit: June 08, 2013, 09:15:42 PM by yeemartin |
|
I think TSMC won't even talk to KNC if they have less than 50 million USD upfront for 28nm process.
Who can afford to use 28nm? Samsung, Intel, AMD, Qualcomm, Altera etc. All are multibillion dollar companies. You think a tiny Swedish company can afford 28nm?
It's so unrealistic that they are aiming for 28nm.
Even 40nm ACIS bitcoin miner seems so unrealistic for me at the moment, let alone 28nm.
|
|
|
|
|
|
KS
|
|
June 08, 2013, 09:15:02 PM |
|
So he's really pro KNCMINER. Why the bashing? **** knows?! Bizarre. easy: when invested into knc it is bad for your roi when too many people also invest in those machines. it is not only "homo homini lupus est" but also jupiter jupitii lupus estyou bitcoinorama actually hurt my roi much more than the knc-haters. How's that?! What could he mean? what could he mean? Wild guess: you being their best sales rep?  An coolio, I'm enthusiastic and just want to see this project happen by a group of acknowledged experts, and those considering investing to understand the pros and cons. I'm fully aware of the risks, but I take no responsibility for others decisions, and I've berated those that look like their investing without adequate research. I've said that enough times before and during opening of payment. Everyone's responsible for their own research and the fact that I've laid a few hundred down on flights there and back in a day means I'm taking mine seriously. That is not a freakin' advertisement, that's just my own bucket list of confidence criteria I have set myself. Their appears to be plenty of dump people with cash. I both; fortunately, and unfortunately fit in the opposite category. Take it as you will. Whoever meets me on Monday will be able to vouch whether I appear sincere in person. I'll be the guy with bloodshot eyes due to the minimal sleep I will have from now, until then. If you want me to upload my boarding pass with pictures from the day, then not a problem. KS, you're still bitching, yet my inbox has still has no further questions to be asked on the day sent to it. Despite numerous offers. Opportunity's there; put up, or shut up. If you won't pull your finger out to attend, I'm more than happy to ask new questions, or revised questions with what you now know, or repeat questions you want more clarity on, on your behalf. Just PM me questions by tomorrow night (UK time) and I'll print out and record the answers. *I* am bitchin'? lol see 600watt's reply, maybe he's bitchin' too? you're stressed out man, take a breather or sth. |
|
|
|
|
|
KS
|
|
June 09, 2013, 10:27:32 AM |
|
Cross posted from newbies with permission of OP (not quoted for readability, but I'll edit if this is confusing) https://bitcointalk.org/index.php?topic=227625.0OP is SwedErik https://bitcointalk.org/index.php?action=profile;u=127256Yes, please crosspost... I still have to be logged in for two more hours...  There is quite a few things I read there that I could/would like to comment on, but it's not possible now.. Found the threads about KNCminer on this forum and since I know a little about this company. Well actually the formal company "kncminer" is soo new so the one I know about is orsoc, which is one of it's two parent companies. The other parent company that is owned by two guys do I not know much about, but they does not seem to be have with anything similar in the past, but has worked in the financial sector... I created an account and will post what I know. I saw a good summary posted by someone who attended the "openday": There's just not much info to tell.
Are ORSoC part of KnCMiner? Yes they are.
Is ORSoC a legit company? Yes it is.
Is ORSoC's area of expertise desgning FPGAs, ASICs and embedded systems? Yes it is.
Is there a working FPGA prototype hashing at 6.2GH/s? Yes there is.
Is there an ASIC prototype? No there isn't.
What's the plan? To make the ASIC from the FPGA, order the chips with preorder money, and send them directly to assembly.
Are there any gerbers for the ASIC PCB's, etc. that can be showed to us? No, they say they ASIC PCBs will be just based on the FPGA PCBs.
Is this feasible? Mmmm, yeah, why not. That was BFL's plan too - right?
Is this extremely expensive, knowing they are planning to do 28nm? Yes it is. VERY expensive.
Is this project at least x10 bigger in terms of revenue streams, etc. compared to anything else ORSoC has done before? Yes it is. It's a huge project for a tiny company.
I can just confirm everything above and start with the "posstive" things. I do not know the relation between ORSoC or KnCMiner, but I can confirm that ORSoC has been around for quite some time and has made profit in the last 3-5 years. In range of $80k-$250k. They do various things but profile themselfe as fpga consultants. I honestly do not belive that they are trying to make an obvious scam (like; scam people and flee the country), but I would NEVER pay a single dollar in advance. See below for reasons; They have a history with the "openrisc" cpu that they have been promoting for many years. People who are paying them for a-yet-to-be-produced asic should know that this is not the first time they are collecting money in advance for an asic design. Previous try was with a promise to produce an openrisc cpu (at a MUCH less advance process than .28), they gathered some money from the community (the "campaign" may still be active, I do not know, google for it!). However, when the project did not take off, I think most of the money just went into their company...  As far as I know there was nu public records for the accounting, allthought it was "pitched" as a "community project". Among other local companies within the same sector, they are quite well known for being "close to impossible to work with" and quite a few companies has bad relations with them. Ask around!!! My personal belief is some of the people have little touch with reality and tends to be better on producing visions rather than code/hw! That's why I said above that I do not thing they are doing an "obvious scam", but I fear that the result may be the same. Among the five people listed I think would say that four of them are 100% "non-technical". I guess others are also working in producing working code. To my knowledge no one has ever been involved in asic production before (I could be wrong on this) and they are now about to go directly to a 28nm tape out... . Yeah right! But since there are quite a few "non-technical" people on the project (at least relativly) I'm not suprised that they do a good job in marketing and come out with new ideas about lotto and other stuff instead of actually showing working stuff. Or is it 100% confirmed that the mars prototype is actually is working a specified speed??! If soo, why is it not up for sale? Some other thoughts and things I would think is an absolute must before even going public with something like this: * I read about canceling mars.... well, if you can not even meet deadlines or results on the fpga, why even bother moving along to an asic? It's a bit contradictive when they claim they basically will use the same pcb for asic as fpga, but they need to focus on asic. To my knowledge it more or less standard to verify that the design is working in fpga (at much lower clock/density) before moving to asic. Why not release this then if it's so magical... Even if the fpga is working PERFECTLY it's a LONG way before doing a tape out (escpecially at 28nm)... * At least a few people with experince of asic production. * SOME budget that at least shows that there is a possibility that they can handle a production run at 28nm.... Are you guys aware of the HUGE costs involved??! My guess would be several millions of USD.... however, I have not either done anything similar so it's a guess, but should at least give SOME indication. * + Some financial backing for doing a tape out at 28nm.... not just relying on massive preorders. * + some proof that the company and owners really risk their own money, not money collected from users preorders * It's not that many FAB:s that do 28nm, is it public who they will use? If not, if this is legit, why should it be keept secret? All chip manufactures are quite open about who produce their devices. Well, these are just my thought, based on what I know and heard from others in the industry. Sure it would be cool if they could produce such a device, but I highly doubt it based on what I know and have read about kncminer. If it sounds to good to be true, it is probably to good to be true. He did not reply to my question regarding his relationship to ORSoC. I'd like a bit more background info on this character. I'm not particularly fond of KNCMINER/ORSoC but posting potentially fake testimonies is not helping to debunk scams. |
|
|
|
|
|
J35st3r (OP)
|
|
June 09, 2013, 10:43:05 AM |
|
He did not reply to my question regarding his relationship to ORSoC. I'd like a bit more background info on this character. I'm not particularly fond of KNCMINER/ORSoC but posting potentially fake testimonies is not helping to debunk scams.
I only cross -posted SwedErik's comments as he was in newbie jail and it looked relevant to this subforum. He's only 13 minutes shy of Jr Member, so maybe he'll come over here and explain. |
1Jest66T6Jw1gSVpvYpYLXR6qgnch6QYU1 NumberOfTheBeast ... go on, give it a try 
|
|
|
|
KS
|
|
June 09, 2013, 01:12:15 PM |
|
He did not reply to my question regarding his relationship to ORSoC. I'd like a bit more background info on this character. I'm not particularly fond of KNCMINER/ORSoC but posting potentially fake testimonies is not helping to debunk scams.
I only cross -posted SwedErik's comments as he was in newbie jail and it looked relevant to this subforum. He's only 13 minutes shy of Jr Member, so maybe he'll come over here and explain. Sure thing, just quoting you FYI. I also asked in newbie jail, no news. |
|
|
|
|
SwedErik
Newbie
Offline
Activity: 12
Merit: 0
|
|
June 10, 2013, 06:52:16 PM |
|
He did not reply to my question regarding his relationship to ORSoC. I'd like a bit more background info on this character. I'm not particularly fond of KNCMINER/ORSoC but posting potentially fake testimonies is not helping to debunk scams.
I only cross -posted SwedErik's comments as he was in newbie jail and it looked relevant to this subforum. He's only 13 minutes shy of Jr Member, so maybe he'll come over here and explain. Sure thing, just quoting you FYI. I also asked in newbie jail, no news. Finnaly got 4hours logged in time and below is the reply I just posted in newbie forum regarding this:
Interesting post. Did ORSoC ever do a press-release regarding the failed project and where the funding went?
As far as I know, there was press release. This was not preorders as in the knc case, it was more of "donations", however the donations went directly to orsoc and it could be that they gave away some pcb:s in the end to a few of the top donnors. Info should be out there. I just think the campaign was quite similar, lot of marketing hype and very few technical details. Also, what is your relationship to ORSoC?
I like to be anonymous, but I can just say that I work in the same field as they and has known about the company and persons from some time, but as I mentioned, they are quite well known for their behaviour in the local industry. According to a very recent local news article, they claim to have received enough money from existing orders to start production....... I have NOT done any .28nm tape outs, but just to get an idea of costs, have a look at page 4 on the following pdf: http://www.arena-international.com/Journals/2012/03/16/s/i/l/3_Xilinx.pdf...sure, there is many ways to reduce the costs and this will be a very simple design, but I would expect them to need AT LEAST _a few_ million USD:s to do this. This makes me very sceptical! What is the harm in being open about these numbers? Anyway, I may soon be out of newbie status and can then post a little in the other forums.... |
|
|
|
|
|
kendog77
|
|
June 10, 2013, 06:56:28 PM
Last edit: June 10, 2013, 07:39:25 PM by kendog77 |
|
High risk, high rewards.
Let me correct this quote. High risk equals low chance of high rewards, high chance of losing your entire investment. |
|
|
|
|
|
blastbob
|
|
June 10, 2013, 07:06:35 PM |
|
Nice PDF, gives a good image of costs. Only problem is that it is a year old. Let me know when you got some better data to work with. And add a in-depth analysis instead of a hunch from the tooth fairy.
|
Bitrated user: blastbob.
|
|
|
SwedErik
Newbie
Offline
Activity: 12
Merit: 0
|
|
June 10, 2013, 07:11:09 PM |
|
Nice PDF, gives a good image of costs. Only problem is that it is a year old. Let me know when you got some better data to work with. And some more in-depth analysis instead of a hunch from the tooth fairy.
Unfortunally there is very little public info/pricing availiable for such advanced processes as 28nm. Yes, it's a year old and most likely cost for producing a dedicated BTC mining chip. But if you look at the pdf and the previous gen. tech you should get an IDEA about current costs. Even if they will be quite a lot lower it still sounds unrealistic to me... and yet they claim that enough money has allready been received to start production... |
|
|
|
|
|
DPoS
|
|
June 10, 2013, 07:17:19 PM |
|
isn't that two years old?
2012 was the transition year and 2013 has been the high output year for 28nm. Coming in late summer should be a good spot for kncminer to get everything since the backlogs of 2012 should be cleared
costs have a lot of variables.. not sure you'll get straight answers on that
|
|
|
|
|
KS
|
|
June 10, 2013, 07:45:16 PM |
|
He did not reply to my question regarding his relationship to ORSoC. I'd like a bit more background info on this character. I'm not particularly fond of KNCMINER/ORSoC but posting potentially fake testimonies is not helping to debunk scams.
I only cross -posted SwedErik's comments as he was in newbie jail and it looked relevant to this subforum. He's only 13 minutes shy of Jr Member, so maybe he'll come over here and explain. Sure thing, just quoting you FYI. I also asked in newbie jail, no news. Finnaly got 4hours logged in time and below is the reply I just posted in newbie forum regarding this:
Interesting post. Did ORSoC ever do a press-release regarding the failed project and where the funding went?
As far as I know, there was press release. This was not preorders as in the knc case, it was more of "donations", however the donations went directly to orsoc and it could be that they gave away some pcb:s in the end to a few of the top donnors. Info should be out there. I just think the campaign was quite similar, lot of marketing hype and very few technical details. Also, what is your relationship to ORSoC?
I like to be anonymous, but I can just say that I work in the same field as they and has known about the company and persons from some time, but as I mentioned, they are quite well known for their behaviour in the local industry. According to a very recent local news article, they claim to have received enough money from existing orders to start production....... I have NOT done any .28nm tape outs, but just to get an idea of costs, have a look at page 4 on the following pdf: http://www.arena-international.com/Journals/2012/03/16/s/i/l/3_Xilinx.pdf...sure, there is many ways to reduce the costs and this will be a very simple design, but I would expect them to need AT LEAST _a few_ million USD:s to do this. This makes me very sceptical! What is the harm in being open about these numbers? Anyway, I may soon be out of newbie status and can then post a little in the other forums.... How about de-anonimyzing yourself to a trusted third party? (hence not publicly) I'm not sure how the pdf relates to the BTC ASICs, the prices are way higher than anyone could possibly fund. We're talking an order or two of magnitude. Also, I would think they're working with Altera and not Xilinx. The ORSoC boards harbor the same Cyclone IV FPGAs (22K for OpenRisc, 115K for the Mars) and the OpenRisc core was developed on Quintus II (Altera), so I would think they banged the code on the Cyclone and used Hardcopy V (Altera) as they said the ASIC "maker" would take the FPGA code and make it into an ASIC, without the intervention of ORSoC. Looks like the Altera process to me (could be eASIC but they aren't talking about their 28nm on the website yet). Not sure about project cost but they obviously wanted at least 1.5M USD to start the Mars, so I would guess they really needed 0.5M (= profit margin - I'm not really sure what kind of price they could get for only 2500 FPGAs and sell the Mars for 2800$ with 48 FPGAs in it, not even counting PCB, assembly, shipping, NRE, other overhead). |
|
|
|
|
Korbman
Legendary
Offline
Activity: 1064
Merit: 1001
|
|
June 10, 2013, 08:06:59 PM |
|
Nice PDF, gives a good image of costs. Only problem is that it is a year old. Let me know when you got some better data to work with. And some more in-depth analysis instead of a hunch from the tooth fairy.
Unfortunally there is very little public info/pricing availiable for such advanced processes as 28nm. Yes, it's a year old and most likely cost for producing a dedicated BTC mining chip. But if you look at the pdf and the previous gen. tech you should get an IDEA about current costs. Even if they will be quite a lot lower it still sounds unrealistic to me... and yet they claim that enough money has allready been received to start production... I'm not sure I see it as unrealistic as others, even after assuming the PDF is accurate to this day. ASICMiner built their chips at 130nm for under $200k; Avalon is at 110nm and they've been able to do just fine on a couple thousand preorders ($1m+ overall value); BFL built on 65nm with preorders valuing $10m+. Naturally as the architecture shrunk, the cost to produce the chips increased (and obviously it depends on how many chips you're looking to produce in the first place). If KnC can maintain a high preorder count (at a minimum $3,000 each), I don't see why they couldn't begin to afford 28nm...and maybe an outside investor could also play a part in subsidizing the cost. |
|
|
|
|
J35st3r (OP)
|
|
June 10, 2013, 08:08:45 PM |
|
I'm not sure how the pdf relates to the BTC ASICs, the prices are way higher than anyone could possibly fund. We're talking an order or two of magnitude. Also, I would think they're working with Altera and not Xilinx.
The ORSoC boards harbor the same Cyclone IV FPGAs (22K for OpenRisc, 115K for the Mars) and the OpenRisc core was developed on Quintus II (Altera), so I would think they banged the code on the Cyclone and used Hardcopy V (Altera) as they said the ASIC "maker" would take the FPGA code and make it into an ASIC, without the intervention of ORSoC. Looks like the Altera process to me (could be eASIC but they aren't talking about their 28nm on the website yet). Not sure about project cost but they obviously wanted at least 1.5M USD to start the Mars, so I would guess they really needed 0.5M (= profit margin - I'm not really sure what kind of price they could get for only 2500 FPGAs and sell the Mars for 2800$ with 48 FPGAs in it, not even counting PCB, assembly, shipping, NRE, other overhead).
(Niggle: Quartus II not Quintus II) So they have gone down the "Structured ASIC" route. It won't give the performance of a full custom (more expensive per chip), but mask costs will be much cheaper since its just the metallization/vias (hence its cheap enough to be relaistic at 28nm process). Honestly I'm surprised no one else went this route. Still its a tradeoff between up-front cost and per-unit cost. |
1Jest66T6Jw1gSVpvYpYLXR6qgnch6QYU1 NumberOfTheBeast ... go on, give it a try
|
|
|
|
hak8or
|
|
June 10, 2013, 08:42:16 PM |
|
Oh wow, I did not realize that there was such discussion here on bitcointalk! If interested, I made a post a few hours ago on reddit about the technicalities kncminer has to deal with, as well as some explanations at what some of the wording means: http://www.reddit.com/r/Bitcoin/comments/1g0yvf/feathercoin_going_through_a_real_51_attack_people/cafpzm9(Niggle: Quartus II not Quintus II)
So they have gone down the "Structured ASIC" route. It won't give the performance of a full custom (more expensive per chip), but mask costs will be much cheaper since its just the metallization/vias (hence its cheap enough to be relaistic at 28nm process). Honestly I'm surprised no one else went this route. Still its a tradeoff between up-front cost and per-unit cost.
For some reason I myself have not come to such a realization, though I had no idea they were using Altera FPGA's. But keep in mind that if they were doing just the FPGA conversion using the HardCopy process from Altera, they wouldn't really need ORSoC, as the conversion is not that complicated. Heck, they even have the correct feature size on their website, "HardCopy V". http://www.altera.com/devices/asic/hardcopy-asics/about/hrd-index.htmlIf this is true, that I have to admit that I am somewhat disappointed since this is not a full blown asic but instead just an FPGA conversion, so we wouldn't be getting anywhere near the performance that could have been possible. But this does satisfy the issue with them having enough money to do a 28nm full custom asic as doing such an FPGA to asic conversion is far far cheaper. |
|
|
|
|
|
KS
|
|
June 10, 2013, 09:24:26 PM |
|
I'm not sure how the pdf relates to the BTC ASICs, the prices are way higher than anyone could possibly fund. We're talking an order or two of magnitude. Also, I would think they're working with Altera and not Xilinx.
The ORSoC boards harbor the same Cyclone IV FPGAs (22K for OpenRisc, 115K for the Mars) and the OpenRisc core was developed on Quintus II (Altera), so I would think they banged the code on the Cyclone and used Hardcopy V (Altera) as they said the ASIC "maker" would take the FPGA code and make it into an ASIC, without the intervention of ORSoC. Looks like the Altera process to me (could be eASIC but they aren't talking about their 28nm on the website yet). Not sure about project cost but they obviously wanted at least 1.5M USD to start the Mars, so I would guess they really needed 0.5M (= profit margin - I'm not really sure what kind of price they could get for only 2500 FPGAs and sell the Mars for 2800$ with 48 FPGAs in it, not even counting PCB, assembly, shipping, NRE, other overhead).
So they have gone down the "Structured ASIC" route. It won't give the performance of a full custom (more expensive per chip), but mask costs will be much cheaper since its just the metallization/vias (hence its cheap enough to be relaistic at 28nm process). Honestly I'm surprised no one else went this route. Still its a tradeoff between up-front cost and per-unit cost. I can't say for sure but Altera FPGAs seem to be on every ORSoC design and the HardCopy process flowchart is consistent with what I gleaned in the "conversations". I could be totally wrong of course, but it looks to fit nicely. (28nm, FPGA, Quartus II, letting the ASIC maker do the ASIC from FPGA). (Niggle: Quartus II not Quintus II)
Yeah, brainfart with all the Roman numerals...  Quartus II should really be called Decimus or Quintus...  |
|
|
|
|
|
KS
|
|
June 10, 2013, 09:35:39 PM |
|
Oh wow, I did not realize that there was such discussion here on bitcointalk! If interested, I made a post a few hours ago on reddit about the technicalities kncminer has to deal with, as well as some explanations at what some of the wording means: http://www.reddit.com/r/Bitcoin/comments/1g0yvf/feathercoin_going_through_a_real_51_attack_people/cafpzm9(Niggle: Quartus II not Quintus II)
So they have gone down the "Structured ASIC" route. It won't give the performance of a full custom (more expensive per chip), but mask costs will be much cheaper since its just the metallization/vias (hence its cheap enough to be relaistic at 28nm process). Honestly I'm surprised no one else went this route. Still its a tradeoff between up-front cost and per-unit cost.
For some reason I myself have not come to such a realization, though I had no idea they were using Altera FPGA's. But keep in mind that if they were doing just the FPGA conversion using the HardCopy process from Altera, they wouldn't really need ORSoC, as the conversion is not that complicated. Heck, they even have the correct feature size on their website, "HardCopy V". http://www.altera.com/devices/asic/hardcopy-asics/about/hrd-index.htmlIf this is true, that I have to admit that I am somewhat disappointed since this is not a full blown asic but instead just an FPGA conversion, so we wouldn't be getting anywhere near the performance that could have been possible. But this does satisfy the issue with them having enough money to do a 28nm full custom asic as doing such an FPGA to asic conversion is far far cheaper. I think Daggeteo said (rather showed, with a ruler) in the KNC Open Day thread that the finished ASIC would be about 75x75mm! They will use 48 in the Jupiter (can't find where I get the number from, save that it's the same as in the dead Mars). The chips would need to be a bit more than 7GH/s each, with a 20W TDP (they claim the Jupiter will use less than 1000W). Given the Cyclone IV E 115K they used in the Mars was only capable of 150MH/s, I wonder how many cores they'll have on that package (and I worry about the yield, give the sheer size of the thing). |
|
|
|
|
Kuma
Member
Offline
Activity: 107
Merit: 10
|
|
June 10, 2013, 09:57:31 PM |
|
If this is true, that I have to admit that I am somewhat disappointed since this is not a full blown asic but instead just an FPGA conversion, so we wouldn't be getting anywhere near the performance that could have been possible. But this does satisfy the issue with them having enough money to do a 28nm full custom asic as doing such an FPGA to asic conversion is far far cheaper.
You are right. But if they are able to reach desired performance/consumption their chips will be one of the best ones. And after that they may be able to get enough funding for full asic. |
|
|
|
|
|
Anenome5
|
|
June 11, 2013, 04:01:09 PM |
|
They are -not- using a structured ASIC:  |
Democracy is the original 51% attack.
|
|
|
|
papamoi
|
|
June 14, 2013, 01:16:52 PM |
|
jut fyi
if u want to use a standard cell,the minimum timing to produce it is around 6 months
and this is the minimum
so i m doubtful about their promise to deliver in sept
|
|
|
|
|
|
