klondike_bar
Legendary
Offline
Activity: 2128
Merit: 1005
ASIC Wannabe
|
|
August 14, 2013, 09:09:35 PM |
|
Apart from the 600GH/s nonsense, this is one of the better ROI evaluations I have encountered recently.
If you can call it that, considering it was done based on 1.5X the expected hashrate ps: do we have any idea how much (or at all) the H-boards can OC? I know we are limited by power, but is there any margin to push it up by 5-10% with little or no change to voltage?
|
|
|
|
creativex
|
|
August 14, 2013, 09:30:20 PM |
|
I seem to recall Dave mentioning 3% a ways back.
|
|
|
|
xzempt
|
|
August 14, 2013, 09:40:09 PM |
|
Why not 50% pre order payment, and the remainder when you have a product to ship? If you do that im in for a couple starter kits....
|
|
|
|
dben428
Member
Offline
Activity: 106
Merit: 10
|
|
August 14, 2013, 09:44:26 PM |
|
Why not 50% pre order payment, and the remainder when you have a product to ship? If you do that im in for a couple starter kits....
With many who have already paid in full for preorders, I see it as unlikely that the preorder payment model would be changed now.
|
|
|
|
newguy05
|
|
August 14, 2013, 09:58:43 PM |
|
Where are you getting 600GH from??
From what I understand, the current board design does not lend itself to substantial overclocking. I would not bank on that in your ROI evaluation.
Apart from the 600GH/s nonsense, this is one of the better ROI evaluations I have encountered recently.
If you can call it that, considering it was done based on 1.5X the expected hashrate
surprised so many people dont like the 600GH number, that was one of the easier assumptions. BuzzDave has said multiple times in this thread and even their website indicates the chip is capable of 2.5-3GH per, the $8000 kit has the per chip hashrate at only 1.5GH. At 600GH it's about a 50% increase from the kit description which rates the chip at 2.3GH per - i thought that was a very reasonable assumption. Or are you guys saying 600 GH is too low and i should increased the hash rate even higher in the calculation? Not clear..
|
|
|
|
dani
|
|
August 14, 2013, 10:03:59 PM |
|
Where are you getting 600GH from??
From what I understand, the current board design does not lend itself to substantial overclocking. I would not bank on that in your ROI evaluation.
Apart from the 600GH/s nonsense, this is one of the better ROI evaluations I have encountered recently.
If you can call it that, considering it was done based on 1.5X the expected hashrate
surprised so many people dont like the 600GH number, that was one of the easier assumptions. BuzzDave has said multiple times in this thread and even their website indicates the chip is capable of 2.5-3GH per, the $8000 kit has the per chip hashrate at only 1.5GH. At 600GH it's about a 50% increase from the kit description which rates the chip at 2.3GH per - i thought that was a very reasonable assumption. Or are you guys saying 600 GH is too low and i should increased the hash rate is even higher in the calculation? Not clear.. and just how are you going to overclock it with every piece of it at it's limit? liquid nitrogen? voodoo?
|
Hai
|
|
|
Cablez
Legendary
Offline
Activity: 1400
Merit: 1000
I owe my soul to the Bitcoin code...
|
|
August 14, 2013, 10:04:11 PM |
|
The boards do not have the power circuitry in place to OC to 600GH/s. You might get 3-5% as was hinted but new boards would need to be developed to really push the chips.
|
Tired of substandard power distribution in your ASIC setup??? Chris' Custom Cablez will get you sorted out right! No job too hard so PM me for a quote Check my products or ask a question here: https://bitcointalk.org/index.php?topic=74397.0
|
|
|
ssi
Member
Offline
Activity: 70
Merit: 10
|
|
August 14, 2013, 10:05:52 PM |
|
I'm designing a miner around these chips, and I'm assuming 2.2GH each as a top-end on the chip. They consume a LOT more power and produce a LOT more heat toward the top end of the performance range. If the power delivery infrastructure on the rig as a whole can't deliver, you won't see the hashrate. That said, you might be able to see 560MH on that rig if every chip is working 100% and you can deliver the power.
|
18xEDfc7y1Nzm2kmLvwYq56xwwEz4Fdh6
|
|
|
DeathAndTaxes
Donator
Legendary
Offline
Activity: 1218
Merit: 1079
Gerald Davis
|
|
August 14, 2013, 10:08:28 PM Last edit: August 14, 2013, 10:23:14 PM by DeathAndTaxes |
|
Where are you getting 600GH from??
From what I understand, the current board design does not lend itself to substantial overclocking. I would not bank on that in your ROI evaluation.
Apart from the 600GH/s nonsense, this is one of the better ROI evaluations I have encountered recently.
If you can call it that, considering it was done based on 1.5X the expected hashrate
surprised so many people dont like the 600GH number, that was one of the easier assumptions. BuzzDave has said multiple times in this thread and even their website indicates the chip is capable of 2.5-3GH per, the $8000 kit has the per chip hashrate at only 1.5GH. At 600GH it's about a 50% increase from the kit description which rates the chip at 2.3GH per - i thought that was a very reasonable assumption. The CHIP may be capable however the BOARD is not. The board takes a 12VDC input however no chip runs at 12 volts, so the board has a regulator (DC to DC power supply) which steps down the 12VDC to the 0.8VDC used by the chip. Like all electrical components it has a limit on current which IIRC is ~30A. That component can't handle the amperage necessary for the overclock and overvolt you would need to gain a 50% boost in performance. It isn't ever going to happen. Even 10% overstock would be pushing the power handling capabilities of the board close to the limit. It might work, and you also might destroy it in a couple of weeks too, or maybe 10% is fine but 12% will kill it. Still 50% is in imaginary land where circuits can use unlimited amounts of power without issue. The board just wasn't designed for that kind of electrical/thermal load. Even a custom built board (S-HASH), designed from the ground up to get more performance out of each chip, which has a rather "beefy" 50 amp DC power supply still doesn't get a 50% gain over stock. To get that much performance out of the chip would require using larger DC power supply AND less chips per board and more boards. In other words buy a reel of chips and start designing something better.
|
|
|
|
DeathAndTaxes
Donator
Legendary
Offline
Activity: 1218
Merit: 1079
Gerald Davis
|
|
August 14, 2013, 10:11:26 PM |
|
Where are you getting 600GH from??
From what I understand, the current board design does not lend itself to substantial overclocking. I would not bank on that in your ROI evaluation.
Apart from the 600GH/s nonsense, this is one of the better ROI evaluations I have encountered recently.
If you can call it that, considering it was done based on 1.5X the expected hashrate
surprised so many people dont like the 600GH number, that was one of the easier assumptions. BuzzDave has said multiple times in this thread and even their website indicates the chip is capable of 2.5-3GH per, the $8000 kit has the per chip hashrate at only 1.5GH. At 600GH it's about a 50% increase from the kit description which rates the chip at 2.3GH per - i thought that was a very reasonable assumption. Or are you guys saying 600 GH is too low and i should increased the hash rate is even higher in the calculation? Not clear.. and just how are you going to overclock it with every piece of it at it's limit? liquid nitrogen? voodoo? Getting the heat out appears to be less of a problem then getting the power in. Even the S-HASH custom board runs without a heat sink despite a significant (but nowhere near close to 50%) increase in clock. The question is more like "how are you going to get 50 amps out of a 30 amp PSU". The short answer is you won' unless you want to see some sparks and a puff of smoke (one time only).
|
|
|
|
newguy05
|
|
August 14, 2013, 10:12:09 PM |
|
2) Per the website: - 25 GH = 16 chips & 400 GH = 16*16 = 256 chips -> they both average ~1.5GH per chip. - But the chip page states 2.7 GH per chip.
Does that mean the 25 GH full kit's real hashrate is ~43GH(2.7 * 16), and the 400GH is ~691 GH(2.7 * 256)?
3) Are the inventory numbers for the Oct fullkits accurate? right now it shows a 378 unit left for the 400GH, if all 378 gets sold will they all be delivered in Oct?
Thanks, i like how this is not dependent on avalon and already has asic chips in hand.
buzzdave, can you please answer the above questions, i am thinking of placing an order for the oct 400gh unit after the july shipment was confirmed, but like to have the above answered. Thanks #2 chips are capable of over 3Gh/s under highly controlled conditions, but our boards are underclocked in order to avoid issues described many times in this thread. Does that mean the 25 GH full kit's real hashrate is ~43GH(2.7 * 16), and the 400GH is ~691 GH(2.7 * 256)? I'm making no promises here, but Tytus is seeing peaks of 130+ Gh/s from his 4 card rig. This could be just a phenomenon of pool stats. I see 101 Gh/s at this time. He's sending to Slush, so I can't provide a link to the stats well this is what buzzdave said, if the chip is capable of 2.5-3GH why would they clock it at only 1.5GH? it makes no sense, also i remember reading that the 400GH in the description is just that a description, the actual hashrate is higher per the chip the actual hashrate capabilities? No? anway all remain same but using 400GH in the calculation over same 12 months, this turns into a money loser generating $7590.89 over 12 months: ~$400 loss
|
|
|
|
ssi
Member
Offline
Activity: 70
Merit: 10
|
|
August 14, 2013, 10:13:43 PM |
|
well this is what buzzdave said, if the chip is capable of 2.5-3GH why would they clock it at only 1.5GH? it makes no sense, also i remember reading that the 400GH in the description is just that a description, the actual hashrate is higher per the chip the actual hashrate capabilities? No?
anway all remain same but using 400GH in the calculation over same 12 months, this turns into a money loser generating $7590.89 over 12 months: ~$400 loss
I suggest you use the nominal hashrate of the rig as a BEST CASE number in your profitability analysis
|
18xEDfc7y1Nzm2kmLvwYq56xwwEz4Fdh6
|
|
|
newguy05
|
|
August 14, 2013, 10:16:31 PM |
|
a few more replies were posted while making my previous post, ok thanks for the explanation that was a bit of a disappointment but glad it was clarified. edited original post. Here's my latest calculation based on current stats for oct end delivery, very difficult decision either way. However credit must be given to buzzdave and team on how they handled the retail sales so far, compared to avalon/bfl/terrahash etc..it's night and day.
1) Assuming a 35% difficulty increase per 2016 block until Nov 1st delivery, this is very reasonable considering the massive amount of hashpower coming online during that time from all the asic hardware delivery from asicminer/bfl and avalon chips etc..not to mention the large farms coming online.
Starting difficulty at Nov 1: 50,000,000 * 1.35 * 1.35 * 1.35 * 1.35 * 1.35 = 224,000,000
2) Assuming btc price remains flat at average ~$110, also a reasonable assumption given the price history for the last few months.
3) Assuming only a 15% difficulty increase per cycle thereafter over the next 12 months from Nov 1 2013 to Nov 1 2014. I think this is a very conservative estimate given the amount of asic hardware/development coming online, lets just assume best case scenario and things will level off in the next 12 months and averages out at 15% per increase
Profitability decline for 12 months: 0.00948604
4) Putting it all into the final calculator with a starting date of Nov 1st per above:
Bitcoin Difficulty: 224,000,000 Hardware Cost: $8000 Hash Rate: see below Profitability Decline: 0.009 Time Frame: 12 months USD/BTC: 110 Ignore all power usage cost etc..
RESULT1 using 600GH: $11345.29, minus the $8000 hardware cost, ~ $3300 profit over 12 months.
RESULT2 using 400GH: $7590.89, minus the $8000 hardware cost, ~$400 loss over 12 months.
Of course it is a given if the difficulty per cycle goes up by even 1-2% profit will be wiped and if actual is lower then more profit, same goes for $ per btc etc..
again this is just my best estimate based on reality not some fantasy numbers...
As i said, very tough decision buy or not...
|
|
|
|
minternj
|
|
August 14, 2013, 10:18:16 PM |
|
The short answer is that the boards were designed to be used for the 100TH mine. So components were picked on costs and efficiency. Lower speed= lower heat = les electricity costs for the 100TH mine. There is some small room for higher clock but will still be near the 400gh range. it's all in this post somewhere. Wait for someone ot sell the s-hash boards if you want to overclock.
|
|
|
|
buzzdave (OP)
VIP
Sr. Member
Offline
Activity: 472
Merit: 250
|
|
August 14, 2013, 10:39:04 PM |
|
Our boards are designed for max efficiency. They are running the chips at exactly the most profitable value point watts/dollar. If you want more hashes you should buy more hardware. Now that being said, a lot of people don't look at it this way - this is a perspective more suited to a very large private mining operation that intends to run far into the future.
Its true that our boards are hardwired to a certain clock frequency. We've got them running about 440Gh/s gross hashrate. After errors, tested nonce rates are seen at 410 - 415Gh/s. That's probably about all you are going to get out of them. You can gain some watts efficency by turning them down...certainly this will become a factor some day, depending on what you pay for power.
I know there is a *lot* of interest for an overclocking rig and I'm working on it. Can't share details yet, as I'm under NDA, but something cool is coming.
|
|
|
|
bkpduke
|
|
August 14, 2013, 11:25:59 PM |
|
Our boards are designed for max efficiency. They are running the chips at exactly the most profitable value point watts/dollar. If you want more hashes you should buy more hardware. Now that being said, a lot of people don't look at it this way - this is a perspective more suited to a very large private mining operation that intends to run far into the future.
Its true that our boards are hardwired to a certain clock frequency. We've got them running about 440Gh/s gross hashrate. After errors, tested nonce rates are seen at 410 - 415Gh/s. That's probably about all you are going to get out of them. You can gain some watts efficency by turning them down...certainly this will become a factor some day, depending on what you pay for power.
I know there is a *lot* of interest for an overclocking rig and I'm working on it. Can't share details yet, as I'm under NDA, but something cool is coming.
Dave, you tease more than the captain of the cheer squad. With a lead up like this, I am expecting great things.
|
|
|
|
tom99
|
|
August 14, 2013, 11:37:31 PM |
|
+1
|
|
|
|
papaminer
|
|
August 14, 2013, 11:46:43 PM |
|
thank you for the updates guys...
I will see how it goes... and will probably order the Oct 400ghz when the Aug guys get their hardware...
|
฿: 1L7dSte4Rs4KyyxRCgrqSWYtkXdAb4Gy1z MORE INFO ABOUT ME: BTC
|
|
|
ssi
Member
Offline
Activity: 70
Merit: 10
|
|
August 14, 2013, 11:54:02 PM |
|
Our boards are designed for max efficiency. They are running the chips at exactly the most profitable value point watts/dollar. If you want more hashes you should buy more hardware. Now that being said, a lot of people don't look at it this way - this is a perspective more suited to a very large private mining operation that intends to run far into the future.
Its true that our boards are hardwired to a certain clock frequency. We've got them running about 440Gh/s gross hashrate. After errors, tested nonce rates are seen at 410 - 415Gh/s. That's probably about all you are going to get out of them. You can gain some watts efficency by turning them down...certainly this will become a factor some day, depending on what you pay for power.
I know there is a *lot* of interest for an overclocking rig and I'm working on it. Can't share details yet, as I'm under NDA, but something cool is coming.
Are your boards using an external clock? (as opposed to the chip internal clock I mean, not a user-provided external clock)
|
18xEDfc7y1Nzm2kmLvwYq56xwwEz4Fdh6
|
|
|
kaerf
|
|
August 15, 2013, 12:35:24 AM |
|
Are your boards using an external clock? (as opposed to the chip internal clock I mean, not a user-provided external clock)
Nope. They are using the internal clock....which scales with voltage (and software settings).
|
|
|
|
|