BFL Labs ASIC release date - no earlier than February 2013

[punningclan]

Hi guys, I basically had a prophetic dream vision last night, that the BFL Labs ASIC won't be shipped until February 2013 at best.

So it wasn't one of the more expensive "actual" prophetic visions?

[bulanula]

I don't understand.

Maybe you are trolling.

On website is says October.

Even with delays I will expect them to ship before Dec / just when reward halvs so

[ice_chill]

They don't have the actual wafer, just a simulation. ASICs are very complicated and sometimes require several re-spins to get right. Even CPU/GPU manufacturers with years of experience end up having to do re-spins on newly released products, such re-spins are called steppings, or require firmware updates to take care of Erratas.
With ASIC it's difficult to fix with firmware update so a re-spin is usually required.

The BFL FPGA only had a custom power phase design from BFL and on the first attempt they didn't get this right (as they said the simulation showed a power consumption of 20w when it turned out to be 80w so power circuitry had to be redesigned)

Right now they are designing an ASIC without prior experience in designing such things (the FPGA chip is off the shelf product so they can't call it their own)

[bulanula]

They don't have the actual wafer, just a simulation. ASICs are very complicated and sometimes require several re-spins to get right. Even CPU/GPU manufacturers with years of experience end up having to do re-spins on newly released products, such re-spins are called steppings, or require firmware updates to take care of Erratas.
With ASIC it's difficult to fix with firmware update so a re-spin is usually required.

The BFL FPGA only had a custom power phase design from BFL and on the first attempt they didn't get this right (as they said the simulation showed a power consumption of 20w when it turned out to be 80w so power circuitry had to be redesigned)

Right now they are designing an ASIC without prior experience in designing such things (the FPGA chip is off the shelf product so they can't call it their own)

You do realise that a double SHA256 ASIC is dead simple to do compared to AMD GPU or Intel CPU right ?

How do you know they have not been working on this since they released the Singles back in January

My point is ASIC is coming sooner than we think ...

[MrTeal]

I'd be surprised if BFL really pushed the limits of whatever process they're making the ASICs on. AMD and NVIDIA are generally running on new processes that aren't well characterized, and ignore the DRC that TMSC gives them in order to make their designs smaller and faster. Barring any actual logical errors, they would likely be okay. If they do have logic errors that would require a respin, I can't imagine the company surviving.

[Syke]

My point is ASIC is coming sooner than we think ...

BFL is still trying to get the minirig out the door. If the ASIC was so close to being done it would be a waste of effort building new FPGA systems.

[P4man]

I'd be surprised if BFL really pushed the limits of whatever process they're making the ASICs on. AMD and NVIDIA are generally running on new processes that aren't well characterized, and ignore the DRC that TMSC gives them in order to make their designs smaller and faster. Barring any actual logical errors, they would likely be okay. If they do have logic errors that would require a respin, I can't imagine the company surviving.

This. Its also worth pointing out, that as far as asic designs go, a bitcoin miner would be about as simple as it gets. 

[Boussac]

Does anyone know about the forecast power consumption of the BitForce SC 40 GH/s ? I could not find it on the BFL web site ..

[bulanula]

Does anyone know about the forecast power consumption of the BitForce SC 40 GH/s ? I could not find it on the BFL web site ..

750W

[crazyates]

Does anyone know about the forecast power consumption of the BitForce SC 40 GH/s ? I could not find it on the BFL web site ..

Should be about the same as the current Single.

[crazyates]

Does anyone know about the forecast power consumption of the BitForce SC 40 GH/s ? I could not find it on the BFL web site ..

750W

Estimates for the 1TH SC MR are @ 750W. He's asking about the 40GH SC Single.

[bulanula]

Does anyone know about the forecast power consumption of the BitForce SC 40 GH/s ? I could not find it on the BFL web site ..

750W

Estimates for the 1TH SC MR are @ 750W. He's asking about the 40GH SC Single.

Realised that. My mistake.

Too bad I can't delete posts.

[Boussac]

Does anyone know about the forecast power consumption of the BitForce SC 40 GH/s ? I could not find it on the BFL web site ..

750W

Estimates for the 1TH SC MR are @ 750W. He's asking about the 40GH SC Single.

Thanks anyway to both you and bunanula, I got my answer.
That's 1333 MH/J, meaning we could run today's entire hashing power of the bitcoin network on 7500 W..

I am growing skeptical.

The last I had heard was 2O MH/J for FPGAs. Where is the truth ?

[crazyates]

Does anyone know about the forecast power consumption of the BitForce SC 40 GH/s ? I could not find it on the BFL web site ..

750W

Estimates for the 1TH SC MR are @ 750W. He's asking about the 40GH SC Single.

Thanks anyway to both you and bunanula, I got my answer.
That's 1333 MH/J, meaning we could run today's entire hashing power of the bitcoin network on 7500 W..

I am growing skeptical.

The last I had heard was 2O MH/J for FPGAs. Where is the truth ?

The current Single is 10MH/W, but is beat in efficiency by others out there - the ModMiner is 42MH/W.

The new Single is 500MH/W. It does sound unreasonable, but I'm waiting to see some actual performance numbers.

[gonella]

do you know the size of BitForce Single ‘SC’?

[GenTarkin]

Even crazier than the diff skyrocketing will be when all these ASIC's are up and running, and something within the protocol will need to be changed and all those ASIC's become useless, bringing power back to GPU & FPGA miners =P
Or btc will die out and a alt currency which employs something different in its algo will also cause BTC ASIC's to become useless. So, hold onto some of your GPU's everyone =)

[crazyates]

Even crazier than the diff skyrocketing will be when all these ASIC's are up and running, and something within the protocol will need to be changed and all those ASIC's become useless, bringing power back to GPU & FPGA miners =P
Or btc will die out and a alt currency which employs something different in its algo will also cause BTC ASIC's to become useless. So, hold onto some of your GPU's everyone =)

People keep saying this, with no understanding of what a "change in the protocol" means. Performing a SHA256 hash is something that will not change.

[Syke]

Even crazier than the diff skyrocketing will be when all these ASIC's are up and running, and something within the protocol will need to be changed and all those ASIC's become useless, bringing power back to GPU & FPGA miners =P
Or btc will die out and a alt currency which employs something different in its algo will also cause BTC ASIC's to become useless. So, hold onto some of your GPU's everyone =)

People keep saying this, with no understanding of what a "change in the protocol" means. Performing a SHA256 hash is something that will not change.

Someday it will change. Hopefully that will be many years in the future. *No* encryption is future-proof.

[crazyates]

Even crazier than the diff skyrocketing will be when all these ASIC's are up and running, and something within the protocol will need to be changed and all those ASIC's become useless, bringing power back to GPU & FPGA miners =P
Or btc will die out and a alt currency which employs something different in its algo will also cause BTC ASIC's to become useless. So, hold onto some of your GPU's everyone =)

People keep saying this, with no understanding of what a "change in the protocol" means. Performing a SHA256 hash is something that will not change.

Someday it will change. Hopefully that will be many years in the future. *No* encryption is future-proof.

I get that, but if it does ever change, it'll prolly be a switch to SHA512. That's the only way I can see an ASIC being outdated.

[DiabloD3]

Even crazier than the diff skyrocketing will be when all these ASIC's are up and running, and something within the protocol will need to be changed and all those ASIC's become useless, bringing power back to GPU & FPGA miners =P
Or btc will die out and a alt currency which employs something different in its algo will also cause BTC ASIC's to become useless. So, hold onto some of your GPU's everyone =)

People keep saying this, with no understanding of what a "change in the protocol" means. Performing a SHA256 hash is something that will not change.

Someday it will change. Hopefully that will be many years in the future. *No* encryption is future-proof.

I get that, but if it does ever change, it'll prolly be a switch to SHA512. That's the only way I can see an ASIC being outdated.

No, SHA3, whichever candidate ends up winning.