What's better? High freq or more ASICs?

[gt_addict]

I'm limited to how much electricity I can use, mainly because of the price but also because of the heat and ventilation of it aswell. So this is a curiousity question more than anything and I'd like to hear people views too.

What's better:-

More ASICs at a lower speed

Or

Fewer ASICs but at a higher speed.

Now I know you can have the same hashrate with both but from my view more ASICs at a lower speed would be better because you've got more little guys working things out and not generating the higher temp (but of course more volume of heat as there are more of them).

What are your views?

[philipma1957]

I'm limited to how much electricity I can use, mainly because of the price but also because of the heat and ventilation of it aswell. So this is a curiousity question more than anything and I'd like to hear people views too.

What's better:-

More ASICs at a lower speed

Or

Fewer ASICs but at a higher speed.

Now I know you can have the same hashrate with both but from my view more ASICs at a lower speed would be better because you've got more little guys working things out and not generating the higher temp (but of course more volume of heat as there are more of them).

What are your views?

Good question.  But lots of variables.

Since it is on asics and not all miners (GPUs)

Since only bitmaintech is selling a modern miner I will answer for the bitmaintech s9 only

    Lower Asic frequency is better within reason  a 650freq board at 550freq is far less likely to die off then setting it for 675freq.   This is only true for the s9  a broad generalization most likely due to the physical chip size.


  As to how it applies for earning coin I always go with the idea of the highest hashrate possible for your gear as long as error rate is low.  A direct conflict with the paragraph above.

I used to overclock all my bitmaintech gear 5%  once the s9 came out I underclock by  5-10%

[gt_addict]

That's one thing I didn't consider, longevity (lifespan) of the gear in correlation with how hard you push it.

But do you think you stand a better chance of hitting a hash if theres more chips, or do you think there is more chance for duplicate hashes to be created, or is that a null point considering a single chip could create duplicates?

[philipma1957]

That's one thing I didn't consider, longevity (lifespan) of the gear in correlation with how hard you push it.

But do you think you stand a better chance of hitting a hash if theres more chips, or do you think there is more chance for duplicate hashes to be created, or is that a null point considering a single chip could create duplicates?

I know it should scale  100th should do exactly 2x that of 50th.  The math says it does.

 But I 'feel'  more is better. I know not very scientific  

[Biodom]

If I am reading Bitmain's warranty correctly, underclocking also waives the 90 day warranty.
Or is this only with respect to bitmainwarranty (US), but not Bitmain proper's warranty?

[Shiroslullaby]

What is the point in underclocking a miner?
Obviously I can see why someone would overclock a device to push it to the max,
but what benefit do you get by slowing it down?
Is it just to reduce the amount of power you are drawing/ heat generated to hopefully increase the lifespan of a device? Or are there other benefits?

[gt_addict]

Personally (as stated in my original post) it's to reduce electricity consumption due to limitations on how much I can use.

[philipma1957]

What is the point in underclocking a miner?
Obviously I can see why someone would overclock a device to push it to the max,
but what benefit do you get by slowing it down?
Is it just to reduce the amount of power you are drawing/ heat generated to hopefully increase the lifespan of a device? Or are there other benefits?

My Kia Forte can do 100 MPH on an empty highway I drive it at 80 mph on en empty highway.

Why ? 1) safety 2) safety 3) safety  4)  it will last longer.


Back to topic.

Most new gear no longer roi's in 80-100 days. Even with cheap power. The game is changed so running gear oc'd is more of a loser then ever.

If you own s9's and oc they burn and die.

I have 26 gpus I slightly lower volts don't push them and lo and behold no gear dies.
My savings on gear damage is huge.

S-3 was a good oc item freq 243 or 250 vs freq  225 easy
s-5 was good
and s-7 was okay.  I have batches 1 and 2 so 612 or 618 vs  600 worked okay

[Shiroslullaby]

Thanks for your answers.
I know the general principles of overclocking/ underclocking computer hardware,
I just wasn't sure if there was some added benefit for bitcoin mining.

[Dalkore]

No free lunch.    Doesn't matter either way.  Either you have less units that are more power hungry or you have more units that draw less power.  Either way to scale up to a large mining operation you will still need a big pipe and either higher density power distribution or more physical power distribution.      Also with higher draws, we are already hitting limits on the size of a PSU so at a point you will need more PSUs because they don't make them any bigger.   We currently are using server-class PSUs or expensive gamer PSUs to keep the number of PSUs to a minimum.   

What determines the direction is if you are going into an existing electrical plan and that will tell you what is most economical.    With a blank slate, it depends on the hardware.     Lots of variables.   I have been doing this for almost 6 years, finally it is getting much easier because I have dealt with most scenarios you will see.   Still no cake walk.


-D

[sidehack]

From a designer standpoint, it's a tradeoff between initial cost and operating cost. More chips at a lower frequency will get you a higher initial cost per hash (chips being the largest single expense on a miner), but at a lower power draw per hash so reduced power cost over time and, in the end, a longer viable life because of this.
Fewer chips at a higher frequency will reduce the cost to build it so in theory the miner is cheaper, which for greedy entities means a higher profit margin when you sell it for the same price. The increased power draw per hash also reduces the viable lifetime, which for greedy entities means faster obsolescence and increased sales of the next generation.

Pretty sure this is was the driving consideration behind the switch from 54-chip to 45-chip boards in the S7, which maximizes profits for the manufacturer at the consumer's expense.

I like going with a middle ground, stack a decent number of chips at less than peak clock and let the user overclock (possible with overvolting and improved cooling) if he wants to to increase initial hashrate vs investment cost, and still have the ability to undervolt and underclock to run chips colder to extend the viable life. That's about the best way I know of to maximize the consumer's returns.

[philipma1957]

No free lunch.    Doesn't matter either way.  Either you have less units that are more power hungry or you have more units that draw less power.  Either way to scale up to a large mining operation you will still need a big pipe and either higher density power distribution or more physical power distribution.      Also with higher draws, we are already hitting limits on the size of a PSU so at a point you will need more PSUs because they don't make them any bigger.   We currently are using server-class PSUs or expensive gamer PSUs to keep the number of PSUs to a minimum.   

What determines the direction is if you are going into an existing electrical plan and that will tell you what is most economical.    With a blank slate, it depends on the hardware.     Lots of variables.   I have been doing this for almost 6 years, finally it is getting much easier because I have dealt with most scenarios you will see.   Still no cake walk.


-D

I just past my fourth year summer 2012.  You are correct. Even when you think you are all good a wire fails or a chip.

I find the 100 day mark of a gear up to be interesting for me.  My garage floor is 6 inches of concrete .  At about 100 days of hard running the heatsink effect of the concrete  reaches a top end the entire slab is warm about 25 by 25 feet and six inches deep.
I usually shut down for 2 days to allow for the slab to get back to normal temps then fire everything up.

[Sierra8561]

A quick note on the need to overclock or underclock. I've personally never found a need to overclock especially with a s9 other than increased hash rates. On the other hand underclocking for me is a must sometimes. In the summer months with 100 degree F weather and air conditioners running at their max, warrants me underclocking my machines for temperature reasons.

[Daisy14]

Having more ASICs is definitely better than increasing the frequency.

The rig will have a shorter lifespan if it is continually overclocked.

[helipotte]

I have found that every piece of equipment seems to have a "maximum clockrate to efficiency" ratio.  Meaning, clock higher and you increase electrical consumption

while gaining less linearity in hashrate.  A good example where the BFL singles.  When I had these I found that by dropping the voltage/clockspeed just a small amount, the

power usage would go WAY down.  The hashrate would drop just a little.  10% less hashrate but 40% less power consumed.  You just have to do the math to find

your optimum ratio.

[PremiumCodeX]

I think ASIC since that is the only kind, which I and my friends have really used and we all had smooth experience with it.

[lottery248]

Having more ASICs is definitely better than increasing the frequency.

The rig will have a shorter lifespan if it is continually overclocked.

more ASICs will require a higher specification of a computer to secure bitcoins, and your network bandwidth will be higher in order to maximize mining efficiency. overclocking will as well losing in warranty. more ASICs means faster but more capital needed. higher frequency requires less voltage but will need more energy and lead that machine to roaring. both cannot get you more bitcoins even if with a higher hash share in such ultimately fierce mining competition.
wow but it needs mining to be secured.

[leowonderful]

Although overclocking and making an ASIC operate at a higher frequency makes it run faster and produce more hashes per second, it also increases its power consumption to a point where it may not be ideal. I'd just prefer a well-researched and thought through ASIC that doesn't require insanely high clock rates to operate at its desired hashrate.

[PremiumCodeX]

Although overclocking and making an ASIC operate at a higher frequency makes it run faster and produce more hashes per second, it also increases its power consumption to a point where it may not be ideal. I'd just prefer a well-researched and thought through ASIC that doesn't require insanely high clock rates to operate at its desired hashrate.

But it does not matter if you have access to free electricity, does it? I would like to get answer to the question more whether an overlocked ASIC has reduced lifetime due to running it at high freq.

[leowonderful]

Although overclocking and making an ASIC operate at a higher frequency makes it run faster and produce more hashes per second, it also increases its power consumption to a point where it may not be ideal. I'd just prefer a well-researched and thought through ASIC that doesn't require insanely high clock rates to operate at its desired hashrate.

But it does not matter if you have access to free electricity, does it? I would like to get answer to the question more whether an overlocked ASIC has reduced lifetime due to running it at high freq.

Oh, sorry. It would likely have a shorter lifetime with more electricity running through it and with higher temps, although with proper cooling temperatures should not be an issue. Either way, with proper cooling installed and overclocking to a safe level with a good voltage level, the lifetime of such a miner should be pretty much the same as a normal one, as long as proper cooling is done, and it's not overclocked too much. Some parts could break running at too high a frequency or voltage, assuming voltage is tweakable on a miner.