Topic: Will scrypt mining via GPU be obselete eventually? (Read 1180 times)

GDDR6 cards starting @ 500GBs+ ram bandwidth would only sink Scrypt asics even further.

Current top cards do 1.5 Mh/s easy so how could enhanced RAM with shortened latency dwindling from nanos to picos factor in?  Hmmm I wonder.  

How could this trend continue:  https://litecoin.info/Mining_hardware_comparison   ?

See the massive jump from 1 Mh/s cards to 1.5 Mh/s cards?  

For the love of God, enough of this nonsense!

Answer the question: do you have a link about AMD planning a 2 Mh/s card, or is it your view regarding the potential of GDDR6 or something else? If you have a link, share it, ffs!

Provide a link for "...the successor of the 7990, well, that will be at twice the price of a 290 or even more, so with the same Kh / $"

Ya dude, speculation.  But... those who don't learn from history...  What isn't speculation is that GDDR6 is next up and any wise person would factor that into planning.

You misunderstand ...

I know very well that GDDR6 is coming. I also don't need you to google stuff for me or educate me on message boards. What I asked and would like to see is evidence that a card allowing 2 Mh/s, with a desirable or advantageous price, will be released.  Where's the link with such evidence? You claimed 2 Mh/s projected. Projected by who? Or is it just speculation at this time?

If you mean the successor of the 7990, well, that will be at twice the price of a 290 or even more, so with the same Kh / $

Just google "GDDR6 release date" and/or "GDDR6 scrypt"  and READ.  Main reason people surf the web/pay attention to message boards: glean info on what's to come especially for info that 's not mainstream...

Here's what I'm seeing:

http://beta.fool.com/piyusharora/2013/07/23/3-reasons-why-amd-could-turnaround/40007/

But, a contrasting view point could be argued that manufacturers will take advantage of the current high demand of product already in existence which could push GDDR6 back years from the projected 2014.  It makes sense to exploit this to maximize profit and then eventually roll into the next product phase of GDDR6.  At least, it would for AMD.  But if NVIDIA gets GDDR6 first it would hearken worse for AMD being behind the curve when AMD has been expected to get into GDDR6 ahead of NVIDIA.   Regardless, GDDR6 will come sooner or later (GPU miners hope for 2014 but business needs could push that back) and will keep GPU farming viable for years to come.  Sometimes capitalism doesn't work like it should.  BUT... if GDDR6 is in competitive hands it could drive the arms race faster which would be a good example of capitalism working the way it should!  But again, if both AMD and NVIDIA decide that it's more profitable to take profits on existing inventory the delay is mutual.  Until one of them decides to they want to be light years ahead of the competition.  Of course, the current boom of GPU sales could just be totally happenstance bringing much joy to mfrs and the GDDR6 timeline may be exactly where it's supposed to be.  For now, I buy GPUs strictly for most bang for buck which means sometimes buying the older models that are resold for less (like a 5870 for $100!!) instead of the latest greatest with the most markup.  But when GDDR6 is mass produced there will be much more opportunity, possibly even with GDDR5 prices dropping.  And who knows, maybe the GPU mfrs will tweak some GPUs specifically for mining purposes... what then?  BOOM!! That's what then.

Link?

This is true, for now. I doubt it will be overlooked the moment it hits the market though, same as with scrypt (which so far failed to impress).

THIS ^^^^^^^^!!!!!!!

It's just funny how overlooked the prospect of improved GPUs is in these discussions.  GDDR6 by EOY and projected 2 Mh/s on these cards as a starting point.

Not really. While ASICs for Bitcoin/SHA2 offer 10x (BFL Japaleno) to 200x (BlackArrow Prospero X1) the hash speed of a GPU for equivalent price, the announced scrypt ASICs cannot even surpass GPUs, per device or in Kh/s  / $.

But but, lower power consumption of the scrypt ASICs, bla bla... Resale value of GPUs and the rest of the rig, and undervolting. Your gpus can work at 0.9 - 0.95V if or when the priority is low power consumption

asics vs gpu is relative to price.  So I giveGPUs 2 years inscrypt coins.  Then they will all move to another algorithm like scryptjane.  With so many GPU miners in an alt algo like scryptjane the price will go up and it what ever coin uses it will be the next LTC.  GPU mining will lever die..It will just wax and wane.

GPU mining will go obsolete for scrypt currencies (LTC, doge, etc) the same way it did for sha-256 (Bitcoin). People will design and manufacture hardware that is meant to do nothing else, and do it quickly, and the difficulties on the altcoins will spike.

There are good odds that it will happen within a year. It appears that everyone who has pre-orders for this hardware are scams, but eventually someone will come along. There's just too much money at play for it to get ignored.

Maybe then there will be a new chain of alt-alt coins with a new algorithm that again can only be done on a CPU. Then someone will figure out how to do it on a GPU, then someone will come out with new hardware for it.

The bigger question is "how many alt coins can the market sustain" and "should any alt coin be worth more than its hashing difficulty in bitcoin". My guesses there are "more than people think" and "no", but we'll see. If I'm right it will mean that every company or famous person or city will have their own cryptocurrency, and that altcoins are in a massive bubble that is going to crash hard.

But then gold has been in a massive bubble for thousands of years so there's no guarantee that the crash will be any time soon.

GPU mining will never go obsolete.  CPU mining?  Yes.  Only because CPU mining can be quickly coded for GPUs when it becomes profitable.  GPU to FPGA to ASIC is more expensive so only algorithm coins that have established themselves a truly long lasting will get the FPGA/ASIC investment.  But every new algorithm is based on previous ones so each iteration will mean that the FPGA/ASIC implementations will become less and less profitable.  Eventually there will be coins that GPUs will mine and it will not be worth it to fabricate FPGA or ASIC chips for them.

Cryptocurrencies are already 4 years old.  GPU mining was expected to die 2 years ago yet it keeps on being profitable.

I'm gonna assume "Hey hashcows, do your thing" is not considered mining smart
Feel free to give some tips.

I get around 650 kh/s average. I've only mined for 2-days non-stop and come up with around .007-.008. So I figure, .003 x 28days x $850 = $70 is my minimum expected if I'm running 28 days worth of mining a month.

your gross should be around 150 per month with that 7950 if you are mining smart.

New miner here,
I seem to be raking in around .003-.004 BTC a day through hashcows using my 7950, which should net me anywhere from $50-$80 a month; pretty sweet change for a 0-effort job (and I'm a lucky no-electric-bill miner)

But I was wondering, will scrypt mining with a GPU soon be obselete, similar to the way it is now when mining BTC directly? I'm excited to think that I'll have a full bitcoin by the end of the year, but I don't want to set my hopes high at free money and just get dissapointed by the emergence of scrypt ASICs and whatnot making mining even more implausible for.non-pros.

If/when GPU mining is obsolete, do we expect other hash algorithms to come out (along with new altcoins) that will keep GPU mining profitable?

I don't plan on investing on any gear beyond my current gaming computer, so I figure if I'm gonna get dissapointed then better now than later.