Ok so lyra2z is hashing at 20 Mh/s. A single RX 580 hashes at 3.2 Mh/s. A single Vega64 hashes over 6 Mh/s.
You therefore need 6x RX 580 for one FPGA. But you can get 6x RX 580 probably for around half the price of a single FPGA.
It is as I thought it would be. They are not so terribly powerfull, but they do have a loooot better hash per W ratio. They are expensive, so basically for those with deep pockets. On the other hand as this is still a fairly new field being born again from the ashes, this means there will be some birthing problems etc...In the end I would still bet on GPUs simply because they are proven 100 times over and flexible.
You all have to also bear in mind theat GPUs today are basically 3-4 year old arhitecture. AMD for instance is going on with the same artitecture simply because they did not need to get out anything new. I am sure they (and NVIDIA) have much more powerfull GPUs already prepared but are holding back. So it can as well happen you pay costly for FPGAs just to have both GPU manufacturers release relly powefull cards at the end of the year for instance. And because of mass production they will always be cheaper.
I am definitely going to wait to see how all this plays out
EDIT:
And I just read about 90 days of waranty. Is this correct? I get 90 days of warranty for a 3K$ worth piece of tech. Em ok. Meanwhile I get years of warranty for GPUs. Man if that thing dies after 6 months thats gonna be really bad.
I don't try to put anyone down here don't get me wrong. You seem legit, I am just that voice of doubt in the back of peoples heads Although I am sure most will ignore me lol.
You therefore need 6x RX 580 for one FPGA. But you can get 6x RX 580 probably for around half the price of a single FPGA.
It is as I thought it would be. They are not so terribly powerfull, but they do have a loooot better hash per W ratio. They are expensive, so basically for those with deep pockets. On the other hand as this is still a fairly new field being born again from the ashes, this means there will be some birthing problems etc...In the end I would still bet on GPUs simply because they are proven 100 times over and flexible.
You all have to also bear in mind theat GPUs today are basically 3-4 year old arhitecture. AMD for instance is going on with the same artitecture simply because they did not need to get out anything new. I am sure they (and NVIDIA) have much more powerfull GPUs already prepared but are holding back. So it can as well happen you pay costly for FPGAs just to have both GPU manufacturers release relly powefull cards at the end of the year for instance. And because of mass production they will always be cheaper.
I am definitely going to wait to see how all this plays out
EDIT:
And I just read about 90 days of waranty. Is this correct? I get 90 days of warranty for a 3K$ worth piece of tech. Em ok. Meanwhile I get years of warranty for GPUs. Man if that thing dies after 6 months thats gonna be really bad.
I don't try to put anyone down here don't get me wrong. You seem legit, I am just that voice of doubt in the back of peoples heads
Although I am sure most will ignore me lol.There are 2 different lyra2zs..
One is 2,8192,256 (zcoin)
The other is 8,8,8 (everything else)
Which one are you talking about? For the 8,8,8 coins the hashrates are significantly lower than 2,8192,256 (zcoin).
I am talking about the 8,8,8 coins I guess. The GIN coin and company. My RX 580 does 3.2 Mh/s rock solid and my RX 470 around 2.6 Mh/s. So i can match the hasrate for half the price with RX 580 cards but for a lot higher wattage. One cards uses around 100W on lyra2z so 600 for 20 MH/s. Although I would have to calculate when I do break even if I would buy a FPGA. Also you then have to calucate in all the potential problems a new product will have and you only give 90 days warranty. We have 0 experience how the cards will work. There is 0 millage on them by the community. On the other hands my RX 580 come with a long warranty the ability to mine any algo today are rock solid and proven 1000 times over. And I can resell them easily.
So as I see it now at least for lyra2z its not worth it at all. Again just stating my opinion, you seem like a good guy senseless. Thank you for showing your identity and thank you for trying to get FPGA to the masses. I know you could easily just skip that so I really don't want you to see this as me attacking you because I am not. I am just trying to be the voice of the opposite side;)
And keep up the good work, I am really interested what the performance on other algos will be
Also note that lyra2z is maybe a little speciall because people found out how to really do a good miner with AMD cards right now.
After looking into it, I see someone released a new optimized kernel that I either didn't see before or was just recently released... Yes, GPU hashrates have increased a bit from where they were when I was doing my calculations. All I can say is, the core is still under development and I'm not going to promise anything I can't deliver.
Yes for AMD, tdxminer does 3.2 MH/s on RX 580. I am using it for the last month and it works rock stable on Linux. There is also another Kernel that supposedly can do even better. The author claimed like 25% better but then never released it because of the tdxminer. Could be just vapor for all we know it. He wanted a 25% fee for the miner lol.
The AMD cards can do wonders on some algos its just that there are no publicly available really optimized miners for them. That was proven first with optimizing x16r for AMD (and I am sure there is still room) and now with lyra2z. For a long time AMD was all ethash and cryptonight but its slowly changing now.