Finally saw this Grin ASIC 
hope to see the miner soon 
(Came across Thomas from MicroComputer, took a few pictures together with this first Grin ASIC in the world 
) Thomas said this chip was produced at 2020 week 18, as labeled on the chip.
Nice little Asic. Come back Gary 
Is this asic presumably compatible with the last mining algorithm change? (ie. it won’t become outdated?)
Well ASICs would be for c31 (or c32?) not c29. C29 changes. C31 doesn’t.
Thanks for the info , Gary!
Could you possibly ask them if the ASICs uses lean or mean (or the hybrid slean) mining?
Note that lean C32 would require 512/k MB SRAM + 512 MB DRAM for a k-pass solver, while a mean C32 would require about 16GB, and slean could use anything in between.
Tagging to keep updated
Thank you, further any more information, please.
Tagging as well. 80* RT2080TI, would be nice if it can deliver on that promise. Less nice for my RTX 2080Ti’s though, which would become obsolete.
What happened to those obelisk GRIN miners?
@betajuice Mining algorithm never changed for ASIC, which was finalised almost since the beginning (a little “change” was to postpone C33 and so on), the changed algorithm is/was for C29 which is for GPU only to avoid ASIC.
This ASIC chip can work for C31 and C32, but since the phase-out date of C31 is coming soon (about 2 month later), only C32 will make sense and C32 will have a very long life. So, this ASIC chip will work long time for years, or forever if C33/C34/etc not taken by Grin.
@tromp Let me ask them and will come back to you soon.
Sure, I would keep eyes on this asic miner and update here.
Thank you for the update! Looking forward to hear more about the development! Grinning
Cancelled almost 1 year ago: GRN1 Cancellation Announcement
It would also be nice to know how much SRAM is on the chip, and how much HBM2 DRAM is in those HBM2 stacks on top.
Q: if the ASICs uses lean or mean (or the hybrid slean) mining?
A: “Lean”
Q: how much SRAM is on the chip
A: “64M”
Q: how much HBM2 DRAM is in those HBM2 stacks on top
A: “we made Mask for our own HBM Die, different with HBM2.
1GB”.
Thanks, gary! So they use an 8x tmto (time-memory trade-off). For each trimming round they would need to read the 512MB edge bitmap 8*2=16 times. That’s a lot of memory bandwidth!
I don’t understand, they are selling GRN1 miners right now:
https://www.obelisksc1immersion.com/shop/obelisk-grn1/
No, they are not, that is a scam website
This is Obelisk’s site, they are not currently selling any miners at this time: https://obelisk.tech/#mining
oh freaking jesus ok, thanks
Cortex, Ravencoin…
