The Bain Capital investment looks to me like they are buying a lottery ticket on Sia's IP, that's all.  3.5MM is nothing to a venture capital firm.  I did not get a sense of optimism after reading about the Bain Capital infusion. 

Visit the AMDMemTweak thread:  https://bitcointalk.org/index.php?topic=5123724.0

I'm running Windows 10.  Vega 56 Reference cards (flashed to Vega 64) most are now getting 2,030-2,050 kh/s with Teamred Miner mining CNVr Monero.

Jump to page 8 and read from there in the thread.  Windows users need to download the .dll file on Github and put in the WinAMDTweak folder.

Create a bat file to run the Help file.

Create a bat file to capture your current card settings.

Create a bat file for your new timings.

My "newtiming" bat file looks like this:

WinAMDTweak.exe -i 0,1,2,3,4,5,6,7 --cl 20 --rp 11 --rc 45 --rfc 250 --rrd 3 --rcdrd 12 --rcdwr 5
pause

Kind of like eth enlargement pill!

[Edit:] this bumped up my kh/s from 1,930 - 1,980 to @ 2,050 on most cards.  So a modest improvement for a short investment of time so far.
[Edit 2:] dropped --rc to 44 still stable, slight improvement

https://gyazo.com/190137e20ae65c81fe07bea1abc16cb6

Used Kearney666 settings but followed RedN0w's advice on the --rc parameter:

--rp 10 --rc 45 --rfc 250 --rrd 3 --rcdrd 12 --rcdwr 5

Win10

Average hash rate for Reference Sapphire Vega 56s flashed to reference 64 comes out to 2012 kh/s per card.  Most cards are somewhere around 2,030 - 2,050 kh/s but two cards aren't registering much of a benefit.

First time I launched the Command console in Admin mode and ran my "newtimings" .bat file, I got a weird Posix video timing Windows error.

I relaunched almost immediately and Teamred 0.4.3 is running fine so far.  So I have some work to do with the two weaker cards, but for very little work a nice bump up.

I haven't connected up a power meter to do before/after power consumption.

Thanks Heliox.

Fantastic!  We should talk as I like 4U Monero builds.  Delta fans in the front (3).   You can use lesser fans, and this is what equalizes the temps across all cards, zip tied to the support beam that you can move towards the back 1/3 of the case in RoseWill cases like the RSV-L4500.

Place an APW3++ on the bottom/back of the case to the side of your motherboard.  Stack something like a HP server PSU on top of the APW3++ with an X-adapter to supply power to a PICO PSU (eBay, 100 watt minimum) and one or two video cards/risers.  Use dual 6+2 splitter cables to extend the length of the APW3++ cables to power all or most of the GPUs.

This Spotswood kit is really helpful for 4U builds, but the screws and parts are tiny (frustrating at first): https://bitcointalk.org/index.php?topic=1731197.0

Your first "Spotswood" build will leave you cursing, the second will be much easier.  If you are of my age, you will need those reader glasses to handle the tiny screws.  But his designs are elegant in some ways.

I had to build some "partner friendly" living room rigs.  As far as she is concerned, they look like the amplifiers I used to have for Hifi.  Sapphire Pulse Vega 56s, especially the Samsung memory ones give good results.

https://gyazo.com/d98e3651c436ab6cde03cd33b6eb7580

Tuning:  run the miner (right now Team Red miner is working well post cnvR fork).  Open up ONT.  Open up GPU-Z, switch to the Sensors tab in GPU-Z.  Change ONT settings on a card by card basis, watch GPU-Z closely.  Save what works as a ONT profile on a card by card basis if you have the time.  I also use and like the Brnstd Excel SoftPowerPlay table generator.  With Team Red, more often than not you need to reduce mV settings rather than increase to get stability/higher hashes.

Edit:  vBios tricks -- You can flash Reference Vega 56s with the same model/make Reference 64 bios, some debate whether this adds anything.  You can flash the Sapphire Reference Vega 64 air bios to a Sapphire Pulse Vega 56 *Samsung memory* not Hynix memory card.  There are at least four variants of the Sapphire Pulse Hynix Vega 56 bios and I would say don't waste your time on flashing/re-flashing these variants.  So there really is not that much to do as far as bios modding on Vega cards.  You will get more bang for the buck with the Brnstead SoftPowerPlay tables and then spending a little time tuning individual cards using ONT saved profiles, if that is an option.  This is not a situation like the RX 480 8GB Samsung cards where, with a bios mod, one minute you were doing 700 kH/s and then the next with the right bios mod you jumped to 860 kH/s on Cryptonight.  I have not seen huge gains on Vega cards with just reflashing BIOS versions.

One month chart is a jaw dropper for Monero difficulty:

https://www.coinwarz.com/difficulty-charts/monero-difficulty-chart

Where have all the ASICs/FPGAs gone?  Long time passing.

How long do we have until the FPGAs start coming back online?  I will say one to three months, with three months marking the outer limit.  But that's okay though.  I will shut off for Summer and wait for the Fall Monero fork.

It might make sense in that you could manually adjust the top SOC clock rate in your SoftPowerPlay mod.

https://i.gyazo.com/4a1ed3ce7cc6919dc301470fe3e313ec.png

If I am recalling correctly from a few years back, this involved a little endian conversion?  Which someone explained how to do using the Windows calculator/some setting.

It's been a while, but I know this is what I did (bumped up the SOC value) when I wanted P3 to stay at 1100 for memory.

No, working fine here.  Have issues with one card that I will need troubleshoot, but working fine on Nanopool after the fork.

Edit: 18.5.1 driver, OCN settings 1407/910 1100/895, Vega 56 cards (reference -> flashed to 64) + some Sapphire Pulse Vegas 1407/900 960/890.  One card acting up, but the others working fine.  1960/1980 kh/s at 15+14

Stop with the mindless shilling.  Just stop.

I was ignoring this whole Grin thing and then I see you in this chat room.  Crap.  I guess I have to read up on it now.

The SIA team does a fair amount of posting, including Mr. Vorick, over on Reddit.  There you will find recent posts where the SIA team acknowledges that the people hosting are doing so at a loss.  Interestingly, the defense of this is that loss is being "mitigated" by home users (not professional data storage nodes) renting out what would otherwise be "unused" space.  I'm not sure this makes much sense, but whatever.

That's what the SIA team has to say about this, not Brad, not myself.

Wander over and check for yourself:

https://www.reddit.com/r/siacoin/comments/a9pn53/why_is_sia_more_costeffective_than_amazon_s3/

The ever hateful "Obelisk" group have announced an ASIC Grin miner.  Doubt it will ship on time, but maybe they worked with the development team so who knows.  I wouldn't buy any more GPUs based on Grin.

Announcement here:  https://mailchi.mp/obelisk/updates-675157?e=20714c4edc

With the DPS2000B, you want to have as many 80 mm fans to push/pull air through those vented holes as possible.  3 Arctic 80mm fans pushing air from one side, and 2 Arctic 80mm pulling air out from the opposite side seems to work very well.  Arctic 80mm 5 packs go on sale every six months or so it seems.  The DPS2000BB will shut off quickly without decent add-on fans to cool it.

I do not know the dimensions for the 6 slot ONDA, but I like this case:

https://www.amazon.com/gp/product/B0091IZ1ZG/ref=oh_aui_detailpage_o06_s00?ie=UTF8&psc=1

it was $99 for the longest time and now the price seems to have jumped.  It is also sold by Walmart/NewEgg, so check around.  And I second the recommendation for the Spotswood frames -- whether you go with an open frame, or go with the 4U build, I think the Spotswood 4U drop in frame is very worthwhile.  Assembly with all of those exquisitely tiny screws and bolts is a bit of a pain the first time around.  But once it is all adjusted and in place, you'll be glad that you added it to your 4U build.  His open frames are also excellent and minimalist in a good way.

No, that's not true.  XMR has been profitable (somewhat so) but something changed recently, I noticed it.  The change is recent, nothing to do with CNV8 fork back in October.

Also, I would point out that the original reference Vega 64 boards are showing up on ebay for $300 give or take, and price seems to be trending down.  If you have the patience to disassemble them and slap on Morpheus II coolers and keep them in an open air case, they are rock steady at 1830-1860 on CNV8 all day, every day.  I'm just saying they may have value relative to the new $700 cards.  If you want the gory details, pm me.

But their PSUs work pretty well on 220V power.

https://smile.amazon.com/gp/product/B0733JRFVL/ref=oh_aui_detailpage_o02_s00?ie=UTF8&psc=1

$75 for the APW3++ right now on Amazon.  Order some decent 6 pin to 6+2 pin adapters and they become useful for GPU builds.  If you order on Amazon, at this price they will typically offer you 3 to 4 years of SquareTrade insurance type coverage for $2-3.  Crazy.

Looking for suggestions on best settings, PPTable, for Sapphire Pulse RX570 ITX 4GB, single fan cards. 

These cards are tiny, draw little power, and before the last fork, with strap pimping they would get 900+ H/s, no sweat.  For CNv8, I've tried Polaris Bios modes, SRBPolaris, and I've tried generating custom PP tables.  I have one card that hits 830-870 H/s, the rest tend to stay at 650-720 on CNv8.  I've tried various ONT settings with P7 adjusted from 1150-1286/900-925; P4 adjusted from 1850-2100/900.

I'm using Cast 1.6.6 which I have used for a long while.  Cast is working just great with my Vega 54 flashed -> 64 cards.

Any ideas?  Anything obvious that I am overlooking?

RX570 ITX 4GB -- any ideas?

BR, I wasn't calling you out for any sort of BS.  I was more just trying to understand your view and your argument.  I am hopeful for mining, I may be wrong, I just don't happen to have a crystal ball.

PM me if you haven't, I have some questions that would be better addressed via PMs.

Since there are batch 2 buyers still waiting on the SC1 Obelisks ordered (and paid for in full, not just a deposit) almost one year ago (Jan. 2017), I don't think there was a deluge of cheap SC1's on eBay.

However, there were many, many batch 4, batch 5, and I think at one point they had even announced a batch 6, Obelisk SC1 miners that no one in their right mind wanted to part with their money for.

Presto!  The "community" decides to fork and now those worthless, parts already ordered, batch 4/5/6 SC1 miners might have some value.

This is Jack's complete lack of surprise.

I doubt your buyer recouped his investment, but then again, I don't know what you charged him.  If he did, he's lucky, Vosk.

Were you a batch 1 buyer for the Sia Obelisk?  Curious as to the when, and to whom, "Nebulous" is shipping out Sia SC1 units.  They are not particularly helpful when I email them.  No chart of shipped units by order number/date of payment on their website so far as I can tell.  Why is that?

There have been so many games played to date by Sia/Obelisk/Nebulous, that I think all of it is doomed:  the Nebulous hardware, and the part I thought worthwhile -- their storage proposition.  Good idea though.  Hopefully someone else will come along.  

I read Vorick's blog pieces and I have to laugh.  You can just quote paragraphs at random to demonstrate what an absolutely dishonest and double-talking project this is.

The blog piece about how ASICs were "necessary" to protecting the SIA network.
The blog piece about how it was immoral for a competitor to not tell you their business plans ahead of time.
The "community" decision to not fork back in March when Nebulous was nowhere near to shipping out units.
The "community" decision to fork once Nebulous had actual units that worked, kind of.
The offer to compensate buyers for lost time once Sia went from $0.01 to something like $0.0022
The offer to sell overpriced "upgrade" boards to users who are still waiting on an actual unit to arrive before this amazing deal was opened up to the public 2 days later

No offense taken, but I will dance on the grave of this project.

Yes, if I recall he spoke out against the Zcash group for saying "fork!" and then "hey, we won't fork."  I understood that better.  I have gotten quite worked up about certain scams.  This doesn't seem to fit in that bucket. That's what I am wondering about.

Edit:  If you buy what d57heinz is saying, we are all wasting our time (and our money).  Whether you are looking at other ASIC hardware, or other FPGA solutions that are shipping, ALL of it will come to nothing-I think that is his point.  But I notice that people are very much exploring other options.  Not much excitement about any GPU that I can find, but all kinds of threads about new hardware.  If it will all prove to be an enormous waste, then I can think of better targets than this one.  So I am left head-scratching.  That's all.