Putting the 4 hashing boards from two S3+'s into a single box - one could mount a second controller board and remove the stick-on cover over the second RJ45 hole. It would be as if running the two S3+'s, two ethernet cables, but the water or liquid cooling would be a plus especially taking advantage of putting the heat outside while the machine is inside.
If you are not afraid to pull your soldering gun out, you can run all 4 boards off one controller. |
|
|
|
You really are an opinionated nicompoop. And that makes you what? What makes you believe that we are all in this to fall within the constraints of your nifty little spreadsheet on whether we'll make returns and by when? And just in case you missed it, bitmain are actually considering an S1 to S5 kit, so there!
As dogie said, they can consider, just as I consider doing a lot of things that I never end up doing. I'm not saying they won't do it either, they'll certainly do it if they feel there are enough "nincompoops" around who will fall for the illusion of saving money by doing an upgrade. I'm just throwing out my advice to help guide anyone who may fall victim to the illusion of saving a few dollars. It is a tempting siren song, after all. Don't feel bad if you were tempted and have now come to your senses, it is completely understandable. Dogie is NOT bitmain, he is paid by them for whatever they pay him for, and his opinion is as good as anyone's (in this case yours, as in worthless!). Suffice to say, the notion has been brought to the attention of the movers and shakers at bitmain, so your input is moot at this point. And don't flatter yourself, you are no where near guessing what I was thinking with regard to the upgrade kits .... but like I said, you are an opinionated nicompoop and just can't help yourself. |
|
|
|
Was the idea to have a channeled heatsink on each board positioned as they are now in the S3's? I confess the sandwiching idea comes from looking at the C1. But that cover doesn't look like it would lend itself to modification with a nibbler for new PCIE positions.
No, the 2 slabs would sit in the middle joined together (just like the C1), so the backs of the boards would have full contact with the un-channeled faces of the slabs. Then over the chips would be copper tubing with a separate routed heatsink sitting atop them. I have to say that with the S1 boards' design, the idea behind the top heatsink had a flaw in that the copper tubing would have to sit atop another copper square the size of the chip to lift it clear of the resistors either side (which would probably have made it more expensive and fiddly). Add to that, there would not be any board contact on that side to the heatsink (a missed cooling opportunity maybe). For the new S5 boards, it seems to be well laid out such that the chips are on one / internal side and the supporting circuitry attachements on the outer side. |
|
|
|
Of course now we see using a half round routing bit to run tracks for the copper tubing in the aluminum slabs. Having the faces flush would really be better so as to reduce warping due to temperature differences that might lift the slap from the PCB.
I was looking to fashion the aluminium slabs on my very own CNC mill and was more confident of achieving the (percieved) required accuracy with square routing, leave alone that I do not have a half round routing bit! Saying that, I was also looking to run the tube over the chips, embedded or not in an aluminium slab, (on the S1 board) and a round tube would not have worked. On having the faces flush, yes I suppose that would reduce the warping, but I never got that stage for that to be an issue. |
|
|
|
Hummm.. S3 and S5 hashing PCB only sale? That may be hard with warranty as improper heat sink or heat management may kill the chips on the PCB. sold as is may be... still, let me send your idea in as well. Thank you for the input!!  No warranty is not a problem at all! I trust bitmain would not sell me a defective board, in the same way that when I buy a complete unit from them, I know the warranty may not cover my overclocking .... I still overclock them! |
|
|
|
Let's see. Two aluminum slabs the dimensions of S3 heatsinks drilled and tapped to fit S3 boards, additional thru holes to sandwich two slabs together. Copper tubing loops dragon-like sandwiched between the aluminum slabs contacting the slabs well with heatsink compound.
Hey! I had a similar design but could not find square copper tubes cheaply and easily (I probably did not look hard enough). My plans were those of milling 1/2 the length of the tubes as tracts on 2 aluminium slabs (thus the square tubes) and laying those tubes inside one slab then laying the other slab on top, and then fashioning my own radiator, again from the copper tubing tied to some mesh but these never left the drawing board for the above reason. EDIT: If bitmain start selling boards on their own, this is one project I will revive as I was planning on doing this with the S1 boards I had left over from my upgraded ones. I now run them underclocked and air cooled ... but would have been a good pilot project for the S3 / S5 boards. |
|
|
|
Its being considered as is every product, including S2 to S3 upgrades <snip> .... waffle .... </snip>
Really? S2 to S3 upgrades? Does that amount to an upgrade or downgrade? Tell me, is it possible to switch chips off on S3 or S5 boards at software level? |
|
|
|
The pricing is public knowledge on their website. We know from the pricing of the S1->S3 upgrade kit that the savings are minimal and are probably negative if you allocate a cost to your time to assemble the thing and it definitely goes negative when you calculate in the cost of the loss of the old miner which could have been sold as well as the loss for the time it take for them to release the upgrade kit (it took them over 2 months (IIRC) to release the S3 upgrade kit), so you lost 2 months of the difference of hashing with an S3 vs. the S1 at a lower difficulty. If you want a project, pursue your upgrades. For Bitmain it probably isn't worth the hassle. I'd rather have them focus on the S6 or S7 than be wasting time trying to save you $5.
Thanks for being the thread troll.
You really are an opinionated nicompoop. What makes you believe that we are all in this to fall within the constraints of your nifty little spreadsheet on whether we'll make returns and by when? And just in case you missed it, bitmain are actually considering an S1 to S5 kit, so there! |
|
|
|
On the viability of S3's, I did not say they were no longer viable, in any case, there's the argument for S1 to S5 kits in the same suggestion (but of-course you conveniently skimmed over that).
I spoke with yoshi (by email) and they are considering an s1-s5 upgrade kit but not until they get some feedback on batch 1. Upgrade kits are not for inexperienced people and poor thermal paste application or nicking something with a screw driver could result in a few loud and disgruntled buyers who can't RMA a diy product that they damaged by improper installation. My guess is upgrade kits will either have no warranty or be provided only in bulk packs of 5+ so that it isn't being sold to people incapable of correct use. That sounds OK and thanks for taking the time to speak with the big man himself. I completely understand that position, aka no warranty and mostly only being available in packs. But I think there is also market for bare S3 boards' packs, and I am sure they've also looked at that. All round, good news and thanks for conveying it on Xmas day. Merry XMAS! |
|
|
|
I think the next logical step for bitmain is to work hand in had with the developers of cgminer so that we can be able to switch off individual chips (that may be malfunctioning) via the webUI. Seeing the chip count per board (which can be up to 256) has gone up to 30, just shy of the S1 count by 2, despite a reasonable leap in the tech, I think the time is ripe for this kind of developement. At the moment, cgminer (or more likely their obfuscated, closed source PIC fimware) decides when to stop sending work to a chip, but I do not think it is as efficient as can be.
Also, now that bitmain have the new chip and have implemented the string design, I call upon bitmain to:
1. Provide an S3 / S1 upgrade kit to the S5
2. Start offering for sale the components of the (soon to be EOL'd S3), i.e the hashing and controller boards.
Happy XMas everyone - bitmain included!
Turning individual chips off is not a software or firmware based function, it requires expensive on PCB hardware to enable that. Honestly, I am aware you get paid to respond in these threads, but the value in that lies in accurate responses not waffle. Now, I'll ask you a straight question: Are you sure it can not be done in software or is that just some more waffle from you? (I know I have done exactly that on my S1 boards using my very own VB.Net miner) EDIT: Just in case you missed the punchline, it is for that reason that I made the suggestion in the first place, having seen the benefits thereof. I'm not going to start another waffle thread with you again. You can look pages back regarding discussion on an S3 to S5 upgrade kit. tldr S3s are still viable, so it would make no sense to upgrade them today. I have NEVER had a waffle thread with you, and do not intend to have one; you must be confusing me with someone else ... and it does not take much to admit that your assertion is incorrect, but as always, the best way is to avoid making false statements. The point at hand was switching individual chips off at software level and your waffle response was that it requires expensive PCB hardware to be able to do that. So you can redeem your self esteem by answering the specific question: Are you SURE it can not be done on a software level? On the viability of S3's, I did not say they were no longer viable, in any case, there's the argument for S1 to S5 kits in the same suggestion (but of-course you conveniently skimmed over that). |
|
|
|
I think the next logical step for bitmain is to work hand in had with the developers of cgminer so that we can be able to switch off individual chips (that may be malfunctioning) via the webUI. Seeing the chip count per board (which can be up to 256) has gone up to 30, just shy of the S1 count by 2, despite a reasonable leap in the tech, I think the time is ripe for this kind of developement. At the moment, cgminer (or more likely their obfuscated, closed source PIC fimware) decides when to stop sending work to a chip, but I do not think it is as efficient as can be.
Also, now that bitmain have the new chip and have implemented the string design, I call upon bitmain to:
1. Provide an S3 / S1 upgrade kit to the S5
2. Start offering for sale the components of the (soon to be EOL'd S3), i.e the hashing and controller boards.
Happy XMas everyone - bitmain included!
Turning individual chips off is not a software or firmware based function, it requires expensive on PCB hardware to enable that. Honestly, I am aware you get paid to respond in these threads, but the value in that lies in accurate responses not waffle. Now, I'll ask you a straight question: Are you sure it can not be done in software or is that just some more waffle from you? (I know I have done exactly that on my S1 boards using my very own VB.Net miner) EDIT: Just in case you missed the punchline, it is for that reason that I made the suggestion in the first place, having seen the benefits thereof. |
|
|
|
do i have to power off and on, or does hitting save and apply enough?
im trying 243 @ 0740
I usually power cycle after I've decided to run a frequency / voltage setting for a long(er) stint. |
|
|
|
Also, now that bitmain have the new chip and have implemented the string design, I call upon bitmain to:
1. Provide an S3 / S1 upgrade kit to the S5
<snip> ... </snip>
Were you even one of us that actually got an S1 upgrade kit (before they discontinued it in fairly short time, and are quite unlikely to offer again)? It did not include the enclosure panels OR the second fan, and was only BTC0.12 cheaper than an S3 complete package. Therefore, even if your S3 was only worth $50, it would still quite likely be a better net cost on the S5 purchase than an upgrade kit, going on precedence. Everything he said was relevant and somewhat accurate, not sure why the retaliation... I wonder, did you even read the original post ...? You missed the bit about the S1 to S5 kit. So, in time honoured fashion, would you care to retrospectively apply your wisom on the worth / viability of that? Again, do not say it if you have nothing to say. |
|
|
|
i been running my s3s at 237.
they are pretty steady at 479.
except one has twice as much hwe at that speed then the other.. .002 vs .001
i was thinking maybe a slight volt tweak would help it..
plus i want to try to get it to 243. but when i set them to 243 they dont get much faster and the hwe goes way up.
If you are trying to run at 243, then start setting your voltage setting from 0720 towards 0750. (look at the 3rd table in the OP of tis thread for example settings at certain freqs) I'd also be tempted to run at freq 250 voltage 0750 (and incrementing depending on result) in your case. |
|
|
|
some reason on my S3s if i type a number in the volts field the hash goes way down.. i dont know what im doing wrong.
ive try 0750, all the way to 0795.
Not all S3's support voltage settings (hardware wise), and please don't ask me which ones do as even bitmain can't tell. On the other hand, set a voltage to its corresponding frequency (0750 is for freq 250 and you can also try freq 262.5); You also have to let it run for a while to tell whether that is the "sweet spot" for your S3. If you get too many HW errors, say in the first hour, then that is not a good voltage setting. NOTE: I'd be patient for another few days /weeks until bitmain have released the new firmware which (seems) to auto set the voltage depending on what frequency you select. |
|
|
|
Also, now that bitmain have the new chip and have implemented the string design, I call upon bitmain to:
1. Provide an S3 / S1 upgrade kit to the S5
Why? Just sell your S3 for $150-200 (current going price on eBay). Then pony up the additional $250-300 (assuming a new direct purchase from Bitmain) for a new S5 which is probably cheaper than the upgrade kit would be. After all, you could only re-use the cheapest parts of the assembly anyway, so not worth the effort. You sell yours if you want. And believe it or not, I actually know what the price of an S5 is, and I am sure you do not do any pricing for bitmain products, so you wouldn't know whether it'd be more expensive. All in all, thanks for nothing .... better still (and I am tiring of saying this in this thread), if you've got nothing to say, do not say it. |
|
|
|
|
I think the next logical step for bitmain is to work hand in had with the developers of cgminer so that we can be able to switch off individual chips (that may be malfunctioning) via the webUI. Seeing the chip count per board (which can be up to 256) has gone up to 30, just shy of the S1 count by 2, despite a reasonable leap in the tech, I think the time is ripe for this kind of developement. At the moment, cgminer (or more likely their obfuscated, closed source PIC fimware) decides when to stop sending work to a chip, but I do not think it is as efficient as can be.
Also, now that bitmain have the new chip and have implemented the string design, I call upon bitmain to:
1. Provide an S3 / S1 upgrade kit to the S5
2. Start offering for sale the components of the (soon to be EOL'd S3), i.e the hashing and controller boards.
Happy XMas everyone - bitmain included!
|
|
|
|
Thanks for the reply.. will said firmware be publicly avail soon? or now? would love to retest the wattage at the wall issue.
Go to the S3 Discussion and Support thread, find one of the BITMAIN_***** people and send them a PM asking to test the new firmware. Don't forget to add your email address to the PM and they'll send it to you by email, no idea when it is due for release though. |
|
|
|
Could you please post how many watts at the wall this is pulling... when I did my tests of voltages I could not get the wattage at the wall to change by changing the volts..
I do not have a tool to measure the pull at the wall, but if your tests showed no change, I have no reason to doubt it. OBITER DICTA: It seems the new firmware that bitmain are testing (I got a copy from bitmain earlier today and you can ask to be sent a copy), handles voltage settings well using the default value of 0000 (I suppose that signifies auto select the voltage). I flashed one of my S3's with the firmware and in the first hour that I was able to observe, it seemed to be handling well. I will not be able to check on the rig as I am away for a couple of days, but will be able to on the 26th or 27th. |
|
|
|
Also, it is actually more expensive, to ship within EU, than from bitmains current location, if you calculate per km shipped, so if they would save 20% on bulk shipping to the eu, then besides warehouse, and staffing, shipping from within eu would inflate the price to a higher level than they already are, i personly see no cost savings, for neither bitmain, nor any eu customer.
Your analysis is perfect for a corner-shop, not an internationally successfull business like that of bitmain. There are financial incentives to having a base in the EU that you never take into account because your analysis is blinded by this overhead / corner-shop mentality. Honestly, I wonder why I keep trying .. |
|
|
|
|
Show Posts
