Sensors in a standard S5 are also cooled by the fans, thus a lower reading, whereas in a water cooled setup, not much air movement is concentrated on the boards (aka sensors), so the readings are going to be higher.
In any case, if this were not the case, then we'd see that reflected in the other stats- hashrate, HW% etc, but we don't.
Cooling isn't there just for the chips, so if you're saying that its okay that the other components are too hot because the chips themselves are cool, you're gonna have a bad time. To answer your other statement, we don't usually see temperatures reflected in those other stats anyway. Plenty of people using low RPM fans run at similar temps without any noticeable effect on the telemetry. Another dogie-waffle response ... The hashing chips are the primary heat generators in the S5, and the water cooling block (like the air cooling) is targeted at the chips. As was pointed out, the sensors are on the opposite side of the board and the reading they give in an air cooled setup is subject to the active air cooling, i.e the fans, which is not the case with a water cooling setup as the fans are mounted on the radiator (that should not be too hard to understand really). If the temps are too high, you'll get HW errors when the chips fail to return a result (which will also reduce the reported hash-rate), and in the extreme case, the unit will stop hashing aka thermal cutoff .... and that is where temps are reflected in the other stats in case you were still trying to figure that out. |
|
|
|
Those are hot temps.
From what I understand the temperature sensors are on the opposite side of the board than the asic chips. Since he is using water blocks to cool the chips, he won't get an accurate reading on temperature. The heat is pulled away from the chips more efficiently with the water block than with the air cooling heat sink. That doesn't really make sense. A water block or an 'air block' would make no difference to how the temperature probe reads in relationship to the actual. It is what it is, try and get your head around it! Sensors in a standard S5 are also cooled by the fans, thus a lower reading, whereas in a water cooled setup, not much air movement is concentrated on the boards (aka sensors), so the readings are going to be higher. In any case, if this were not the case, then we'd see that reflected in the other stats- hashrate, HW% etc, but we don't. |
|
|
|
what do you mean?
I've ssh'd it to run at 250.. isn't that all that is required?
Check if you have an Advanced Settings tab under the Miner Configuration tab, if so, you need to click on it and set the frequency there. Latter firmware also does not read frequency settings from that file ... rather another one. It worked!!! thanks!!!!  now if I can overclock the C1 miner  Good. Same thing for C1. |
|
|
|
What frequency are you running that thing at? Also, I cannot see you having more than 2 blades, but I may be wrong, how many blades are those?
2580GH/s .... ? Care to post a screenshot of the webUI?
frequency 400 2630 GH/s 1375W Silwerstone 1500W PSU with soldered EVGA double cables ( SST-ST1500 ) <snip> http://www.perearstike.ee/C1/C1.400.2.jpg </snip> Impressive .... I suppose the real test on that setup would be the longevity of that PSU and whether you'd need a 1700W / 2200W (80% / 60% load respectively). But yeah, that's approx 690W for a " standard" S5. |
|
|
|
3. Server PSUs are a who other matter. I wouldn't even bother with a 550W server PSU, since you can find 800W server PSUs for that price just as easily. You just need to be willing to do some extra work to get them working.
On the mark there ..... just happen to have a ton of them 550W types in my shed that were powering my S3's that I have now disposed of. (they actually worked well on the S1's as well!) In that case, if you have a bunch already and they're all wired up, you might as well just use them. I have a bunch of 450W server supplies kicking around that I used for S1s as well, but they just have wire leads and I'm not going to bother putting PCIe plugs on them. I suppose you are right ... still, I am getting 1350GH/s average (as opposed to the 1312GH/s I got from a single 850W server PSU), so never know, tweeking up the voltage (a simple pot twist on the 550W's!) may yield a bit more yet! Still think 2 550W's are more efficient than one 750W / 850W even 1100W powering an S5 .... and over 6 months, that may be something. Gotta get that killawat. |
|
|
|
3. Server PSUs are a who other matter. I wouldn't even bother with a 550W server PSU, since you can find 800W server PSUs for that price just as easily. You just need to be willing to do some extra work to get them working.
On the mark there ..... just happen to have a ton of them 550W types in my shed that were powering my S3's that I have now disposed of. (they actually worked well on the S1's as well!) |
|
|
|
My water-cooled S5 with selected miner blades works.with EVGA G2 1300
1295W 2580GH/s 0.502W/GHs
I have to try a more powerful power supply unit.
What frequency are you running that thing at? Also, I cannot see you having more than 2 blades, but I may be wrong, how many blades are those? 2580GH/s .... ? Care to post a screenshot of the webUI? |
|
|
|
Between 50-85% there is usually low single digit efficiency differences. Not enough to care about really.
Do not feel obliged to comment. I already got a sensible response. |
|
|
|
Just putting this out there, would you not get better efficiency if you run the S5 off 2 PSU's, e.g a 550W for each blade than run the whole rig off a single 750W/850W/1100W PSU?
Depends. Generally within the same product line a 550W running at 50% load will be about as efficient as a 1100W running at the same 50% load. More practically, the 550W units would tend to be easier to find and cheaper, but often the cheap ones you find aren't as efficient or well built as the 1100W units. You then also have another full ATX supply to deal with, and if you have more than a couple of these it gets to be a pain. I actually thing the sweet spot is probably 750-850W. You can get decent well built gold PSUs in that range dirt cheap, running a 750W PSU at 550W (DC) won't really get it into the bad part of it's efficiency curve, and they're still super easy to source. 1. Throw into that mix the 550W PSU being a server one - smaller (yes), less efficient than a 1100W (no) 2. If we get to the sweet spot being between 750W-850W for the S5, then I can not see how either would be more efficient than 2 550W PSU's, assuming same product line. 3. Then there will be price, I know you can pick up a multi rail 750W PSU for around GBP 50.00 (could be cheaper depending), but you can also pick up a single rail 550W server PSU for GBP 10.00 I think it is both cheaper and more efficient to run an S5 that way .... I am just not sure how the draw at the wall would differ (and that's why I put it out here, been tumbling about in my head for a while now!) |
|
|
|
|
Just putting this out there, would you not get better efficiency if you run the S5 off 2 PSU's, e.g a 550W PSU for each blade, than run the whole rig off a single 750W/850W/1100W PSU?
|
|
|
|
what do you mean?
I've ssh'd it to run at 250.. isn't that all that is required?
Check if you have an Advanced Settings tab under the Miner Configuration tab, if so, you need to click on it and set the frequency there. Latter firmware also does not read frequency settings from that file ... rather another one. |
|
|
|
|
Little orange devil is flashing regularly on first .....
|
|
|
|
Other options for getting an S5 variant for those upset about the price increase ..... https://bitcointalk.org/index.php?topic=936343.0Also, just in case bitmain are reading this thread, can we have a price for ordering plain boards (does not matter whether S5 or S3 boards) just like syscooling seem to have? |
|
|
|
Answer to your psycology question is simple, I wonder why you could not figure it out. (and there-in lies your answer, thing is, can you figure that out?)
.... Help me understand the psychology here. "If I was hesitant purchasing something at $341, why should I be less hesitant to purchase it at $370 ? Why would I have more motivation to purchase at $370 when it was just $341 earlier ? $310 a day before that."Furthermore, BTC appears to be trending down. How should one feel if they purchased at $370 right now, and later today, the price is reduced to $360 ? Like I suspected, you could not figure out your own conundrum, so I'll bow to your request and get you out of that misery. 1. You should be less hesitant to purchase at the current higher price because it can go higher. But you have options right now, as you pointed out earlier, in the shape of the SP20, till they re-instate their price in sympathy with bitmain, then the proverbial crap will have hit the fan for you. 2. It's your call whether to purchase now at $370 or hedge your bets in anticipation of a price drop later today to $360. But be prepared to live with your call should you choose not to buy now and the price later today staggers to $371. Oh, I nearly forgot, you should feel as you please whatever call you make and refrain from blaming bitmain (or real life as you prefer to call it) for your decision. Forgive the verbosity, just in response to your cry for help. |
|
|
|
So why mourn at bitmain? Mourn at real life ...... it's the one that distracted you from the bargain of the century .... Yeah.... no. Are you suggesting the competition raise the prices of their goods from $479 to $529, and we should be OK with that ? Psychology Question for you: If I was hesitant purchasing something at $341, why should I be less hesitant to purchase it at $370 ? Why would I have more motivation to purchase at $370 when it was just $341 earlier ? The competition do not set the price for bitmain. In any case, in a perfect market, buyers will choose the lower price (and for some reason I believe bitmain know this obscure fact), so go ahead and buy from the competition. Answer to your psycology question is simple, I wonder why you could not figure it out. (and there-in lies your answer, thing is, can you figure that out?) |
|
|
|
I was about to purchase two at $310, communicated with Yoshi to resolve a coupon discrepancy, got distracted with real life, come back and the price was raised.
So why mourn at bitmain? Mourn at real life ...... it's the one that distracted you from the bargain of the century .... Yeah.... no. |
|
|
|
|
OK, so you put off your buying and now BTC has experienced a fillip and all of the sudden bitmain are greedy?
When BTC was falling, everyone with a keyboard was shouting at bitmain to tag their BTC price to USD so they could buy in fiat, and also asked for fiat outlets e.g Amazon et al, to which they obliged. They went further and offered a 15% discount for a period after setting up the fiat channels.
Still some jombas put off their buying decision. Now the discount is gone, BTC is rising against USD ..... you only have yourselves to blame!
But all we have in this thread now is mourning and whining .... guys, what do you really want? You want to see bitmain go bust?
|
|
|
|
I know they have removed any references to the 9v under-volt from the product description now, but on release they said you can undervolt to 9v to achieve better efficiency, thus the quote for the rated voltage is for running the unit at standard settings, but I was refering to over-voltage here.
So the question remains, what is the safe overvolt voltage for the S5?
OBITER DICTA:
1. I have run my S5 with just 1 PCI-E connector to each board off a 550W PSU at stock freq and it run OK, though clearly the PSU was being overloaded.
2. Up to this day, I run all my overclocked S3's with just 2 PCI-E connecotrs without any issue despite it being recomended to use all 4.
Are you not under powering the boards by not powering both connectors? Or overloading the current 12v rail? For the S5, going by the HW%, I was not underpowering the boards at all running at stock freq, however, I was clearly overloading the 12v rail (I've never known that particular PSU blow so much hot air, literally!) For the S3 I am neither under-powering the boards nor overloading the the 12v rail. |
|
|
|
And here-in lies the real question. Would I be able to achieve a better hash-rate (and efficiency) if I over-volted and under-clocked? e.g 13.5v @ 375
there's no point in doing that. Your hashrate might stay about the same but your efficiency would go from 0.5w/gh to ~0.6w/gh, and likely exceed the current amount of cooling you provide. I personally was able to get a stock unit with 5C intake to max out around 412.5MHz (going to 425MHz resulted in hashrate loss). The best ive seen is guys in below-zero (celcius) garages/sheds getting to 425-437.5MHz (1.4TH) before anything higher is a hashrate dropoff. in both cases actual chip temperatures dont seem to be the limiting factor as long as its below 65C so if you can get 412.5MHz at the most right now, supplying 13V might let you run at the same frequency and push out ~5% more hashrate at the expense of ~10% more power draw, which seems like a lot of work to squeeze out 70w/50gh more by running it outside of specs, not to mention that power efficiency on a 13V PSU is typically lower than a 12V-gold psu. Thanks for the detailed response and explanation behind the logic. I think my initial follow-up question was answered here: From https://www.bitmaintech.com/files/download/BM1384_Datasheet_v2.1.pdf : "...Recommended Operation Conditions Symbol Parameter Min. Typ. Max. Unit VDD Core Voltage 0.65 0.8 1.1 V ..." So ...1.1V x 15 chips = 16.5V MAX Voltage for Overclocking ...by the official Datasheet...IMHO, to be on the save side ...no more than 13.5V - 14.0V will be the MAX...that I will be brave to try...  Experiment on your own risk...!!! Now I know the max voltage I can pump into an S5 is 16.5v .... I am not going to run this anytime soon as I do not have a PSU that can push out that much voltage, but I have one that can do a tad over its standard 12v, so I may be tempted to try that in the near future ... |
|
|
|
You are correct. The web scraping isn't going to work well for non english browsers. I have done nothing to make it locale aware.  That's a task I'm not sure I can get into. What locale do you have your browser set to? I might be able to look at this, but I have a short amount of "spare" time, and it's needed elsewhere mostly. What miner is this coming from? .NET assumes the current UI culture as the culture to run the thread on, if not set otherwise, or will revert to an invariant culture (read US culture), if the code is inside a module that has no culture setting on the thread. I use a general, simplistic workaround on numeric values, i.e I try to parse them three ways, (1) by UI culture (2) by US culture (3) by a french / german culture. Here's a function that I use, it is in VB.net but if you write in C# you can convert easily. you can also substitute decimal for any numeric type ..... integer, double etc ... Dim providerUS As IFormatProvider = CultureInfo.CreateSpecificCulture("en-US")
Dim providerDE As IFormatProvider = CultureInfo.CreateSpecificCulture("de-DE")
Function parseDecimal(ByVal value As String) As Decimal
Dim trnAmt As Decimal = CDec(0.0)
If Decimal.TryParse(value, System.Globalization.NumberStyles.AllowDecimalPoint Or NumberStyles.AllowLeadingSign, CultureInfo.CurrentUICulture, trnAmt) Then
Return trnAmt
Else
If Decimal.TryParse(value, NumberStyles.AllowDecimalPoint Or NumberStyles.AllowLeadingSign, providerUS, trnAmt) OrElse Decimal.TryParse(value, NumberStyles.AllowDecimalPoint Or NumberStyles.AllowLeadingSign, providerDE, trnAmt) Then
Return trnAmt
Else
Try
Return Decimal.Parse(value, NumberStyles.AllowLeadingSign Or NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture)
Catch ex As Exception
Return CDec(0.0)
End Try
End If
End If
End FunctionEDIT: I think the issue with this one is that the culture of the rigs is not the same as the culture of the computer your program is running on, in the case of the post above, his rig is set as a US type culture while his PC is a fr-FR culture. in any case, that function would deal with that specific issue and similar (but not all scenarios). |
|
|
|
|
Show Posts
