Bitcoin Forum
December 06, 2016, 08:17:24 AM *
News: To be able to use the next phase of the beta forum software, please ensure that your email address is correct/functional.
 
   Home   Help Search Donate Login Register  
Pages: « 1 2 [3]  All
  Print  
Author Topic: Best mining card?  (Read 8772 times)
AzN1337c0d3r
Full Member
***
Offline Offline

Activity: 238

★YoBit.Net★ 350+ Coins Exchange & Dice


View Profile
June 07, 2012, 11:23:51 PM
 #41

To stay on topic, I have a 5870 overclocked to 1200 mV and 950 MHz, I get ~425 Mh/s. It's great if you can get one used for cheap on eBay or Craigslist.

Holy small texts, Batman! How much are those going for nowadays? $180?

1481012244
Hero Member
*
Offline Offline

Posts: 1481012244

View Profile Personal Message (Offline)

Ignore
1481012244
Reply with quote  #2

1481012244
Report to moderator
Bitcoin mining is now a specialized and very risky industry, just like gold mining. Amateur miners are unlikely to make much money, and may even lose money. Bitcoin is much more than just mining, though!
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction. Advertise here.
Gladamas
Sr. Member
****
Offline Offline

Activity: 294


Bitcoin today is what the internet was in 1998.


View Profile
June 07, 2012, 11:29:06 PM
 #42

To stay on topic, I have a 5870 overclocked to 1200 mV and 950 MHz, I get ~425 Mh/s. It's great if you can get one used for cheap on eBay or Craigslist.

Holy small texts, Batman! How much are those going for nowadays? $180?

 Cheesy I quoted Bulanula, this was a joke: https://bitcointalk.org/index.php?topic=78386.msg937603#msg937603

1GLADMZ5tL4HkS6BAWPfJLeZJCDHAd9Fr3 - LQ6Zx8v7fHVBiDX5Lmhbp6oEDB7dUFjANu
GPG 0xF219D5BB3C467E12 - Litecoin Forum
zvs
Legendary
*
Offline Offline

Activity: 1386



View Profile WWW
June 08, 2012, 02:39:19 AM
 #43

7970 is probably the best right now, not just because it uses less power than other cards to mine, but also because its the current top-card from AMD and will retain its value longer and better than older cards like the 5870 and 5970.

If electricity cost is not an issue.. well.. I've personally hit 750+MH/s @ 320W on air with one of my 7970's. Cheesy

top-end stuff doesnt retain value for long, it drops in value the fastest.  boggle.  this is a pure $ thing, not %    (ed: tho in this case, both would apply, when comparing vs something like, say, a 5830, 5870, or 5970)

the non-enthusiast savvy buyer always buys one or two techs down

Dacentec, best deals for US dedicated servers. They regularly restock $20-$25 Opterons with 8-16GB RAM & 2x1-2TB HDD's (ofc, usually lots of other good stuff to choose from).  I did a Serverbear benchmark of one of my $20/mo Opteron (June last year), it's here.  Have had about a half dozen different servers with Dacentec, & none have failed to sustain at least 40MB/s (burst higher). My favorite is a 12-month rent-to-own ZT Systems 2XL5520 16GB 2x2TB SATA for $40/month (got lucky with the 'off-brand', haven't seen a RTO 2xL5520 for under $50/mo since -- at least for monthly contracts).  wholesaleinternet.com has some ancient 2-core intel CPUs @ $10/mo sometimes (I got an Intel Core 2 6300 @ 1.86GHz, with a 250GB HDD with 46000 hours on it, LOL. $20 @ Dacentec is much better, if you can grab one). joesdatacenter.com (same location as Wholesale Internet) also occasionally has specials (or if you don't want to wait, it has an AMD Opteron 170 @ $16/mo).
scifimike12
Sr. Member
****
Offline Offline

Activity: 304



View Profile
June 08, 2012, 02:48:45 AM
 #44

the non-enthusiast savvy buyer always buys one or two techs down

One of the main reasons I opted for 5770's when bitcoin mining took off last year.  Everyone was paying $$$ for 5870's and 5970's while 5770's remained near MSRP or lower.  5830's were also a good investment but supply was meager.

ssateneth
Legendary
*
Offline Offline

Activity: 1288



View Profile
June 08, 2012, 06:25:25 AM
 #45

My no-graphics card installed System Idle Power is 294W.

294W idle? Dude, there's something seriously wrong here ...

Look at sig dude. It's a 3930K @ 1.52V with 4 mechanical hard drives, 2 SSDs, and 50W in fans alone.

You could probably tweak that OC a little... I have the same cpu and waterblock, and honed in to 46 x 104 for 4784MHz @ 1.405v. Just cranking up the multiplier sometimes doesnt work. I was getting "A clock interrupt was not received on a secondary processor..." BSODs at 47x and higher at 100 bclk rarely, but more often with LinX/IntelBurnTest runs with 8GB memory. Tried up to 1.5v and no dice, got really uncomfortable with temps. For some reason though, it likes 46x max and allows a little bclk increasing for a net increase in cpu speed while also allowing lower vcore and a little memory speed bump to boot. Passed 20 passes of 6/12 thread LinX with 8GB memory used (8GB seems to peak gflops and wattage for me, even though I have 64GB)

Your proc 40W (very pessimistic assuming heavy overclocking).

Have you ever overclocked a 3930K to 4.6 GHz before? Most of the time you have to turn off all the power-saving features. I assure you, it does not idle at 40W. Try 100+.

I beg to differ. less than 26 watts Wink

AzN1337c0d3r
Full Member
***
Offline Offline

Activity: 238

★YoBit.Net★ 350+ Coins Exchange & Dice


View Profile
June 08, 2012, 09:58:41 AM
 #46

I beg to differ. less than 26 watts Wink

You have EIST and Turbo on. Like I said, I had to disable power-saving features (which doesnt matter much to me, my CPU rarely idles).


ssateneth
Legendary
*
Offline Offline

Activity: 1288



View Profile
June 08, 2012, 10:11:09 AM
 #47

I beg to differ. less than 26 watts Wink

You have EIST and Turbo on. Like I said, I had to disable power-saving features (which doesnt matter much to me, my CPU rarely idles).



AFAIK, I have Turbo off. IIRC, Turbo is pretty much just a stock clocks thing with a relatively low, but variable, multiplier, which would be normally 32x for the 3930k and -up to- 38x on a single core and smaller multipliers if load is on multiple cores. Mine is either 12x multiplier on no load or 46x with any load. I highly doubt your cpu "rarely idles" unless you are cpu mining or folding @ home with a cpu or something like that. Nothing really uses all 6 cores all the time unless its an ongoing project limited only by cpu speed; Even encoding live video wouldn't use that much, however encoding already recorded video to another format is another matter. Most games use, at most, 2 busy threads (and a number of other threads which don't contribute much to keeping the cpu busy).

But yes, I did enable most power saving features, because I believe any performance hit from turning them on would be humanly imperceptible.

AzN1337c0d3r
Full Member
***
Offline Offline

Activity: 238

★YoBit.Net★ 350+ Coins Exchange & Dice


View Profile
June 08, 2012, 10:46:12 AM
 #48

I beg to differ. less than 26 watts Wink

You have EIST and Turbo on. Like I said, I had to disable power-saving features (which doesnt matter much to me, my CPU rarely idles).


Nothing really uses all 6 cores all the time unless its an ongoing project limited only by cpu speed;
...
But yes, I did enable most power saving features, because I believe any performance hit from turning them on would be humanly imperceptible.

Disabling the C-power states and EIST and Turbo are about stability, not perceptibility.

My system runs a 24/7 continuous compilation and automated testing of several large software packages, hence the hex-core CPU, large amount of RAM, and lots of hard drives.

cmg5461
Sr. Member
****
Offline Offline

Activity: 374



View Profile
June 08, 2012, 11:15:10 AM
 #49

I beg to differ. less than 26 watts Wink

You have EIST and Turbo on. Like I said, I had to disable power-saving features (which doesnt matter much to me, my CPU rarely idles).


Nothing really uses all 6 cores all the time unless its an ongoing project limited only by cpu speed;
...
But yes, I did enable most power saving features, because I believe any performance hit from turning them on would be humanly imperceptible.

Disabling the C-power states and EIST and Turbo are about stability, not perceptibility.

My system runs a 24/7 continuous compilation and automated testing of several large software packages, hence the hex-core CPU, large amount of RAM, and lots of hard drives.

If your computer runs 24/7, how did you get youd "idle" power?  I mean, wouldn't it be fluctuating?

If I've helped: 1CmguJhwW4sbtSMFsyaafikJ8jhYS61quz

Sold: 5850 to lepenguin. Quick, easy and trustworthy.
AzN1337c0d3r
Full Member
***
Offline Offline

Activity: 238

★YoBit.Net★ 350+ Coins Exchange & Dice


View Profile
June 08, 2012, 11:47:58 AM
 #50

If your computer runs 24/7, how did you get youd "idle" power?  I mean, wouldn't it be fluctuating?

The obvious answer would be I killed the scripts running the compile...

mrb
Legendary
*
Offline Offline

Activity: 1120


View Profile WWW
June 10, 2012, 09:53:51 AM
 #51

How'd you do that? Can you post your core/mem/V/kill-a-watt numbers?

More than happy to:

My no-graphics card installed System Idle Power is 294W.

Using Afterburner I set my 3x 7970 system to the following:

Core Voltage: 825mV
Core Clock = 460 MHz
Memory Clock = 685 MHz
Memory Voltage = 1.5V <--- this doesnt seem to change my power consumption, so I doubt it is actually working in Afterburner. I include this here just for reference's sake.

My Kill-A-Watt reads: 380W. Minus the 294W for the system idle makes 86W for 3 cards.

Diablominer is showing 275 MH/s per card, so total of 825 MH/s.

825MH/s / 86W = 9.59 MH/J. Not exactly 10, but meh, close enough.

Your measurements are wrong.

What is happening, most likely, is that you are measuring a 294W wall consumption when the system is hung during POST because it cannot detect or initialize graphics adapters. (You don't notice it because you have no monitor.) BIOSes run busy loops on the CPU when waiting for devices to initialize (or when hung). This causes an unusually high power consumption, which invalidates your assumption that 294W is the true "idle" consumption.

The only correct way to measure power consumption of a graphics card is to use a clamp meter on a PCIe extender with isolated 12V wires, and on each PCIe power connectors:


In reality, a 7970 at 825mV and 460MHz does about 6 to 7 Mh/J depending on junction temperature, ASIC quality, and fan speed. (I measured this myself with a clamp meter under these settings.)

Theory validates this number: a stock 7970 does roughly 3 to 3.5 Mh/J. Efficiency decreases by the square of the voltage. So 825mV, down from the default 1175mV, would make the efficiency about  3/((825/1175)**2) = 6.1 Mh/J to 3.5/((825/1175)**2) = 7.1 Mh/J.

This number can also be infered by correcting your inaccurate measurements. I would venture that a busy loop on a CPU like yours (i7-3930K overclocked to hell Smiley) consumes roughly 30W. Which means your true idle power consumption is 294-30 = 264W. Your 3 cards actually use 380-264 = 116W. And 825 Mh/s / 116W = 7.1 Mh/J
AzN1337c0d3r
Full Member
***
Offline Offline

Activity: 238

★YoBit.Net★ 350+ Coins Exchange & Dice


View Profile
June 10, 2012, 12:13:23 PM
 #52

BIOSes run busy loops on the CPU when waiting for devices to initialize (or when hung). This causes an unusually high power consumption, which invalidates your assumption that 294W is the true "idle" consumption.

No, I was idled at Windows desktop for 5 minutes which I checked by logging in via Remote Desktop Connection.

Quote
In reality, a 7970 at 825mV and 460MHz does about 6 to 7 Mh/J depending on junction temperature, ASIC quality, and fan speed.

It is possible your 7970 leaks a lot more than mine. I am under water which lowers leakage and dont have a fan to power as well.

Quote
Theory validates this number: a stock 7970 does roughly 3 to 3.5 Mh/J. Efficiency decreases by the square of the voltage. So 825mV, down from the default 1175mV, would make the efficiency about  3/((825/1175)**2) = 6.1 Mh/J to 3.5/((825/1175)**2) = 7.1 Mh/J.

Current is not related to the square of voltage at 28 nm. This rough squared check is true if DYNAMIC leakage dominates your overall IC leakage. Starting at around 65 nm, ASICs tend to be dominated by the static leakage, which is governed by Shockley's Diode Equation (P ~ V * e ^ cV)

I'm not going to type my whole previous response to this misinformation so read this for more information: https://bitcointalk.org/index.php?topic=84298.msg938166#msg938166

Quote
I would venture that a busy loop on a CPU like yours (i7-3930K overclocked to hell Smiley) consumes roughly 30W.

A 3930K that's busy draws way more than 30W. Hell someone's idling 3930K earlier in this thread showed 25W... thats with power-savings features turned on.

But my CPU is not busy as I'm logged into Windows and idling at desktop, so that's moot.

mrb
Legendary
*
Offline Offline

Activity: 1120


View Profile WWW
June 10, 2012, 03:12:25 PM
 #53

No, I was idled at Windows desktop for 5 minutes which I checked by logging in via Remote Desktop Connection.

It is possible your 7970 leaks a lot more than mine. I am under water which lowers leakage and dont have a fan to power as well.

Ok. That makes sense; cool temps + no fan reduces power consumption a lot. But then your "9.59Mh/J" number is misleading IMHO as it does not include power consumption of the water pump + radiator fans which are required to run the GPUs.

Current is not related to the square of voltage at 28 nm. This rough squared check is true if DYNAMIC leakage dominates your overall IC leakage. Starting at around 65 nm, ASICs tend to be dominated by the static leakage, which is governed by Shockley's Diode Equation (P ~ V * e ^ cV)

I'm not going to type my whole previous response to this misinformation so read this for more information: https://bitcointalk.org/index.php?topic=84298.msg938166#msg938166

Interesting.


Quote
I would venture that a busy loop on a CPU like yours (i7-3930K overclocked to hell Smiley) consumes roughly 30W.

A 3930K that's busy draws way more than 30W. Hell someone's idling 3930K earlier in this thread showed 25W... thats with power-savings features turned on.

I was speaking of a busy loop consuming 30W on a single core. Multiply by 6, for the 6 cores = ~180W for the whole CPU (which is about right given the overclocking).
aqrulesms
Sr. Member
****
Offline Offline

Activity: 265


Cryptocurrency trading and mining


View Profile
June 10, 2012, 03:15:40 PM
 #54

if electricity is an issue, go FPGA

if not, 5970 is the best mhash/$.

5830 and 5850 are the best in terms of mhash/w.  7970 falls into this category too, but it is expensive in mh/$.

Although, I reccommend the 7970 if you game Cheesy (I have one and I love it Cheesy)

You can always sell the 7970s when you're done mining as well. It will continue to be a popular mining card through December because of its efficiency.
AzN1337c0d3r
Full Member
***
Offline Offline

Activity: 238

★YoBit.Net★ 350+ Coins Exchange & Dice


View Profile
June 10, 2012, 04:42:39 PM
 #55

I was speaking of a busy loop consuming 30W on a single core. Multiply by 6, for the 6 cores = ~180W for the whole CPU (which is about right given the overclocking).

180W might be right for a CPU at around 1.35-1.4V which is about as far as most people will take it but note in my sig that the CPU is @ 1.52V.

It appears that my CPU leaks a lot more than the typical 3930K although it still has the same 4.6-4.8 GHz wall everyone else does. I can assure you the load on this CPU is much much larger than 200W. I would estimate it to be 350ish.

mrb
Legendary
*
Offline Offline

Activity: 1120


View Profile WWW
June 10, 2012, 09:35:59 PM
 #56

180W might be right for a CPU at around 1.35-1.4V which is about as far as most people will take it but note in my sig that the CPU is @ 1.52V.

It appears that my CPU leaks a lot more than the typical 3930K although it still has the same 4.6-4.8 GHz wall everyone else does. I can assure you the load on this CPU is much much larger than 200W. I would estimate it to be 350ish.

That's easy to measure. Just put a clamp meter on the ATX12V connector.
AzN1337c0d3r
Full Member
***
Offline Offline

Activity: 238

★YoBit.Net★ 350+ Coins Exchange & Dice


View Profile
June 11, 2012, 04:22:18 AM
 #57

180W might be right for a CPU at around 1.35-1.4V which is about as far as most people will take it but note in my sig that the CPU is @ 1.52V.

It appears that my CPU leaks a lot more than the typical 3930K although it still has the same 4.6-4.8 GHz wall everyone else does. I can assure you the load on this CPU is much much larger than 200W. I would estimate it to be 350ish.

That's easy to measure. Just put a clamp meter on the ATX12V connector.

If I had one, I'd go do that. Since I dont, I'm going to go with what my Kill-A-Watt tells me and it tells me that just the idle-load (Prime95) delta is >250W.

Pages: « 1 2 [3]  All
  Print  
 
Jump to:  

Sponsored by , a Bitcoin-accepting VPN.
Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!