Topic: Common Mining Cards Comparison Chart (Read 7992 times)

I don't mess with litecoins, but there is a great deal on a 7850 right now:

http://www.newegg.com/Product/Product.aspx?Item=N82E16814131482

Promo Code AFFW85VGA

with a $30 mail-in rebate, makes it $100 with free shipping

i have a powercolor 7950 that i used to use to mine bitcoins (the "one fan" design), it could run at 1250/160, temp never went above 70oC

that is the crappy fan design one..  Basically, that card will run about 20 degrees hotter than the ones with either the fan on the end which is twice as many RPM or the dual and tri fan designs.

I'd say its worth about 190 as low as you have to keep the core to keep it below 90degrees..  just an educated opinion for anyone considering.

As it stands, I understand some people still mine BTC on GPUs, and that some people already own, or have recently bought the USB eruptors...but simple match shows that they'll NEVER pay off the investment for either mining BTC alone, so, in a thread discussing GPU mining, I assume most will be talking about alternate currencies...I know this is bitcointalk.org - but there's plenty of threads and even whole subforums dedicated to talking about alt coins on here, so I didn't think that would be out of place.

depreciation on 7xxx cards should be a lot less than the eruptors as well.  i don't suspect crazy people will be around paying $50 for them still in a few weeks

btw, 7950 for $203 AR @

http://www.superbiiz.com/detail.php?name=MSI-7950B3&c=CJ

ed:  $190 now, @ TigerDirect

http://www.tigerdirect.com/applications/SearchTools/item-details.asp?EdpNo=8078201&CatId=7387

I forgot that I was posting on litecointalk.org

AFAIK the eruptors, like any other ASIC, are SHA256-only, and being that he measured his machine in 'khash/s', he's doing an sCrypt coin - meaning, your USB eruptors are useless for 10% the cost.

I hope you are using it for gaming also since you could buy USB Eruptors for 10% of the cost of that computer and do the same amount of work.

Yeah, that would be a probably fantastically accurate reading...for YOUR setup. The problem is even if you know that with your PSU, motherboard, & card/cooling setup, you can run @ 315 KH/s on LTC for 86 W per card, for a 24hr average of 3.6628 KH/s/W ...if I live in a cold climate, or have a slightly more efficient PSU, or if my driver setup is slightly worse or slightly better, or if my motherboard is mismanaging my power or whatever, let's say I do the same test and I get 307 KH/s/W @ 91 W per card for a total of 3.3736 KH/s/W... or I could get 320 KH/s @ 80 W for 4(!!!) KH/s/W.

Once you have your own rig up and running and have done your own measurements you can surely get super-accurate numbers on your own rig. But this chart is mainly for helping compare cards and setups that you don't actually have hands-on with, so that you can try to determine possible average performance and effectiveness of different setups before purchasing.

Don't get me wrong though - I'd be super-interested in your numbers and your settings.


No real harm done. Sorry for the somewhat harsh response. Just been real busy lately and it was frustrating to get that sort of reply when I spent a good portion of the small amount of free-time I have right now to do that.

Yeah, the spreadsheet is linked in the OP, so if you need to update pricing, or even change to a different currency or something, you can definitely do so.

Sorry I didn't mean to be rude. I just pointing out the fact that the data are mostly useful for you and the people in your area.

The good point is that we can update the data ourselves so the most important part for us is the worksheet calculation and presentation in itself.

amazon warehouse deals has a bunch of 7970's @ $250 right now

i guess i could take a kill-a-watt to that computer, leave it for about an hour w/ just one gfx card and idling....  then put in the other cards, start them all mining, leave it there for 24 hrs..  then that reading should be 99% accurate.  i guess if you have a motherboard that has an onboard gpu, it'd be a bit less watts (re: i'd have to have one 7790 in there for my computer to work.  i don't imagine they use much idling though)

MH/s/W can swing pretty quick. For instance, in your case, if you card is between 310 and 315, and between 80-90 W, that's a range of 3.4-3.9 MH/s/W.

I wonder if there's any better way to gauge this? Maybe something bigger? Like perhaps MH per Kilowat Hour? I don't know.

i don't recall exactly since they're in another room, but i think i run my 7790's at 1200-125 and that's about 310-315mhash for bitcoins.  they were pulling about 85 watts each, so similar numbers

i have a powercolor 7950, it can run stable at 1250-160 and gets about 640mhash.  you can get one for $230 off of tigerdirect right now with free shipping (http://www.tigerdirect.com/applications/SearchTools/item-details.asp?EdpNo=8078201&Sku=M452-7952)...  or open box $190 from newegg http://www.newegg.com/Product/Product.aspx?Item=N82E16814131478R... or new from newegg @ $220 http://www.newegg.com/Product/Product.aspx?Item=N82E16814127737

not that i'd be buying any cards for mining anymore, just throwing some numbers out there

haha, nm, open box one is out of stock now

there was an open box 7970 for $255 too, but i bought one and posted it on slickdeals and now it's out of stock 

Yeah, definitely. I'm all for examining the numbers and trying to dig up more info and I wasn't really referring to you. It just seems like some people's comments are less about that and more about telling me I've done it wrong.

Hey, relax... I think its great, but I was just pointing out some of the numbers look very strange.  I thought there was a near linear hashrate to shaders formula in play, but after I suggested the 78XX numbers looked very odd in a few places, someone said they could hit one of them.  I don't want to nitpick, but I just would love to hear if people can really hit over 400 with that number of shaders... it seems not to make sense and I'd love to see more confirmations.

I don't know how much more explicit I can be here, people. If you want to come in here and nitpick every little thing to death with the "I get 5 less W on that card" "I get 50 less KH/s on that card." "That card only goes up to XXX MH/s" "That card's less expensive/more expensive in my area", I guess that's cool...

This sheet isn't supposed to be a be-all end-all of 100% accurate-in-every-case exhaustive listing of every possible value for every possible card, with every possible setup, in every possible region on Earth.

I mean, is that ACTUALLY what you were expecting? Like honestly? You though I'd come up with a tool that would automagically discern every possible value for your specific case and explicitly tell you your exact input and out for every possible hardware configuration AND tell you which would be best for you?

And you thought I did this in my spare time, for free, and released it to you for no charge?

And I came up with this in only a couple days, then added several requested features, and still charged nothing?

I mean, if that's what you're expecting, I really can't help you. I'm sorry.

I just hope the for the people who WEREN'T expecting some sort of techno-miracle, this was at least somewhat helpful.

It is interesting but it is all wrong if you consider the fact that you use the best Mhash performance to get the best Mhash per watt.

To get the best Mhash per watt you have to experiment and test yourself different voltage/clock/memory and depending of the brand manufacturer you get different result.

Every card can improve the Mhash per watt but what's the point of paying 150$ more per card for a little improving ?

So the only concrete point of comparison is the price you pay for the card at your location.
I say at your location because for me that live in Europe, none of the price match.

The special 7790 driver nowadays isn't needed anymore, you can use just the actual driver. I'm using A10-5800K APU, HD 7950 and HD 7790 together without problems.

Using the Bitminter Client (Java Based) I get about 310.6 MHash per second. I get about 2% variance form machine to machine -- this is average.

I am sitting at the least efficient machine -- the systems draws an extra 95 watts when the client is running at 310 Mhash. The boards cost me about $145 plus the 11% tax (federal plus provincial)

A 7990 for comparison could run up to 900 MHash (or faster) but will draw about 390 Watts and will cost me about $1100 + tax.


A 7950 will be about 510-605 MHash and 185 - 370 Watts depending on overclock settings. Price -- $330 to $350.

Pricing is from the Toronto Canada area fwiw in Canadian Dollars (Loonies)

hth

so what khash/s does your 7790 dualx achieve?

Yeah, that's definitely the important subtext here, there's a lots os variables to consider and of course your mileage may vary. Even if this is indicative of 'TYPICAL' results, your actual figures on your own rig could be way more or way less. H/s/W is going to change a lot if you're under or over-volting, if you're under or over-clocking memory or GPU, it's all variable. This chart won't tell you EVERYTHING you need for anything, but hopefully it gives you a head-start in the right direction.