Bitcoin Forum

If the BTC community buys me a 7970, I will benchmark and tune DiabloMiner and cgminer for it. No one has released valid benchmark numbers, and I'm tired of the bullshit, and I expect everyone else is too.

Send BTC to 1DbeWKCxnVCt3sRaSAmZLoboqr8pVyFzP1 and help make this happen.

http://media.bestofmicro.com/Y/O/320064/original/itmining.png

If you want you can send me the Bitcoins you collected. :)

Since the 79xx series does not grow on trees, I shall be donating a few coins to help the cause.

I read that article, and they didn't seem to know shit about bitcoin mining.

Read the article, they say shit like "bitmining" and "server wouldn't verify so we had to solo mine". I have no clue what they were smoking, but all I wanted was answers like which miner they used, what args they used with it, and etc etc.

tl;dr: fucking mass media, how does it work

Edit: Also, their testing methodology is bullshit. From my calculations, a 5870 should be around 390 mhash at stock.

Fuck the mass media, let's collectively purchase a 500EUR graphics card to some internet stranger so that he can reap it's benefits.

its* (auto-correction)

lolwut
This is the guy that wrote DiabloMiner, remember?

I'm sorry, I meant internet-supercelebrity. What's your point again?

Such rage.

The 7970 doesn't produce valid shares with any miner out right now.

..............
Your asking the community to buy you a brandnew Top of the line (most likely not even out on the market yet) video card.

"for legit 7970 test/bench(s)"

Dude your a moderator, what the hell is this

My 5870 with stock speeds (875/1225) and ATI 11.12 ~375MHash

Or contact the manufacturer with your credentials ask them for a card for evaluation and benchmarking purposes, go through their verification process and receive one shipped directly from them. Though your reviews and results into a coherent and consistent blog, get several hundred thousand subscribers and get similar cards from lots of manufacturers, along with mobo's, psu's and all sorts of other goodies for evaluation.

It worked for Tom's Hardware, and is a very traditional path to working with pre-release evaluation copies of products.

@LoupGaroux
Yep, this is a good idea :)

This thread is weird.   No offence, but no I'm not going to help buy you a free video card.    Anyone could have made this thread and said the same thing (and I've never used DiabloMiner so that doesn't particularly sway me).

I'm quite happy to wait for release when we'll get a million reviews from people who buy the cards on this forum anyway.

Have you mined Bitcoin on your GPU before?  Did you write your own mining software from scratch, or did you use one that a skilled developer wrote for you?  Did you pay them?

Yes.  Anyone could have.  Anyone can write a GPU miner.  My grandma has written two of them.  Blindfolded.

Yes.  The card will be released, people will buy them, and then all mining software will magically just start working with the new architecture.  Honest.


This type of mentality pisses me off.  If you head over to a place like xda-developers (Android phone developers forums), you'll find people who regularly kick in a few bucks to a skilled developer to get them a phone because they'd love to see a decent ROM made for their device.  A skilled developer makes a similar post on the Bitcoin forums and people just shit on him.  If you're currently mining and you've made a fair donation to the dev who wrote your miner, I'll gladly eat crow.  Otherwise, do the community a favor and stop posting selfish crap like this.

I use cgminer which has a built in donation option that I've chosen to enable.  The Dev gets a regular percentage of all my mining activity.  Eat away.

I guess I'm not in such a desperate rush to get results for this card when its just going to happen naturally anyway.    Diablo is just some random on the inernet to me, but if others are desperate enough to buy him a new card, then good for him I suppose.

Happy Christmas.

just donated 0.000001 to the cause..... hope it helps, i think the transaction fee was over 1000 times that but there's only so much to go around right?

Don't forget that to get this results the voltage is over 250W vs the 5870 190w.

Doesn't matter since their benchmark is bogus. None of the numbers are right, and they apparently couldn't even get the miner to produce valid shares.

Well as hopefull one could be with the new Tech from AMD, results like the above make me think that one should pickup some of these babies.

http://www.memoryexpress.com/Products/MX32649 at $669 each seems tempting lol.

Yuck.  Get a used 5970 for half that price (or less).

Yeah, but what if this thing can do 700-800 mhash/sec overclocked?

I'm sure Tom's hardware just pulled 414.7 MH/s right out of their ass!

Yes, the same place they magically pulled their tuned-for-79xx miner from.

I never go to Tom's Hardware, but if the 7970 can do 500MH/s or more I'll never ever go there.

Heh, after they got caught taking bribes from Intel to make AMD look bad back in the K6/K7 days, they appeared on my permanent shit list. I do not trust a damned thing they say, ever.

BULL fcking shiet!

Show me how the 5870 stock MH/s is wrong? Within a tolerance error, 376 was spot on

Dont act like your miner is a magic.

Hey dumbass, Toms was sold and owned by different ppl now. You dont know jack enough to have any credit here.

I agree with the general consensus in this thread.

I'm definitely not going to contribute to buying you a free video card just so we can get some early stats when we can just wait for people to buy them themselves and upload their stats for free.

I think I'll put my contribution towards a video card of my own, thanks.

Who cares whether you coded DiabloMiner or not? You get donations for that separately. Why not put those towards buying the card?

I think this is a good opportunity for everyone that codes miners. Do not publicly release kernels optimized for GCN. Time to monetize the effort.

So how are they going to upload stats? Using an untuned kernel on a miner that does not work on 79xx at this time? Who is going to fix the miners so they work? Without a 79xx, it surely is not going to be me.

Also, how many donations do you actually think I've gotten? Its about $30 USD total over the past year.

The 5870 hash rate looks about right.  But how did they get 392 MH/s out of a 6970 without overclocking it, given the fact that it has even less stream processors than a 5870?

Get a job?

If my miner is magic, then so is cgminer. I get roughly 384 on my 5850 at 930, so that should be about 390 on a 5870 at stock 850. I don't see what the issue is.

It would be interesting to see how well the 7970 can mine but IMHO until some months after release we will never really know. We need optimized kernels and OC.

One person can't bring all that I think. Maybe make a common fund and buy a card so that each dev gets some time with it etc. ?

It took basically three people to make GPU mining what it is now. You seriously underestimate the power of a single developer.

I'd like to contribute my time/effort to testing a HD7970 if I could find a way to get my hands on one via hte common fund perhaps? or sort out funding another way (jobs,selling stuff maybe).I'll have to upgrade PSU as well as HD7970 needs more power than the PSu can handle.I'm thinking a partial subsidy is best rather than fully subsidising other people if the 'common fund' is used to make the most of the 'common fund'.I do like the idea as I need a piece of hardware to review for my blog as well.(the HD7970 could fit the bill nicely)

Wrd, if anything that 7970 should go to ArtForz first, as far as development is concerned.

Sure, but he seems to have quit development. That, and he has already left GPU mining and has funded his own ASIC production thanks to his gigantic GPU farm, so I'm pretty sure he could afford a few dozen 7970s if he so wanted them.

Con will fix cgminer as its his baby.  He gets plenty of donations as plenty of people use his miner.

I support the idea of passing around the card to other developers for their own development, benchmarking, and testing. Perhaps they could all be asked nicely to always leave it at stock clocks.

Then, once the 79xx series is sold out due to demand, sell the card used and distribute the profits  ;D

i think I may just wait for reliable results to be published.I then hope to republish them onto my blog so a hardware review can be done without too much trouble (due credit from the original review will of course be given) if he/she oks me republishing it.

I wonder how review sites can get a free GPU for testing purposes as that would help with the hardware review on my blog?

shiiipiiinnnggg cooossssttttssss

+1. Once again, why should we pay for you to get a new video card? Kernels will be released which are optimized for the 7xxx series whether you code them or not. I'm sure you could buy one with your own money if you really wanted to support the community.

I think it's a good idea not to deliver kernels optimized for GCN at all. Since there appear to be no more than 7-8 people here that can code that, it won't be that hard :D

What makes you so fanatically sure of this? I'm sure the coders capable of this realize the potential benefits of an architecture-optimized kernel, and it isn't unfathomable that they would keep it to themselves.

For one thing, the benefits more than likely won't be great enough to warrant keeping it to themselves.

because we are all beging for a 7970 now. I would also like one on someone elses dime. All donations to buy me a 7970 should sent to Theymos with the message "This is a donation for the forum" attached so I know what to reply with.

First thing is that _NO_ miner would work on GCN without modifications. The reason is simple: the BFI_INT replacement routine. GCN is a completely different architecture and so is the opcode. So no miner would produce correct results (if they produce any results anyway).

This needs to be changed.

Then GCN architecture puts an end to the need to vectorize. Any current kernel would be inefficient on GCN without a rewrite.

In fact without a rewrite, you are not getting slower, unoptimized version of the miner for the GCN hardware. You are not getting a working miner at all.

Every miner I know has an option to disable BFI_INT patch otherwise they wouldn't work for Nvidia cards.

So while results may not be optimized the 7970 would work just fine "out of the box" with most (all?) current miners.

Nice, that would explain why experimental results are just a bit better than 6970.

One more reason why optimized code for GCN should not be released in public :)

Detecting VLIW5/4 and only running bfi_int on those is simple. Plus, 2.6 is supposed to fix the bfi_int bug, although I have not tested it, but I've been told it should actually work correctly with bitselect().

GCN does not end the need for manual vectorization, it just makes the helical wound uint2 vectorization we use now less optimal. Most likely DiabloMiner already has an (more) optimal vectorization already available, but without a GCN card myself, theres no way to find out.

So if you truly believe you have a better moustrap- invest in yourself and buy a card, and sell the resulting improvements as a new commercialized version of your miner. The world will beat a path to your door and provide you with a revenue stream to keep up support and future improvement on your product.

Or, as suggested before, if you want an evaluation copy for legitimate evaluation and benchmarking purposes- provide suitable credentials and documentation to the manufacturers and they will ship you a card for testing and developmental work. In fact, you could probably get on the bleeding edge and receive red cards for testing of release candidates. You might even have something valuable to suggest to them in the testing process that would make their card better.

Acting petulant about getting the community to buy you one is probably not the best strategy. With the card in the wild it will be a matter of days before one of the enormously talented coders tweaking in the community fine-tunes the code anyway.

It does not (tried personally) - bitselect is still not mapped to BFI_INT. Although it is now exposed at IL level at last.

Heh, still, we won't need it on 79xx, the existing BFI_INT code can stay.

What makes you think you won't need it?

No bfi_int instruction on that arch, apparently. Unless there is a useful replacement, we use the normal code path.

You are wrong about that. There is a BFI intruction in the GCN architecture - v_bfi_b32. Strange though, there is no "s_" equivalent on the scalar unit. But it is there. I have occasionally seen that in my Tahiti kernels and you can also grep it from libaticaldd.so.

Now bitselect() does not map to v_bfi_b32 for sure, yet some patterns of the kind (a&b)|(~a&c) where part of the variables are constants do produce bfi code.

I cannot patch the binary of course because the architecture is different and the opcodes are different as well. Trying to patch the binary the same way I do for VLIW kernels does nothing. I guess until the 79xx ISA reference comes out, using bfi on GCN hardware would not be possible. Unless they finally map bitselect to bfi of course.

I'll have to keep that in mind, but I hope they fix the driver and/or the SDK to emit BFI for bitselect(). It would make a lot of people's lives easier, and speed up existing unoptimized code.

http://forums.amd.com/devforum/messageview.cfm?catid=390&threadid=143488&highlight_key=y

That URL implies BFI will be output for bitselect() on SDK 2.6. I wonder what happened

Yay, we are all waiting for more than a year :) I constantly have problems with BFI patching on my kernels and besides that, there are some rare cases that amd_bytealign would do a good job for me but it's clobbered in a way due to that patching, grr!

But truth is it did not appear in SDK 2.6. I am not acquainted with AMD's OpenCL on Windows, but with linux since recently they started shipping the opencl library as part of the Catalyst suite rather than part of the SDK package. That means with each new driver you get new surprises, hopefully pleasant ones :)

However both 11.11 and 11.12 do not map bitselect to BFI_INT. Neither does the "preview" OpenCL 1.2 driver you can download from their site. It's annoying to say at least :(

<Krusty the clown voice>I heartily endorse this event or product.</>
http://www.youtube.com/watch?v=lsiuxim8vsM

Donation current count: 7.01002 out of 150

+1

I do aswell, Really do, But it is indeed with the Krusty The Clown style

Don't understand why everyone is shitting on Diablo.

He. Created. A. Freaking. Miner. That is not a small achievement.

The fact the he was nice enough to release it for free the first time is AMAZING. There is every incentive in the world to keep it to himself, as once the code got into the wild, everyone else started mining on GPUs. He could have kept it to himself and had his single GPU mine at incredible speeds compared to everyone else's CPU.

And everyone shitting on him now should know that when he does "get a job" and buy one for himself, he'll probably sit on the code personally if it ends up being a great improvement in hash rate. And you will be the reason.

It honestly astonishes me that he only received about $30 in bitcoins for developing the diablominer. Hell, I personally sent like $20 worth of btc to the guiminer developer. People are ungrateful.

No people need an incentive.  Many in the OpenSource community fail to understand that.  Bounties are effective tool.  Giving something away for free and expecting retroactive compensation is not.

The bounty project I (and others) started to add RPC support to cgminer paid out over 125 Bitcoins for a relatively (compared to say a complete miner) simple addition. 

Maybe Diablo should look at it as a business proposal rather than charity. 

1) Raise funds to purchase 7970.
2) Calculate time value of coding optimizations for 7970
3) Start a bounty funds to collect funds in exchange for optimized 7970 code.
4) When bounty is paid distribute funds to himself and those who fronted the capital.
5) Release code

Bounties are not, and never have been, a viable way of FOSS donations. People constantly renege on bounty promises, and often had no intention of paying up to begin with. And I don't know about you, but I don't have $500 in cash laying around. I doubt most people do.

Is it worse than doing work for free, then begging for donations after the fact and getting bitter than you earned less than minimum wage?

OTC ratings can be used for bounty promises. 
OTC can be used for bounty promises.  Still payout rate for the RPC project was over 87% just based on promise to pay.

You are asking for donations why not ask for investors? (i.e. step #1 in the quote you quoted).

Because that makes no sense. I believe I personally have invested a lot of time and energy into Bitcoin, and people serve to make a lot of money if GCN turns out to be the next ultimate mining arch after 5xxx+SDK 2.1. There are people out there who make 150btc a day and use DiabloMiner, yet I have not heard a peep from them.

I don't get why I am getting so much resistance on this. People have requested optimized GCN support, and I obviously cannot provide this without having a GCN myself, nor can I afford one out of pocket. If people want this support, they will donate towards it. For them, this IS an investment.

Well you obviously got this figured out which is why you were well compensated for your prior work and almost at your donation goal.  I will stop talking about things which don't make sense.  :)

The people are outraged that someone would want a free 7970, but early testing and tuning the Diablominer to it would earn the community a ton of bitcoin in the future if he could make a 7xxx purr like a 5xxx.  I think it's unlikely but he's right, y'know.  Pay the man, he'll pay you later when you buy twenty 7970s from the store for yourself (you greedy little bastards).

I don't get what the huge outrage is about.
Author of FOSS miner offers to work on improved 7970 kernel if he gets enough donations towards buying a 7970.
That's pretty much the same way you get your personal pet hardware supported by FOSS firmware. Guess next you'll be boycotting dd-wrt and the likes.

Since I'm out of n00b jail, I'll make my first free man post since my return to this place.
I cannot believe the attitude here.

It's absolutely disturbing.

There's going to be god damn flying 12 monkeys up in the streets in 12 months with food riots and you people are bitching about helping out someone whose work benefits everybody?
Don't give me that investor bullshit. Lead a path to his door? Where do you see these magic droves of customers coming from? If you haven't been paying attention 9/11, the bank failouts, the Breivik shit, and the prospect of either a war banger or a nasty war banger as president, these tragedies of the past, present, and future have sucked the optimism out of most people. We are all mostly burnt out and could use a little spontaneous vitality.

Holy fucking shit.

Just for that I'm donating the .00017760 someone sent me once to DiabloD3. Sorry man, I hardly have any, but I thought I'd pay you the same compliment someone once paid me under my previous account (I no longer have access to that email) a long time ago.

Also, for anyone here still fantasizing about the economy and telling people to get jobs: Are you hiring? If not stop projecting your forced optimism from your own insecurity on someone else.

Also, cocks.

Welcome to the internet, enjoy your stay.

LOL. Yeah. Thanks.

You buy me a Ferrari, I'll work some on making a miner for the car radio, that's for sure!

15M1Qb128LtaaGG2YLV8zpjb3BwsuNi6S7

Because that comparison was definitely apples to apples.

Thanks a lot. Its the thought that matters.

Total is now: 7.0101976

Total is now: 8.0101976

Thats the thing; I doubt anyone is planning on buying a farm of 7970s for bitcoin mining. The card is just too expensive and its untapped performance potential almost certainly quite limited. So almost no one has an incentive to pay for a 7970 optimized miner.  Even more so as  Diablo seems no longer the miner app of choice.

Im pretty sure it will be easier to amass donations to, say, port cgminer to that butterfly labs fpga board. Anyone planning on buying those in any quantity may find that a worthwhile donation. If I where diablo, Id offer to do that.

Rasengan added 10 btc to the effort. Thanks man.

Current total: 18.0101976.

You may be right about most people, but I do. But before I give it up I'd be looking to get a reasonable return on my "invested" capital. I also understand that my capital could "magically" disappear. That's the risk. Once the code is developed I don't want the card, it wouldn't do me any good. I'd like my capital and a fair return back so I can move on to the next investment. I don't have enough cash on hand to be an angel investor.

Why don't I want the card after the development is complete? It's because I don't have the hardware and cheap enough power to run a GPU mining rig with a high enough hash rate that can pay for itself in some reasonable amount of time given the current difficulty and payout. Given my base this does not benefit me directly. I'm too late to the Bitcoin mining scene to earn a reasonable return.

This work is going to benefit the community of miners who already have extensive, and hopefully profitable, GPU mining operations and who are looking to upgrade to improve their hash rate enough to offset the costs of buying the new GPU and the power consumption. In fact those are the people I'd be approaching first. Scope all the pools and see which miners have the highest hash rates and see if they are willing to fund you.

- Zed

Woohoo, my first post as a freed noob.

That is some premo 'first post out of noob jail'

Spot on.  However I believe Luke Jr has already put together some sort of cgminer based BFL software...

I'm not sure when "toms hardware" became an authority on bitcoin mining but the fact is a 5870 gets 440 mhash easily, consumes much less wattage than a 7970 and cost about $150

I just don't see the plausibility in anyone making any kind of major mining operation based on these cards , there's absolutely no profitability and if we use the direction AMD has taken with their drivers as any kind of indicator they are certainly not making an effort to purposely gear their cards towards mining.

Diablod3 is certainly a well known early adopter and programmer but personally I would much rather donate to a project like a miner that can squeeze more mash out of an fpga board

To me It's almost like investing in oil rigs when there are some kick ass electric cars out there.

Even if he doesn't create a miracle driver, he could add a few handy lines of code that make it into the eventual best miner created by another author.  And the total cost of the hardware isn't too much for the community as a whole.

Only if you buy it used and overclock the shit out of it. He is talking about stock numbers here.

no no no

not overclocking the "shit out of them"

Im talking clocks at 950/180 - with 440mhash  no big deal at all. the 5870 is a superior mining card to the 7000 series we have seen so far

the fact of the matter is AMD is NOT designing their cards with mining in mind and thats a fact we just have to accept.

a 7000 series mining farm will NEVER be profitable unless of course you get free electricity or some other bizarre circumstance

TOMS REVIEWED CARDS @ STOCK.

This isn't a hard concept.  You can't compare one card @ stock to another card overclocked.  950MHz is an 11% overclock.

Look I also think the 7970 is a dud but to compare apples to apples you need to look at stock vs stock.  So throwing out overclocked numbers is worthless.   It looks like the 7970 can overclock 20%+ at stock voltage. 

your right and your wrong

it depends on the context of why we are comparing

if we are comparing what cards can mine faster at stock clocks then yes the 7xxx wins (according to toms hardware and his dubious unproved claims)

but if we are comparing what is going to be a more profitable card than 5870 wins hands down

so apples to apples - and oranges to oranges - we all need to realize 7xxxx series will never be profitable

and we need to take into consideration that BTC blocks will be getting halved soon

and we need to realize that FPGA is the most realistic next step - for sensible and profitable miners

If you count the 5870 at 950mhz then you should count the 7970 at 1125mhz (at which it should do close to ~500mhash judging by the numbers). And that's with an unoptimized miner for a barely available card that was released, like, five days ago, smack in between christmas and new years eve celebration seasons.

Assuming that, with an optimized miner, the 7970 can do the same speed PER SHADER as the 5870, then the card could do ~520mhash on stock and ~620mhash overclocked. This is without looking into further enhancements made possible by the new architecture itself, if there are any.

The thing is, there is, what, about a thousand GPU miners who just own a single GPU and alternate between mining and gaming? If even just every other one of them donated just 1 btc to this, I wouldn't have to worry about development costs ever again. One person doesn't need to fund this, a hundred could. 150 btc is a lot to one person, but its not much to a thousand people.

I agree with Toms Hardware not being an authority, however, they used stock cards. What damns them is a stock 5870 can get 394 mhash/sec, theirs wasn't even close. They did not perform a useful test, and they didn't even say which miner with which settings.

Also, FPGAs do not seem to be entirely worth it. There is a high buy in cost and a very low resale value. Unless your power is extremely expensive, I don't see the point of owning one unless you're going to buy a 100 or more of them at a time.

Most people who mine own a single GPU, and this sits in their desktop and they sometimes game. This is where the vast majority of hash power comes from, and DiabloMiner serves people like that.

Or I find an absurdly easy optimization that puts 600-700 mhash/sec on stock settings.

Neither is mine, nor any 5870 Ive seen here when adjusted for overclock:
https://en.bitcoin.it/wiki/Mining_hardware_comparison

To be clear, stock speed of a 5870 is 850 MHz.

You can get slightly higher speed than what THG got out of a 5870, but it will require, among other things downgrading drivers and OpenCL. They seem to have used something newer even than 11.12 . Experimenting with older drivers on an older card to provide a comparison point for a new card is not something you can reasonably expect them to do.

Its also not reasonable to expect the 7970 will gain more speed than a 5870 using older drivers (it probably doesnt even work), or when using a better miner than what they used (Im guessing they used GUIminer)

As such their numbers are completely reasonable for an apples to apples comparison. There might be room for improvement for either card, but its just not likely that it will change the basic equation.

I use SDK 2.1, which is what is recommended to use on 5xxx hardware. If you want strict oranges to oranges numbers, 2.6 is really really bad on 5xxx. On my 5850@960 on 11.12, under 2.1 I get 400, 2.4/2.5 I get 387, under 2.6 using the same settings as 2.1 and 2.4/2.5 I get 357, and under adjusted settings for 2.6 I get 365.

As usual, YMMV.

So THG is getting substantial better results than you, since its clear they are not using the old SDK 2.1. That might be because of the drivers and SDK they used are better for mining than the available (and terrible) 11.12/2.6, but the point still stands. Their results are entirely reasonable and a fair point of comparison. If anything it shows the 7970 is a bigger dud than you might think looking at the chart, because you can use older drivers to boost 5870 performance, you probably cant use them for the 7970.

7970 requires at least 11.12 and at least SDK 2.6. It looks like they used 11.11 + SDK 2.5 or similar for these tests (maybe pre-release 11.12 + 2.6?). Also, I'm pretty sure 2.6-fail is a bug and nowhere near the intended behavior, so I'm going to assume their 375 is really supposed to be 380, and their 392 on the 6970 is supposed to be 397. This would mean, completely unoptimized, the 7970 is 420.

Which, really, isn't all that horrendous. 1.68 mhash/watt vs 6970s' 1.58 mhash/watt vs 5870s' 2.02 mhash/watt (or vs 2.09 on SDK 2.1). And thats unoptimized, I'm really thinking 7970s at stock can push 500 mhash which pushes us back up to 2 mhash/watt which brings us back to the glory age of 58xx cards.

The problem with 58xx is *we can't buy them anymore*. Used cards off Ebay is iffy at best, and even the $599 price of a 7970 isn't all that much when you realize 5870 and 6970s were in the same ballpark ($499 and up). If I don't get enough donations for a 7970, I can always buy a 7950, and if I can't get that, I can always wait for the inevitable price drop to drop it down even more.

50 btc at today's prices is still $200. 7950s may eventually drop to that, but I'm not sure if everyone wants to wait that long for support. The donation serves purely as a "does the community want this or not" thing.

Spiccioli just threw in a BTC. Thanks man.

Current total: 19.0101976.

That is hardly a problem.  The prices are so discounted <$150 for 5870 and $300 for 5970 that hell if one in 5 fails you still come out way ahead.  Ebay has a 30 day buyer protection.  Of 13 5970s I bought on ebay.  1 I returned (no issue) because its second core seemed very unstable in first two weeks.  The other 12 are running fine 9 months later all purchased at a >50% discount to new price.  The failure rate after 30 days seems to relatively low (note it doesn't need to be 0% because cards are massively discounted).  So far I had one card "fail" when a fan went bad.  Some cleaning, and replacement fan and it is back in operation.

Comparing the price of a card released 3 years ago is dubious.  Moore's law marches on.  The issue in price isn't nominal $ but MH/$.  So say optimized you could get 500 MH/s from a 7970 for $550.  It still would be garbage compared to used 5970 pushing 730+ MH/s for $300.

thogar threw in 1 btc.

Current total: 20.0101976.

Releasing a GCN-optimized miner would be a HUGE mistake.

Really, efforts should monetize. Just stop it before it's too late - you are turning the bitcoin mining community into an even more fucked up version of the SL3 one for yet another generation of GPU chips. Be responsible :)

what?

Just another raving lunatic.

I have added a much easier way of donating to this. I have added a donation mode to DiabloMiner, download the most recent binary and use the -b flag.

problem is.... this card cannot be bought anywhere yet.

Yet, but soon. We already know its going to be between $400 and $600. Donations do not happen over night.

i will be pretty upset if the 7xxx comes out only marginally better for mining. Because i just sold both 5970 at $300 a pop on ebay getting ready for the 7xxx series. Miner is down right now waiting for the new card to release.

you...

...you sold 5970 based on a rumor and news on 79xx?

???

Why would you do that?  The discussion has been for months that GCN architecture may not be mining friendly.  

Your sold 5970s deliver 1.4 to 1.5GH.  You do realize that it is a pipe dream to think the 7970 will be ANYWHERE near that.  Worse at launch the card is likely going to sell at $549 and even then availability may be difficult. 

Disablo may be able to pull some respectable performance out of the card (eventually) but your sale was downright foolish.  If the 7970 could someday deliver 500 MH that would be decent.  600 MH+ would be amazing.  It isn't going to deliver 1.5GH.  You can't optimize something out of nothing.

I'll have to agree here. Unless I hit it out of the park and find, like, some hidden single cycle "hash ALL the coins" function, I'd still use a 5970 over a 7970 especially since they're two different products... what you want is a 7990 (the dual 7970 board, matching the 6990 and the 5970), which will have a retail price of approximately $9001.

He did not say 7970.  He could wait for a 7890 or something that uses VLIW4 which are better at mining. 

Well a couple problems w/ that.

1) 7890 or 7870 or 7850 will all be relatively expensive at launch.
2) Launch for the 78xx series isn't until Feb or March and that might be pushed back further.  What is the oppertunity cost of 2x 5970 for 4 months?
3) There is no guarantee that 7870 will use VLIW4 or be any better at mining.  Later reports indicate the 78xx will be GCN also and only the entry level OEM trash (76xx) will be die shrinked VLIW4.  

Like I said all this has been extensively discussed for last month or so.  


Now IF the 78xx series has good performance and IF someone like diablo can tweak mining kernel enough to unlock that performance AND if the cards launch on time and IF AMD is aggressive in pricing then MAYBE the 7850 will be a decent mining cards ... in 3 months.  To sell hands down the best mining card for a hope is .... silly.

78xx is most likely going to be the existing 69xx arch fabbed at 28nm (since it was originally meant for 28nm anyhow). 69xx still doesn't beat SDK 2.1-able VLIW5. That isn't a good trade unless you can get a 6970-like card for under $150 (and then he can go buy four of them).

well my reasoning was the new gfx card will be released in the next week or two, causing the older models to drop in price significantly as this is a full refresh not just an upgrade.  So if 7xxx isnt all that hyped up to be for bitcoins, i should be able to score some of the older cards at much lower price otherwise i will just get a pair of 7xxx. The one thing i didnt want to do was holding onto an aging pair of 5970 that will be two generations old.  

If it was some cheaper $150 card i would probably just kept it, but at $300 a pop they can still lose quite a bit of value quickly.

I'm pledging 5btc to Diablo.  Will either send it tonight or tomorrow when I get home (if I forget (I have kids - they steal your memory) Diablo has permission to publicly call me out until I pay up. :)  )

6 x 7670 should get 1.2 ghsh/motherboard, might be worthwhile.

If you have a lot of space.  :)

3x 5970 gets 2.2 to 2.3GH/ rig and are pretty cheap to assemble (no extenders needed w/ right MB).

Depending on what that series ends up being, sure. AMD needs to stop with the chimera families. I liked it better when every member of the family was some variant of the main arch.

According to this thread https://bitcointalk.org/index.php?topic=56630.0 someone is cracking 568 MH/s on a mildly overclocked 7970 already.

For those wondering, DiabloMiner already only runs BFI_INT on cards that can do it: it will not do it on 79xx (at least, not until we can change the kernel binary, the instruction is different on GCN).

I was right then, I think. I said 7970 should exceed 500 stock properly tuned, and that guy is doing it with a minor overclock.

well I wouldn't call a 20%+ overclock "mild".  :)

Well, what do you call my 5850, stock 725, at 960? ;)

Well good news, looks like with 2.6, bitselect() is mapped to the corresponding bfi instruction on GCN archs. But not on VLIW ones. That is nice. So the bfi problem is solved.

A small donation sent. I hope the 7970 card is worth it! =)

This is extremely good news. Less ugly code is always better.

+5 from tonto
+0.11 from burger

Thanks guys.

Current total: 25.1201976.

20BTC sold for around $4.75
5BTC sold for around $6.50

where's my credit for .00001 bitcoin?

maybe i got the decimals wrong..... sorry =/

There is two for 0.000001, both are in the current total.

Thanks a lot, though. Every btc counts.

You could have gotten more than 4.75 for the BTC! 

Well, yes, we know that now. Hindsight is 20/20, and no one expected BTC to hit $7 out of nowhere.

sent you a satoshi  8)

Current total: 25.1201977

650mhash/sec now

Nope, 666 now.

LOL. We are doomed. Why all the signs of the antichrist over in the BTC world ?

$6.66 prices
666 hashrates

What is the meaning ? ;D

How about you get to share that 7970 with others through a pooled fund or something so we can get most optimizations done as a community ?

Because its too difficult to physically ship the card. There also is maybe three people in the world that would benefit from this, and one of them quit working on GPU miners altogether, and the other can probably afford to buy one straight out. Seriously, there simply is not that many people in the world that understand mining well.

Well their up for sell in Canada

http://www.memoryexpress.com/Products/MX37298

Newegg has them for $550 now.

OP - How about a virtual contribution?  I have a 7970.. just got it today..  I might be willing to give you a linux box with full root access with 1 7970 installed for optimization purposes..  Is this something you would / have considered?

teek

+1

that's a good idea!

BREAKING NEWS: I now have all the donations needed to grab that 7970, I should be ordering it within the next week give or take.

The last donation was big enough to help me cover that gap, and I think this wonderful person wishes to remain anonymous due to the size of the donation.

Sweet! Looking forward to what GCN can achieve  :D

Gee, wish I got free stuff.

I'm pretty sure Diablo has put hundreds of hours into his mining code that most people use without giving him anything.  If you consider a person's time free, then yeah... I guess it's free.  If you are at all intelligent though, you would realize Diablo is someone who deserves donations.

Great, I look forward to a new kernel for cgminer... though I expect I'll end up doing that myself based on your work  :P

Spend hundreds of unpaid hours messing with internals of GPU programming (I guess at this point DiabloD3 knows parts of the ATI hardware better than the AMD engineers). Write one of the most used miners and a mining core that's used in other very popular miners as well. Also be supportive of it - answer questions, test bugs, do fixes and continue improving it until you squeeze another 1% performance, and another 1%, and another 1%. All this makes money for the miners using it and you're not getting paid.

Then you will get "free" stuff.

No one pays ME in gum :'(
 ;D :D ;D :D ;D :D ;D :D
Looking forward to see what you come up with Diablo  ;)

Not better than the engineers, not even close. The problem is, theres stuff that just isn't documented on how Radeons work, and most of what I know is either ArtForz or I threw shit at the Radeon to see what would stick, and then mixed in some moon dust from there.

The GCN, though? Lemons.

I wish I had stuck with assembly programming.. I totally rocked assembler on my Apple ][e  :)
 
But alas, life took me a different direction.. I'm thinking had I stuck with it, I could be squeezing some more performance out of a miner by redoing it in assembly. :)

The CPU miners are written heavily in assembly in some areas just to get maximum speed.

DiabloD3

You should take a look on this results: https://bitcointalk.org/index.php?topic=57410.0

They are really impressive and there will be an improvement on scripts and everything.

A photo of 3 7970s in their naked and shiny glory? Let me save that to my porn folder....

This(!):

Efficiency-wise, it's awesome!  But $1/MH sucks :(

Agreed. There are better mining solutions out there. No one will be swarming to get 7970s for mining unless someone pulls a rabbit out of that GCN architecture.

I recommend you guys click the link to my thread and not rely on the quoted info from it as I keep updating the results.

Curious about how much room for optimization there is, I made some calculations to see how close Diablominer is to the maximum theoretical performance of the 7970 (see here (https://bitcointalk.org/index.php?topic=58511.msg696537#msg696537)) and unless I'm missing something in my assumptions I found it to already be within ~2% of the maximum theoretical performance.

Edit: Fixed link.

I finally had the time to install a pair of Gigabyte 7970s last night.

I tried a few of the settings in the other thread and also found the best I could get out of cgminer was around 475Mh/s per GPU.

I downloaded diablo miner, oc'd the cores to 1100mhz on stock voltage and dropped the memory down to 600mhz.  Started up and bam, 1300Mh/s.   It was very late and I had no time to do any tinkering, but 650Mh/s per gpu out of the box on stock voltage with a nice oc is pretty awesome.

That'd be depressing if all I can get is another 2%. I'm trying to get the damned Kernel Analyzer to work in either Win in VM or in Wine, both is a no go, has to have a real running copy of the drivers, and the native Linux version locks up soon as DM calls an a CL function.

Damnit AMD

Your link had a link to https://bitcointalk.org/index?topic=7964.msg550288#msg550288 which server dies on.  Can you find a link to original 3375 ops calculation.

Add .php after index.

Thanks, I fixed the link.

Yeah, it kinda takes the fun out of optimizing if all you can get is 2%.

Any idea if there's a GCN feature that we can exploit for more performance like fixed function hardware or new instructions/amd-specific extensions? I initially thought that using size 16 vectors would help, thinking that the GCN SIMD cores were akin to x64 SIMD registers, but it seems that the card exploits its SIMD cores by running a scalar instruction on 16 threads at once.


Don't even bother until the release a new version because Kernel Analyzer won't even correctly list the kernel stats for tahiti. Running the profiler via command line does give some useful output however, and I have .il, .cl, .isa  and a comma separated value file of a profile run with -v 1 from my 7970 if you're interested.

I'm more interested in it for generic 2.6 optimization. I'm hoping I can get 2.6 running as fast as 2.5 on my 5850 and _IF_ I can get it running as fast as 2.1, I can abandon 2.1 support (but I don't see how 2.6 could get that fast)

I find it funny how so many people criticised giving you a 7970 to do diablominer tuning to and now you have delivered a miner that like EVERYONE is using for their own gain.

Yeah, and now cgminer also has the kernel among the ones it supports. I think I delivered on the promise as best as possible, although I'm still looking for new ways of beating the compiler.

To be fair, AMD's GCN compiler (not the same as the 5xxx/68xx/69xx compiler(s), seems to be ground up brand new) is doing a pretty good job, I haven't been able to find any useful optimizations after I did that two week brutal death march of tuning for 2.6.

I think all in all, the community benefited from their investment.