The fastest HD 69xx miner. 250 BTC.

[mrb]

Faster than Phoenix 1.3 by 6.3%[1] on HD 69xx.
Faster than ArtForz's private calminer on HD 69xx.
hdminer is a miner I developed for AMD Radeon cards using the low-level CAL interface (as opposed to OpenCL). Its target market is cluster owners who operate 20+ GPUs and are power or cooling constrained, yet need that +6.3% perf improvement without increasing power consumption or cooling needs.

• 840 Mhash/sec on a Radeon HD 6990 (BIOS switch at "overclocked" position 1; with "aticonfig --odsc=960,1260" to further overclock the GPU to 960 MHz and mem to 1260 MHz)
• 802 Mhash/sec on a Radeon HD 6990 (BIOS switch at "overclocked" position 1; with "aticonfig --odsc=915,1260" to further overclock the GPU to 915 MHz and mem to 1260 MHz)
• 746 Mhash/sec on a Radeon HD 6990 (BIOS switch at "overclocked" position 1) -- this is higher than the theoretical 723 Mhash/s I would expect for this clock because hdminer leaves some TDP headroom which allows AMD PowerTune to dynamically adjust the clock to a value averaging more than 880MHz
• 708 Mhash/sec on a Radeon HD 6990 (BIOS switch at "default" position 2) -- this speed has been measured with Catalyst 11.1. Catalyst 11.2 through 11.4 contain a performance regression affecting my compute shader that downgrades the speed to 683 Mhash/s. However because aticonfig in Catalyst 11.1 does not support the HD 6990, I advise users to initially install Catalyst 11.4 or later, run aticonfig to generate xorg.conf, then downgrade to 11.1 for operating hdminer on a day-to-day basis.
• 569 Mhash/sec on a Radeon HD 5970 (stock clock 725MHz)

On the HD 69xx series this is about 6.3%[1] faster than the fastest publicly available miner (Phoenix). (On the 5xxx series, hdminer is less optimized and appears to be currently 5% slower than Phoenix.) I am selling it for 250 BTC. Email me m.bevand@gmail.com. Check my user rating on irc/freenode: my nick is "mrb_". What you get:

- Source code (compiles for Linux)
- Supports AMD Radeon HD 5000/6000 series only (HD 4000 and earlier not supported)
- No 100% CPU busy loop! Most combinations of OpenCL miners/Catalyst drivers/SDK versions suffer from a "busy event loop" bug, but not hdminer because it is written in CAL, so does not use OpenCL. A pinky single-core AMD Sempron 140 2.7GHz is amply sufficient to handle the speed of 3 x HD 6990 for example.
- Internal SHA-256 implementation uses BIT_ALIGN and BFI_INT instructions (the first miner ever to do so; read about what is and how I added BFI_INT support to another GPGPU app I wrote: http://blog.zorinaq.com/?e=43)
- Supports as many GPUs as the Catalyst driver can support (currently AMD supports 8 GPUs on Linux, such as 4 x HD6990, or 4 x HD5970)
- Simple command line interface (see screenshot below)
- Connects to bitcoin via the standard RPC 'getwork' interface (compatible with bitcoin version 0.3.19 and up)
- Compatible with mining pools

[1] HD 6990 stock GPU 830MHz clock, stock memory clock, stock GPU VDDC, stock memory voltage: hdminer 708 Mhash/s vs. Phoenix 666 Mhash/s source: http://bitcointalk.org/index.php?topic=2949.msg106238#msg106238

Screenshot of it running on a 4 x dual-GPU HD 5970 computer (8 GPUs total):


Dead easy guide to set it up from scratch from a fresh Ubuntu 10.04 install:

Code:

These steps have been tested on a fresh Ubuntu 10.04 (lucid lynx) install.

1. Install the dependencies for compiling the Catalyst driver and hdminer:
  $ sudo apt-get install build-essential linux-headers-generic dkms libqtcore4 lib32gcc1 libglu1-mesa libxrandr2 libxinerama1 libcurl4-gnutls-dev

2. Download the Catalyst driver (eg. ati-driver-installer-11-3-x86.x86_64.run):
  http://support.amd.com/us/gpudownload/Pages/index.aspx
  And install it:
  $ sudo sh ati-driver-installer-11-3-x86.x86_64.run
  (Select all default options.)

3. Create /etc/X11/xorg.conf (and fix the wrong perms set by aticonfig):
  $ sudo aticonfig --initial -f --adapter=all
  $ sudo chmod 644 /etc/X11/xorg.conf

4. Reboot to make sure that the Catalyst kernel module fglrx.ko loads correctly
  and that X11 starts properly configured:
  $ sudo reboot

5. Download the APP SDK (eg. AMD-APP-SDK-v2.4-lnx64.tgz):
  http://developer.amd.com/gpu/AMDAPPSDK/downloads/Pages/default.aspx
  And extract it in /usr/local:
  $ sudo tar -C /usr/local -xzf AMD-APP-SDK-v2.4-lnx64.tgz

6. In hdminer, in the Makefile, edit SDK_PATH to point to the SDK:
  SDK_PATH = /usr/local/AMD-APP-SDK-v2.4-lnx64
  And compile it:
  $ make

7. Mine! (By default, hdminer connects to the bitcoin JSON-RPC endpoint on
  localhost, on the default port 8332, and authenticates with the user/password
  bitcoin/password):
  $ ./hdminer
  To specify non-default values (eg. to mine on a pool):
  $ ./hdminer -s servername -p 8332 -a user:password
  See help:
  $ ./hdminer -h

[1714733834]

1714733834

[1714733834]

1714733834

[davout]

Would you open source it for a 400 BTC bounty ?

[mrb]

Would you open source it for a 400 BTC bounty ?

Well, out of respect for previous buyers who spent their hard-earned BTC on it, no, I should not...

[jwalck]

Would you open source it for a 400 BTC bounty ?

Well, out of respect for previous buyers who spent their hard-earned BTC on it, no, I should not...

I'd rather be inefficient than proprietary. The client looks nice though! Is it using OpenCL?

To be honest, I think if you get youtipit or a web page describing your effort with the repository and a donate address, you'd make plenty too. And + on happiness and goodness in the world.:)

[jorgen]

Can you implement the miner for PS3 console?
If yes, could you make it open source and post it here for reasonable bounty?

[mrb]

I do feel that selling proprietary software (the first time for me) is somewhat contrary to my own values as everything else I write and use is open source :-)

hdminer is written in CAL IL (Compute Abstract Layer - Intermediary Language), a pseudo assembly language.

The PS3 is not good for mining. 7 usable cores (6 SPU + 1 PPU) with 128-bit registers at 3.2GHz makes it only about as fast as a cheap 6-core AMD Phenom. Both are handily beaten by low-end GPUs.

[mrb]

I changed my mind.
I am now selling the source code of hdminer as well.
Price remains the same: 400 BTC.

(However it does not mean it is open source. Again people have paid money for it, I don't think they would appreciate it to suddenly become open source.)

[hacim]

Maybe if the people who paid for it don't want it to be freely available could offer you some amount to keep it closed. Then the bidding war can begin.

Another option would be to offer a certain amount to those who purchased it to accept the opening of the source, so they don't feel bad.

In any case, those who did purchase it have benefited from having it far earlier than anyone else.

[ElectricGoat]

You could also open-source it, but not release it for free. I'm not convinced people who pay for a slightly faster miner would rush to distribute it for free if they were allowed to.

[mrb]

Actually davout's idea would work: a bounty of about 2500 BTC for me to open source hdminer would be high enough that I could refund most of the money to the buyers who already paid 400 BTC.

[jorgen]

Can you compile a windows version of your hd miner?

[mrb]

If more than 1 user is interested in Windows, I might consider compiling for Windows.

[comboy]

Please share donation address. Your blog is fuck*** awesome.

[Garrett Burgwardt]

I'd love a windows client! Would be excellent

[Mahkul]

I'd love a windows client! Would be excellent

I would love one too!

[kiba]

Please share donation address. Your blog is fuck*** awesome.

Or maybe he could put up ads space for bitcoin with http://www.operationfabulous.com

[mrb]

Ugh. I wasn't expecting that much Windows demand.
Oh well I guess I'll have to set up that Windows dev environment which I am not looking forward to

[bitjet]

so whatever happened to the super miner? poclbm seems to match these numbers.

[dingus]

With poclbm I get ~270 m/h/s with both cores at stock speeds. I am not sure this is worth 400 btc.

[bitjet]

yep. Im getting around 315 per core overclocked.

[RulerOf]

yep. Im getting around 315 per core overclocked.

Overclocked to what speeds?  I see about 300 per GPU at 810 MHz, but I'm on f=120 right now.  f=1 seems to be worse.

[sc8nt4u]

yep. Im getting around 315 per core overclocked.

Overclocked to what speeds?  I see about 300 per GPU at 810 MHz, but I'm on f=120 right now.  f=1 seems to be worse.

310M @ 825 MHz, 320 @ 850Mhz both @ -f 5.

[nster]

At STOCK speeds he's getting about 285 per core, so this miner is about 5~6% faster than poclbm. Not worth 400BTC now though

[qed]

With poclbm I get ~270 m/h/s with both cores at stock speeds. I am not sure this is worth 400 btc.

Are we talking about 569 Mhas/s insted of 540 Mhash/s garated by poclbm? It's 5.3% improvement for 400 btc...

[nster]

With poclbm I get ~270 m/h/s with both cores at stock speeds. I am not sure this is worth 400 btc.

Are we talking about 569 Mhas/s insted of 540 Mhash/s garated by poclbm? It's 5.3% improvement for 400 btc...

5.4% xD well 5.37 that round up to 5.4% if you wanna keep one decimal, but I'm just nitpicking

[qed]

With poclbm I get ~270 m/h/s with both cores at stock speeds. I am not sure this is worth 400 btc.

Are we talking about 569 Mhas/s insted of 540 Mhash/s garated by poclbm? It's 5.3% improvement for 400 btc...

5.4% xD well 5.37 that round up to 5.4% if you wanna keep one decimal, but I'm just nitpicking

This is making an huge difference having not less than 1Mhash/s as absolute error, due to the presenation of the hashrate, and there is an ~ in front of it.... Oh look, someone just lost the chance to shut up. 

[grue]

how much improvement (in %) does this software have? by the replys im seeing, it seems to be very minor, and i think the buyer should lower the price significantly. maybe 20 BTC?

[nster]

how much improvement (in %) does this software have? by the replys im seeing, it seems to be very minor, and i think the buyer should lower the price significantly. maybe 20 BTC?

You guys do realise this thread is 1 month and a half old? back when 569Mh/s for a 5970 at stock was HUGE?

[mrb]

hdminer is only ~5% faster than the best publicly available miners nowadays, so in a sense, it is not worth 400BTC if you are a small-scale miner (less than a handful of GPUs). However I plan to implement optimizations soon (have been wanting to for a while), which should bring another +5% improvement or so.

Bumping this thread because I tested it on HD 6990 (see original post for perf numbers).

[mrb]

I actually measured 746 Mhash/sec on a Radeon HD 6990 at "880MHz" (BIOS switch at position 1). This is higher than the theoretical 723 Mhash/s I would expect for this clock due to AMD PowerTune constantly dynamically adjusting the clock which averages higher than 880MHz.

[Grinder]

Very few people would benefit enough from a 5% increase to justify paying 400 BTC. I think the only viable business modell is to set up you own pool and lock your client to it. If it really is 5% faster you could take a 4-5% fee and people would still earn more than from the other pools.

[mrb]

I am happy with my current business model & price.

[Raulo]

There is no "5% increase over best publicly available miners" anymore, at least for 5970, because with m0mchil's poclbm git HEAD version, I get 569-570 Mh/s on stock 725 Mhz (-f 1 -v -w 128, Linux, Catalyst 10.12, SDK 2.1). It needn't to be HEAD, the improved OpenCL kernel was introduced in early February. There are other reports of 560-565 Mh/s for this card at stock frequency.
https://en.bitcoin.it/wiki/Mining_hardware_comparison

[nster]

If mining solo or pool, you'd break even only at 8000BTC mined, if benefit is 5%. Not even Slush mined that much.... He actually only mined HALF of that.

Now that difficulty is over 82K, this would take a long time... Even if we say in a month it will not rise, to do 8000 BTC in a month, it would take 22Gh/s. And this is being on for 24/7 no fault, and assuming the difficulty will stay stable. All this just to break EVEN = 0 PROFIT. 1 month of 24/7 mining with 22Gh/s.

That would take 30x 6990s, divided into at least 8 PC's. Each with either MONSTER PSUs or 2 BIG PSUs. 4* 6990s on 880Mhz stock would take about 1500W alone, not even including the rest of the system.

[Littleshop]

Noting that, for someone with two 5970's that better miner software would be worth maybe an increase of 20 to 30 coins a month.  So if you charged 50btc the buyer MIGHT get most of that back after two months.  Or you could have sold it for 10btc and sold many many of them. 

[Grinder]

That's the problem with selling it, there's really no way to set a good price. Running a pool is more work, but he could probably earn 100-200 BTC each day if it really is 5% faster. Slush's pool gets 2000 BTC each day. If he charged 4% or less he could probably get a much higher hash rate than Slush, because both those who already use a pool and those who currently mine solo would earn more by switching.

[nster]

That's the problem with selling it, there's really no way to set a good price. Running a pool is more work, but he could probably earn 100-200 BTC each day if it really is 5% faster. Slush's pool gets 2000 BTC each day. If he charged 4% or less he could probably get a much higher hash rate, because both those who already use a pool and those who currently mine solo would earn more by switching.

look at poclbm-mod from bitcoinpool... everyone can use it anywhere.... if he makes everyone in his pool use it, everyone could use it elsewhere...

[Grinder]

Not if he locks it down and uses a proprietary interface.

[bitcool]

Is there a comparison matrix for all miner programs, running on different OS, cards?

[nster]

Ok I get what you are saying... but will people accept the double fee? I guess they would but ya never know. And couldn't someone just leak the miner and kill that for slush? so there are risks. and would the creator accept that the largest pool shares his miner with so many people with paying more than 400 BTC? if he would then slush should do it... HOWEVER other big pools would probable DO THE SAME in which case, idk if he could even raise his fee.

So atm, lets say slush is getting 40BTC per day (2% fee)

Now if everyone uses this 5% increase miner (I'd like to note that we do not know the increase really. Someone has to try 6990s on the poclbm), The total pool would generate 5% more faster. Problem is with that, slush is a quarter to a third of the total mining power of BTC AFAIK. Meaning bitcoin as a whole will be increased by ~1.32%(if we take 160Gh/s number) , which WILL be translated in a 1.32% difficulty increase.

So while he is getting 5% more, it will be 1.32% harder to mine... ie: 105/101.32= 3.6% more as a total.

so 2000*1.036= 2072 BTC that the pool will generate. Here's the thing, if he keeps his fee at 2%, he'll be getting 1.44BTC a day more, so it would take 278 days to get back his BTC invested in this, not worth it. an extra 1 % would make him pay it off in 20 days, and 2% would need 10 days. If deepbit does not do the same for perhaps 2 weeks, then slush is fine, but if he has competition, it's not a good thing for him!


I would like to know if the creator would let the big 2 pools use it, provided they pay him 400BTC ea.

[mrb]

Currently, hdminer is not redistribuable, as specified by its license. However if a pool operator would have such special needs, I am willing to arrange a special deal.

It is true that hdminer's target market is not the individual with 1 or 2 GPUs. Many of my customers are, or at least appear to be large-scale miners. Of course, I cannot divulge any details.

[nster]

well, the pool operators do not seem interested, even at 400 BTC.

So far, about half the bitcoin hashrate is in pools. The people inside the pools are far from the 22Gh/s I mentionned earlier.

Even if all the solo miners had 22Gh/s, that would make 13 people who cold actually make money off of this

[mrb]

Yes, sounds about right. Except I started selling in January when the market of interested buyers was much larger.

[mrb]

I lowered the price from 400 BTC to 350 BTC.
hdminer is currently the fastest miner for the HD 6990 at 708 Mhash/s at stock 830 MHz clock, compared to Phoenix's 650 Mhash/s, a difference of 9%.

Future reply to the random guy who will reply "but I get 7xx Mhash/s with $MINER when overclocking" -> I said at stock clocks !

[drgr33n]

350 bitcoins !!! Man that's a lot of money 1 btc = 1.20 sterling. So you think your miners worth £420 per license for an extra what 40MH/s lol !!! I think I'll stick to my regular free miners and donate. This is just plain greedy Well my opinion anyway.

[mrb]

Fair enough 

PS: 1 Bitcoin is 1.44 GBP 

[drgr33n]

PS: 1 Bitcoin is 1.44 GBP 

lol really I was just going by britcoins converter. Be honest how many people have bought hdminer

[Loader009]

Why don't you make a donation-address and say "If I get about 1500 BTC on that, it will be open source"?
Yes it's much, but if more and more spends money on that, this program could be open source in a few months
Greetings

[ghost]

I lowered the price from 400 BTC to 350 BTC.
hdminer is currently the fastest miner for the HD 6990 at 708 Mhash/s at stock 830 MHz clock, compared to Phoenix's 650 Mhash/s, a difference of 9%.

Future reply to the random guy who will reply "but I get 7xx Mhash/s with $MINER when overclocking" -> I said at stock clocks !

Just for reference, I get 714 Mhash/s at 830 MHz clock running phoenix 1.3 with VECTORS=on AGGRESSION=10 WORKSIZE=128 BFI_INT. At 900 MHz I get 774 Mhash/s.

EDIT - Actually, that is with the overclock switch turned on. Otherwise, at stock speeds, I only get 668 MHash/s. Could you post numbers for these other scenarios?

[nster]

I lowered the price from 400 BTC to 350 BTC.
hdminer is currently the fastest miner for the HD 6990 at 708 Mhash/s at stock 830 MHz clock, compared to Phoenix's 650 Mhash/s, a difference of 9%.

Future reply to the random guy who will reply "but I get 7xx Mhash/s with $MINER when overclocking" -> I said at stock clocks !

Just for reference, I get 714 Mhash/s at 830 MHz clock running phoenix 1.3 with VECTORS=on AGGRESSION=10 WORKSIZE=128 BFI_INT. At 900 MHz I get 774 Mhash/s.

EDIT - Actually, that is with the overclock switch turned on. Otherwise, at stock speeds, I only get 668 MHash/s. Could you post numbers for these other scenarios?

His "stock clock" is with 830 clock (ie: switch turned to Overclocked)

[mrb]

Just for reference, I get 714 Mhash/s at 830 MHz clock running phoenix 1.3 with VECTORS=on AGGRESSION=10 WORKSIZE=128 BFI_INT. At 900 MHz I get 774 Mhash/s.

EDIT - Actually, that is with the overclock switch turned on. Otherwise, at stock speeds, I only get 668 MHash/s. Could you post numbers for these other scenarios?

As the first post says: with the o/c switch at position 1 (880MHz) = 746 Mhash/s. This means hdminer is faster than Phoenix by 4.5% assuming you did not change anything else (PowerTune settings, memory clock, voltages, etc). Is it the case? Some resident o/c tools automatically changes the PowerTune settings for improved performance for example, even when running at "stock" clock.

[mrb]

His "stock clock" is with 830 clock (ie: switch turned to Overclocked)

No, 830 is not overclocked.
Overclocking with the switch brings it to 880 MHz.

Damn it. Why is it that 50% of the posters in this thread make mistake when comparing o/c vs non-o/c card? This is what I am saying, all comparisons should be done at stock clocks to prevent mistakes and misinterpretations

[drgr33n]

how many bitcoins did he give you to say that hahaha Only kidding

[travex]

I'm willing to pay 35 BTC for HDminer , 350 no thanks

[error]

I lowered the price from 400 BTC to 350 BTC.
hdminer is currently the fastest miner for the HD 6990 at 708 Mhash/s at stock 830 MHz clock, compared to Phoenix's 650 Mhash/s, a difference of 9%.

Future reply to the random guy who will reply "but I get 7xx Mhash/s with $MINER when overclocking" -> I said at stock clocks !

The numbers don't work for miners with a single 6990, or even four of them. With 10, it starts to look attractive. If I had 20 6990s, I'd buy without a second thought.

That said, your price severely limits your sales volume to those who are setting up large mining operations. As an example, I would venture to guess that at 35 BTC, you would sell far more than ten times as many copies since it would then be attractive to every AMD GPU user here mining with two or more cards, and to many of them with single cards. I'm sure you've already considered how to price your miner, though I'm not sure you considered it from your potential customers' perspective.

[nster]

I lowered the price from 400 BTC to 350 BTC.
hdminer is currently the fastest miner for the HD 6990 at 708 Mhash/s at stock 830 MHz clock, compared to Phoenix's 650 Mhash/s, a difference of 9%.

Future reply to the random guy who will reply "but I get 7xx Mhash/s with $MINER when overclocking" -> I said at stock clocks !

The numbers don't work for miners with a single 6990, or even four of them. With 10, it starts to look attractive. If I had 20 6990s, I'd buy without a second thought.

That said, your price severely limits your sales volume to those who are setting up large mining operations. As an example, I would venture to guess that at 35 BTC, you would sell far more than ten times as many copies since it would then be attractive to every AMD GPU user here mining with two or more cards, and to many of them with single cards. I'm sure you've already considered how to price your miner, though I'm not sure you considered it from your potential customers' perspective.

at that volume, it may get leaked pretty quick

[Syke]

• 708 Mhash/sec on a Radeon HD 6990 (stock clock 830MHz: BIOS switch at position 2) -- this speed has been measured with Catalyst 11.1. Catalyst 11.2 and 11.3 contain a performance regression that downgrades the speed to 683 Mhash/s. However because aticonfig in Catalyst 11.1 does not support the HD 6990, I advise users to install Catalyst 11.3 or later, run aticonfig to generate xorg.conf, then downgrade to 11.1 for operating hdminer.

Very interesting downgrading Catalyst. Using Catalyst 11.3 and Phoenix I get 670 Mhash/s vs. your 683 Mhash/s. Sounds to me that hdminer is only 2% faster, and that the bulk of your performance improvements are in system configuration, and not miner optimizations.

[mrb]

I am certaintly interested in the feedback that you guys give me about pricing, thanks. Predicting the demand curve for a given price is certainly not trivial. As nster said, pricing it too low may cause it to be leaked and pirated. I am not going to divulge how many buyers have purchased hdminer so far, but I do think that 350 BTC is a right balance at this moment.

Syke: it is the first time I hear 670 Mhahs/s at 830 MHz. Did you modify other hw settings such as memory clock, memory voltage, gpu voltage?

[Syke]

I think I got 2 extra MH/s from downclocking the memory. Everything else is stock.

[nster]

I am certaintly interested in the feedback that you guys give me about pricing, thanks. Predicting the demand curve for a given price is certainly not trivial. As nster said, pricing it too low may cause it to be leaked and pirated. I am not going to divulge how many buyers have purchased hdminer so far, but I do think that 350 BTC is a right balance at this moment.

Syke: it is the first time I hear 670 Mhahs/s at 830 MHz. Did you modify other hw settings such as memory clock, memory voltage, gpu voltage?

I do however think that the price adjustment was not enough... going from 320$ to over 1000$ is silly

I reasonable price would be anything between 100 and 200 BTC. I could see it working better at 150 BTC

[mrb]

Thanks reeses for these very precise numbers, and quality of the report. (Compared to Syke, I don't know if he was rounding up to 670 Mhash/s, and he doesn't seem 100% sure of the "2 Mhash/s" impact of memory downclocking).

I assume your numbers are without memory clock or voltage changes, right? If yes, I will update my claim to 708/666 = 6.3% faster than Phoenix as I want to convey exact numbers.

[dingus]

I would be interested in a 5970 optimized miner, and I know there is such a thing  .

[Syke]

I optimized my Phoenix settings for my 6990 and installed the new 11.4 driver. At stock card settings I now get 680 MH/s, which compares very favorably with your 683 MH/s using 11.3. I have yet to try the downgrade to 11.1 trick.

[mrb]

I optimized my Phoenix settings for my 6990 and installed the new 11.4 driver. At stock card settings I now get 680 MH/s, which compares very favorably with your 683 MH/s using 11.3. I have yet to try the downgrade to 11.1 trick.

Is 680 Mhash/s rounded up by a few Mhash/s?

Are you saying that you, just now, improved your performance from 668 Mhash/s (as you claimed in your previous posts: "670 Mhash/s" minus the "2 Mhash/s from memory downclocking") up to 680 Mhash/s, this time without memory downclocking?

[mrb]

No problem.  Stock voltage, gpu, memory clock settings.

Thank you. Post updated.

[Syke]

I didn't have the optimal Phoenix settings before. And I was rounding down. Here's an actual shot:

681.06

[mrb]

Is this 681 Mhash/s with or without memory downclocking?

[Syke]

681 with stock everything. 830 MH/s, etc.

[zippa]

I have 3 ATI 5970's  with standard clock.  It is doing 1704 Mhashes/second

[nster]

the price is now OVER 2500$++ for this? I find that ridiculous. needs another price adjustment

[Sukrim]

The price is exactly the same as always, only the dollar fell against the bitcoin recently...

[mrb]

Exactly

More seriously, regardless of the BTC/USD exchange rate, my miner still produces 6.3% more Bitcoins than the 2nd fastest miner, so it makes sense for my price to be relative to the Bitcoin instead of the Dollar.

(Feels weird to say "the Bitcoin").

[mrb]

BTW, Syke, I couldn't reproduce your numbers with Phoenix. I obtained 664 Mhash/s, which is slightly lower than reeses.

[nster]

the profit margin of bitcoin mining did not going up as much as the exchange rate has, so yes, this needs a price adjustment

[Sukrim]

the profit margin of bitcoin mining did not going up as much as the exchange rate has, so yes, this needs a price adjustment

If you look at the profit per 100MH/s, the price would rather go up than down...
http://bitcoin.atspace.com/income.html

[gigitrix]

the profit margin of bitcoin mining did not going up as much as the exchange rate has, so yes, this needs a price adjustment

If you look at the profit per 100MH/s, the price would rather go up than down...
http://bitcoin.atspace.com/income.html

Awesome site, I'm apparently making a lot more than I initially thought, and the yield is majorly more stable than I first thought...

[Sukrim]

...you did NOT do these calculations yourself before starting to mine?!

[mrb]

I am slashing the price down to 250 BTC.

I measured my miner at 802 Mhash/sec on a properly overclocked Radeon HD 6990 (BIOS switch at "overclocked" position 1; with "aticonfig --odsc=915,1260" to further overclock the GPU to 915 MHz and mem to 1260 MHz). And 840 Mhash/sec at 960 MHz, although the hardware wasn't very stable at this speed.

[grue]

I am slashing the price down to 250 BTC.

your original price was 400 btc, which was less than $20(February prices). now you're selling it at 250, which is approximately $2100.

[lvlrdka22]

At stock:
Your miner numbers: 30 days 12 hours=732 hours per block
Your Phoenix numbers: 32 days 11 hours=779 hours per block
According to another guy's 6990 at stock, 680 MH/s: 31 days 18 hours=762 hours per block

People buying/using your miner will need 250 BTC=5 blocks of extra work to pay for the investment in your miner.

With your Phoenix numbers:
1/732 blocks per hour*x-5 blocks=1/779 blocks per hour*x
x=60662.55 hours=6.925 years

With 680 MH/s:
1/732 blocks per hour*x-5 blocks=1/762 blocks per hour*x
x=92964 hours=10.61 years

Congratulations, you just made a nice investment that takes more than 6 years (or 10 years) to make back, if you're going with single 6990 (assuming difficulty stays constant)!
To be able to recuperate the costs under 1 year, you'll need to be running 7x 6990s (or 11, according to another guy's 680 MH/s stock 6990).
That's not counting electricity bills.

[mrb]

To be able to recuperate the costs under 1 year, you'll need to be running 7x 6990s (or 11, according to another guy's 680 MH/s stock 6990).

Yes. As you guessed, my buyers are large miners. (I wouldn't trust that 680 Mhash/s number -- couldn't reproduce it.)

grue: $200, not $20. As I explained earlier, it makes sense for my price to be indexed on BTC, not USD.

[rowbot]

Congratulations, you just made a nice investment that takes more than 6 years (or 10 years) to make back, if you're going with single 6990 (assuming difficulty stays constant)!
To be able to recuperate the costs under 1 year, you'll need to be running 7x 6990s (or 11, according to another guy's 680 MH/s stock 6990).
That's not counting electricity bills.

Why would difficulty stay constant?  

I don't have the time to crunch the numbers, but you should assume anywhere from a 10% - 50% increase in difficulty which is compounded roughly every 2 weeks. Those kinds of diminishing returns should certainly be factored in. A 6990 may not be worth running in the near-to-distant future due to difficulty increases, thus rendering the miner moot, and perhaps making the investment a net loss.

[jgraham]

Why would difficulty stay constant?  

It shouldn't. It should go up as the block size increases.   This was part of the design to account for increases in performance.

Couple of things about this super snazzy software.   I'm using vanilla poclbm and each of my 6990 GPUs (stock clocks) seem to be registering 339 and 340 MH/s depending on when I look at them (flags are -w 64 -a 5 -f 1).   This is with 11-5 drivers and SDK 2.4 - so assuming that these have the same problem that 11-4 has this makes hdminer a less than 1% advantage.   Hard to cost justify the $4000 it would cost me today to get it.  This is assuming it gives any real advantage at all. (cue shock and awe)

Consider for a moment that all we appear to be doing here is comparing the MH/s statistics output by these two programs.  
Anybody else see how we could just as easily be seeing some kind of measurement error?  I'm not trying to be a jerk but really small gains should always be viewed with the utmost skepticism.  So if you're a fan of Ioannidis the corollary to "Most published research is false" would be "Most small gains are noise" ;-)

that +6.3% perf improvement

Aleged +6.3% improvement is a better term.   I find it amazing that people who do software development for a living still are kind of retarded when it comes to math.

Using poclbm (guiminer) and the newest catalyst (11.5) and app sdk (2.4), I am getting 711 mhash on stock settings - 830mhz, bios switch in the default position
if hdminer only gets 708 mhash at stock, wouldn't it actually be a performance regression to switch to it?

Sorry for the continued edits.  I'm still a noob by board standards.  I just tried your settings and now I'm within 2% of the hdminer's scores for clocks of 915/1260.  A very similar gap to what I came up with under stock clocks.   Your observation is correct and it's contrast with mrb's numbers illustrates a kind of a pet peeve of mine. mrb made a reasonable assumption.  Poclbm and other programs have overhead (python interpreter, OpenCL) before they can even get the code executing on the GPU.  So writing the whole app "closer to the metal" seems reasonable.  So he re-wrote it and saw an increase and....assumed that ALL of that - down to one tenth of one percent - was due to his optimizations.  In other words he is assuming that beyond that ALL VARIANCE is at 1/100 of a percent or less.   Which is the kind of thinking that strains if not breaks a sanity check.  Generally speaking it's hard to get a general case which is that accurate.

Not to mention things like Ahmdal's law which IMHO is just a way of saying that optimizations are rarely as fast as you think they are. ;-)

[Dirt Rider]

I get 810 MH/s out of a water-cooled 6990 running at 950mhz using poclbm.  For 250 BTC I can buy a nice 2 X 6990 fully water cooled rig that'll give me 1620 Mh/s.  I suspect I must be missing something here and would appreciate some enlightenment

[mrb]

Yep you are missing something.

hdminer's target market is not you the home user, but cluster owners who operate 20+ GPUs and are power or cooling constrained (eg. power circuits at 80% utilization per the US National Electric Code), yet need that +6.3% perf improvement without increasing power consumption or cooling needs.

[Luke-Jr]

Does this support free Linux (ie, no fglrx)?

[mrb]

AFAIK it is not yet possible to mine without fglrx (until someone works on this bounty).

[djex]

I get 810 MH/s out of a water-cooled 6990 running at 950mhz using poclbm.  For 250 BTC I can buy a nice 2 X 6990 fully water cooled rig that'll give me 1620 Mh/s.  I suspect I must be missing something here and would appreciate some enlightenment

My thoughts exactly.

[Syke]

To be able to recuperate the costs under 1 year, you'll need to be running 7x 6990s (or 11, according to another guy's 680 MH/s stock 6990).

Yes. As you guessed, my buyers are large miners. (I wouldn't trust that 680 Mhash/s number -- couldn't reproduce it.)

My guess is you have no interest in trying to reproduce the 680 MH/s (also verified by jgraham), because they significantly reduce your claimed benefits.

[happyland]

Using poclbm (guiminer) and the newest catalyst (11.5) and app sdk (2.4), I am getting 711 mhash on stock settings - 830mhz, bios switch in the default position
if hdminer only gets 708 mhash at stock, wouldn't it actually be a performance regression to switch to it?

[Sukrim]

Using poclbm (guiminer) and the newest catalyst (11.5) and app sdk (2.4), I am getting 711 mhash on stock settings - 830mhz, bios switch in the default position
if hdminer only gets 708 mhash at stock, wouldn't it actually be a performance regression to switch to it?

Settings (switches like -v -w 128...), memory speed, core voltage?

I would be more interested in a comparison with hashkill though...

[happyland]

memory speed and core voltage are both stock (default bios switch)
settings are -v -w128, the standard for 6xxx series

[Luke-Jr]

AFAIK it is not yet possible to mine without fglrx (until someone work on this bounty).

A while back, I recall seeing an example CAL program for free Linux...

[FlipPro]

I get 810 MH/s out of a water-cooled 6990 running at 950mhz using poclbm.  For 250 BTC I can buy a nice 2 X 6990 fully water cooled rig that'll give me 1620 Mh/s.  I suspect I must be missing something here and would appreciate some enlightenment

THIS

[Jack of Diamonds]

Do you want to make money by adjusting price, or stubbornly price a 6.3% performance increasing miner at BTC worth $2,000 dollars (or two full rigs with 2x 6990) ?

I operate ~40 GPUs in datacenter and wouldn't pay even 20BTC for this software based on current exchange ratio to fiat.

Your software is good.
Don't ruin your chances to profit from it by setting a ridiculous price or refusing to accept current economical realities; It will take years to pay off the investment even for huge clusters.

[jollyjim]

From comparing the speeds I'm getting to hdminer, the miner isn't 6+% faster, it's only 1.2-2% faster.  I'm using Catalyst 11.4 without downgrading and underclocking the memory so the difference could be even less than that (11.5 is supposed to give it a few Mh boost and not underclocking it would increase hashing rates by 0.5-1.5Mh/s).

Here are the rates (per core) I get at various speeds in the overclocked position using the phoenix 1.48 miner with the poclbm kernel:

880 - 379.8
900 - 387.0
905 - 389.6
910 - 391.9
915 - 393.4
920 - 396.0
925 - 398.2
950 - 408.5
955 - 410.8
961 - 412.6
966 - 414.6

Those numbers are near the higher end of the spectrum but not the very top.  The numbers are sustainable and not just one time blips that timing/thread switching might inflate.  At 880, hdminer was actually 1.8% slower (759 vs 746).  For the 915 and 960 settings, hdminer was faster (@915, 786 vs 802 = ~2.0%, @960, 824 vs 840 = ~1.94%).  These are using numbers that would favor hdminer more (lower hashing rates than optimal and rounding my rates down).  More optimal numbers would probably put it closer to a 1.2% difference.

Using $17 as the value of a BTC, 250 BTC = $4,250.  If the cost of a 6990 were $740, that'd be 5.74 6990s.  Assuming hdminer was 2.0% faster, you'd need over 295.6 6990s to make a profit.  So unless you're running an operation that exceeds 248GH/s, you shouldn't buy hdminer.  I'd be surprised if there are several people with that much capacity, let alone even one.  It'd be even worse if hdminer were only 1.2% faster.

[jollyjim]

hdminer's target market is not you the home user, but cluster owners who operate 20+ GPUs and are power or cooling constrained (eg. power circuits at 80% utilization per the US National Electric Code), yet need that +6.3% perf improvement without increasing power consumption or cooling needs.

I don't know what speeds they're getting at what temps but I'm able to do 2.43GH/s with 3 6990s at 1300w (10.83A) after 15 mins when the fans have been running at 100% for some time and temps are higher.  Temps are all below 80C on air cooling on all cores except for a defective one that's always 12-20C higher than the other core on the same card (I limit the overclocking on that core so temps remain in the low 80s; perhaps I only need to reapply some thermal paste on it but I'm RMAing it to avoid any warranty issues).  At times (probably when the AC kicks on), temps will be in the high 60s to low 70s.  The thermostat is in another room and set to 76F in a very hot climate (currently over 100F outside) so it'll go on and off at various times.  This is with an enclosed HAF 932 case with one 200mm fan removed and 5 120mm fans added.

What seems to makes a big difference is the PSU used.  10.83A on a 15A breaker (it's actually on a 20A breaker but I'm treating it as 15A for safety reasons) means I still have 1A to overclock it to possibly 2.50GH/s.  I've also got plenty of room to add more cooling if necessary to bring the temps down.  I'm comfortable with them running in the 70s and spiking into the low 80s for a short period of time.  I've used a cheaper PSU and could only do 2.38GH/s while pulling over 1380w.  I use two PSUs but only swapped out the main one.  Mind you, the better main PSU isn't high end at all so if you're willing to pay a hefty premium, perhaps it'll be able to do better.  The secondary PSU is on the cheaper side so that could be improved as well.  The extra fans I'm using are cheap Yate Loons and not something expensive like the jet propelling Deltas that can pull in over 2x as much air flow.  I don't even have a big external fan blowing on it.

I'm also running Linux, which means I'm more limited on the things I'm able to tweak (but with the custom tools I've made, I think I'm very close to what someone on Windows is able to do).

If your customers are getting the same rates or less than what I am with those constraints, they're spending a lot of money for things they can tweak on their own.  At 1300w for 2.43GH, that's an efficiency of almost 1.87MH/J for the entire machine.  When the time comes, I can make it go slower and be above 2.1MH/J (likely more) for the entire machine or above 2.4MH/J per card.

Also, assuming that hdminer is 6.3% faster, that'd mean they need to be doing at least 76.2GH/s to do better than spending the same amount of money on extra 6990s.  I agree with Jack of Diamonds that you're pricing yourself out.  If people do the proper research, they wouldn't be paying 250 BTC, let alone 100 BTC for something that is 1.2-2% faster. At 50 BTC and 2%, you'd want to do at least 49.7GH/s to do better than adding the same amount in extra hardware.  I also suspect that it being faster means it will probably use a little more power, though not necessarily in the same proportion as the increase.

[Fiyasko]

The price your asking is fucking rediculos. 40more mh/sec per 6990 for Two Hundred and Fifty bitcoins?
Thats Fucking Rediculos

250btc@16=$4000

[jgraham]

From comparing the speeds I'm getting to hdminer, the miner isn't 6+% faster, it's only 1.2-2% faster.  I'm using Catalyst 11.4 without downgrading and underclocking the memory so the difference could be even less than that (11.5 is supposed to give it a few Mh boost and not underclocking it would increase hashing rates by 0.5-1.5Mh/s).

Have you used the modified phatk kernel yet.   Some have been showing it to have a 2%-3% increase.  I rather suspect that eliminates the differences between this client and phoenix.   mrb even acknowledged that the recent change was one that hdminer already had.    Which since, like you I suspect that the difference is closer to 2% rather than 6% makes hdminer obsolete.

[jollyjim]

Have you used the modified phatk kernel yet.   Some have been showing it to have a 2%-3% increase.  I rather suspect that eliminates the differences between this client and phoenix.   mrb even acknowledged that the recent change was one that hdminer already had.    Which since, like you I suspect that the difference is closer to 2% rather than 6% makes hdminer obsolete.

I get better hash rates with the poclbm kernel on the 6990.  I tested a few of the optimizations people were trying out recently but those made no difference or made things worse.  On the 5970, the modified phatk is much better.

However with the recent Ma changes and some other tweaks, I get better rates than the ones shown for hdminer.  At 915, I get 817.6 compared to hdminer's 802.  At 960, I get 857.8 compared to hdminer's 840.  It doesn't necessarily mean that hdminer is slower since I don't have hdminer to try out but it does show that you can get better numbers than the published hdminer ones and anyone thinking of getting it should think twice about doing so.  Perhaps there'd be another 2-3% increase if the tricks hdminer uses were used as well, as the Ma one was only about half of the 6% difference.

Upgrading to Catalyst 11.5 did increase hash rates a bit over 11.4.  11.7 is supposed to be even faster but it's still beta and I heard it was buggy.

[Syke]

Last and final bump... Using the latest kernel at http://forum.bitcoin.org/index.php?topic=25860.0 at stock 6990 settings I get 352 mhps vs hdminer's 354. Hdminer is obsolete. End of thread.

[film2240]

Where can I download this miner? As I wish to boost my performance as well on my HD6950 card.is this free? If not what's the charge? Why charge for this app?I'm sure that someone with the know how can probably replicate this app and then it'll be pointless to charge for this if successfull.Here's an idea,can someone with the know how, create a free version of this app with similar performance characteristics please?

Thank you

[mrb]

It is not free. I still charge for hdminer. And I am still faster than today's best on HD 69xx. I benchmarked cgminer w/phatk at 691 Mh/s on a stock 6990, compared to my 708 Mh/s.

However my speed advantage is now so thin, that I doubt anyone is interested in spending 250 BTC on hdminer... The rich feature set of cgminer probably outweighs any performance advantage I have anyway.

[mb300sd]

It is not free. I still charge for hdminer. And I am still faster than today's best on HD 69xx. I benchmarked cgminer w/phatk at 691 Mh/s on a stock 6990, compared to my 708 Mh/s.

However my speed advantage is now so thin, that I doubt anyone is interested in spending 250 BTC on hdminer... The rich feature set of cgminer probably outweighs any performance advantage I have anyway.

Since theres not much of an advantage over phatk anymore, why not open source it and allow others to have a try at optimizing it even further? I'd love to try it just to get rid of the 100% cpu bug.

[film2240]

It is not free. I still charge for hdminer. And I am still faster than today's best on HD 69xx. I benchmarked cgminer w/phatk at 691 Mh/s on a stock 6990, compared to my 708 Mh/s.

However my speed advantage is now so thin, that I doubt anyone is interested in spending 250 BTC on hdminer... The rich feature set of cgminer probably outweighs any performance advantage I have anyway.

So you're the guy who thinks it's smart to charge for things? Even though the 'speed boost' is so thin that it should either be free or expect a backlash like being pressured and forced to pay for a TV licence in the UK.I for one will refuse to pay until it is freely accessible like the TV licence.I don't watch TV yet I'm forced and threatened to buy one.

I don't mind being charged about 5BTC but 250BTC is taking the mick.If you want my respect,you'll rectify the pricing or give it for free if you can't find a reasonable price people are willing to pay. That's my 2 cents :-)

How well will this perform on my HD6950 card (shader unlocked,Oc'd to 902MHz,320MHz mem clock,overvolted by an extra 0.005mV)?I get 402MHash/s atm.I want to know what kind of speeds I can get with your app.

[mrb]

I will eventually open source hdminer, but not until I test a few more optimization ideas. The high price tag is mostly because I can't afford to have its source code leaked today. On your 6950, in its current shape, you can expect a speed +2.4% faster than cgminer/phatk.

[Syke]

It is not free. I still charge for hdminer. And I am still faster than today's best on HD 69xx. I benchmarked cgminer w/phatk at 691 Mh/s on a stock 6990, compared to my 708 Mh/s.

You clearly don't know how to run and optimize other miners. I get 704 on stock 6990s using Phoenix/Phatk. If you'd like to send me a copy of your miner I'll post an independant performance comparison.

[terrytibbs]

It is not free. I still charge for hdminer. And I am still faster than today's best on HD 69xx. I benchmarked cgminer w/phatk at 691 Mh/s on a stock 6990, compared to my 708 Mh/s.

You clearly don't know how to run and optimize other miners. I get 704 on stock 6990s using Phoenix/Phatk. If you'd like to send me a copy of your miner I'll post an independant performance comparison.

Independant? What makes you think the hash-rate increases/decreases based on whether the person sitting in front of the computer is biased or not?

[Syke]

Independant? What makes you think the hash-rate increases/decreases based on whether the person sitting in front of the computer is biased or not?

Because the hash-rate is dependent on many variables including OS, SDK version, Driver version, and numerous miner versions and settings. I get about 2% higher performance with phatk than he is claiming, so something isn't quite right.

[terrytibbs]

Independant? What makes you think the hash-rate increases/decreases based on whether the person sitting in front of the computer is biased or not?

Because the hash-rate is dependent on many variables including OS, SDK version, Driver version, and numerous miner versions and settings. I get about 2% higher performance with phatk than he is claiming, so something isn't quite right.

Fair enough.

mrb, clear things up for our friend Syke over here, will 'ya?

[mrb]

Perhaps Syke is taking cgminer's best 5s peak averages instead of the all-time average? Perhaps not and he is right. I don't care enough to argue about 2.4% to send him a free copy of my miner... *shrugs*

[mrb]

Some people might look at the thread title and think "what is an altcoin miner doing on this subforum"! But it is just a blast from the past... 6 years later, just for kicks, I put the source on github as a historical artifact: https://github.com/mbevand/hdminer

[okg]

Did you keep all them Coins? 

[vm1990]

Did you keep all them Coins? 

if only we knew then how much BTC would be worth, i doubt anyone would have spent there coins

[scox]

Please, just stop. Stop beating the damn decomposed carcass... It's dead alright!

[Saitekman12345]

Holy crap. I wish I took bitcoin seriously back then.