Topic: Noise from the "Process-invariant hardware metric" thread. (Read 1072 times)

Actions speak louder than words. You can say you are not, that does always not make it so.

You just got here, so I don't expect you to have any knowledge of BFL's past indiscretions. Don't make generalizations about subjects for which you neither possess knowledge of nor did research for and I won't point out that you are mistaken. BFL was anything but transparent before April 2013.

Avalon does have a few stuck orders in batches 1 & 2, and those look as if they can't be resolved (payment dispute, bad address, shipping issues) with engineering. Avalon is very late with Batch 3 and that is very disappointing (although from a mining point of view, I stand to gain from that). ASICminer is actually the most successful product imo, not Avalon and not BFL. There is just very little to criticize about them other than their prices. But as the only market entrant that can ship an order immediately, they can set the price.

Avalon hasn't had any problems with yield that have come to light. We will have much better insight into that when their chip orders ship.

Ok, one thing i'd like to make absolutely clear. I am not a BFL supporter. I am a BFL customer and I am not a BFL hater. Their issues particularly surrounding communication have been bad for everyone, and I like many customers wish I had my devices some time ago. I'm pleased delivery is finally happening though.

There is a reason I generalised in my comment, through use of the word basically. I did not do exhaustive research on the topic, however I did point out a clear bias in your postings of being completely anti BFL and apparently pro Avalon (feel free to point out where you haven't been). There have been some Batch 1 and 2 orders which are still unfulfilled for various reasons too, yet I see them getting a lot less grief than BFL, which despite their poor track record has been pretty transparent in everything else but timelines (two more weeks).

As for chips, I'm well aware of the comparison aspects with the chips, in terms of complexity and cost of production, power density, etc. I suppose one last bit to throw in on a comparison might be reliability and fault tolerance. In a BFL chip if an engine doesn't work properly for some reason it can be disabled and the chip will continue to function, albeit at a lower output rate. Could someone comment on the fault tolerance of Avalon chips? I haven't had a chance to read their info on their chips...

Fool! The thread is about chips not devices!
Try taking Josh's privates out of your mouth so you can see what you type!

Why am I surprised that a BFL supporter did a cursory inspection of the facts and then misrepresented them here?
Here are 4 threads that I dumped a few posts each into in the last 4 days.
https://bitcointalksearch.org/topic/m.2632649
https://bitcointalksearch.org/topic/m.2632481
https://bitcointalksearch.org/topic/m.2623786
https://bitcointalksearch.org/topic/m.2622547
Yes there has been a lot of BFL flying around lately, but if BFL fanbois could do math I wouldn't have to spend 50 posts explaining how bitcoin profit calculators work to mouth breathers like minternj. If they could do math I imagine there would be a lot less waste in general. When Avalon lies to their customers for 6 months, I will jump on their case too.

Yep. ASICMiner and Avalon nailed this one. That is why small, cheap, and fast to market can beat complicated, expensive, and slow to market. That doesn't show in a glossy brochure, but some light is shed on it with a metric like this one.

These belong to the bang for the buck category, and all things being equal they drive the buying decisions for mining hardware. However, when balanced against availability, bang for the buck does not always win. Especially when there is risk the company will not deliver late or at all.

True. But this thread is probably not aimed at the average customer, so if you are average you can safely ignore it. The process size matters because that is a driver of power efficiency and cost. The number of engines matters because that increases the size, cost and complexity of the chip and makes it more difficult to cool because of increased power density. One is not "better" than the other, they are simply different.

Seems like time for the thread to wander off to the off-topic section like the last few have, its always the same time and time again. Out of curiosity, I went and had a look at k9's post history, and I was actually surprised to see basically nothing but BFL hate and praise for avalon. Yet the BFL hate crew go around calling everyone BFL shills?

Crazy....

As far as chips go, there are a bunch of factors to consider:
Production scalability - how fast you can make heaps of them
Production cost - per unit or thousand units usually
Power expenditure - how much each chip eats in watts at full capability
Hashing output - how many GH/s the chip can put out

The average customer doesn't really care about much else... Not how many engines are on a chip, nor the die size or even how many nm were used on the fabrication process. They just want to see how much power goes in and how many hashes come out, and how easy and costly it is to get them.

Moving along now...

Troll harder.

Others might actually want to see progress on the metric. Why don't just stop posting and go away.

Fool - the Avalon requires 240 chips to make ~70GH/s
The BFL SC Single requires 16 chips to make ~60GH/s

Try putting your brain in gear before typing.

Ok i think this is the lowest you can go i think. You sir have earned your idiot title from me. Please get out your head out of your ass and stop lying yourself. How can the number of chips delivered be not relevant at all? Are you really implying that for the end customer the hash-meters per second is a lot more important than the actual availability of the chips? Wow! Just wow!

You said all that already ........ saying it twice doesn't suddenly make it true.

Lulz the number of chips delivered is not relevant at all.
You could say the number of GH/s delivered or you could say the number of devices delivered, but the number of chips is completely irrelevant.
So if someone comes up with a device that needs 10,000 chips to do 10GH/s and delivers 100 of them, they're the best? LOLOL
Avalon uses 240 ASICs per 3 board device ...

Bigot? False.

As for the typical comment of people regarding my interest in BFL, as I have already said many times before, their ability to deliver is indeed crap.
But they have delivered now.
Their chip, on the other hand, is indeed better than anything else anyone has so far released, which, as I have said before, is my reason for liking their hardware.

If someone releases something better, then BFL won't have the best chips. Until then, they do.

Of course we could (as I have already said) go to the next step and look at the implementations.
The Avalon really does suck so badly there also. The chips constantly loop through the work provided to them, so e.g. if you don't provide work to it in the time frame of a nonce range, it will repeat doing the work it has already done. The MCU protocol is ... non existent. Even the damn USB chips have blank iManufacturer and iProduct fields so you can't tell it's an Avalon without trying to mine on it. The FPGA firmware in it is badly buggy also. But ...... what would I know I simply wrote the original cgminer usbutils library we use to talk to it ...

The Avalon is a smaller and cheaper chip that got to market earlier and earned its user base a ton of cash and earned Avalon a $5 Million chip order.
BFL is a larger and more complicated chip with some yield and design issues. Yes, they put 16 engines on the chip instead of 1 but couldn't manage to run them at a higher clock rate than Avalon even though they are at 65nm instead of 110nm. The only reason BFL uses a bit less power is because they are at 65nm instead of 110nm, not because of some awesome design.

Again, the number shows your hated Avalon compares with BFL so you would never concede there is anything to it.


We are talking about chips. Avalon has delivered far more chips than BFL, and Avalon has way more chips on order still. That is not off topic, that is directly on topic. You actually refuse to speak of the design and just say over and over "BFL BETTAR!". Then again, you don't know anything specific about either chip so from your point of view BFL is better because you got it free of charge before anyone else. You are still earning your Jalapeno today in this thread.

Yep, you are still a bigot. Blinded by nerd rage and in love with Josh.
https://bitcointalksearch.org/topic/m.1511889

Wrong again. Is English your fifth language? Your reading comprehension is terrible.

Read this thread too!
https://bitcointalksearch.org/topic/m.1511889
Kano hates Avalon. Kano gets free BFL device. Kano shills for BFL on these forums.
No great mystery there.

LOL.
You post that point 1) saying I suck off Josh and then complain when I ignore it and also complain when I don't ignore it.
You're a waste of time and indeed a nobody.

And yes feel free to read all that thread ... it is indeed quite clear that Avalon lied about quite a few things including their excuse for not releasing the GPL cgminer source code. The git dates clearly show that. GitSyncom just thought he could lie as he felt like and that the git log wouldn't show that he was lying. Avalon also do not give anything but crap software support for their devices. They charged ckolivas $5000 for an Avalon rather than giving him one, and now after he's had one for a while, he's got them hashing at over 80GH/s. Yes look at what has happened before reading that thread ... then think about that while reading it ... anyone got a Batch #3 yet? 8 Avalons a day since January ... even if you try week days only ...

Now you resort to Ad Hominem. You do this because you have already been wrecked on this subject and hope I don't bring it up again.

Kano vs Bitsyncom
https://bitcointalksearch.org/topic/kano-vs-bitsyncom-142083
Now everyone can read why you hate Avalon so much and why you go out of your way to post negative things about them.

My comments are on topic because you used this thread to attack Avalon (yet again) and the OP's metric just because it shows Avalon in a posititve light.
My comments are not false and anyone can read your tantrums about Avalon.

No, just that you are a nobody here and there is no point responding to false off topic comments from a nobody.

Glad you finally concede point 1.

You keep using that word "available", I do not think it means what you think it means. I can understand your confusion, you got your unit free and before everyone else so you wouldn't understand how for 90% of BFL's order book their chip is not "available".


I guess you are afraid the metric shows if Avalon ported their design to 65nm it would probably beat out BFL. But Avalon doesn't need to do that, BFL is already beaten because they cannot deliver product. Avalon has to worry about KNC and Bitfury if they are going to continue in the ASIC mining marketplace. Or maybe they made enough money already and won't bother with a second generation product.

And yes even though they missed their specs they are still better specs than anything else available ... lol

Correct, this drivel makes the crappy Avalon chip compare favourably.
Exactly why it is drivel.

I didn't say the metric was good. I said you were an idiot for not understanding what the metric was. The OP explained in great detail what the metric was.

As for the rest of your drivel:
1) You really hate Avalon because they wouldn't give you a freebie. Nobody should trust a word out of your mouth (when Josh's privates are not in it) about Avalon.
2) You shouldn't be praising BFL's chip because they barely got it working at all. They missed their clock rate AND power targets AND delivery date by a country mile. But they gave you a free unit before their customers go them so we are inflicted with your sycophancy.
3) BFL maybe more efficient in hashes "per chip" but that is a useless metric. BFL uses 3.5 times the die size in a process that gives 4 times the density for 14 times as much logic. .275 x 14 = 3.850. That is why Avalon compares favorably to BFL using the reported specs.

Clap clap clap... "The metric is useless except it seems to apparently show the low tech chips pretending to look better than the low tech crap they are." You seem to get what the metric could show or in your mind SHOULD show... nice piece of deduction there. Funny how you want to just cherry pick what supports your favorite flavor of chip though. Which is pretty transparent in most of the threads in which you post. Your use of adjectives about Avalon certainly is telling and puts you into a category of people who are clearly disinterested in facts and more interested embellishment. That is great for story telling but holds little weight in analysis of pretty straight forward tech and design. I have yet to see any chip PRETEND it was anything other than a chip... nice turn of phrase personifying the chip and Avalon as a company but invariably in creating metrics like this it is unnecessary you would have to agree? We want a measure or metric not a story filled with words from the Grade 3 thesaurus you seem use for your prose.

Clearly this a metric in progress and there is probably more that can be done to define it so it can shed light on different designs. Seeking out more information and being academically honest about it is what we need. It be nice for once to come to thread like this one and have people drop the puffery for BFL or AVALON chips etc and take an honest appraisal of the tech.

I am pretty certain the intention is not to create a metric to bash BFL or anyone else. It was done with academic integrity. Why must people lower the level of debate and obfuscate the benefits of such a metric?


I applaud the effort eldentyrell. Can the metric say be bench marked against say past CPU chips to see how it compares existing tech development over the years and possibly show it is a reliable metric? That way we can pretty much shut down the hype-filled and adjective wielding fanboys on both sides, but particularly Kano as he seems to really take offense to anything that shines negatively on BFL for some unknown reason, and get to the meat "the quality of the underlying designs."

Too bad you can't read, otherwise you would understand the metric being used.
Or can't you see the screen with your head up Josh's... 

So ...  I come back again and find ...

The Avalon chip that hashes slower than the chip used in the old BFL FPGA, and uses at least 1.5 times the power of a BFL SC (per MH/s) and requires ~15 times the number of chips compared to a BFL SC (per MH/s) and a box somewhere between 5 and 10 times of a BFL SC Single ... rates:


The Avalon above the BFL SC

Not only that, but the BFL SC is pure custom ASIC, whereas the Avalon seems more and more each day to be a quick a dirty hack implementation.

Again these numbers are irrelevant to anyone but someone who wants to name a new number and pretend it's important.