So what exactly is a "block"?

[Lagardo]

The client tells me "such-n-such blocks left in test". But what is the meaning of a "block"? It seems to be about a percent of a full test, but that is all I understand here...

Can that be communicated to a layman?

[MAD-ness]

While I thought that I had heard it explained how blocks are measured/defined, I couldn't dig up any info directly form Louie using the search here.

However, I did find several instances of users saying that blocks were the result of tests being divided up in "250k cEMs" sections.

Whether or not this is true, I don't know. The way blocks are currently used is primarily to allow the client to report to the server that it is still working on the proth test and to get more frequent reports of progress for stats.

Again, I am not positive on this, just my understanding of the situation.

As always, hopefully someone who knows what they are talking about will be by shortly to explain things.

[Mystwalker]

That's right, 1 block = 250k cEMs. (it was 100k cEMs once)

That's why there are more and more blocks in one test when n increases and less and less time it takes to compute one block.

[MAD-ness]

Yet another faq worthy tidbit.

hehe.

[Troodon]

AFAIK a block= 250 M cEMs. Otherwise your computer would complete a block in some seconds .

[MAD-ness]

I think we may have run into the "cEMs = cEM(s)" vs "cEM(s)/second" issue again. Hard to tell which is being refered to (rate or work done) by looking at them or when people talk about them.

[Troodon]

For me, cEMs=the plural of cEM and cEMs/s or cEM/s=cEM per second.
AFAIK, the international norms for units say that a unit should always be referred in singular, so "cEMs" isn't correct.

[Lagardo]

Well, whether it's a plural or not -- I do more than 250 kilo-cem per second and thus in less than one second I should have completed a block if it were 250 kilo-cEM, so 250McEM makes more sense to me than 250kcEM.

And the metric prefixes (like kilo- and mega-) already imply a plural, namely 1000, so it shouldn't be neccessary to use a plural-s; however the use is quite mixed: "one kilometer" (1000 meters, but the plural for meter is implied) but "five kilometers" (where kilometer is now the new unit and we have a plurality of them). (cf. "kilobyte" etc.)

[Troodon]

When I said "a unit should always be referred in singular" I was meaning the symbol, not the word/name.
For exmpale: "Kilometers" is correct, but "kms" not.

[MAD-ness]

My point was that not everyone is a scientist or a mathematician and even the simplest things can confuse us.

I wasn't intending to argue standards or nomenclature with anyone, especially not people who actually know what they are talking about (everyone but me).

[Cmarc]

cEMs is indeed not correct, strictly speaking. then again I don't usually speak very strictly.
To get back to the block size, Louie posted recently a formula for calculating the cEM value based on n value and # of iterations. roughtly a test size in cEM would be about n^3/10E9 or for current tests 27GcEM. Most of my tests these days have around 120 blocks so 250McEM seems about right, considering this formula is somewhat aproximative and n value has been rounded down to the closest million.
Cheers,
Marc

[Mystwalker]

Yepp, of course 250 McEM...

But as an excuse: Louie made that mistake, too, when he wrote about the size increase of a block (100 McEM --> 250 McEM).

[MAD-ness]

Well, glad we got that figured out.