29 December 2017

Spinning Wheel [update]

It's deja vu all over again. Anandtech reports on a "new" Seagate HDD with heads split between two arm sets. The comments don't think much of the idea. Nor do I. If only because it's a lousy implementation of an ancient innovation. My lower brain stem memory told me that mainframe DASD routinely had multiple arm sets, way back when.

So, off to the innterTubes for some documentation. A few minutes found this paper, from 2008.
A later work [45] explored the possibility of having multiple arms that are capable of moving independently, and the IBM 3380, which was a 4-actuator drive released in 1980 for the IBM System/370, embodied this feature.
[this is the note: [45] A. J. Smith. On the effectiveness of buffered and multiple arm disks. In Proceedings of the
International Symposium Computer architecture (ISCA), pages 242-248, 1978.]

It's not unimportant that the IBM mainframes use CKD storage protocol at the time such drives were in use (now emulated on hard-formated PC HDD).

[update]
And, here's the Godzilla of HDD. I knew there was such a beast. With CKD, head per track has lots o benefit.
Read/write heads were fixed in position over each track. That eliminated seek time and contributed substantially to system performance. Data could be written at rates up to 3 million bytes per second.

That rate was in 1970.

I Told You So - 29 December 2017

Many's the time I've asserted that Orange Julius Caesar is just a creeping dictator. Or, creepy dictator. Well, now he admits it:
I know the details of taxes better than anybody. Better than the greatest C.P.A. I know the details of health care better than most, better than most.
"Nobody knew health care could be so complicated."
Lying on demand is a dictator's best quality.

"I have absolute right to do what I want to do with the Justice Department."

Spoken like a true dictator.

28 December 2017

Thought for the Day - 28 December 2017

Blockchain??? Well, just MVCC, but no VACUUM. How much fun will that be?

22 December 2017

Now, It's Your Turn

One might fairly say that Orange Julius Caesar has had a year, nearly, of riding on Obambi's economic coattails. So, here's what briefing.com had to say just now (8:33 am):
Just in, personal income climbed 0.3% in November (Briefing.com consensus +0.4%) following an unrevised increase of 0.4% in October. Meanwhile, personal spending rose 0.6% in November (Briefing.com consensus +0.4%), up from a revised increase of 0.2% in October (from 0.3%).

The PCE Price Index increased 0.2% in November (Briefing.com consensus +0.3%), while the core PCE Price Index, which excludes food and energy, increased 0.1% (Briefing.com consensus +0.1%). Year-over-year, the core PCE Price Index is up 1.5%.

November durable goods orders increased 1.3%, which is worse than the 2.1% increase expected by the Briefing.com consensus. The prior month's reading was revised to -0.4% from -1.2%. Excluding transportation, durable orders decreased 0.1% (Briefing.com consensus +0.4%) to follow the prior month's revised increase of 1.3% (from 0.4%).

So, giving away the Damn Gummint's tax money to the billionaires will surely cause the economy to rocket. Won't it????

Dat slope, it be slippery when dares nuttin but greased kiddie slide betwixt youself and dat Pit. And dat Pendulum, it do be swingin. Best keep yo neck out de way.

Or as the Wicked Witch of The West put it: "I'm melting!!"

21 December 2017

Bayes at the Moon - part the fourth

Gentle reader likely recalls the many times these missives have warned that Bayes is fundamentally evil, and should never be used in medicine. Luckily the latest scam has failed.
... BAN2401, an anti-amyloid beta protofibril antibody, did not meet the criteria for success based on a Bayesian analysis at 12 months as the primary endpoint in an 856-patient Phase II clinical study (Study 201).

Which report includes this gem
"By using Bayesian statistics in this uniquely-designed trial we had hoped that it would enable us to demonstrate clinical success faster than more traditional study designs. We now await the final study analysis which will be conducted after 18 months of treatment, which represents an amount of treatment time that is considered as appropriate for assessing efficacy in disease modifying agents for Alzheimer's disease," said Lynn Kramer, MD, Chief Clinical Officer and Chief Medical Officer, Neurology Business Group, Eisai.
[my emphasis]

Now, why would Bayes analysis generate stat sig faster? After all, the real efficacy is purely a function of the compound, not the method of analysis. In particular, there's a finite amount of information in the trial data. Bayes can't (well, not legitimately anyway) add any information. Well, investigators can fiddle with the priors, naturally. Obviously, that didn't work this time.

QED

20 December 2017

Darwin In Action

The Tea Baggers incessantly bray that the "free market" solves all ills. So, today we find that DeVos is paying back her donors.
Students who attended for-profit colleges filed nearly 99 percent of the requests for student loan forgiveness alleging fraud, according to [The Century Foundation] the [sic] a liberal-leaning think tank.

Education should be priced at the students' additional life time earnings, discounted.

19 December 2017

Octoberfest

Bust Your Stein!!

During last night's MSNBC pundit marathon, one of them off-camera announced that Jill Stein had come into view. There, again, was the still photo from the 2015 RT meet with Putin and Flynn, with Stein's profile in the foreground. Today, there's real news. The main point made by the pundit was that Stein cost Hillary the election, since the difference between Hillary and Orange Julius Caesar in the three Blue states (PA, MI, WI) was less than Stein's votes. Well, yes. But...

Not quite. Here's the Wiki table of the results. Note that Johnson polled about 3 times as many votes as Stein nationally and he got more votes than she did in each state. The net (or gross, as you prefer) is that Trump got ~63,000,000 and NotTrump got ~71,000,000. And most of that difference was Johnson+Stein (Johnson alone eclipsed Hillary's margin), not Hillary.

What would have happened if there were no Johnson or Stein? As a first approximation, let's assume that Johnson voters would vote for Orange Julius Caesar and Stein for Hillary. Some of both would have not voted, naturally, but this is a speculative thought experiment.

So far as the Three Blue States go, yes, Stein "took" enough away from Hillary to make up the difference with Orange Julius Caesar. But, of course, Johnson polled about 3 times the vote as Stein in those states, so having no third-party votes wouldn't have changed the result.

The following states would have flipped.

To Orange Julius Caesar:
Colorado
Maine (at-lg.)
Minnesota
Nevada
New Hampshire

To Hillary:
none

In the end, Hillary was, net, the benficiary of third-party, just not enough.

15 December 2017

Deja Vu All Over Again - part the second

As regular reader may remember, I've told the tale of my eclectic journey through the data worlds. One of those legs involved an ancient (even at the time I used it in the late 80s) TI-990 based VAR software for the construction industry. Even more odd, this company also had a side-line business in mechanical contracting parts, aka plumbing, which ran an application on a separate machine run by TI's then single chip version of that same 990. The contracting software vendor is defunct, while the wholesale application kind of still exists; the company's been bought and the software converted to normal chips and languages. I eventually migrated the company to (again) two applications running on an RS-6000/AIX and Progress. Since Progress wasn't/isn't especially relational while its 4GL is COBOL/BASIC-ish, and thus schemas were subordinate to code, integration wasn't as simple as falling off a log, but it worked. And last I talked with the folks, still running.

So, what has all of that got to do with today, you might ask? As well you might. Well, here's some of the news.
There are dozens of companies experimenting with ISC and early results look quite promising: offloading select tasks from CPU to SSDs can reduce latencies by a factor of 2-3 while also decreasing power consumption. The key purpose of ISC is to reduce (or even avoid) "expensive" data transfers from a storage device to a processor by performing computing operations on the former. Latency reductions will be crucial in the looming 5G era, especially for edge computing environments.

So, we have non-volatile DIMMs (somewhere, sometime in the future) and now a CPU that talks to SSD/foo skipping memory altogether. And that's not sort of new.

Here's the point of the 990 (from the Wiki):
The TI-990 had a unique concept that registers are stored in memory and are referred to through a hard register called the Workspace Pointer. The concept behind the workspace is that main memory was based on the new semiconductor RAM chips that TI had developed and ran at the same speed as the CPU. This meant that it didn't matter if the "registers" were real registers in the CPU or represented in memory. When the Workspace Pointer is loaded with a memory address, that address is the origin of the "registers".

There are three hard registers in the 990; the Workspace Pointer (WP), the Program Counter (PC) and the Status register (ST). A context switch entailed the saving and restoring of only the hard registers.

The reason this even made sense was that, in the 70s and 80s when the machine was in wide use, CPU and memory ran about the speed, so skip the middle man. It was a very successful machine among VARs for quite a while. But, as one might suspect, when chips and memory progressed to the point where load/store RISC architectures took off, the 990 was doomed. But the idea of compact circuitry talking directly to its data is not such a bad one. Back to the future?

13 December 2017

Thought for The Day - 13 December 2017

Friday the 13th falls on a Wednesday this month, and offers us this observation:
There's still much that remains unknown about the [Greenland] ice sheet, which at roughly 650,000 square miles is more than twice the size of Texas. The sheet, up to two miles thick, contains enough ice that, if it all melted, would raise oceans around the world by 24 feet.

Can you swim? Not a problem for you or your kids, but what about their kids?

09 December 2017

A Bit of a Problem

Bitcoin's tulip bubble price rise is in the news of late. While mentioned before, some points bear repeating, therefore.

1 - bitcoin, unless those in control change their minds, has a 21 million unit fixed limit. Recent reporting puts total mined ~17 million.
2 - bitcoin availability (per unit time) is halved every 4 years, approximately.
3 - the cost/amount of electricity needed to mine is going up extremely non-linearly.

Put them together, and what do we get? For one thing, some in the techno-geek world think the instant deflation that would ensue when all bitcoin have been mined (again, assuming that the powers that be don't bump the size of the lode) is a good thing. Given the asymptote situation, below, we may get the effect long before all bitcoin are mined.
Once all 21 million bitcoins have been mined, the supply cannot increase — regardless of growing demand. The result of this discrepancy between the supply of and demand for money is a steady and gradual decrease in the general price level, which equates to an equally steady and gradual increase in the purchasing power of money. Therefore, as Bitcoin miners collect transaction fees over time, no matter how large or minute, the funds gain value.

Even math oriented econ types know that deflation is not a benign experience.

Put another way, bitcoin is near/at the flat plane of its asymptote:
It is quite interesting to think about how far bitcoin has come since its inception. With a hard limit of 21 million BTC to be generated by 2140, a lot of people assume there are still a lot of coins to be mined for the next few years. While that is true up to a certain extent, we are getting closer to 80% of the finite supply being brought into circulation already. Said milestone will take place roughly 365 days from now [now being 1 February 2017].

The point being, if you do some back of the napkin arithmetic, that halving the number of new bitcoin/unit time, means you get to really sparse supply for way most of the time they're available. Remember that old conundrum: "if you step half-way to a wall, how many steps does it take to get to the wall? Infinity. And you never get there."

Some have ascribed the recent explosion in bitcoin $$$ value to creeping towards the end of supply. Could be. But, since the rule is to limit, in current period, the number of bitcoin released no matter the number/compute-power-used of miners (and that number is supposed to decrease on a published schedule: halved, appx. each 4 years and that hasn't happened this year), that seems unlikela. The improvement of compute power is also subject to the asymptote of progress, as the ability to create more compute cheaper as we near the limit of node size wanes. Either way, the bubble has been going bonkers recently. At least with gold, it doesn't get harder (mostly) as one empties the lode. The lode just goes dry all at once.

Imagine if we'd spent the money educating the spawn of the Red states, instead? They'd know better than to elect those out to make their lives worse.

If nothing else, bitcoin proves the foolishness of "store of value" in useless "things", including gold. Unless, of course, you want catastrophic (for the 99%, naturally) deflation. Just look at 19th century USofA. Mostly in recession or depression. The Garden of Eden of Freedom.

And it's not trivial to note that security of virtual money may well be (IMHO, is) much less than real. Yes, central banks can debase their paper money by printing with abandon, but that's a purposeful act. Hacking away data, on the other hand, isn't from the point of view of the money. You don't need a fleet of 18-wheelers to spirit away billions of virtual bucks the way you would real bucks or gold. And given the mono-culture that pervades the innterTubes, I expect we'll see yet more of it.

07 December 2017

You Don't See This Every Day

Yes, you don't:
In the trial, treatment for 14 days with SAGE-217 was associated with a statistically significant mean reduction in the Hamilton Rating Scale for Depression (HAM-D) 17-Item total score from baseline to Day 15 (the time of the primary endpoint) of 17.6 points for SAGE-217, compared to 10.7 for placebo (p<0.0001).
[my emphasis]

Now, that's a p-value to love.

06 December 2017

New VT-220

Gentle reader has seen here the epithet, New Gold, referring to the US Buck and its effect on all things data. Well, all things in the end. Long ago, in the first essays here, gentle reader was offered the notion that Organic Normal Form™ databases and high bandwidth innterTubes connectivity mean a return to the thrilling days of yesteryear: the central host with all the data and logic over a wire to a (semi?) dumb terminal (VT-220 as example). The real question is whether networks/servers/engines can, in fact, support thousands+ of persistent connections. If so, then just call me Marty McFly. OLTP, modulo Amazon, why yes. Yes, of course.

So, today's news of Always Connected PC battle may mean just that. And I couldn't be happier. The revenge of RDBMS and 5th NF.

04 December 2017

The US of Mississippi

During the fiasco of the Leona Helmsley Memorial Tax Cut for the 1%, I heard some Red State Congressman answer the question about why remove SALT deductions with the bill, which would disproportionately hurt Blue States. His reply was that may be such states shouldn't tax so much. IOW, the whole country should be just like Mississippi: uneducated, unskilled, unhealthy, but God fearing and multiplying like rats. Such a model worked in the 19th century plantation South. Not so much today.

Let's consider what that situation might mean. If you look at this graph, you can see that the Red States all lag the national average in per capita GDP. Not only would the "average" go from $50,000 to $31,000 (that's a 40% drop), but it also means that said Red states lose their market.

Back before the Civil War, there was Bleeding Kansas, what most historians consider the real beginning of the Civil War. It pitted Free Soilers in the soon-to-be state of Kansas against Slavers. The Kansans understood that slaves were cheaper farm help than Freemen, so they went to war over it. The exact same effort is going on now: the Red State Slavers, aka Republicans, intend to impose their model on the rest of us. Since the US Buck is New Gold, even if they win, they lose; who will buy?

03 December 2017

Minority Report, part the fifth

In the continuing saga of creeping dictatorship, we finally get j'accuse from the likes of the American Enterprise Institute.
The failure of Republican members of Congress to resist the anti-democratic behavior of President Trump — including holding not a single hearing on his and his team's kleptocracy — is cringe-worthy. A few Republican senators have spoken up, but occasional words have not been matched by any meaningful deeds. Only conservative intellectuals have acknowledged the bankruptcy of the Republican Party.

So, yes Virgina, there is a USofA in your future. It's just all going to look like Mississippi. One wonders how the bible thumping Red Staters imagine they'll be able to sell all that stuff when there's no body in Blue States with any more moolah than they have. The Achilles heel of slave wage exporting autocracy is finding markets with currencies that are stable and (near?) par with its own. As these essays have said, and more recently so have some mainstream pundits described here, the US Buck is New Gold, so everybody else manages their currency to maximize against it. That's hard to do within the US of Mississippi.

According to a reporter (didn't note the name), the Senate is in the state (yes, a pun) where 40% of the population commands a super-majority. That's already minority rule. Better read your Bible.

02 December 2017

My, How Times Have Changed, Again

That didn't take long. AnandTech reports on Really Bigly DRAM.
The key advantage of 128 GB LRDIMMs is their density. For example, a dual-socket Xeon Scalable platform using the -M suffixed processors, featuring 12 memory slots, can expand the maximum memory size by 2X to 1.5 TB from 768 GB by using 128 GB LRDIMMs over 64 GB LRDIMMs. For DRAM-dependent applications, such as large databases, holding everything in memory is the most important thing for performance.
[my emphasis]

Not that I expect anyone's going to pay me enough to equip my development machine thusly. But still?? What the hell is wrong with 5NF at this point??

01 December 2017

Dirty Harry Speaks

"It's a tax bill for middle class"
-- Orange Julius Caesar

what it actually means
An updated Senate plan to overhaul the U.S. tax code could dramatically raise taxes on households earning between $10,000 to $30,000 starting in 2021, according to new findings released Thursday by the Joint Committee on Taxation.

But being this is a Billionaire's Tax Cut, the most powerful giveaway in the world and would blow your head clean off, you've gotta ask yourself one question: "Do I feel stupid?" Well, do ya, punk?
-- Dirty Donald/2017