Dan's Data letters #15
Publication date: 4-Dec-2002.Last modified 03-Dec-2011.
One... MIIILLION... dots per inch!
I've been approached with a question at work and I hoped I could get your opinion.
I refer to this page on Hewlettpaq's wonderfully informative interweb thingy.
999,999dpi enhanced res is quoted for all the scanners on this page. I assume they talk about interpolated resolution... and then when I consider a piece of software adding in just under one million approximated pixels in between the pixels that were actually scanned I have to wonder what you'd end up with. A big fat mess? A really really really stupidly big file?
Because I always like to consider the possibility that I'm actually retarded, I figure there are two possibilities.
1 - I missed something or enhanced res is not interpolated res as I assume.
2 - HP think the reason you should buy their scanners is that they are capable of doing things that are absolutely useless.
What's your two cents? Should I tell my questioner that 999999 is, in fact, the exact amount in kilos of cocaine consumed by the HP marketing team on a yearly basis?
Will
Answer:
Well, I don't know about HP marketing peoples' exact recreational pharmaceutical
preferences, but yes - their scanner driver scales the image up, and some
witless marketroid
is responsible for the sliders going all the way to 999,999dpi.
Interpolation is a handy thing when your scanner has different optical and mechanical resolutions - HP quote resolution in both directions as "optical resolution" on that page, but real optical resolution is the horizontal resolution of the scanner, across its sensor. It's determined by how many cells per inch the sensor has. Mechanical resolution is vertical, and is often coarser than optical resolution; it's determined by the smallest step by which the scanner can move its sensor. The 2300c, for instance, has 600dpi mechanical and 1200dpi optical resolution. To get the best possible resolution out of that scanner, you have to set it to 1200dpi, in which mode it'll use full sensor resolution horizontally, and 2X interpolation vertically, to make up the extra pixels needed to make the vertical resolution equal to the horizontal.
Interpolation higher than your scanner's real resolution in both directions is useless from an information theory viewpoint - you get nothing that you wouldn't get from a lower res scan - but it may be handy if you for some reason need a particular size for the final image. Interpolating up to almost a million dots per inch, though, is just insane.
A one inch square scan at 999,999dpi will give you 999,998,000,001 pixels, which at three bytes per pixel (24 bit colour) is 2,794 gigabytes of data, before compression. The image would compress pretty darn well, since even if it came from one of the 2400 by 2400dpi scanners there'd only be one real scanned pixel worth of data in the image for every 173,611 output pixels, which means it'd be astoundingly fuzzy. But the saved file size would still be astronomical, and there's no way in the universe you could ever actually do such a scan; it's much too big for any Intel-compatible-CPU machine to hold it in memory.
Even a one square millimetre scan at this resolution would be 4.33Gb, which all but the most completely stacked PCs would flog themselves to death doing. You couldn't scan a full stop!
For reference, you can get around 40,000 real dpi by shooting with a half-decent digital camera through a microscope; I did that here. I'd be able to do better with my current camera, though the microscope's quality would probably be the limiting factor. The most I've managed so far is below 3500dpi, using a macro lens.
Dedicated film scanners have higher resolution than flatbeds, but fine-grained film only has about 6000dpi of detail capacity in it, so I don't think you'll find film scanners that approach microscope magnification capabilities.
More power!
My new PC is based around the Lian Li PC-60 case with a Gigabyte GA-8IHXP motherboard, 512Mb PC800 RDRAM, etc. I was originally aiming for a P4 2.4b, but settled for a P4 Celeron 1.7. This way I can work my way up all the way to 3.06GHz in the future. That's the plan anyway.
All the bits have been ordered now, so I'm ready to get everything installed. But now I just read that the P4 Celerons based on the Willamette core need 1.75V from the Motherboard. Somewhat nervous I surfed around and what do I see to my astonishment? The GA-8IHXP motherboard only goes up to 1.7V!
My question now is, will it work? And if not, can I do anything to increase the voltage?
Michel
Answer:
There's no such limitation. The
HardOCP piece
you read didn't say that the ceiling voltage was 1.7 volts; it said that
the most adjustment you can get is +0.2V, which with the 1.5V CPU they were
using meant 1.7V max. If you plug a 1.75V CPU into the board, it should
power it fine.
The manual supports this view. It only says that CPU voltage is adjustable up to 0.1V above the original value, not 0.2V, but it doesn't mention a ceiling voltage anywhere.
If you ever do run a CPU below its rated voltage, by the way, it may work fine, or it may be flaky, or it may not even make it through the Power On Self Test (POST). Winding down the Front Side Bus (FSB) speed (and thus the CPU core speed) can allow CPUs to run at less than their rated voltage - the GA-8IHXP, though, doesn't seem to have any below-stock FSB options, though, so you can't "underclock" with it.
Steamin' P4
Is 58 degrees too hot for a P4 processor ?
I bought the Cool Tank Breeze that you reviewed, to try and keep temperatures down. I have used Artic Silver 3 as the TIC and have followed your directions regarding application method (and thickness applied).
My case shouldn't be suffering by underventilation with two 80mm fans at the front, one 80mm fan at the rear (all three fans hover around 3000rpm) and an Antec 480W PSU (with 2 fans), but my processor STILL runs in excess of 48 degrees Centigrade at idle and 58 degrees when playing games etc (at an ambient temperature of about 30 degrees).
Is this too hot? The mobo has software that allows me to set "temperature warning buzzers" if the monitored component temperatures get too high, and by default the CPU warning temperature is 90 degrees! Call me a big girl but that seems WAY too hot for my liking.
My father has a P4 2GHz system in which the CPU doesn't get over 40 degrees. Yes I know his processor is slower and that mobo temperature monitoring probes are prone to inaccuracy but when I touch the side of the heatsinks, I can well believe that there is over 15 degrees difference and that mine is over 55 degrees, as advised.
Rick
Answer:
The danger temperature for P4s (which they ought never to reach, thanks
to their automatic speed throttling if they get very hot) is above 80 degrees
Celsius. Yours is fine.
Note that your dad's CPU has a Thermal Design Power (broadly representative of its power consumption and heat output when working hard, though not a worst-case figure) of less than 55 watts; your faster P4's TDP is 81.8 watts. That probably accounts for the difference.
Spying time
I have been asked a question by a workmate, and I'm stuck.
His partner is worried that one of her flatmates is using her computer to view objectionable material, the sort of stuff that not only worries her but might be of interest to certain legal authorities. The problem is they always clean up after themselves (no pun intended) and she cannot catch them at it. I need to locate a utility that would track their internet use, preferably without them knowing about it, and my mind is a blank. Could you point me in the right direction?
Patrick
Answer:
There are quite a lot of "activity loggers" out there that'll let you do
this on a standalone computer. Here are a few options.
The free trial version of the inventively named Activity Logger may do nicely.
iCom Stealth looks promising, too - though, perhaps unsurprisingly, the Web site tries to install Gator spyware when you visit it.
This slightly outdated version of SafeNet may cut it, as well; the free trial version of the current SafeNet release has a two week time limit.
And then, there's this, which apparently has an icon in the System Tray, which is no good. But maybe you can turn that off. And it's a free download.
Grab that screen
I was wondering if you know of a way to do screen captures from full-screen programs like 3DMark. I tried to use the video-out on my video card, but then the resolution sucks. I also have dual monitor output on my card, but I can't think of a way to utilize this (none of the hardware I use has a VGA-in, only SVIDEO in).
Tarek
Answer:
You can do screen-grabs in the various Mad Onion benchmark programs, including
3DMark, by pressing F12 (see
here
, for instance). Many games work that way too, though they may use a different
key. You probably won't be able to copy the screen contents to the clipboard,
though, so many outboard screen-grab programs like the excellent
GrabClipSave won't
work, if there isn't a screen grab key built into whatever you're trying
to grab from.
You can, however, use more powerful screen-grab software, like Hyperionics' HyperSnap-DX, which can grab frames from hardware-3D applications. Fraps also comes highly recommended.
Wanted: Theoretical particle
Is it possible to get a magnet with single polarity?
Sansuma
Answer:
If you mean a magnet with only one pole, no. If we ever manage to discover
magnetic monopoles then it might be, and you might even be able to buy
a block of them the size of a pinhead for less than a trillion dollars!
How much for the little girl (or a good review)?
Do you always remain objective in your reviews? You are quite obviously affiliated with a few retailers (the most obvious one being Aus PC Market), and others would no doubt often give you a bit on the side (not in a sexual context, but for humour value you may like to replay that line in your head repeatedly).
I mean, I read your "this is a great thing" review and immediately I am presented with a link to buy it. Part of me likes this convenience, but another part of me thinks that perhaps this would cause one to take a slightly different tone when doing reviews.
I lied, I have another question - do you find that yourself writing something and then rephrase it to better suit your sponsors, or are you one of those people that prides themselves on their journalistic integrity?
I'm not having a go by the way, as I can't really detect bias in the things that you do... but it has always been something that unnerves me a little.
Russell
Answer:
Nobody's perfectly objective, but I do my best
Aus PC Market do, indeed, pay me for clickthroughs from my "you can buy this from..." links.
This doesn't change the content of my reviews, though, if only because there's really no point talking up a product just because my friends at Aus PC are trying to shift it. Never mind whether I'm inherently honest or not (I am, but I would say that, wouldn't I); writing biased reviews just isn't a good long-term strategy. If I were to say something was great when it wasn't, people who bought it and discovered the truth wouldn't believe me next time. Reviewers who want to make a career of it, as I do, can't get away with selling their favours.
Aus PC (and occasional other dealers) also try out new products on me before they decide whether to stock them or not. If the product's good, I tell them, and they make sure they've got enough stock to satisfy demand when the review goes up. If the product's not so good, I tell them, put up a review that's not going to attract a whole lot of customers, and they don't get more stock in. In the occasional case when a product's pure rubbish, and I discover this before I've put much work into the review, I tell the supplier, and don't bother to write the review at all. I've got lots of stuff to review, and I'd rather spend time telling people about good products than about awful ones.
I still write awful-product reviews when I find them amusing, mind you.
I've got a couple of other affiliate deals - Backyard Artillery and Puttyworld, plus a couple of others that pay pretty much nothing at all, but that's it for my review-product-supplier-connected revenue.
Some people who've sent me review gear have also sent me money after I wrote the review, but that's very much the exception rather than the rule, it's usually because I've done something else for them - like provide them with the high-res originals of all of the pictures I took of their product, for instance - and there's never any pre-arranged deal. Nobody's ever even offered me money for a good review, possibly because they've read this.
I usually do get to keep the review gear people send me, which is nice. But, by and large, that's all the compensation I get from them.
As far as rephrasing-stuff-to-suit-the-sponsors goes, all I do in that department is often, in a review of a not-so-good product from Aus PC, mention something better that they sell. I commonly mention better products that they don't sell as well, though. So you can't pin the rap on me, see.
