Atomic I/O letters column #81
Originally published in Atomic: Maximum Power Computing Last modified 16-Jan-2015.Welding helmet probably not needed
Playing the likes of Oblivion a year ago wasn't a very visually exciting affair, since I was still running a dusty and rusty GeForce 4 MX440. But now I glee with happiness, for I have my new ATI X1950 Pro 512Mb.
Well, sort of glee anyway. After only an hour or so of playing, with HDR enabled of course, I noticed the in-game sun. I found it rather disturbingly bright - realistically bright.
We're told not to look at the sun in real life because it'll cause permanent damage, but this got me thinking about how these new lighting techniques in games might affect gamers.
I get headaches from Oblivion HDR, so I choose to look down and not gaze at the sunlit skies, or turn HDR off. But that kind of ruins the other lighting in the game.
Is this some placebo trick my mind is playing on me, or can HDR game lighting actually seriously discomfort people?
Jason
Answer:
No software can make your monitor brighter than normal. Run a word processor, open a
new blank white document, and that right there is the brightest white your monitor can
display, with its current brightness setting.
(Colour calibration can mess around with this a bit, but it's not relevant to peak brightness.)
HDR rendering in games calculates the on-screen image using a larger dynamic range than any monitor can actually display. Cunning algorithms are then used to map the result onto the monitor's actual colour gamut, to give a more pleasing image.
If the rendering engine decides that the best way to go is to make the whole bright area of sky around the sun into one big white blob, then the in-game "sun" may look more striking. But none of that white blob is any brighter than a blank word processor page.
Lots of people do, however, suffer from eyestrain when they're using their computer. The single most common monitor setup mistake used to be leaving a CRT screen set to 60Hz; today, it's setting the brightness too high. This may be the source of your problem.
Modern LCD monitors, in particular, often have very high maximum brightness - much brighter than most CRTs could manage. This super-bright backlighting is handy if you have to use the computer in a sunlit room, but for the average indoor computer user, even during the day, it's far too bright.
Your monitor should actually be no brighter than a well-lit book. For a lot of modern monitors, that means turning the brightness as far down as it'll go. Even that may not be quite enough!
There are plenty of Web sites that'll help you set up your monitor properly; try starting here, for instance.
It's also not crazy for ordinary home computer users to get a proper hardware monitor calibrator. Calibrators used to cost several hundred dollars, but now you can get a Pantone "Huey" calibrator for less than a hundred bucks.
You don't need a calibrator if you're just playing games. But if you're even a casual digital photographer, a calibrated monitor can make a big difference.
VESA Local Bus cards also a problem
I've heard that a PCIe 2.0 motherboard is backwards compatible with PCIe 1.0 video cards. My question is, does a PCIe 1.0 motherboard support PCI-E 2.0 video cards?
Steve
Answer:
A PCIe v1.0 motherboard won't work with a PCIe 2.0 video card.
A PCIe v1.1 motherboard, on the other hand, should be fine. It'll only allow the v2 card to communicate at v1 speed, but like every other bus speed bump since the days before AGP, this won't make much difference.
Generally speaking, recently-released motherboards should all be either PCIe 2.0 or PCIe 1.1. There are some oddities, though, like Asus' recent P35-chipset enthusiast boards, only one of whose two full-length PCIe slots is PCIe 1.1 compliant. The other one's only PCIe 1.0.
Combined diagonals: 80 inches
Can you use a dual DVI link (as needed for monster screens) and the TV out on a video card at the same time?
Mark
Answer:
Yeah, I think so. You can't have two monitors and a TV all running at once from a standard
two-DVI-plus-TV graphics card (maybe if two of the displays are in "clone" mode and
both showing the same image), but dual-channel DVI output only counts as one monitor.
This also, of course, means that if you have a graphics card with two DVI outputs on it, and one of them is hooked up to a giant monitor that requires dual-link DVI, you can still use the other output for a second monitor. Your video card may or may not be able to do dual-link DVI on both outputs, but if your second monitor is a normal single-link or "VGA" screen, this won't be a problem.
I still use Professional File System
What should I format a Gigabyte i-RAM Solid State Disk as? I formatted mine as FAT32, but should I have used NTFS instead?
What's the overhead difference between FAT32 and NTFS? I know FAT... has a FAT. But I never really found a good explanation on how an NTFS partition is structured.
Yorrel
Answer:
If all you care about is performance, the limiting factor for an i-RAM is almost always
going to be the speed of the SATA interface.
(NTFS is less prone to corruption if a write's interrupted than FAT, but if you're only putting temp and swap files on your i-RAM, this doesn't matter.)
NTFS does still have a Master File Table that's somewhat analogous to the File Allocation Table of the FAT filesystems. NTFS is a lot more flexible, though, and supports a bunch of UNIX-y features like disk quotas and hard links. The Wikipedia article on NTFS is a good jumping-off point if you'd like to learn more about the format.
I don't think any of the many technical differences between NTFS and FAT are likely to matter for most i-RAM users, though. Even stuff like file fragmentation (considerably worse with FAT32 if you're creating and deleting files all over the place) shouldn't make any difference, since the i-RAM's seek speed approaches zero.
The SSD for the rest of us
I've got a stack of 16/32/64/128Mb SD and MMC memory cards - that are, for the lack of a better word, useless.
I've searched for months for a multi-SD-card reader, where you can plug in 10+ cards and the little box will RAID/JBOD them all. No luck, obviously. I'd try to build one myself but I'm clueless about circuits.
I'm certain there's a market for a device like this - think about all the digital cameras and mobile phones sold with puny useless cards. Why do they even bother?
SD card readers are selling online for $2, so the cost of basic parts might make this viable. I'm just not sure how much the RAID hardware would cost.
Calvin
Answer:
A device that at least approximates this functionality
does exist.
That one's far too expensive, though, which is probably why nobody seems to be selling it any more. Here's another one, which converts up to three CompactFlash cards (so it'd be no use to you - but high-capacity CF cards are only slightly less astonishingly cheap on eBay these days as are cards in other formats) into one SATA device. It costs only two hundred bucks. I don't know of any others.
Even if you combined eight 128Mb cards, though, you'd only end up with 1Gb total; you could just buy a 1Gb CompactFlash card for $notmuch and plug it into an IDE cable with a cheap adapter instead. Adapters like this one I reviewed ages ago are now close to free on eBay.
In theory, a multi-card device could have significantly higher transfer rates than a CF card in an adapter. In practice, it might not. Even purpose-built one-piece Flash-RAM Solid State Drives are still not actually very fast, and a multi-card device is unlikely to perform any better.
Old low-capacity CompactFlash cards still have a role as boot devices for things like do-it-yourself Linux network appliances. Again, you just plug 'em into a cheap ATA plug adapter, and then they're an ATA device as far as the computer's concerned (you can get adapters that give you a laptop-drive ATA connector, too; this can be a pretty nifty way to resurrect an old laptop or what-have-you). Old cards won't support DMA data transfers and will also very probably appear in Windows as a "Removable" device, rather than a "fixed disk", which means you can't make them part of a RAID array, but this doesn't matter if you just want to move a floppy disk worth of Linux boot code onto something more reliable.
(The removable-disk problem also prevents you from using memory cards in a USB reader to make a RAID array, under Windows. Both Mac OS and Linux let you do that, though, as well as more perverse things, like the famous USB-floppy-drive RAID. If you want to actually install Windows on a CompactFlash card and have the resultant computer run reasonably quickly then you'll need a "fixed disk" CompactFlash card that also supports DMA transfers, which is a surprisingly rare beast; I've been working on a reply to someone's letter about that for an awfully long time. It seems that CompactFlash card makers jealously guard the simple utilities that flip the fixed/removable disk bits in their card firmware.)
Apart from that, old small cards are indeed pretty much useless these days. It's not as if they weigh much, though, so you might as well keep 'em in your camera bag for the few extra photos you can fit on them in an emergency.
Photojournalists today are also used to carrying "sacrificial" memory cards to hand over to people with guns who want to confiscate their camera's "film". It's a decent bet that uniformed checkpoint bandits won't be able to tell an 8Mb card from an 8Gb one.
This, though, is unlikely to be a significant selling point for most users.
Microsoft.sucks.com is also taken
I came across this just today while searching for some info on exmerge for Microsoft Exchange. I typed "exmerge" into Google and clicked the top link, as it looked to be a Microsoft site.
Firefox came up with a certificate warning saying that I was attempting to contact a domain called "thesource.ofallevil.com", that was using a certificate belonging to Microsoft!
Trusting my antivirus and firewall settings, I proceeded to click to see what was going on. Sure enough it looks like a valid MS download page, but all the links pointed to thesource.ofallevil.com. I looked up this domain and it came back as what would seem to be a totally crap registration.
I just thought it was odd that this obvious fake was so high in the Google hits.
Seth
I wonder how many people actually do "notify the site administrator".
Answer:
Thesource.ofallevil.com looks like a Microsoft site... because it is a Microsoft
site!
You're just reaching it via a different DNS entry, which is what causes the certificate mismatch.
Whoever runs ofallevil.com just pointed the DNS info for his various subdomains to the sites of major companies. The DNS record for thesource.ofallevil.com, for instance, points to www.microsoft.com. So thesource.ofallevil.com appears to have every Microsoft page on its own "site".
There's also theroot.ofallevil.com, which points to VeriSign. Only if you ask for www.ofallevil.com or ofallevil.com will you get a site (and a blank Web page) actually belonging to this prankster.
Whoever runs ofallevil.com doesn't even know when you've visited one of these prank "sites". Your browser asks DNS what IP address thesource.ofallevil.com is, and DNS tells it that it's 207.46.193.254 (or whatever other Microsoft server comes up in the round-robin sequence), just as it would if you'd asked for www.microsoft.com. Then your browser starts asking that IP address for Web pages, and getting data from the Microsoft server. The ofallevil.com guy is never part of the conversation.
This is only a security concern if the ofallevil.com guy takes advantage of his high search engine position to start delivering pages from his own server that look like the genuine article but aren't. He got his high Google PageRanks because lots of people absent-mindedly linked to ofallevil.com URLs rather than the genuine ones. He's missed his chance to capitalise, though; just recently, Google noticed him and trashed his PageRanks again; as I write this, they're still all zero.
He had a pretty good run, though. The ofallevil.com prank is more than five years old, now!
