- Get a motherboard bundle from a place that puts it together and tests that the CPU/RAM/MB are compatible and that it boots
- Buy a good quality case and you'll still be using it 10 years from now. A cheap case or proprietary case is just landfill fodder. I just put a brand new set of components into my original ATX tower case from 1997 (with a new PSU). That new system will probably run until at least 2015.
- You get a better quality system. You get control over what quality components go into the build instead of relying on Dell not to sacrifice quality in the name of profits. Less likely to have no-name parts inside the system.
I build a dozen or two machines per year. I never have random crashes - because I'm smart enough to test my system during the build. (All of my builds run Prime95 and other burn-in tools for at least a day to make sure that memory timings are correct and that the system doesn't fail under load.)
Yes, you run risks and you have to deal with warranties yourself. But it's very rare that you have a failed part arrive out of the box anymore. Imaging products like Acronis, Ghost, or NTFSClone take away the pain of a failed hard drive.
Here's our stock build for office PCs. Dual-core AMD X2s, 2GB RAM, RAID1 SATA drives, integrated video, WinXP Pro, Office 2003 Pro. Prices on RAM/CPU may be a bit dated (they've dropped recently). Part #s are from MWave.
$0132 AA15070 WindowsXP Pro OEM
$0299 AA24200 Microsoft Office Pro 2003 OEM
-----
$0431
$0097 BA30107 Antec Sonata II case w/ 450W PSU
$0008 AA36600 NEC 1.44MB BLACK FLOPPY FD1231H-302
$0032 AA52110 NEC AD7170A 18X BLACK DVD MULTI WRITER
$0125 (2) SATA drives (200GB+)
-----
$0262
~$1100/unit
We've put a number of these in already and the users are very happy with them. Our goal is that they last at least 6 years and possibly as long as 12. That should be possible since they are dual-core.
I built a game machine for around $1500 last month, including SLI.
Only if you go for lowest cost. I have a few brands that I've worked with over the past 5 years that I trust. And one motherboard manufacturer where I've used a few dozen of their motherboards in systems with no issues.
I also plan for failures. That means RAID1 for work machines (the newer boards include RAID on the motherboard and a 2nd hard drive isn't that expensive) along with imaging the system periodically.
The newer motherboards (Asus M2N-E) use heatpipe and cooling fins instead of tiny fans that fail. That removes a big issue in longevity. And there are fanless video cards (or just go with an integrated video like the GeForce 6150 boards).
My experience been with the AMD Athlon CPUs where a screwdriver and significant amount of force is needed to put the clip where it belongs. It's no fun when the screwdriver slips and you think you just wiped out the motherboard.
True for the old Socket A chips. Not so true for the more modern 939/940/AM2 chips. They use a cam lever to get the necessary force, no screwdriver needed.
Personally, I still prefer to buy motherboard bundles where the stock heatsink has already been fitted. Saves me 5 minutes of horsing around (plus I know that the CPU is compatible with the board and they've at least booted the board once). Well worth the $9 charge for assembly and testing.
Last time I looked, 2.5" drives run on about 1W compared to 10-12W for 3.5" drives. That includes the newer 200GB 2.5" drives.
Your POV though, is correct when talking about using older 3.5" drives. A lot of folks try to save money by building a large RAID out of older 80-120GB drives. So they put 6-10 of these together at 10W each, when they could've simply used a pair of 500GB / 750GB drives in RAID1. Less noise, less heat, faster transfer rates are just a few advantages by using newer drives.
I have a firewall server that runs on a VIA EPIA and a pair of laptop drives. Not the least expensive solution in the world, but it's extremely low power and quiet (no fans).
If you had said 512MB, I'd have (mostly) agreed with you. Otherwise I think you're engaging in very wishful thinking for anyone other then a home user that doesn't do much more then read e-mail. As soon as you start getting into situations where the users are running 2-3 programs at the same time and keeping other applications in the system tray, that 400-800MHz system is going to be a dog. Especially in a work situation where you're interfering with productivity.
If it was a pair of 400MHz CPUs (dual-CPU setup), I'd give it more leeway.
There's a point where too cheap is wasteful. A modern, low-power, reasonably efficient dual-core system can be bought for $450 to $600. Plus fees for the OS and software.
Definitely nicer to work in a small company in that regard. I try to make sure the user's have what they need, we don't lock down the machines and we spent the $$$ on Acronis or Ghost to make periodic images of the systems. Most of our users are well-behaved enough now that I don't worry as much. The ones that aren't, I make sure I always know where the "gold" image is for their machine.
As for the firewall we simply tell them: "we know what you're doing".
Now, generally, I don't bother watching the logs to track packets or websites or whatever. But if bandwidth spikes or we start maxing out our line to the public network, the users know that I'm going to start looking for people to name and shame. (I'll actually address the issue privately first.)
In a small company, you can also say directly to the CEO - we can bump up the per-message limit on e-mail to 20MB, but it's going to require us to spend $10k on new server hardware within the year. So do you want us to spend the $10k or do you want to sit down and have a chat with Jane, Bob and Frank?
A very handy feature. After all, if you're going to have a RAID1 + hot-spare setup, there's some advantages to just making it a 3-way active mirror instead. When one of the primary drives dies, you don't have a recovery window while the hot-spare syncs up with the array.
The only downside would be that the hot-spare drive gets used and may suffer from wear-and-tear more then a hot-spare drive that spends its time powered down.
On the flip side... you don't have to sit and wonder if that hot-spare drive will work correctly when it's finally needed.
I suspect that absolute temperatures, within certain limits, are not important to drive reliability, but that temperature variation is.
That's my suspicion as well. Minimizing the amount of thermal cycling is probably important. This means adequate cooling to keep drives at a steady temperature state, no matter if they are spinning idle or operating under a heavy seek load. It may also mean that spinning drives down to save on power may cause premature failure due to the thermal variation.
I guess it would depend on how many times per hour you spin drives down or cycle them from 25C to 45C.
Please, mr dell, start a price war between RAM manufacturers next! I live in perpetual obsolescence thanks to the dramatic cost of DDR and DDR2! Won't someone think of the child processes!
Memory prices are already down to about where they were last July (2006). That's about 30-40% less then prices were in Nov/Dec 2006.
FarCry is one of my favorite FPS for single-player. Devilishly tricky in places, but it gave you lots of ways to approach situations. You could even shake off your pursuers if there was enough cover around that you could hide in. Not to mention the ever-present supply of rocks that you could throw to distract the AI.
One fun way to clear out the dock area at the end of an early mission where you're trying to meet up with the doctor: Take out the rocket-shooting bugger on the hilltop tower. Then climb up the tower, pick up the rocket and zero in on the docks that are about 1-2km away. Shoot a single missile. Watch as all the AIs down at the docks go and investigate the explosion area. Shoot the rest of the missiles and watch the carnage.
It's a pretty wild game if you play in a manner more like a covert ops might. Getting a vantage point, scouting the opposition, drawing them off into an ambush. Dealing with the AI as it tries to flank you. Listening in on conversations, hearing the AI give orders and information to each other.
Dang... now I may go fire it up and blow my Saturday afternoon on it. (grin)
Yes, it gives them an advantage (in material and power). But I don't think they'll manage to take over all of 0.0 without falling apart. The odds are highly against them because of the following:
1) They have to keep growing their borders and taming more corporations to act as pets. If they don't grow or constantly engage in wars, their members may get bored.
2) At the same time, they have to keep reign on the existing pets. As the size of the pet population starts to outnumber the owner population, this will become more difficult. Pet corporations eventually will start asking themselves why they pay a fee to live in BoB space.
3) BoB no longer holds the moral high ground. This means that instead of attracting players with integrity, they may start attracting only the sociopaths. (Which is a bit of a black/white simplification.) While sociopaths may make good PvP'ers, I'd argue that they don't make good corp mates.
So I'd put odds of them managing to conquer the entire 0.0 space at about 10:1. They are very organized about it and it is their stated goals. But I don't think they can ultimately pull it off.
Maybe... they could've also had the pilots submit daily / weekly / monthly XML dumps of their character information from the "My Character" XML export:
What I'm not sure of is whether one player can look at another player's skilltree on the website, or whether it's locked down. (Pretty sure it's locked down.)
The XML file wouldn't be as trustworthy as getting it from the source (myeve.eve-online.com server), but if a pilot was slacking it would quickly be apparent in-game.
Alternately, the person in charge could require screenshots from the in-game character window that shows the skills.
Look at the area of control for "BOB" down in the lower-left. I wouldn't call it 30-40%. Not even of 0.0 space (which is the area outside of empire influence). Well, maybe 30% of empire space. The area they just conquered is the region in the lower-right (now colored blue).
They do control a sizeable junk of 0.0 space though. And most of the rest of the corps in 0.0 space consider them a threat.
And while there are no current DX10 games, they'll be coming within the year. So the answer is: XP for games now, Vista by the year's end.
I find that overly optimistic.
While some game companies may release DX10 games this year, there's no way in hell that DX10 will be required for any titles until 2009-2010. Not unless the game company wants to sabotage their sales. While gamers tend to upgrade quicker, I'd still be surprised at faster then a 25-30% uptake per year.
Memtest86 / Memtest86+ are not very good at finding memory issues. They work, but only in cases where the memory is completely bad (an actual failed section).
What they don't do well is find RAM that is marginal, overlocked or misclocked.
Because the only way to find such issues is to load the system down (CPU *and* RAM simultaneously). And Memtest doesn't do much with the CPU.
For real RAM testing, try running Prime95 in torture test mode for a few days. (Folks have been doing this for about 10 years now.) If you can manage to run Prime95 for a few days without any errors, you can call your system stable.
Or run the Gentoo server as a Xen guest OS. Prior to running the update, take it down and snapshot it. Do the update and you can roll back if needed.
Which is how I prefer to approach just about any OS. Be prepared to roll back on a minute's notice even if your test server said it was a no-risk upgrade.
1) I'd bet that the AMD chip is more picky about the memory that you have installed. Which is probably why you had reboot issues. It's one of the few downsides of having the memory controller on the CPU.
2) AFAIK, all AMD chips that came out after the AthlonXP series have on-chip sensors and thermal throttling (AMD Cool-n-quiet?). At the time of the infamous Tom's Hardware test, Intel chips had thermal throttling and AMD chips didn't. And up until Core and Core 2, Intel chips were affectionately known as space heaters, taking back the space heater crown from the old AMD Athlon chips.
3) I call bullshit on a CPU failure taking out other components.
I did not post about it but as you have identified one of SOE's problems in EQ was that their "Play Nice" policies were enforced in such a way that they really only applied to those who followed them and became disadvantaged as a result.
That's one of the reasons I stopped playing the original EQ.
Back in the day, if you wanted to travel long distances, you had 3 or 4 options:
1) Hoof it - Basically you spend time on travel.
2) Pay for a port - Spend plat/gold to reduce the time needed.
3) Befriend a porter - Get your travel for free, but you pay for it with other responsibilities.
4) Become self-sufficient - Level up a druid/wizard. Your travel is now free, but you put time into getting to that point.
When travel was more difficult, EQ was a nicer place. Folks didn't move around willy-nilly to the latest "hot" XP place. You tended to simply adventure wherever you happened to be instead of flocking to the uber-zone-of-the-week. This meant that you tended to meet and know most of the locals. Folks who were new in the zone tended to stand out. Troublemakers were quickly outed and they had trouble getting ahead of their reputations. Real troublemakers would even have difficulty getting ports (the community was able to punish troublemakers by denying them access to a player-provided service).
So you could cause trouble. But it would cause trouble for you and make life difficult for you.
Fast forward until after SoL and PoP were released. Travel has now become absolutely trivial. Because everyone is moving around willy-nilly, nobody knows who the regulars are. The community can no longer punish troublemakers by denying them easy transportation. Everyone flocks to the uber-zones, making 80-90% of the other zones ghost towns. And the perceived uber-zones are overcrowded hot spots of conflict. Troublemakers can more easily outrun their reputation.
The advantage in EVE is that it's wide-open PVP. (I've played both EQ, EQ2 and Eve.) So, if someone ticks you off, you can go hunting for them, or wardec their corporation, or hire mercenaries to hunt them down or make life difficult for them. There are ways to retaliate against griefers and idiots.
The other half of it is that Tech 2 isn't "all that". On average, I'd say tech 2 items are maybe 20% more powerful then Tech 1. It's not an "I Win" level of difference. So flying them isn't a super advantage, but being able to produce them using a BPO can make you rich (players will pay for even a small advantage). Yeah, Tech 2 items are nice to have... but they won't make you unkillable.
Best part about Eve is the fairly level playing field (a 2 week old rookie can still be useful in a fleet op) and the training system that doesn't require grinding. The only thing I have to grind is if I want to make ISK quickly. The skills I can just login for 5 minutes, set my next skill to start training, then logoff again.
Okay... a fan of that size usually is in the 2000-4500 RPM range. Which means 33-73 revolutions per second. So it would seem like a 1/60 sec shutter speed would involve blur.
Well it would, except that the discharge time for a Xeon strobe (a.k.a. the flash) is a lot less then 1/60th of a second. (The primary reason, I think, that 1/60th second is used was that on old, mechanical cameras, it was the easiest way to make sure the strobe fired while the shutter was open. Without requiring super expensive timing circuits or spending more on precision engineering. In fact, if you mistakenly left the shutter speed at 1/125 or 1/250, you would see correct exposure on half the frame with the other half blocked by the shutter due to mis-timing of the strobe.)
Unfortunately, I don't have hard numbers on discharge time. But since the fan appears to be stopped in flash photos for fans that spin at 2000-4500 RPM, you can figure that it's probably no longer then 1/1000 sec or even an order of magnitude faster.
(All educated guesses. I could be and probably am wrong.)
Makes me wonder just how much trouble the US or international financial community would be in if an adversarial organization cracked a major security encryption and didn't politely announce it, but instead kept their achievement secret. And then either cracked mountains of banking/military data at a leisurely pace, selling it piecemeal to finance rogue networks OR timed a widespread release of the crack algorithm for a catastrophic hit upon (inter)national security. What steps are being taken to combat this from eventually occurring?
Read back through CRYPTO-GRAM and the like (search for "munitions" and PGP).
This is why crypto folks were so hot under the collar back in the late 80s / early 90s about the munition laws. Due to the munition laws and other attempts to control encryption, it made it difficult to study existing algorithms for weaknesses. Or to develop new ones.
It's a bit like open-source vs closed-source software. The only reasonably secure encryption algorithm is one that has been peer-reviewed by a wide gamut of poeple. Such as the AES finalists or the ones recommended by NSA (which tend to be very closely examined by the crypto community), etc.
The government keeps wanting to revert to a closed-source model (Clipper, key-escrow, limiting the # of bits you can use, mandating the use of "weak" or secret algorithms).
They don't. It's simply the poor file system in Windows that doesn't support hardlinks (completely and/or correctly - take your pick). In unix / linux / BSD / Solaris (and OS X?) you can use hard links to make a file appear in multiple directories at the same time. One file, multiple locations, but without using up extra space.
Now, users might find that a bit confusing for the first year or two. But I think they'd understand it better then shortcuts that break all the time. Imagine if instead of "create shortcut" option we had "create linked copy" option in Windows.
Slashdot Mirror
Advantages to building it yourself:
- Get a motherboard bundle from a place that puts it together and tests that the CPU/RAM/MB are compatible and that it boots
- Buy a good quality case and you'll still be using it 10 years from now. A cheap case or proprietary case is just landfill fodder. I just put a brand new set of components into my original ATX tower case from 1997 (with a new PSU). That new system will probably run until at least 2015.
- You get a better quality system. You get control over what quality components go into the build instead of relying on Dell not to sacrifice quality in the name of profits. Less likely to have no-name parts inside the system.
I build a dozen or two machines per year. I never have random crashes - because I'm smart enough to test my system during the build. (All of my builds run Prime95 and other burn-in tools for at least a day to make sure that memory timings are correct and that the system doesn't fail under load.)
Yes, you run risks and you have to deal with warranties yourself. But it's very rare that you have a failed part arrive out of the box anymore. Imaging products like Acronis, Ghost, or NTFSClone take away the pain of a failed hard drive.
Here's our stock build for office PCs. Dual-core AMD X2s, 2GB RAM, RAID1 SATA drives, integrated video, WinXP Pro, Office 2003 Pro. Prices on RAM/CPU may be a bit dated (they've dropped recently). Part #s are from MWave.
$0132 AA15070 WindowsXP Pro OEM
$0299 AA24200 Microsoft Office Pro 2003 OEM
-----
$0431
$0138 MB-BA23082 AMD ATHLON 64 X2 3800+
$0084 Asus M2NPV-VM AM2 MicroATX Motherboard
$0150 Corsair 1024MBx2
$0009 Assemble/Test Bundle
-----
$0381
$0097 BA30107 Antec Sonata II case w/ 450W PSU
$0008 AA36600 NEC 1.44MB BLACK FLOPPY FD1231H-302
$0032 AA52110 NEC AD7170A 18X BLACK DVD MULTI WRITER
$0125 (2) SATA drives (200GB+)
-----
$0262
~$1100/unit
We've put a number of these in already and the users are very happy with them. Our goal is that they last at least 6 years and possibly as long as 12. That should be possible since they are dual-core.
I built a game machine for around $1500 last month, including SLI.
Only if you go for lowest cost. I have a few brands that I've worked with over the past 5 years that I trust. And one motherboard manufacturer where I've used a few dozen of their motherboards in systems with no issues.
I also plan for failures. That means RAID1 for work machines (the newer boards include RAID on the motherboard and a 2nd hard drive isn't that expensive) along with imaging the system periodically.
The newer motherboards (Asus M2N-E) use heatpipe and cooling fins instead of tiny fans that fail. That removes a big issue in longevity. And there are fanless video cards (or just go with an integrated video like the GeForce 6150 boards).
My experience been with the AMD Athlon CPUs where a screwdriver and significant amount of force is needed to put the clip where it belongs. It's no fun when the screwdriver slips and you think you just wiped out the motherboard.
True for the old Socket A chips. Not so true for the more modern 939/940/AM2 chips. They use a cam lever to get the necessary force, no screwdriver needed.
Personally, I still prefer to buy motherboard bundles where the stock heatsink has already been fitted. Saves me 5 minutes of horsing around (plus I know that the CPU is compatible with the board and they've at least booted the board once). Well worth the $9 charge for assembly and testing.
According to hearsay (no source) - Enterprise drives (SCSI/FC) are tested individually while consumer drives are only spot-checked as part of a batch.
I have no idea if that's true or not.
Last time I looked, 2.5" drives run on about 1W compared to 10-12W for 3.5" drives. That includes the newer 200GB 2.5" drives.
Your POV though, is correct when talking about using older 3.5" drives. A lot of folks try to save money by building a large RAID out of older 80-120GB drives. So they put 6-10 of these together at 10W each, when they could've simply used a pair of 500GB / 750GB drives in RAID1. Less noise, less heat, faster transfer rates are just a few advantages by using newer drives.
I have a firewall server that runs on a VIA EPIA and a pair of laptop drives. Not the least expensive solution in the world, but it's extremely low power and quiet (no fans).
If you had said 512MB, I'd have (mostly) agreed with you. Otherwise I think you're engaging in very wishful thinking for anyone other then a home user that doesn't do much more then read e-mail. As soon as you start getting into situations where the users are running 2-3 programs at the same time and keeping other applications in the system tray, that 400-800MHz system is going to be a dog. Especially in a work situation where you're interfering with productivity.
If it was a pair of 400MHz CPUs (dual-CPU setup), I'd give it more leeway.
There's a point where too cheap is wasteful. A modern, low-power, reasonably efficient dual-core system can be bought for $450 to $600. Plus fees for the OS and software.
Definitely nicer to work in a small company in that regard. I try to make sure the user's have what they need, we don't lock down the machines and we spent the $$$ on Acronis or Ghost to make periodic images of the systems. Most of our users are well-behaved enough now that I don't worry as much. The ones that aren't, I make sure I always know where the "gold" image is for their machine.
As for the firewall we simply tell them: "we know what you're doing".
Now, generally, I don't bother watching the logs to track packets or websites or whatever. But if bandwidth spikes or we start maxing out our line to the public network, the users know that I'm going to start looking for people to name and shame. (I'll actually address the issue privately first.)
In a small company, you can also say directly to the CEO - we can bump up the per-message limit on e-mail to 20MB, but it's going to require us to spend $10k on new server hardware within the year. So do you want us to spend the $10k or do you want to sit down and have a chat with Jane, Bob and Frank?
A very handy feature. After all, if you're going to have a RAID1 + hot-spare setup, there's some advantages to just making it a 3-way active mirror instead. When one of the primary drives dies, you don't have a recovery window while the hot-spare syncs up with the array.
The only downside would be that the hot-spare drive gets used and may suffer from wear-and-tear more then a hot-spare drive that spends its time powered down.
On the flip side... you don't have to sit and wonder if that hot-spare drive will work correctly when it's finally needed.
I suspect that absolute temperatures, within certain limits, are not important to drive reliability, but that temperature variation is.
That's my suspicion as well. Minimizing the amount of thermal cycling is probably important. This means adequate cooling to keep drives at a steady temperature state, no matter if they are spinning idle or operating under a heavy seek load. It may also mean that spinning drives down to save on power may cause premature failure due to the thermal variation.
I guess it would depend on how many times per hour you spin drives down or cycle them from 25C to 45C.
Please, mr dell, start a price war between RAM manufacturers next! I live in perpetual obsolescence thanks to the dramatic cost of DDR and DDR2! Won't someone think of the child processes!
Memory prices are already down to about where they were last July (2006). That's about 30-40% less then prices were in Nov/Dec 2006.
(nods)
FarCry is one of my favorite FPS for single-player. Devilishly tricky in places, but it gave you lots of ways to approach situations. You could even shake off your pursuers if there was enough cover around that you could hide in. Not to mention the ever-present supply of rocks that you could throw to distract the AI.
One fun way to clear out the dock area at the end of an early mission where you're trying to meet up with the doctor: Take out the rocket-shooting bugger on the hilltop tower. Then climb up the tower, pick up the rocket and zero in on the docks that are about 1-2km away. Shoot a single missile. Watch as all the AIs down at the docks go and investigate the explosion area. Shoot the rest of the missiles and watch the carnage.
It's a pretty wild game if you play in a manner more like a covert ops might. Getting a vantage point, scouting the opposition, drawing them off into an ambush. Dealing with the AI as it tries to flank you. Listening in on conversations, hearing the AI give orders and information to each other.
Dang... now I may go fire it up and blow my Saturday afternoon on it. (grin)
Yes, it gives them an advantage (in material and power). But I don't think they'll manage to take over all of 0.0 without falling apart. The odds are highly against them because of the following:
1) They have to keep growing their borders and taming more corporations to act as pets. If they don't grow or constantly engage in wars, their members may get bored.
2) At the same time, they have to keep reign on the existing pets. As the size of the pet population starts to outnumber the owner population, this will become more difficult. Pet corporations eventually will start asking themselves why they pay a fee to live in BoB space.
3) BoB no longer holds the moral high ground. This means that instead of attracting players with integrity, they may start attracting only the sociopaths. (Which is a bit of a black/white simplification.) While sociopaths may make good PvP'ers, I'd argue that they don't make good corp mates.
So I'd put odds of them managing to conquer the entire 0.0 space at about 10:1. They are very organized about it and it is their stated goals. But I don't think they can ultimately pull it off.
Maybe... they could've also had the pilots submit daily / weekly / monthly XML dumps of their character information from the "My Character" XML export:
s p?characterID=XXXXXXXXXX
http://myeve.eve-online.com/character/skilltree.a
What I'm not sure of is whether one player can look at another player's skilltree on the website, or whether it's locked down. (Pretty sure it's locked down.)
The XML file wouldn't be as trustworthy as getting it from the source (myeve.eve-online.com server), but if a pilot was slacking it would quickly be apparent in-game.
Alternately, the person in charge could require screenshots from the in-game character window that shows the skills.
Here's the Current Political Map
Look at the area of control for "BOB" down in the lower-left. I wouldn't call it 30-40%. Not even of 0.0 space (which is the area outside of empire influence). Well, maybe 30% of empire space. The area they just conquered is the region in the lower-right (now colored blue).
They do control a sizeable junk of 0.0 space though. And most of the rest of the corps in 0.0 space consider them a threat.
And while there are no current DX10 games, they'll be coming within the year. So the answer is: XP for games now, Vista by the year's end.
I find that overly optimistic.
While some game companies may release DX10 games this year, there's no way in hell that DX10 will be required for any titles until 2009-2010. Not unless the game company wants to sabotage their sales. While gamers tend to upgrade quicker, I'd still be surprised at faster then a 25-30% uptake per year.
Memtest86 / Memtest86+ are not very good at finding memory issues. They work, but only in cases where the memory is completely bad (an actual failed section).
What they don't do well is find RAM that is marginal, overlocked or misclocked. Because the only way to find such issues is to load the system down (CPU *and* RAM simultaneously). And Memtest doesn't do much with the CPU.
For real RAM testing, try running Prime95 in torture test mode for a few days. (Folks have been doing this for about 10 years now.) If you can manage to run Prime95 for a few days without any errors, you can call your system stable.
Or run the Gentoo server as a Xen guest OS. Prior to running the update, take it down and snapshot it. Do the update and you can roll back if needed.
Which is how I prefer to approach just about any OS. Be prepared to roll back on a minute's notice even if your test server said it was a no-risk upgrade.
1) I'd bet that the AMD chip is more picky about the memory that you have installed. Which is probably why you had reboot issues. It's one of the few downsides of having the memory controller on the CPU.
2) AFAIK, all AMD chips that came out after the AthlonXP series have on-chip sensors and thermal throttling (AMD Cool-n-quiet?). At the time of the infamous Tom's Hardware test, Intel chips had thermal throttling and AMD chips didn't. And up until Core and Core 2, Intel chips were affectionately known as space heaters, taking back the space heater crown from the old AMD Athlon chips.
3) I call bullshit on a CPU failure taking out other components.
I did not post about it but as you have identified one of SOE's problems in EQ was that their "Play Nice" policies were enforced in such a way that they really only applied to those who followed them and became disadvantaged as a result.
That's one of the reasons I stopped playing the original EQ.
Back in the day, if you wanted to travel long distances, you had 3 or 4 options:
1) Hoof it - Basically you spend time on travel.
2) Pay for a port - Spend plat/gold to reduce the time needed.
3) Befriend a porter - Get your travel for free, but you pay for it with other responsibilities.
4) Become self-sufficient - Level up a druid/wizard. Your travel is now free, but you put time into getting to that point.
When travel was more difficult, EQ was a nicer place. Folks didn't move around willy-nilly to the latest "hot" XP place. You tended to simply adventure wherever you happened to be instead of flocking to the uber-zone-of-the-week. This meant that you tended to meet and know most of the locals. Folks who were new in the zone tended to stand out. Troublemakers were quickly outed and they had trouble getting ahead of their reputations. Real troublemakers would even have difficulty getting ports (the community was able to punish troublemakers by denying them access to a player-provided service).
So you could cause trouble. But it would cause trouble for you and make life difficult for you.
Fast forward until after SoL and PoP were released. Travel has now become absolutely trivial. Because everyone is moving around willy-nilly, nobody knows who the regulars are. The community can no longer punish troublemakers by denying them easy transportation. Everyone flocks to the uber-zones, making 80-90% of the other zones ghost towns. And the perceived uber-zones are overcrowded hot spots of conflict. Troublemakers can more easily outrun their reputation.
Dunno if all that makes sense or not.
The advantage in EVE is that it's wide-open PVP. (I've played both EQ, EQ2 and Eve.) So, if someone ticks you off, you can go hunting for them, or wardec their corporation, or hire mercenaries to hunt them down or make life difficult for them. There are ways to retaliate against griefers and idiots.
The other half of it is that Tech 2 isn't "all that". On average, I'd say tech 2 items are maybe 20% more powerful then Tech 1. It's not an "I Win" level of difference. So flying them isn't a super advantage, but being able to produce them using a BPO can make you rich (players will pay for even a small advantage). Yeah, Tech 2 items are nice to have... but they won't make you unkillable.
Best part about Eve is the fairly level playing field (a 2 week old rookie can still be useful in a fleet op) and the training system that doesn't require grinding. The only thing I have to grind is if I want to make ISK quickly. The skills I can just login for 5 minutes, set my next skill to start training, then logoff again.
Okay... a fan of that size usually is in the 2000-4500 RPM range. Which means 33-73 revolutions per second. So it would seem like a 1/60 sec shutter speed would involve blur.
Well it would, except that the discharge time for a Xeon strobe (a.k.a. the flash) is a lot less then 1/60th of a second. (The primary reason, I think, that 1/60th second is used was that on old, mechanical cameras, it was the easiest way to make sure the strobe fired while the shutter was open. Without requiring super expensive timing circuits or spending more on precision engineering. In fact, if you mistakenly left the shutter speed at 1/125 or 1/250, you would see correct exposure on half the frame with the other half blocked by the shutter due to mis-timing of the strobe.)
Unfortunately, I don't have hard numbers on discharge time. But since the fan appears to be stopped in flash photos for fans that spin at 2000-4500 RPM, you can figure that it's probably no longer then 1/1000 sec or even an order of magnitude faster.
(All educated guesses. I could be and probably am wrong.)
If you don't like the moderations... make sure you meta-mod whenever possible (which undoes some of the damage).
Makes me wonder just how much trouble the US or international financial community would be in if an adversarial organization cracked a major security encryption and didn't politely announce it, but instead kept their achievement secret. And then either cracked mountains of banking/military data at a leisurely pace, selling it piecemeal to finance rogue networks OR timed a widespread release of the crack algorithm for a catastrophic hit upon (inter)national security. What steps are being taken to combat this from eventually occurring?
Read back through CRYPTO-GRAM and the like (search for "munitions" and PGP).
This is why crypto folks were so hot under the collar back in the late 80s / early 90s about the munition laws. Due to the munition laws and other attempts to control encryption, it made it difficult to study existing algorithms for weaknesses. Or to develop new ones.
It's a bit like open-source vs closed-source software. The only reasonably secure encryption algorithm is one that has been peer-reviewed by a wide gamut of poeple. Such as the AES finalists or the ones recommended by NSA (which tend to be very closely examined by the crypto community), etc.
The government keeps wanting to revert to a closed-source model (Clipper, key-escrow, limiting the # of bits you can use, mandating the use of "weak" or secret algorithms).
So why do directories have to be hierarchical?
They don't. It's simply the poor file system in Windows that doesn't support hardlinks (completely and/or correctly - take your pick). In unix / linux / BSD / Solaris (and OS X?) you can use hard links to make a file appear in multiple directories at the same time. One file, multiple locations, but without using up extra space.
Now, users might find that a bit confusing for the first year or two. But I think they'd understand it better then shortcuts that break all the time. Imagine if instead of "create shortcut" option we had "create linked copy" option in Windows.