RAM is actually sold at a loss as often as not. Manufacturers temd to go through cycles where they'll bleed huge amounts of money for a couple of years, then make a nice profit for a few years.
As far as pricing right now goes, last summer memory prices were the lowest I can remember having seen. I bought a stick of 512mb DDR for $58. Then Hynix got slapped with a major tariff by the US and EU for receiving multi-billion dollar subsidies from the Korean government, and prices shot up. IE: The market was flooded with not low margin, but significantly below cost RAM at the time. 3 months after I bought that $58 stick, the same stick was going for $120. Today it goes for about $80, so there's definitely been a price drop.
With regards to SDRAM being relatively expensive even though it's older technology, this is because manufacturing has shifted away from SDRAM so there's a reduced supply. Manufacturers can get better use of their facilities by producing DDR and flash memory (noticed how much cheaper flash cards and jump drives have gotten?). Expect SDRAM to continue to increase in price - the same thing happened to 72 pin EDO when SDRAM was introduced.
Barracudas are great in respect to noise. Samsung Spinpoints are even better. I have 4 160GB Spinpoints in various machines, and the only way I ever know they're working is when the HD LED on the case is on. They're as close to silent as I've ever had in a hard drive.
The SATA versions are slightly noisier though. I've noticed that if it's quiet enough, you can barely hear the head on the SATA drive seek on occassion.:-)
A couple of things here. Sticking with 2.4 is reasonable, but running an old version of 2.4 is a bad idea IMO. There are enough security vulns fixed every few releases that I'd seriously consider patching, if I were you. Know how we all pick on Windows kiddes for not updating? Linux doesn't give you a free pass to run unsecure versions either.
Second, even if a particular kernel has issues on your machine, there is *no* reason you would have to reformat. Simply create a new entry in your bootloader and leave the old Kernel as an option. That way if you forget to compile something you need in, you still have the old kernel to fall back on. This is the reason why when my laptop boots up, GRUB offers me a choice of the stock Slackware 2.4 kernel, and 4 or 5 2.6 revisions. HD space is cheap and kernel binaries are small - there's no reason not to.
Re: Multicore chips - AMD's actually in a better position with this too. The K8 was designed from day one with multicore in mind, with a P4 or PM type design, it'll just be hacked on. K8 was most definitely more forward looking in design in more than one way.
As was pointed out to me in another thread, the probability of failure is (1-P)^N. (I initially screwed up and said that it was P^N, which is actually the probability of non-failure. Oops.:-)).
At any rate, if you don't believe me, graph it. It may appear linear for samples as small as a RAID 0 array is likely to be, but the system is most definitely exponential.
It's a matter of probability. When you add drives to an array, your probability of failure goes up exponentially with each drive.
So yes, as far as losing data is concerned, it's the same both ways. However, the chances that you'll lost that data in the first place are much higher.
Here's the whole thing - I *have* tried it. If your workload involves lots of long, sequential reads, it's a great thing. I've personally got 2 machines running drives in RAID 0 as they get used for working with files in the 1.5-2GB range. It makes a difference here.
The whole point of SR and AT's articles, however, is that for most desktop systems, RAID 0 is pretty much a bad idea. You'll see marginal improvement on more random data sets, but you've spent four times as much, and, more importantly in my mind, your probability of failure has increased from P to P^4.
So really, I can see some applications where RAID 0 can be useful - I fit one of them. But for most desktop systems, it's not worth the cost. For systems with more than 2 drives anyway, it seems like a patentedly Bad Idea(TM). You really should've gone with RAID 5 - you'd still have striping, but you don't risk losing everything to a single faulty drive.
Yeah, I did an absolute double take when I got to that part. They spend an entire article bashing the two of the most methodical sites out there on methodilogy and then try to use a completely unscientific poll as backing evidence to their claim? Let alone a poll that's naturally pre-biased to a particular conclusion. It really puts the validity of the rest of the article into question. If that's acceptable evidence, what other shoddy methods are acceptable to them?
If you've spent the extra money on RAID 0, you're going to believe there's a difference going in. Hell, I've done it myself - I have 2 machines with RAID 0 setups, but that's because they're commonly used for working with multi-gig sized files in photoshop - IE: I actually need the strong sequential speed.
For normal desktop setups, I'd absolutely agree with AT and SR on this one. Unless you're doing massive amounts of large sequential reads/writes, you're just not going to see a difference in speed worth the cost of another drive and the major increase in potential failure and data loss. Remember, by adding that second drive, your chance of failure goes up *exponentially* which is something a lot of hardcore "tweakers" forget.
Actually, EMT64 is an incomplete clone of x86-64 by most reports, and doesn't appear to be binary compatible with x86-64. The x86-64 bit Linux distros are having to hack in support for the CPUs that essentially still does paging in software.
On top of that, all the ALUs on the CPU are still 32 bit, and it does not support the NX bit. There's a reason why Intel is only touting it as an "extended memory" architechture. It's an incomplete hack on top of the existing 32 bit chips that seems like nothing more than an attempt to save face by Intel.
I pretty much agree with you as far as Asus and Gigabyte boards go, but thought I'd throw in that if you're ever building a budget machine, seriously consider Biostar. I took a chance on some of their boards a couple of years ago because of the price, and have been nothing but impressed. Very sane board layouts (connectors are grouped in a reasonable manner, making routing through the case easy), fairly solid, and unbelievable customer support when something goes wrong.
I had a nForce2 IGP board with a dying northbridge fan, so I email customer support, expecting a long, frustrating ordeal which would probably end in me just buying a fan as I could just see them making me RMA the whole board. Instead, I get an email back asking me to fax a copy of the invoice from when I bought the board. I do that, and within *45 minutes* of me sending the fax in, I had a tracking number for a replacement fan - no questions asked. I had already been throroughly impressed with the quality of the boards, but that sealed my good impression of them.
Oh yeah, they run Linux perfectly too.:-) Just thought I'd mention them as a lot of people shy away due to them being so cheap, assuming they're of poor build quality, when I've found they're actually some of the best boards you can get for someone that doesn't need a $130 board with RAID, etc.
MSI boards are generally more expensive feature-for-feature than comparable Asus, Gigabyte, or ABit boards, all of which I generally consider to be of a higher quality anyway. In fact, the board you're talking about is the exact model that I've run into major driver issues with two different machines now. To make matters more interesting, the MSI GF4 Ti4200 in one of them *also* had driver issues.
Combine this with the fact that I've had nothing but bad luck with their optical drives - ie: a CD burner that burned more coasters even with buffer underrun protection than every other drive I've ever owned combined and a DVD drive that wouldn't read half the discs I threw at it... well, I guess you could see why I don't particularly like MSI.
Even if their equipment did work perfectly every time, I wouldn't really be a fan, since their motherboards have traditionally been more expensive than even Gigabyte, who built some of the best boards available as far as I'm concerned. Ah well, glad you had good luck, just hasn't been my experience at all.:-)
All I can say is you got lucky, and the chipsets must have been similar. I work in a PC repair shop and most of the time a motherboard swap with XP means a repair install. Not bashing, just saying your situation is not the norm.
I'd heard about trouble in particular with Asus nForce boards, but I'm not going to deny there may be problems either. I just know that my experience has been that the (Biostar) nForce boards I've got running Linux have been running flawlessly from day one. Slackware 9/10 with 2.6 kernels.
In the past, I'd agree with you - chipsets were the sticking point for AMD, but nForce2 is coming up on 2 years old this winter, and that was the turning point. Outside of one odd implementation (an MSI board that doesn't even use the standard drivers, but I dislike MSI anyway), I have yet to see an nForce2 machine with stability problems.
Of the 3 nForce2 based machines I own and all the ones I've built for other people, I've yet to come across a piece of hardware that didn't just work. Time to bring your notions about AMD out of 2001.:-)
Except AMD's roadmaps has shown that they'll continue supporting Socket 754 for the immediate future. By time it's discontinued, you'd probably be looking at a new motherboard to keep up with "modern" features anyway.
And really, since nVidia stepped in with the nForce series of chips, bringing the unified driver system from their video cards over, upgrading isn't much of an issue anymore. A friend of mine recently upgraded from an nForce2 based Athlon system to an nForce3 based Athlon 64 and it didn't require so much as a reload. Swapped the board and CPU, plugged in his drives and that was it.
That's my perspective on it anyway. Right now the price differential between Socket 754 and 939 just isn't worth it, especially given that once 939 becomes the normal commodity part, you'd probably be able to upgrade the CPU and mainboard for *less* than the price differential you'd pay now, and come out with more modern equipment.
In fact, for functions that don't need 64 bitness, 32 bit mode is preferable on the G5. 64 bit mode will actually be a little bit slower for code that doesn't require it.
Like you said, lots of people get confused by x86-64 bringing such a performance jump, but that's because x86-64 brings some major additions to the architechture. With the G5, 64 bitness mearly means it can natively do 64 bit math.
Because Apple believes that the PDA and cell phone markets will likely converge in the near future, and aren't really interested in competing in the cell phone market right now. Steve Jobs has said as much in numerous interviews.
Here's the thing: The Mac/Gnome button ordering comes specifically from usability testing. Apple didn't just arbitrarily come up with their HIG, they spent a lot of money on usability testing to arrive at those choices.
That said, Sun has also provided funds for usability testing, which went directly into Gnome's own HIG. It's my understanding that the Gnome project has been working towards adopting the recommendations of the HIG, so GoneME just really strikes me as a bunch of stubborn geeks who want things the old way simply because that's the way they learned it.
Usability testing reports can be absolutely fascinating, and after reading some, you'll start to find that some of the "quirks" of usability-centric UIs like MacOS and Gnome actually have a lot of thought and very sound logic behind them. Some people just won't allow themselves to move beyond "but the buttons are backwards from Windows/KDE!"
Right click on a folder, and select "Browse Folder." All of the sudden you have a tree view. The best of both worlds are available without changing a single setting.:-)
Every one hates spatial? News to me. In fact, I'd say there are many rather strong proponents of a spatial interface, that Ars article just being the tip of the iceberg.
Personally, I've found Spatial Nautilus to be excellent for situations where you have a relatively shallow directory tree - it lets you treat files more like objects than abstract collections of data, which is the whole point of a spatial interface. It has its weaknesses, sure, but a less than perfect initial implementation doesn't mean the whole concept is flawed.
Either way, it'd help people take you more seriously if you weren't essentially jumping up and down screaming "Look at me! I'm flamebait!"
You're a little less than coherent, but either way, you're a bit off track here.
XP was fairly stable at release. You have to keep in mind that NT was totally new ground for Microsoft. By time XP rolled around, they had not only a year of experience from Win2K, but all the years of experience from NT. They've had almost 3 years since then to further improve the system. Assuming SP2 doesn't introduce any major new problems, stability isn't really a concern for XP as it's been stable for quite some time.
Slashdot Mirror
Right click My Documents --> Properties --> Move --> click on the partition of your choice. Done.
;)
Or you could try ripping it out from under Windows and hoping things don't break, but somehow I think the actual way MS provides is better.
RAM is actually sold at a loss as often as not. Manufacturers temd to go through cycles where they'll bleed huge amounts of money for a couple of years, then make a nice profit for a few years.
As far as pricing right now goes, last summer memory prices were the lowest I can remember having seen. I bought a stick of 512mb DDR for $58. Then Hynix got slapped with a major tariff by the US and EU for receiving multi-billion dollar subsidies from the Korean government, and prices shot up. IE: The market was flooded with not low margin, but significantly below cost RAM at the time. 3 months after I bought that $58 stick, the same stick was going for $120. Today it goes for about $80, so there's definitely been a price drop.
With regards to SDRAM being relatively expensive even though it's older technology, this is because manufacturing has shifted away from SDRAM so there's a reduced supply. Manufacturers can get better use of their facilities by producing DDR and flash memory (noticed how much cheaper flash cards and jump drives have gotten?). Expect SDRAM to continue to increase in price - the same thing happened to 72 pin EDO when SDRAM was introduced.
Barracudas are great in respect to noise. Samsung Spinpoints are even better. I have 4 160GB Spinpoints in various machines, and the only way I ever know they're working is when the HD LED on the case is on. They're as close to silent as I've ever had in a hard drive.
:-)
The SATA versions are slightly noisier though. I've noticed that if it's quiet enough, you can barely hear the head on the SATA drive seek on occassion.
A couple of things here. Sticking with 2.4 is reasonable, but running an old version of 2.4 is a bad idea IMO. There are enough security vulns fixed every few releases that I'd seriously consider patching, if I were you. Know how we all pick on Windows kiddes for not updating? Linux doesn't give you a free pass to run unsecure versions either.
Second, even if a particular kernel has issues on your machine, there is *no* reason you would have to reformat. Simply create a new entry in your bootloader and leave the old Kernel as an option. That way if you forget to compile something you need in, you still have the old kernel to fall back on. This is the reason why when my laptop boots up, GRUB offers me a choice of the stock Slackware 2.4 kernel, and 4 or 5 2.6 revisions. HD space is cheap and kernel binaries are small - there's no reason not to.
Re: Multicore chips - AMD's actually in a better position with this too. The K8 was designed from day one with multicore in mind, with a P4 or PM type design, it'll just be hacked on. K8 was most definitely more forward looking in design in more than one way.
As was pointed out to me in another thread, the probability of failure is (1-P)^N. (I initially screwed up and said that it was P^N, which is actually the probability of non-failure. Oops. :-)).
At any rate, if you don't believe me, graph it. It may appear linear for samples as small as a RAID 0 array is likely to be, but the system is most definitely exponential.
It's a matter of probability. When you add drives to an array, your probability of failure goes up exponentially with each drive.
So yes, as far as losing data is concerned, it's the same both ways. However, the chances that you'll lost that data in the first place are much higher.
Here's the whole thing - I *have* tried it. If your workload involves lots of long, sequential reads, it's a great thing. I've personally got 2 machines running drives in RAID 0 as they get used for working with files in the 1.5-2GB range. It makes a difference here.
The whole point of SR and AT's articles, however, is that for most desktop systems, RAID 0 is pretty much a bad idea. You'll see marginal improvement on more random data sets, but you've spent four times as much, and, more importantly in my mind, your probability of failure has increased from P to P^4.
So really, I can see some applications where RAID 0 can be useful - I fit one of them. But for most desktop systems, it's not worth the cost. For systems with more than 2 drives anyway, it seems like a patentedly Bad Idea(TM). You really should've gone with RAID 5 - you'd still have striping, but you don't risk losing everything to a single faulty drive.
Yeah, I did an absolute double take when I got to that part. They spend an entire article bashing the two of the most methodical sites out there on methodilogy and then try to use a completely unscientific poll as backing evidence to their claim? Let alone a poll that's naturally pre-biased to a particular conclusion. It really puts the validity of the rest of the article into question. If that's acceptable evidence, what other shoddy methods are acceptable to them?
If you've spent the extra money on RAID 0, you're going to believe there's a difference going in. Hell, I've done it myself - I have 2 machines with RAID 0 setups, but that's because they're commonly used for working with multi-gig sized files in photoshop - IE: I actually need the strong sequential speed.
For normal desktop setups, I'd absolutely agree with AT and SR on this one. Unless you're doing massive amounts of large sequential reads/writes, you're just not going to see a difference in speed worth the cost of another drive and the major increase in potential failure and data loss. Remember, by adding that second drive, your chance of failure goes up *exponentially* which is something a lot of hardcore "tweakers" forget.
XP 2600+
Radeon 9600 Pro
1GB PC2700
160GB Samsung Spinpoint
1024x768, medium settings, 2x AA
Looks great, runs smooth
Actually, EMT64 is an incomplete clone of x86-64 by most reports, and doesn't appear to be binary compatible with x86-64. The x86-64 bit Linux distros are having to hack in support for the CPUs that essentially still does paging in software.
On top of that, all the ALUs on the CPU are still 32 bit, and it does not support the NX bit. There's a reason why Intel is only touting it as an "extended memory" architechture. It's an incomplete hack on top of the existing 32 bit chips that seems like nothing more than an attempt to save face by Intel.
I pretty much agree with you as far as Asus and Gigabyte boards go, but thought I'd throw in that if you're ever building a budget machine, seriously consider Biostar. I took a chance on some of their boards a couple of years ago because of the price, and have been nothing but impressed. Very sane board layouts (connectors are grouped in a reasonable manner, making routing through the case easy), fairly solid, and unbelievable customer support when something goes wrong.
:-) Just thought I'd mention them as a lot of people shy away due to them being so cheap, assuming they're of poor build quality, when I've found they're actually some of the best boards you can get for someone that doesn't need a $130 board with RAID, etc.
I had a nForce2 IGP board with a dying northbridge fan, so I email customer support, expecting a long, frustrating ordeal which would probably end in me just buying a fan as I could just see them making me RMA the whole board. Instead, I get an email back asking me to fax a copy of the invoice from when I bought the board. I do that, and within *45 minutes* of me sending the fax in, I had a tracking number for a replacement fan - no questions asked. I had already been throroughly impressed with the quality of the boards, but that sealed my good impression of them.
Oh yeah, they run Linux perfectly too.
Your case is definitely an exception then. :-)
:-)
MSI boards are generally more expensive feature-for-feature than comparable Asus, Gigabyte, or ABit boards, all of which I generally consider to be of a higher quality anyway. In fact, the board you're talking about is the exact model that I've run into major driver issues with two different machines now. To make matters more interesting, the MSI GF4 Ti4200 in one of them *also* had driver issues.
Combine this with the fact that I've had nothing but bad luck with their optical drives - ie: a CD burner that burned more coasters even with buffer underrun protection than every other drive I've ever owned combined and a DVD drive that wouldn't read half the discs I threw at it... well, I guess you could see why I don't particularly like MSI.
Even if their equipment did work perfectly every time, I wouldn't really be a fan, since their motherboards have traditionally been more expensive than even Gigabyte, who built some of the best boards available as far as I'm concerned. Ah well, glad you had good luck, just hasn't been my experience at all.
All I can say is you got lucky, and the chipsets must have been similar. I work in a PC repair shop and most of the time a motherboard swap with XP means a repair install. Not bashing, just saying your situation is not the norm.
I'd heard about trouble in particular with Asus nForce boards, but I'm not going to deny there may be problems either. I just know that my experience has been that the (Biostar) nForce boards I've got running Linux have been running flawlessly from day one. Slackware 9/10 with 2.6 kernels.
You mean the same warranty program AMD offers on their processors?
In the past, I'd agree with you - chipsets were the sticking point for AMD, but nForce2 is coming up on 2 years old this winter, and that was the turning point. Outside of one odd implementation (an MSI board that doesn't even use the standard drivers, but I dislike MSI anyway), I have yet to see an nForce2 machine with stability problems.
:-)
Of the 3 nForce2 based machines I own and all the ones I've built for other people, I've yet to come across a piece of hardware that didn't just work. Time to bring your notions about AMD out of 2001.
Except AMD's roadmaps has shown that they'll continue supporting Socket 754 for the immediate future. By time it's discontinued, you'd probably be looking at a new motherboard to keep up with "modern" features anyway.
And really, since nVidia stepped in with the nForce series of chips, bringing the unified driver system from their video cards over, upgrading isn't much of an issue anymore. A friend of mine recently upgraded from an nForce2 based Athlon system to an nForce3 based Athlon 64 and it didn't require so much as a reload. Swapped the board and CPU, plugged in his drives and that was it.
That's my perspective on it anyway. Right now the price differential between Socket 754 and 939 just isn't worth it, especially given that once 939 becomes the normal commodity part, you'd probably be able to upgrade the CPU and mainboard for *less* than the price differential you'd pay now, and come out with more modern equipment.
In fact, for functions that don't need 64 bitness, 32 bit mode is preferable on the G5. 64 bit mode will actually be a little bit slower for code that doesn't require it.
Like you said, lots of people get confused by x86-64 bringing such a performance jump, but that's because x86-64 brings some major additions to the architechture. With the G5, 64 bitness mearly means it can natively do 64 bit math.
Because Apple believes that the PDA and cell phone markets will likely converge in the near future, and aren't really interested in competing in the cell phone market right now. Steve Jobs has said as much in numerous interviews.
Dropline Gnome isn't a fork, it's a meta-distribution on top of Slackware. There's a world of difference between the two.
A fork is when you're working from two distinct and seperate codebases with a common origin.
A (meta) distribution is a bundle of software packaged together and generally set up to be as cohesive as possible.
Here's the thing: The Mac/Gnome button ordering comes specifically from usability testing. Apple didn't just arbitrarily come up with their HIG, they spent a lot of money on usability testing to arrive at those choices.
That said, Sun has also provided funds for usability testing, which went directly into Gnome's own HIG. It's my understanding that the Gnome project has been working towards adopting the recommendations of the HIG, so GoneME just really strikes me as a bunch of stubborn geeks who want things the old way simply because that's the way they learned it.
Usability testing reports can be absolutely fascinating, and after reading some, you'll start to find that some of the "quirks" of usability-centric UIs like MacOS and Gnome actually have a lot of thought and very sound logic behind them. Some people just won't allow themselves to move beyond "but the buttons are backwards from Windows/KDE!"
Share a "secret" with you?
:-)
Right click on a folder, and select "Browse Folder." All of the sudden you have a tree view. The best of both worlds are available without changing a single setting.
Every one hates spatial? News to me. In fact, I'd say there are many rather strong proponents of a spatial interface, that Ars article just being the tip of the iceberg.
Personally, I've found Spatial Nautilus to be excellent for situations where you have a relatively shallow directory tree - it lets you treat files more like objects than abstract collections of data, which is the whole point of a spatial interface. It has its weaknesses, sure, but a less than perfect initial implementation doesn't mean the whole concept is flawed.
Either way, it'd help people take you more seriously if you weren't essentially jumping up and down screaming "Look at me! I'm flamebait!"
You're a little less than coherent, but either way, you're a bit off track here.
XP was fairly stable at release. You have to keep in mind that NT was totally new ground for Microsoft. By time XP rolled around, they had not only a year of experience from Win2K, but all the years of experience from NT. They've had almost 3 years since then to further improve the system. Assuming SP2 doesn't introduce any major new problems, stability isn't really a concern for XP as it's been stable for quite some time.