Linus' tactic of not doing much to improve multiprocessing support hurt the high end, at the dubious claim that it had to hurt the low end.
I've thought for a long time that you want different schedulers available for different scales of multiprocessing. Heck, even Windows has different "drivers" for uniprocessor and dual processor machines for W2K Pro.
There are a couple on-board header-type jacks that aren't accounted for, very close to the I/O jack area.
I agree. RS-232 is great, and it is simple. With some simple programming trickery, any $2 microcontroller can do it. With a little less trickery and using in-built hardware on said chip plus a few cents on a MAX232 chip, and with a $100 parts & programmer budget, one can make a very nice peripheral that is easy to program for. This is not anywhere nearly as cheap or easy as USB, PCI or any other I/O standard that I know.
There were times where I wanted more than four ports, simply because I had a video projector control, microchip programmer, RS-232 Palm dock plus my hardware project test unit connected, as well as wanting a real hardware modem connected.
For industrial uses, you want a connector that you can screw into for a truly positive vibration resistant connection. Actually, it's too bad that there wasn't any USB / ADB / PS/2 port standard that had a positive hold. Screws or thumb-spring latches would do.
It depends which circuit they are in, the DMCA subpoena process was upheld against Verizon and I don't think the Supreme Court bothered to pick up the appeal, so they just might see it as a moot point, a losing battle.
Bummer, you have to have a support contract. I have a Digital Alpha and a couple Compaq Xeons. The Alpha is probably completely out of support due to age, 1997 build I think.
Re:this is great but...
on
Athlon 64 Debuts
·
· Score: 3, Interesting
I don't get this. Aren't you missing the point? The idea is that it is supposed to be forward compatible. Simply because there were no 32 bit apps at the time didn't mean that people didn't buy 386s and 486s.
If you are waiting for everything to catch up, then there is currently little advantage to x86-64.
Most apps don't need this power, the ones that do will be rewritten as the need arises, everything else can still run in 32 bit mode.
Think mass market and not the techno-gadget of the month.
Try fitting "QWERTY" on a mobile device. It would require a longer and skinnier pad, or in the case of that phone, you'd be making it shorter and wider.
The world doesn't need another typewriter keyboard crammed into a tiny device.
Believe it or not, fewer people have learned to type than you think, and not everyone bothers. For a device like this, I think it would be better, it's not as if there would be any easy way to standardize on any one key arrangement because the needs of and form factor of each device type are different.
I think the fourty five degree corners is about as much of a compromise as one can get without being too expensive with routing.
Another problem is that the speed of memory itself isn't that great unless you want to spend a _lot_ of money, to the tune of $50-$100 per megabyte as we see in advanced processor caches, and the faster it is, the more very power inefficient it becomes, maybe to a sizeable fraction of a watt per megabyte.
100MPH is all most cars really need, and IMO the power is where it counts.
Beyond 100MPH for a car meant for highway is pretty much in the range of felonious speeding anyway (assuming there is a felony range for speeding), rather than just civil infraction if you go 5-15 over.
I agree one sub-thread up that "zero emissions" still doesn't address the the problem of manufacture and disposal of batteries, and doesn't address how the electricity gets there. With every source of electricity so far, there is a negative environmental impact even if no smog chemicals are made.
I don't think that will happen for a _long_ time. For one, the screen size dictates the available space as it dictates the package size. Power consumption dictates power supply size, which takes away from available montherboard / battery / drive space. At the moment, drive sizes are pretty stable and that's about it.
There are simply too many trade-offs. AT/ATX lasted so long because they had so much room and you didn't have to integrate monitor, mouse, keyboard and a battery into an ultra compact (and shrinking) and ultra light (and lightening) package.
Everything still points to custom job, I don't think it will improve jack.
I don't see a problem, there are plenty of two way ATX boards, and you can easily get ATX power supplies that go up to and over 600 watts, I see no issue with it happening with BTX.
I don't know if there is anything of a standardized case for big iron.
I would expect that Athlon64 and Opteron systems (up to 2-way) would use BTX. Four-way systems of any kind really stretch the space, particularly considering that each A64 needs its own RAM bank(s) I am betting that is heading into the proprietary-only region.
One thing to keep in mind is that when PCI came around, people had a hard time fathoming what could take advantage of it. Initially, AGP showed little advantage over PCI. There may be something yet, I wouldn't discount it just yet.
I hate that too, most hardware sites do that. The sad part is that the side bar menus have too much crap on them too. At least Anand and a few others have a print version that strips most of the ads and stupid menus. So far that I've found, Tom's and many others don't have this courtesy.
That is nice but I prefer to use two drives. It helps for copying discs. One can go external, but usually that's an extra $50-$100 to do so. I would want an extra cheaper CD writer so that I'm not putting extra wear and tear on an expensive drive to write a CD.
Having only two internal hard drive slots is also limiting. Heck, I know one video guy that puts three drives in a system - one system drive and two data drives in a stripe. Can't do that here.
I like the G5 systems, but they seem too limited in internal expansion for a true power user, thus relegating them to buying expensive external devices. IMO, non power users don't even remotely need that kind of a system yet.
In a way, I think AGP really didn't need to happen. AGP mostly added a pipelined transfer system, the double and quadruple data rate and the ability to use system memory for texturing. I think they could have done this by adding these features to PCI. To the OS, an AGP bus looks like another PCI bus anyway because of how the bridge operates.
I think PCI-X is much like a merging of AGP concepts and adding clocks and I think an optional wider & faster data path. What is nice is that from what I've read, each slot is essentially its own bus, so there is no contention between devices, potentially increasing the total system I/O rates.
It is fine if people want to make up asinine units of measurement. It is entirely another thing to try use them outside of the technical fields where their use makes sense.
I would rather not have any consumer device present numbers of this kind to a user simply because it will not help their ability to use the system. Just because it is important to the technical people doesn't mean normal people should be subjected to it. Few put up with the details of accountants, physicians, politicians, mechanics, etc and I don't think they should have to deal with it from computer people either, because it is simply another point of alienation.
People that did their research would know this. Anyone that has simply read the retail box for a hard drive upgrade would know this.
They aren't making your icecream disappear or giving you less gasoline. I think it is the system designer's fault for showing the number based on powers of two. The computer gigabyte (1024^3) is in some ways hiding extra space that you really got anyways, just over seven percent.
Even if the management were better, one still has the quandry of only something like a 3% installed base for desktop systems. Games are complex and a total rework to expand your market by a few percent simply doesn't make sense.
My impression is that the P4 Xeon has only had two sockets so far, 603 and 604. The P3 Xeon at least had slot 2, I don't know if they had a socket or not. If you bundled all of Xeon together, then yes, they've had at least three sockets.
I never really expected it to be compatible, and there's so little information on what is in Prescott, I simply assumed it to be a P5 anyway.
Cache on that order of size only helps for memory or bandwidth intensive applications. If the data and instructions for any particular thread fit in 512k, then you get no benefit for that thread.
Slashdot Mirror
Linus' tactic of not doing much to improve multiprocessing support hurt the high end, at the dubious claim that it had to hurt the low end.
I've thought for a long time that you want different schedulers available for different scales of multiprocessing. Heck, even Windows has different "drivers" for uniprocessor and dual processor machines for W2K Pro.
There are a couple on-board header-type jacks that aren't accounted for, very close to the I/O jack area.
I agree. RS-232 is great, and it is simple. With some simple programming trickery, any $2 microcontroller can do it. With a little less trickery and using in-built hardware on said chip plus a few cents on a MAX232 chip, and with a $100 parts & programmer budget, one can make a very nice peripheral that is easy to program for. This is not anywhere nearly as cheap or easy as USB, PCI or any other I/O standard that I know.
There were times where I wanted more than four ports, simply because I had a video projector control, microchip programmer, RS-232 Palm dock plus my hardware project test unit connected, as well as wanting a real hardware modem connected.
For industrial uses, you want a connector that you can screw into for a truly positive vibration resistant connection. Actually, it's too bad that there wasn't any USB / ADB / PS/2 port standard that had a positive hold. Screws or thumb-spring latches would do.
"The RAM slot looks like it takes laptop ram, not stadard desktop DIMMs."
Yes, that's what a SODIMM is, as shown in what the slashdot 'story' said "moved to mini-PCI and SODIMMs"
I don't know but I really doubt they need special drivers unless you call OHCI special, which it is not.
True. Tere is one such four drive / dual chain enclosure:
http://www.cooldrives.com/qubayredk1st.html
I've seen an eight drive enclosure too.
There are brainboards with backing plates that that are specifically intended to replace the SCSI ports on old SCSI enclosures.
I must say, after looking at that site for a second, especially seeing a part of the, the um, picture, I have no interest in ever trying it.
It depends which circuit they are in, the DMCA subpoena process was upheld against Verizon and I don't think the Supreme Court bothered to pick up the appeal, so they just might see it as a moot point, a losing battle.
Bummer, you have to have a support contract. I have a Digital Alpha and a couple Compaq Xeons. The Alpha is probably completely out of support due to age, 1997 build I think.
I don't get this. Aren't you missing the point? The idea is that it is supposed to be forward compatible. Simply because there were no 32 bit apps at the time didn't mean that people didn't buy 386s and 486s.
If you are waiting for everything to catch up, then there is currently little advantage to x86-64.
Most apps don't need this power, the ones that do will be rewritten as the need arises, everything else can still run in 32 bit mode.
Think mass market and not the techno-gadget of the month.
Try fitting "QWERTY" on a mobile device. It would require a longer and skinnier pad, or in the case of that phone, you'd be making it shorter and wider.
The world doesn't need another typewriter keyboard crammed into a tiny device.
Believe it or not, fewer people have learned to type than you think, and not everyone bothers. For a device like this, I think it would be better, it's not as if there would be any easy way to standardize on any one key arrangement because the needs of and form factor of each device type are different.
I think the fourty five degree corners is about as much of a compromise as one can get without being too expensive with routing.
Another problem is that the speed of memory itself isn't that great unless you want to spend a _lot_ of money, to the tune of $50-$100 per megabyte as we see in advanced processor caches, and the faster it is, the more very power inefficient it becomes, maybe to a sizeable fraction of a watt per megabyte.
100MPH is all most cars really need, and IMO the power is where it counts.
Beyond 100MPH for a car meant for highway is pretty much in the range of felonious speeding anyway (assuming there is a felony range for speeding), rather than just civil infraction if you go 5-15 over.
I agree one sub-thread up that "zero emissions" still doesn't address the the problem of manufacture and disposal of batteries, and doesn't address how the electricity gets there. With every source of electricity so far, there is a negative environmental impact even if no smog chemicals are made.
I don't think that will happen for a _long_ time. For one, the screen size dictates the available space as it dictates the package size. Power consumption dictates power supply size, which takes away from available montherboard / battery / drive space. At the moment, drive sizes are pretty stable and that's about it.
There are simply too many trade-offs. AT/ATX lasted so long because they had so much room and you didn't have to integrate monitor, mouse, keyboard and a battery into an ultra compact (and shrinking) and ultra light (and lightening) package.
Everything still points to custom job, I don't think it will improve jack.
I don't see a problem, there are plenty of two way ATX boards, and you can easily get ATX power supplies that go up to and over 600 watts, I see no issue with it happening with BTX.
I don't know if there is anything of a standardized case for big iron.
I would expect that Athlon64 and Opteron systems (up to 2-way) would use BTX. Four-way systems of any kind really stretch the space, particularly considering that each A64 needs its own RAM bank(s) I am betting that is heading into the proprietary-only region.
One thing to keep in mind is that when PCI came around, people had a hard time fathoming what could take advantage of it. Initially, AGP showed little advantage over PCI. There may be something yet, I wouldn't discount it just yet.
I hate that too, most hardware sites do that. The sad part is that the side bar menus have too much crap on them too. At least Anand and a few others have a print version that strips most of the ads and stupid menus. So far that I've found, Tom's and many others don't have this courtesy.
That is nice but I prefer to use two drives. It helps for copying discs. One can go external, but usually that's an extra $50-$100 to do so. I would want an extra cheaper CD writer so that I'm not putting extra wear and tear on an expensive drive to write a CD.
Having only two internal hard drive slots is also limiting. Heck, I know one video guy that puts three drives in a system - one system drive and two data drives in a stripe. Can't do that here.
I like the G5 systems, but they seem too limited in internal expansion for a true power user, thus relegating them to buying expensive external devices. IMO, non power users don't even remotely need that kind of a system yet.
In a way, I think AGP really didn't need to happen. AGP mostly added a pipelined transfer system, the double and quadruple data rate and the ability to use system memory for texturing. I think they could have done this by adding these features to PCI. To the OS, an AGP bus looks like another PCI bus anyway because of how the bridge operates.
I think PCI-X is much like a merging of AGP concepts and adding clocks and I think an optional wider & faster data path. What is nice is that from what I've read, each slot is essentially its own bus, so there is no contention between devices, potentially increasing the total system I/O rates.
Was this default share = on intended to be devious? I've never run the program so I really don't know.
It is fine if people want to make up asinine units of measurement. It is entirely another thing to try use them outside of the technical fields where their use makes sense.
I would rather not have any consumer device present numbers of this kind to a user simply because it will not help their ability to use the system. Just because it is important to the technical people doesn't mean normal people should be subjected to it. Few put up with the details of accountants, physicians, politicians, mechanics, etc and I don't think they should have to deal with it from computer people either, because it is simply another point of alienation.
People that did their research would know this. Anyone that has simply read the retail box for a hard drive upgrade would know this.
They aren't making your icecream disappear or giving you less gasoline. I think it is the system designer's fault for showing the number based on powers of two. The computer gigabyte (1024^3) is in some ways hiding extra space that you really got anyways, just over seven percent.
Even if the management were better, one still has the quandry of only something like a 3% installed base for desktop systems. Games are complex and a total rework to expand your market by a few percent simply doesn't make sense.
My impression is that the P4 Xeon has only had two sockets so far, 603 and 604. The P3 Xeon at least had slot 2, I don't know if they had a socket or not. If you bundled all of Xeon together, then yes, they've had at least three sockets.
I never really expected it to be compatible, and there's so little information on what is in Prescott, I simply assumed it to be a P5 anyway.
Cache on that order of size only helps for memory or bandwidth intensive applications. If the data and instructions for any particular thread fit in 512k, then you get no benefit for that thread.