I run AMD on my highest demand boxes because the Hypertransport bus kicks the everloving crap out of a slow FSB for memory sharing between processors. I'm in good company, the NYSE runs much of their infrastructure on HP DL585's running linux. For high throughput processing on commodity hardware it's hard to beat. Intel might catch up around 2H'09 when they bring out the multiprocessor version of the i7.
Anywhere with decent PC pricing, but they don't have the Canadian media tax on them so I guess I can understand why they are more expensive up there even if I disagree with the reason. Even so, at a buck a piece they are no more expensive than a VHS tape and barring little ones dragging them across the floor should last much longer.
An hour of video at much better than VHS quality is only ~700MB using H.264, with 1TB HDD's costing about $100 you get 1,300 hours of recording for a pittance. Even if you use a cheaper recorder that does MPEG2 you are still looking at over 500 hours (almost a month) for that price.
VHS tapes will last for many many recordings whereas DVDs do not (cheaply, at least).
Uh, a spindle of 50 DVD+-RW's costs about 10 bucks, each can be recorded to many, many times with zero loss in quality. That's significantly better than VHS. As to your moving tapes comments, get a couple Tivo's, then you can share between them and even with your PC or portables.
Quantum has drivers for XP for the later Traven drives, there are IDE/SATA bridge chips that will work with the oldest IDE drives, parallel SCSI cards are still available (the classic for tape drives is the Adaptec 29160 which also has an internal 50 pin header which will work with SCSI-1 drives). The only common media I'm aware of that you can't easily read on a modern PC is the 8" floppy and MFM/RLL HDD's. Everything else going back 30 years can be read today.
There are patches out there to disable Macrovision on ATI cards, the first link on that page is to one such tool. Not sure if it works on current drivers but you could always downgrade while ripping.
Get better HP contracts and it will be Atlanta and they won't mess around (especially with 6hr call to repair contracts, only about 20% more expensive than a normal 4hr onsite contract but they really seem to pay attention when you mention the 6hr ctr). Also I've yet to have a significant issue fixed by TAC India, my best luck has been with the Aussies.
Like Crysis at 1600*1200 at 20fps? Yeah, I think most people will take that "low" resolution and slow gameplay. Oh and Sacred 2 at 30-40fps at 1080p, it's so damn annoying playing an incredible looking game at smooth framerates...
Oh, and I don't buy Cisco gear for performance, I buy it for reliability. Their chassis based switches are simply tanks and I have a very good track record with them, but they've always been overkill for the environments I've used them in, not sure how they fare when you really push them to the wall.
The difference is in an ethernet/TCP network you generally just get poorer performance unless you are really pushing things, in a FC environment things fall over when delivery can't be made. Our 9140's fell over at about 1.7Gbps max total throughput, they claimed to support 2Gbps on the first 8 ports and 4Gbps on each group of 4 for the other 32 ports. They tried to blame it on our storage array even after being shown the proof from their own tech captures and the detailed logging from our storage controllers. We replaced them with Brocades and all of our problems went away.
You already have that with enclosure like the HP C-class, only it involves well engineered and supported blades and your choice of fiberchannel switches and/or Infiniband. They even offer storage and backup blades so you can do a datacenter in a box for midsized remote offices.
Their Fibre Channel switches SUCK. I had the most downtime I've ever had in my career recently due to their sucky designs. They massively oversubscribe things and they have way too few hardware B2B credits. Their big switches are an absolute joke. Look at Brocade's DCX vs a 9513, the Brocade can do full 8Gb full duplex bandwidth between every port on a 48 port blade and has 256Gbps of backplane which doesn't even get touched if your storage and hosts are on the same blade because the blades do local switching. By comparison the brand new gen 2 9513 fabric has 96Gbps and no local switching between ASIC's! In the real world even the old 48000 series is better with "only" 64Gbps of backbone due to the local switching. After having been bitten by Cisco it will be a loooong time before I even reconsider them.
p.s.
For a good comparison see this page, yeah I know it's from Brocade but it's all truthful information.
Actually 2003's scheduler is Hyperthreading aware which is why it's not a problem leaving in enabled under 2003, under 2000 it's often recommended to disable the feature in BIOS for maximum performance. 2003's scheduler is also NUMA aware and will try to keep threads on a processor close to the memory they are allocating/have allocated. Just because you are not aware of the advances doesn't mean they haven't been made.
WRONG, the 9600GSO will play modern games just fine and it maxes at 45W power draw, combine it with a 45W AMDx2 and an efficient motherboard and you are looking at a peak draw under 150W, assuming it has decent 12v rails a 350W power supply will be fine with this setup. Mine has a 450W but that's because I wanted an 80+ PSU with a 120mm fan and the only affordable unit I could find two years ago was a 450W one. You won't find this setup from your local OEM, but it's entirely possible to build it =)
Crysis isn't THAT hard on the GPU, my lowly 9600GSO can play it at around 20fps at 1080p resolution, a GTX260 can do 26fps min 40fps avg. That's a $250 card running the game perfectly fine at the max resolution that many people's monitor supports.
The Windows NT kernel was doing SMP when trying to run Linux on an SMP boxes was experimental and very immature, I should know I tested both back then. The problem isn't the OS, it's the way games are designed. If game developers did things like split off AI, physics, networking, etc into their own threads you would likely see some improvement on multicore systems (though you would have to evaluate message passing overhead vs extra resource availability). The fact is game development for almost all shops is a lower margin business so it's just not worth it to them to add the development and testing complexity.
Back in 1998 I knew someone that bought a pair of ~$500 (even more inflation adjusted) Voodoo2 graphics cards to play GLQuake at the crazy resolution of 1024*768 =)
Today I have a 350GFlop GPU that maxes out at 45W of power draw. That's comparable to a top of the line Cray T3D from 1993 which would have had 2048 processors and used hundreds of KW of power =)
It doesn't matter, under the Berne Convention and the Berne Convention Implementation Act of 1988 you do not need to register your work to hold copyright and have protections under copyright law. The only remedy not open to Apple is attorneys fees and statutory damages if they didn't register, since their main goal is a permanent injunction I don't think they care.
Look into Xen/KVM for your open source hypervisor needs. If you want a supported solution most of the application houses have one or you can go with Citrix Xenserver for generic ESX style virtualization. We use the HP branded Xen Express because it gives us a nice virtual KVM (not the same as the KVM app, stupid opensource people reusing an already popular TLA) through the iLo for our Windows guests.
As you can see even going to 440 isn't going to get you anywhere. Oh and then you have to multiply times ~1.1 for PF correction as I'm sure it doesn't look like a resistive load.
Never, in any context is it ok to write incorrect information
WRONG, language is about communications and so long as the point was conveyed correctly and the meaning non-ambiguous there is no problem messing up the case. This is especially true when a bunch of lay-people are discussing a technical area. People like you need to adjust their attitude to realize that human languages are NOT the same as programming languages or scientific equations, some of the time getting it close enough really is ok.
Right because we have enough non-fossil plants today to meet transportation demands. Oh and there's that problem with distribution, we already have a blackout about once a decade in the NE, can you imagine if everyone decides to recharge their cars before their summer vacation?
Abit was probably a wholly owned subsidiary or some equally fun legal construct so while the profits flowed to the parent company I would bet they are absolved of any liability for Abit products.
Slashdot Mirror
I run AMD on my highest demand boxes because the Hypertransport bus kicks the everloving crap out of a slow FSB for memory sharing between processors. I'm in good company, the NYSE runs much of their infrastructure on HP DL585's running linux. For high throughput processing on commodity hardware it's hard to beat. Intel might catch up around 2H'09 when they bring out the multiprocessor version of the i7.
Anywhere with decent PC pricing, but they don't have the Canadian media tax on them so I guess I can understand why they are more expensive up there even if I disagree with the reason. Even so, at a buck a piece they are no more expensive than a VHS tape and barring little ones dragging them across the floor should last much longer.
An hour of video at much better than VHS quality is only ~700MB using H.264, with 1TB HDD's costing about $100 you get 1,300 hours of recording for a pittance. Even if you use a cheaper recorder that does MPEG2 you are still looking at over 500 hours (almost a month) for that price.
VHS tapes will last for many many recordings whereas DVDs do not (cheaply, at least).
Uh, a spindle of 50 DVD+-RW's costs about 10 bucks, each can be recorded to many, many times with zero loss in quality. That's significantly better than VHS. As to your moving tapes comments, get a couple Tivo's, then you can share between them and even with your PC or portables.
Quantum has drivers for XP for the later Traven drives, there are IDE/SATA bridge chips that will work with the oldest IDE drives, parallel SCSI cards are still available (the classic for tape drives is the Adaptec 29160 which also has an internal 50 pin header which will work with SCSI-1 drives). The only common media I'm aware of that you can't easily read on a modern PC is the 8" floppy and MFM/RLL HDD's. Everything else going back 30 years can be read today.
There are patches out there to disable Macrovision on ATI cards, the first link on that page is to one such tool. Not sure if it works on current drivers but you could always downgrade while ripping.
Get better HP contracts and it will be Atlanta and they won't mess around (especially with 6hr call to repair contracts, only about 20% more expensive than a normal 4hr onsite contract but they really seem to pay attention when you mention the 6hr ctr). Also I've yet to have a significant issue fixed by TAC India, my best luck has been with the Aussies.
Like Crysis at 1600*1200 at 20fps? Yeah, I think most people will take that "low" resolution and slow gameplay. Oh and Sacred 2 at 30-40fps at 1080p, it's so damn annoying playing an incredible looking game at smooth framerates...
Oh, and I don't buy Cisco gear for performance, I buy it for reliability. Their chassis based switches are simply tanks and I have a very good track record with them, but they've always been overkill for the environments I've used them in, not sure how they fare when you really push them to the wall.
The difference is in an ethernet/TCP network you generally just get poorer performance unless you are really pushing things, in a FC environment things fall over when delivery can't be made. Our 9140's fell over at about 1.7Gbps max total throughput, they claimed to support 2Gbps on the first 8 ports and 4Gbps on each group of 4 for the other 32 ports. They tried to blame it on our storage array even after being shown the proof from their own tech captures and the detailed logging from our storage controllers. We replaced them with Brocades and all of our problems went away.
You already have that with enclosure like the HP C-class, only it involves well engineered and supported blades and your choice of fiberchannel switches and/or Infiniband. They even offer storage and backup blades so you can do a datacenter in a box for midsized remote offices.
Their Fibre Channel switches SUCK. I had the most downtime I've ever had in my career recently due to their sucky designs. They massively oversubscribe things and they have way too few hardware B2B credits. Their big switches are an absolute joke. Look at Brocade's DCX vs a 9513, the Brocade can do full 8Gb full duplex bandwidth between every port on a 48 port blade and has 256Gbps of backplane which doesn't even get touched if your storage and hosts are on the same blade because the blades do local switching. By comparison the brand new gen 2 9513 fabric has 96Gbps and no local switching between ASIC's! In the real world even the old 48000 series is better with "only" 64Gbps of backbone due to the local switching. After having been bitten by Cisco it will be a loooong time before I even reconsider them.
p.s.
For a good comparison see this page, yeah I know it's from Brocade but it's all truthful information.
Your rep sucks, our orders both directly from Cisco and from CDW arrive in a normal timeframe of several days.
Actually 2003's scheduler is Hyperthreading aware which is why it's not a problem leaving in enabled under 2003, under 2000 it's often recommended to disable the feature in BIOS for maximum performance. 2003's scheduler is also NUMA aware and will try to keep threads on a processor close to the memory they are allocating/have allocated. Just because you are not aware of the advances doesn't mean they haven't been made.
WRONG, the 9600GSO will play modern games just fine and it maxes at 45W power draw, combine it with a 45W AMDx2 and an efficient motherboard and you are looking at a peak draw under 150W, assuming it has decent 12v rails a 350W power supply will be fine with this setup. Mine has a 450W but that's because I wanted an 80+ PSU with a 120mm fan and the only affordable unit I could find two years ago was a 450W one. You won't find this setup from your local OEM, but it's entirely possible to build it =)
Crysis isn't THAT hard on the GPU, my lowly 9600GSO can play it at around 20fps at 1080p resolution, a GTX260 can do 26fps min 40fps avg. That's a $250 card running the game perfectly fine at the max resolution that many people's monitor supports.
Windows isn't built for more than one core CPU
The Windows NT kernel was doing SMP when trying to run Linux on an SMP boxes was experimental and very immature, I should know I tested both back then. The problem isn't the OS, it's the way games are designed. If game developers did things like split off AI, physics, networking, etc into their own threads you would likely see some improvement on multicore systems (though you would have to evaluate message passing overhead vs extra resource availability). The fact is game development for almost all shops is a lower margin business so it's just not worth it to them to add the development and testing complexity.
Back in 1998 I knew someone that bought a pair of ~$500 (even more inflation adjusted) Voodoo2 graphics cards to play GLQuake at the crazy resolution of 1024*768 =)
Today I have a 350GFlop GPU that maxes out at 45W of power draw. That's comparable to a top of the line Cray T3D from 1993 which would have had 2048 processors and used hundreds of KW of power =)
I guess that's why Oracle costs so damn much, it's quite happy using as many cores/cpu's as you've got.
It doesn't matter, under the Berne Convention and the Berne Convention Implementation Act of 1988 you do not need to register your work to hold copyright and have protections under copyright law. The only remedy not open to Apple is attorneys fees and statutory damages if they didn't register, since their main goal is a permanent injunction I don't think they care.
Look into Xen/KVM for your open source hypervisor needs. If you want a supported solution most of the application houses have one or you can go with Citrix Xenserver for generic ESX style virtualization. We use the HP branded Xen Express because it gives us a nice virtual KVM (not the same as the KVM app, stupid opensource people reusing an already popular TLA) through the iLo for our Windows guests.
You would need a really freaking big plug! My numbers:
52.22kWh=187,992kWs
187,992kWs=18,799,200Ws
18,799,200Ws=18,799,200VAs
18,799,200VAs=220V*xA*300s
x=2,8484.73
As you can see even going to 440 isn't going to get you anywhere. Oh and then you have to multiply times ~1.1 for PF correction as I'm sure it doesn't look like a resistive load.
Never, in any context is it ok to write incorrect information
WRONG, language is about communications and so long as the point was conveyed correctly and the meaning non-ambiguous there is no problem messing up the case. This is especially true when a bunch of lay-people are discussing a technical area. People like you need to adjust their attitude to realize that human languages are NOT the same as programming languages or scientific equations, some of the time getting it close enough really is ok.
Right because we have enough non-fossil plants today to meet transportation demands. Oh and there's that problem with distribution, we already have a blackout about once a decade in the NE, can you imagine if everyone decides to recharge their cars before their summer vacation?
Abit was probably a wholly owned subsidiary or some equally fun legal construct so while the profits flowed to the parent company I would bet they are absolved of any liability for Abit products.