Invest in your own servers (about US$ 5-8K) and then you'll find a world of options opening up to you as you look for colocation companies. We're on EGI Hosting which costs around $700 a month for an 95%tile 100Mbps ( on a 1Gbps connect ) pipe.
If you do go for a colo, I also recommend EGI. I did a 8 month stint there with some servers, and the NOC guys were great, facilities were good, uplink was awesome.
There are reputable, stable companies out there that won't flake out, ie. BuyVM (http://buyvm.net).
For 25$ a month, you get 70GB disk space, 4TB bandwidth, its on a gigE link (I just pulled at 379.2mbit/s from cachefly), and I suffered an hour of downtime when they were physically moving datacenters a few months back, other than that, none at all.
I run a lot of little hosting projects all on VPSes, and I think my aggregate bandwidth usage is around 9TB a month, and I never really run into issues (I've actually gotten two 2TB+/mo boxes from different companies and tested how much bandwidth I can use, never got complains).
You can also research alternatives on lowendbox.com. You won't find cheap tier 1 bandwidth, but you will easily find cheap bandwidth.
First of all, if Intel + AMD's integrated graphics on the CPU suck, then NVIDIA's niche will not be only high end, it will be mid range too. But AMD is definitely going to support its discrete business, its making them money, its a great product at the moment. Intel, well its safe to say all their products absolutely suck. But, Intel has massive "persuasion" when it comes to spending millions "convincing" companies to support Intel products.
Secondly, while integrating CPUs and GPUs will be possible, you will not get two high performance parts combined into one, ever. NVIDIA has a current answer to that, GPU in the chipset (the venerable ION/MCP7A chipset driving many Atom-based systems, and _every single shipping Apple machine_ (paired with dGPUs in higher end parts), and while in the future they will have issues on the Intel front with legal issues, they will definitely find ways around it (pushing Tegra probably).
Why is NVIDIA's outlook not too bright? In all honesty, how often does not having an open source driver impact your average Joe? Virtually never. How useful will an entirely open driver actually be for the majority of Linux users? Have you ever looked into a graphics driver? Think its easy to tweak and fix?
Their driver support on Linux right now is pretty damn solid with their blob. In most cases, it just "works". And yes, there are cases where it does not, but you give me any piece of software for Linux, and I'll give you a case where it doesn't work. Opening the source will not magically solve this issue.
The real message to gather from TFA is all those wonderful references to the fact that WORKSTATION LINUX IMPLEMENTATIONS MAKE NVIDIA MONEY. Lots of money. A lot more money than your friend buying a 9400GT to throw in a *nix-based HTPC. They are going to go after that market, first and foremost. Guess what? It's working. The Quadro lines are the only viable workstation choice for Linux CAD/DCC, and CUDA on Linux is sweet too.
AMD/ATI tried to push this entire open source movement, but its really hurting them. If they put half of that effort to making their driver rock solid in linux as a BLOB, they could really push into the market with their amazing 5xxx line of cards.
This has been addressed before. nVIDIA cannot open up the source to their drivers because of source code included from their 3DFX acquisition. They have stated time and time again, they would LOVE to open it up, but they legally cannot.
That is miles more than ATI have provided for Linux. Hopefully when the NDA on the contract runs out we will see open drivers actively supported from nVIDIA.
Everyone here is constantly saying "Oh its an Intel system, built by an Intel team, vs. an AMD system, built by an Intel team... I'll trust the reviews when independant people get them."
If you looked a little you would see, that there are already lots of people with the Conroe in their hands. And it has shattered every PI, 3DMark, world record there is. We are talking about 10s 1M SuperPi runs, and if you know anything about that benchmark you will know that is absolutely crazy. Why not read some forums, like XtremeSystems or more specifically some benchmarking threads where the world record was broken on air w/ Conroe, but now its under LN2 for some other people (including coolaler) and is holding the world record.
First of all the substance is Gallinstan, if it was pure Gallium it would corrode away whatever they put it in unless it was teflon coated, but then you can kiss your head conductivity goodbye.
Water has the highest heat capacity of any natural substance, there are others that serve different purposes (I use Dowfrost in my loop), but water is still the best in most cases.
Just because the head conductivity is 65x (Sapphire-speak) its thermal capacity is horrible (12x less than water). Now whoever said that capacity is not very important is wrong. These cards pump out heat like there is no tomorrow, and you would have to 1) Pump 12x as fast to get to WATER-comparable temperatures EXTRACTED from the core, and you would have to have a fan running fast in order to cool that liquid as its flying through the loop (we are talking over a 1000gph is necessary for effective cooling of Ga).
Also, this cooling is not that much "better" than water because you will still never acheive below ambient temperatures, it is a physical impossibility, so whats the point on spending so much on this junk (it would cost you in the realm of 500$ to fill a loop with it) when water will work JUST as well, even better, and you still have silent cooling available (MCP 350 anyone?) and its dirt cheap (200$ will get you an entire quality loop, easy!).
These are just heatpipes on steroids, you can't use a pelt because it would freeze the metal, the pump is expensive and needs rare earth metals in order to create enough of a field to actually spin and pump this stuff through...
All in all its just marketing. Gallium is a super cool metal to play with, but cooling with substances other than water or water based products is still pointless, as water has the highest natural heat capacity.
Slashdot Mirror
It's coming soon, it's free, open source, and is shaping up to be a really kickass video editor: http://www.openshotvideo.com/ Kickstarter was here: https://www.kickstarter.com/pr...
Invest in your own servers (about US$ 5-8K) and then you'll find a world of options opening up to you as you look for colocation companies. We're on EGI Hosting which costs around $700 a month for an 95%tile 100Mbps ( on a 1Gbps connect ) pipe.
If you do go for a colo, I also recommend EGI. I did a 8 month stint there with some servers, and the NOC guys were great, facilities were good, uplink was awesome.
I recommend going down the VPS route.
There are reputable, stable companies out there that won't flake out, ie. BuyVM (http://buyvm.net).
For 25$ a month, you get 70GB disk space, 4TB bandwidth, its on a gigE link (I just pulled at 379.2mbit/s from cachefly), and I suffered an hour of downtime when they were physically moving datacenters a few months back, other than that, none at all.
I run a lot of little hosting projects all on VPSes, and I think my aggregate bandwidth usage is around 9TB a month, and I never really run into issues (I've actually gotten two 2TB+/mo boxes from different companies and tested how much bandwidth I can use, never got complains).
You can also research alternatives on lowendbox.com. You won't find cheap tier 1 bandwidth, but you will easily find cheap bandwidth.
First of all, if Intel + AMD's integrated graphics on the CPU suck, then NVIDIA's niche will not be only high end, it will be mid range too. But AMD is definitely going to support its discrete business, its making them money, its a great product at the moment. Intel, well its safe to say all their products absolutely suck. But, Intel has massive "persuasion" when it comes to spending millions "convincing" companies to support Intel products.
Secondly, while integrating CPUs and GPUs will be possible, you will not get two high performance parts combined into one, ever. NVIDIA has a current answer to that, GPU in the chipset (the venerable ION/MCP7A chipset driving many Atom-based systems, and _every single shipping Apple machine_ (paired with dGPUs in higher end parts), and while in the future they will have issues on the Intel front with legal issues, they will definitely find ways around it (pushing Tegra probably).
Why is NVIDIA's outlook not too bright? In all honesty, how often does not having an open source driver impact your average Joe? Virtually never. How useful will an entirely open driver actually be for the majority of Linux users? Have you ever looked into a graphics driver? Think its easy to tweak and fix? Their driver support on Linux right now is pretty damn solid with their blob. In most cases, it just "works". And yes, there are cases where it does not, but you give me any piece of software for Linux, and I'll give you a case where it doesn't work. Opening the source will not magically solve this issue.
The real message to gather from TFA is all those wonderful references to the fact that WORKSTATION LINUX IMPLEMENTATIONS MAKE NVIDIA MONEY. Lots of money. A lot more money than your friend buying a 9400GT to throw in a *nix-based HTPC. They are going to go after that market, first and foremost. Guess what? It's working. The Quadro lines are the only viable workstation choice for Linux CAD/DCC, and CUDA on Linux is sweet too.
AMD/ATI tried to push this entire open source movement, but its really hurting them. If they put half of that effort to making their driver rock solid in linux as a BLOB, they could really push into the market with their amazing 5xxx line of cards.
This has been addressed before. nVIDIA cannot open up the source to their drivers because of source code included from their 3DFX acquisition. They have stated time and time again, they would LOVE to open it up, but they legally cannot.
That is miles more than ATI have provided for Linux. Hopefully when the NDA on the contract runs out we will see open drivers actively supported from nVIDIA.
Yes, but clearly pen missle or even tiger claw will oust the rock in the end.
Ask Aziz.
Aziz knows.
Jurassic Bark was easily one of the most touching comics I have seen. It's right up there with the Pinky and the Brain Christmas Special.
Everyone here is constantly saying "Oh its an Intel system, built by an Intel team, vs. an AMD system, built by an Intel team... I'll trust the reviews when independant people get them."
If you looked a little you would see, that there are already lots of people with the Conroe in their hands. And it has shattered every PI, 3DMark, world record there is. We are talking about 10s 1M SuperPi runs, and if you know anything about that benchmark you will know that is absolutely crazy. Why not read some forums, like XtremeSystems or more specifically some benchmarking threads where the world record was broken on air w/ Conroe, but now its under LN2 for some other people (including coolaler) and is holding the world record.
Than H2O.
First of all the substance is Gallinstan, if it was pure Gallium it would corrode away whatever they put it in unless it was teflon coated, but then you can kiss your head conductivity goodbye.
Water has the highest heat capacity of any natural substance, there are others that serve different purposes (I use Dowfrost in my loop), but water is still the best in most cases.
Just because the head conductivity is 65x (Sapphire-speak) its thermal capacity is horrible (12x less than water). Now whoever said that capacity is not very important is wrong. These cards pump out heat like there is no tomorrow, and you would have to 1) Pump 12x as fast to get to WATER-comparable temperatures EXTRACTED from the core, and you would have to have a fan running fast in order to cool that liquid as its flying through the loop (we are talking over a 1000gph is necessary for effective cooling of Ga).
Also, this cooling is not that much "better" than water because you will still never acheive below ambient temperatures, it is a physical impossibility, so whats the point on spending so much on this junk (it would cost you in the realm of 500$ to fill a loop with it) when water will work JUST as well, even better, and you still have silent cooling available (MCP 350 anyone?) and its dirt cheap (200$ will get you an entire quality loop, easy!).
These are just heatpipes on steroids, you can't use a pelt because it would freeze the metal, the pump is expensive and needs rare earth metals in order to create enough of a field to actually spin and pump this stuff through...
All in all its just marketing. Gallium is a super cool metal to play with, but cooling with substances other than water or water based products is still pointless, as water has the highest natural heat capacity.