Autodesk dominates the closed source 3D market much like Adobe dominates the closed source 2D market. Other people don't try to enter the market because it is bloody hard to write software which can compete with these products. Even Microsoft bought Softimage at a point then gave up and sold them off to Avid. Today Softimage is owned by... Autodesk. Autodesk, not content with having 3D Studio, bought Softimage and Maya. Today there is little competition left. There are lots of patents in the area but I don't think that is the main problem.
There has been no strong push to provide alternatives for the Adobe-applications, so why would there be anything such now all of a sudden? I do not see the situation changing for years to come.
What? Never heard of the GIMP for bitmap drawing or Inkscape for vectorial drawing? IMO Inkscape is superior to Illustrator. I even use it in Windows.
The situation is more dire for video editing where there are several programs with somewhat laughable feature sets like you would see in entry/mid level video editing programs in other platforms. The interesting thing is the best library for multimedia is probably libavcodec and its open source. The best video players are open source as well.
Games are a bit of a mess. You have access to all the emulators so if you play old games (consoles, DOS) you are covered. There are a couple of nice open source games like Warzone 2100, Wesnoth, Frets on Fire. However if you want to play recent PC games like Mass Effect, Call of Duty, Dragon Age, Starcraft II you are pretty much screwed. If one of those fabled Linux consoles ever came out perhaps we could have platform commonality and get more games via that route. Otherwise I do not see companies like EA or Activision doing Linux ports.
I suspect eventually Wine will run DirectX 9 games better than Windows itself does but that isn't much consolation since you are following a moving target.
Actually the maximum pressurized cargo capacity for Dragon is 3310 kg which you can read about in the COTS-2 press kit. This resupply mission is not using up the full capacity of the capsule in fact SpaceX even used up the spare capacity to launch a 150 kg ORBCOMM satellite. If you are launching an 80 kg astronaut with a Sokol suit (10 kg) how much do you need extra for life support anyway?
That article flies in the face of AMD's recent statements regarding Steamroller where they claim they are moving from doing layout by hand to maximize performance and density as practiced in Bulldozer to using a high-density cell library for layout.
Shown above is a portion of the chip's FPU. The top image comes from a current Bulldozer chip, which employs the hand-drawn custom logic that's generally used in high-end x86 CPUs. The lower image comes from a potential future chip that uses a more automated high-density cell library. On the same 32-nm process node, the high-density library purportedly crams the same logic into 30% less area, with 30% less power use. As the slide notes, gains on this order would usually come from the transition to a newer, smaller fabrication process. We'd expect the more automated approach to design to reduce AMD's time to market, as well.
IBM will sell and maintain for you anything you want to buy or service even if it comes from a competitor. They earn most of their money from services. Intel's manufacturing edge at the moment is not something easy to surpass. It's not like the other non-Intel manufacturers are doing any better. IBM does have some unique technology like large low-latency on-chip eDRAM which is used in their higher end processors. IBM usually does not optimize their processors for mobile applications and I doubt Apple was willing to pay IBM for doing a custom design with such capabilities. Motorola would have been more inclined to do so since they sold for embedded markets however once again Apple was unwilling to fund the design by themselves. So they had to use a higher volume processor from Intel which means someone else is paying for the R&D as well not just them. Personally I would take the more automated approach for CPU design since it provides good results at much lower R&D costs at this moment. I am unaware of how Qualcomm does things despite having met some of their hardware design people once (I'm a software guy). There are rumors that the 22nm Intel process power savings are indeed lackluster as the performance comparisons between Sandy Bridge and Ivy Bridge processors will show but like you said it means they can get more chips for wafer so they do it.
Actually the Russians have more advanced nuclear powered spacecraft than the US. B1B? Sucks compared to a Tu-160 which has twice the speed and twice the payload. C17? Sucks compared with an An-225 Mriya and their Il-76 do the job just fine. B-52s? They are okay but the Russians also have the Tu-95 Bear. I doubt there is anything in the F-16 or F-18 that they don't have already done better. B-2s are good to bankrupt yourself.
F-22s and F-18 Super Hornet have some interesting technology which they are attempting to compete with the PAK FA.
Tanks is one area where they have been doing some large improvements over their past models but their doctrine is different since they prefer to have more tanks of lower quality than less more expensive higher quality tanks. They used to have two tank models: a cheaper one for mass production (T-72) and a more expensive one for quality (T-80). However their experience in Chechnya showed the T-80s were not worth the cost since they were just as vulnerable. Russia has a larger land border than the US to defend. They could do better in optics, or gun stabilization, or armor but it would require doing things differently. Missiles? They have S-400 Triumf, the Vympel R-73, R-77, P-800 Oniks, etc. The US has somewhat more advanced air to ground weapons but they also have a fairly diverse catalog they can use.
IBM has lower power processors as well like the ones they use in Blue Gene/Q. A Blue Gene/Q system called Sequoia is currently #1 in the TOP500 list and those guys care about performance/watt. All current generation consoles use IBM designed processors be it XBox 360/PS3/Wii. IBM don't sell really low power processors but if you want that you are better off using ARM anyway.
Every time a satellite is built they usually build a spare in case the launch goes wrong. Once they decide they have enough satellites these spares can either be discarded or repurposed. That is probably what happened here.
I'm old enough to remember when MMUs and FPUs were sold as separate chips. The additional integration will happen in due time. It always does. In fact it is happening now across the board. Whether it is Intel, AMD, or ARM everyone has a CPU+GPU solution. You can reduce bandwidth requirements by increasing on-chip cache and using memory compression. If eDRAM is used the on-chip cache will be large enough to fit the pixel buffer and z-buffer in core with the same area used for L3 cache in present processors.
This problem happened because AMD was caught by surprise with AVX doubling the register size over SSE once the Bulldozer design was well advanced. Then Intel pulled their little FMA4 to FMA3 switcheroo on top of it ensuring that code compiled for future Intel processors such as Haswell (FMA3) won't run at top speed on Bulldozer (FMA4). Piledriver supports FMA4 and FMA3. AMD could double the number of FP units but changing the decoder and load/store units to cope is non-trivial. Things will only get worse once Haswell comes out since it not only has FMA3 but gather-scatter support like the old Cray vector processors or GPUs.
A lot of people diss Bulldozer but the fact is making the leap to a four issue processor is bloody hard. It hadn't been done before and both Intel and AMD were the first to implement this in an actually working and available processor. It is hard to extract ILP to keep such a processor fed and it is hard to design the decoders. Compare an Alpha 21264 to an AMD Bulldozer block to see the difference. Look at the number of arrows coming in and out of the instruction decoder. The processor is flexible enough that a single-threaded application can use both integer cores in Bulldozer, while a multi-threaded application may use one integer core per thread.
No kidding. A lot of it I think is the politicians are scared that by dismantling the megacorps the US will be more vulnerable to offshore competition. However in the long run it will only lead to stagnation and losing to the same competition. It has happened elsewhere before.
Also you are telling bullshit. AMD hand designs their chips where it matters just like Intel does. IBM are the ones using high-level design tools and they are quite successful at it. POWER7+ is the highest performing processor right now. IBM manages to design a lot of good performing processors with a lot less people than either Intel or AMD.
IMO AMD has a more advanced core in Piledriver than Intel has with Ivy Bridge. However AMD's cores are unbalanced and perform poorly on legacy apps. Just look at the CPU design blocks or the supported instructions.
The performance difference can be explained by a lot of things. Bad OS scheduling of tasks, applications not compiled to use the fused-multiply-add ops which Intel is only going to have when Haswell comes out (2x FP performance), and CPU scheduling issues. FWIW I'm probably going to buy AMD once Vishera comes out in the end of this month using the Piledriver core. I only need high performance in my Linux development platform and I compile the high-performance apps myself. The remaining applications are not performance critical.
Actually you are wrong. AMD needed new CPU fabs since their old Austin fab did not have enough production to fulfill demand. At one point AMD was severely capacity constrained manufacturing Thunderbird. They could sell three times their production if they had it. Still they did well enough that even after the Dresden expansion they still had the cash reserves to build a new fab in New York state and get to the point where they could supply one of the three big OEMs (HP, Apple, Dell) all by themselves. This is important since often AMD lost business because even if a vendor wanted to use AMD CPUs on a certain product then Intel would come around and charge extra for any CPUs they sold to that vendor to make the business unprofitable. In order to break their first deal with a major OEM AMD had to give CPUs for free to Compaq I kid you not. Despite AMD having a superior product at the time it would have cost that much to Compaq to buy AMD from Intel price hikes. The AMD management issue was Hector Ruiz used up all these cash reserves, enough for a new fab, buying ATI stock over the market price just before the stock market crash and put AMD on the edge of bankruptcy. Then he sold the fabs to Arab oil businessmen who wanted to get into the semiconductor business and jumped ship.
NOR flash gave more profits at the time than NAND flash manufacturing. Once the NOR flash market started going under AMD spun off those operations. There was talk at a point of converting one of those plants to CPU production, which would have been great at the time, but it seems the manufacturing tools weren't similar enough to make such a choice viable. NAND flash manufacturing is a cut-throat business where players like Samsung and Hynix dominate. There simply was not enough margin for AMD to enter that market. For Samsung at least it makes more sense since they can include NAND into their consumer electronics products. Intel went into NAND flash by a joint venture with Micron so it is really not that much of a big win for Intel and I doubt they get a lot of profit from it.
I can understand why AMD used a shared FPU in Bulldozer. With their Fusion strategy the CPU and GPU are supposed to be on the same chip. The GPU simply kills the CPU in FP compute performance. They probably thought they would simply eliminate the FP units in the long term or that they would fall into disuse. Their new designs still have twice the theoretical peak FP compute performance of their old designs. The issue is most applications simply are not compiled to use the new FMA instructions and they probably won't be until Intel's Haswell comes out.
There are corporations working on it precisely because they now know there is a market. Virgin Galactic (suborbital flight), SpaceX, Boeing (orbital flight), Bigelow (orbital habitats), etc. Sure they don't do it at the moment but I have little doubt knowing people are willing to pay so much for a space flight helped provide the drive to do it.
The Soyuz rocket has had multiple upgrades including full digital avionics and the capsule was modified to carry 3 large passengers (the Soviet Union had limits on cosmonaut height so they could carry more of them). The money had to come from somewhere. Even if this just saves the cost of a single flight it was worth it. Sure there is endemic corruption on any system where there is no effective democratic term limit. If you keep the same people for too long on the same position it is inevitable that abuses of power will happen. However I think it is overly cynical to think all the money will be diverted to something other than the Russian space program.
It's plainly obvious you know nothing about it since you called Soyuz a "Russian Freight". Progress capsules do most of the cargo carrying to the Russian segment of ISS so those are the "Freight". Soyuz capsules are transportation vehicles with minimal cargo capacity. Nitwit.
Slashdot Mirror
Autodesk dominates the closed source 3D market much like Adobe dominates the closed source 2D market. Other people don't try to enter the market because it is bloody hard to write software which can compete with these products. Even Microsoft bought Softimage at a point then gave up and sold them off to Avid. Today Softimage is owned by... Autodesk. Autodesk, not content with having 3D Studio, bought Softimage and Maya. Today there is little competition left. There are lots of patents in the area but I don't think that is the main problem.
There has been no strong push to provide alternatives for the Adobe-applications, so why would there be anything such now all of a sudden? I do not see the situation changing for years to come.
What? Never heard of the GIMP for bitmap drawing or Inkscape for vectorial drawing? IMO Inkscape is superior to Illustrator. I even use it in Windows.
The situation is more dire for video editing where there are several programs with somewhat laughable feature sets like you would see in entry/mid level video editing programs in other platforms. The interesting thing is the best library for multimedia is probably libavcodec and its open source. The best video players are open source as well.
Games are a bit of a mess. You have access to all the emulators so if you play old games (consoles, DOS) you are covered. There are a couple of nice open source games like Warzone 2100, Wesnoth, Frets on Fire. However if you want to play recent PC games like Mass Effect, Call of Duty, Dragon Age, Starcraft II you are pretty much screwed. If one of those fabled Linux consoles ever came out perhaps we could have platform commonality and get more games via that route. Otherwise I do not see companies like EA or Activision doing Linux ports.
I suspect eventually Wine will run DirectX 9 games better than Windows itself does but that isn't much consolation since you are following a moving target.
Actually the maximum pressurized cargo capacity for Dragon is 3310 kg which you can read about in the COTS-2 press kit. This resupply mission is not using up the full capacity of the capsule in fact SpaceX even used up the spare capacity to launch a 150 kg ORBCOMM satellite. If you are launching an 80 kg astronaut with a Sokol suit (10 kg) how much do you need extra for life support anyway?
POWER7+ was discussed at Hot Chips and released a couple of days ago.
Dear moderators: You are overrated! And I don't like hearing the truth != Troll.
That article flies in the face of AMD's recent statements regarding Steamroller where they claim they are moving from doing layout by hand to maximize performance and density as practiced in Bulldozer to using a high-density cell library for layout.
Shown above is a portion of the chip's FPU. The top image comes from a current Bulldozer chip, which employs the hand-drawn custom logic that's generally used in high-end x86 CPUs. The lower image comes from a potential future chip that uses a more automated high-density cell library. On the same 32-nm process node, the high-density library purportedly crams the same logic into 30% less area, with 30% less power use. As the slide notes, gains on this order would usually come from the transition to a newer, smaller fabrication process. We'd expect the more automated approach to design to reduce AMD's time to market, as well.
IBM will sell and maintain for you anything you want to buy or service even if it comes from a competitor. They earn most of their money from services. Intel's manufacturing edge at the moment is not something easy to surpass. It's not like the other non-Intel manufacturers are doing any better. IBM does have some unique technology like large low-latency on-chip eDRAM which is used in their higher end processors. IBM usually does not optimize their processors for mobile applications and I doubt Apple was willing to pay IBM for doing a custom design with such capabilities. Motorola would have been more inclined to do so since they sold for embedded markets however once again Apple was unwilling to fund the design by themselves. So they had to use a higher volume processor from Intel which means someone else is paying for the R&D as well not just them. Personally I would take the more automated approach for CPU design since it provides good results at much lower R&D costs at this moment. I am unaware of how Qualcomm does things despite having met some of their hardware design people once (I'm a software guy). There are rumors that the 22nm Intel process power savings are indeed lackluster as the performance comparisons between Sandy Bridge and Ivy Bridge processors will show but like you said it means they can get more chips for wafer so they do it.
F-22s and F-18 Super Hornet have some interesting technology which they are attempting to compete with the PAK FA.
Tanks is one area where they have been doing some large improvements over their past models but their doctrine is different since they prefer to have more tanks of lower quality than less more expensive higher quality tanks. They used to have two tank models: a cheaper one for mass production (T-72) and a more expensive one for quality (T-80). However their experience in Chechnya showed the T-80s were not worth the cost since they were just as vulnerable. Russia has a larger land border than the US to defend. They could do better in optics, or gun stabilization, or armor but it would require doing things differently. Missiles? They have S-400 Triumf, the Vympel R-73, R-77, P-800 Oniks, etc. The US has somewhat more advanced air to ground weapons but they also have a fairly diverse catalog they can use.
US cars suck. Why would I want to copy those? Construction equipment? The Japanese have better.
IBM has lower power processors as well like the ones they use in Blue Gene/Q. A Blue Gene/Q system called Sequoia is currently #1 in the TOP500 list and those guys care about performance/watt. All current generation consoles use IBM designed processors be it XBox 360/PS3/Wii. IBM don't sell really low power processors but if you want that you are better off using ARM anyway.
Every time a satellite is built they usually build a spare in case the launch goes wrong. Once they decide they have enough satellites these spares can either be discarded or repurposed. That is probably what happened here.
Check when the Kiev was built. It was laid down in 1970 and entered service in 1975.
I'm old enough to remember when MMUs and FPUs were sold as separate chips. The additional integration will happen in due time. It always does. In fact it is happening now across the board. Whether it is Intel, AMD, or ARM everyone has a CPU+GPU solution. You can reduce bandwidth requirements by increasing on-chip cache and using memory compression. If eDRAM is used the on-chip cache will be large enough to fit the pixel buffer and z-buffer in core with the same area used for L3 cache in present processors.
This problem happened because AMD was caught by surprise with AVX doubling the register size over SSE once the Bulldozer design was well advanced. Then Intel pulled their little FMA4 to FMA3 switcheroo on top of it ensuring that code compiled for future Intel processors such as Haswell (FMA3) won't run at top speed on Bulldozer (FMA4). Piledriver supports FMA4 and FMA3. AMD could double the number of FP units but changing the decoder and load/store units to cope is non-trivial. Things will only get worse once Haswell comes out since it not only has FMA3 but gather-scatter support like the old Cray vector processors or GPUs.
A lot of people diss Bulldozer but the fact is making the leap to a four issue processor is bloody hard. It hadn't been done before and both Intel and AMD were the first to implement this in an actually working and available processor. It is hard to extract ILP to keep such a processor fed and it is hard to design the decoders. Compare an Alpha 21264 to an AMD Bulldozer block to see the difference. Look at the number of arrows coming in and out of the instruction decoder. The processor is flexible enough that a single-threaded application can use both integer cores in Bulldozer, while a multi-threaded application may use one integer core per thread.
No kidding. A lot of it I think is the politicians are scared that by dismantling the megacorps the US will be more vulnerable to offshore competition. However in the long run it will only lead to stagnation and losing to the same competition. It has happened elsewhere before.
Also you are telling bullshit. AMD hand designs their chips where it matters just like Intel does. IBM are the ones using high-level design tools and they are quite successful at it. POWER7+ is the highest performing processor right now. IBM manages to design a lot of good performing processors with a lot less people than either Intel or AMD.
AMD has another processor design for mobile called Bobcat.
IMO AMD has a more advanced core in Piledriver than Intel has with Ivy Bridge. However AMD's cores are unbalanced and perform poorly on legacy apps. Just look at the CPU design blocks or the supported instructions.
The performance difference can be explained by a lot of things. Bad OS scheduling of tasks, applications not compiled to use the fused-multiply-add ops which Intel is only going to have when Haswell comes out (2x FP performance), and CPU scheduling issues. FWIW I'm probably going to buy AMD once Vishera comes out in the end of this month using the Piledriver core. I only need high performance in my Linux development platform and I compile the high-performance apps myself. The remaining applications are not performance critical.
NOR flash gave more profits at the time than NAND flash manufacturing. Once the NOR flash market started going under AMD spun off those operations. There was talk at a point of converting one of those plants to CPU production, which would have been great at the time, but it seems the manufacturing tools weren't similar enough to make such a choice viable. NAND flash manufacturing is a cut-throat business where players like Samsung and Hynix dominate. There simply was not enough margin for AMD to enter that market. For Samsung at least it makes more sense since they can include NAND into their consumer electronics products. Intel went into NAND flash by a joint venture with Micron so it is really not that much of a big win for Intel and I doubt they get a lot of profit from it.
I can understand why AMD used a shared FPU in Bulldozer. With their Fusion strategy the CPU and GPU are supposed to be on the same chip. The GPU simply kills the CPU in FP compute performance. They probably thought they would simply eliminate the FP units in the long term or that they would fall into disuse. Their new designs still have twice the theoretical peak FP compute performance of their old designs. The issue is most applications simply are not compiled to use the new FMA instructions and they probably won't be until Intel's Haswell comes out.
AMD supposedly got some next-gen console wins. That should help them get volume and some sort of steady income source.
The Soyuz rocket has had multiple upgrades including full digital avionics and the capsule was modified to carry 3 large passengers (the Soviet Union had limits on cosmonaut height so they could carry more of them). The money had to come from somewhere. Even if this just saves the cost of a single flight it was worth it. Sure there is endemic corruption on any system where there is no effective democratic term limit. If you keep the same people for too long on the same position it is inevitable that abuses of power will happen. However I think it is overly cynical to think all the money will be diverted to something other than the Russian space program.
It's plainly obvious you know nothing about it since you called Soyuz a "Russian Freight". Progress capsules do most of the cargo carrying to the Russian segment of ISS so those are the "Freight". Soyuz capsules are transportation vehicles with minimal cargo capacity. Nitwit.
Welcome to the Web.
At least they are spending the money on something. If enough rich people did it there would be more flights by now.