Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Mac OS 10.5.2, Mac Pro Update At Macworld?
- Thread starter Death Warrant
- Start date
- Sort by reaction score
MacPro should be updated often
Intel has been on a roll, recently updating by increments that make life a lot easier for Dell and HP but Apple has decided to mostly do nothing. I bought an HP Blackbird 002... yes it's a gaming rig. It sports two Nvidia 8800 Ultra graphics cards in SLi format, with a third part software nHancer you can get 1.5GB of GPU power to work for QT, After Effects and Premiere and more. The motherboard is capable of taking the new Penrynn ... indeed the whole unit owes a lot to Apple, clean inside, thumb/push install hard drives and cards. Special liquid cooler that does not remotely leak or require attention.
I'm trying to point out that Apple is silly to think that all its MacPro customers will wait. I wanted a gaming rig that's true but the iMac does not fit that and I also wanted four cores to use when rendering video...
Leopard is so close to Vista Ultimate that it's no problem to switch. Even if Vista ultimate is a rip off of OSX it's not bad and in some places even better. For example what it can do with 3.5GB of usable RAM puts to shame OSX's ram hogishness.
I will buy another MacPro but instead of having 2, I will only buy one.
The HP Blackbird 002 is the highest ranked/graded desktop on CNET and some PC mags. It's great to use for work and play. Apple should not dismiss gaming. The resources needed for gaming can be used for high end video and DVD production. I can slide an extra Blue Ray burner in... there's no need for me to wait for HP as the whole point of a blackbird is to offer off the shelf upgrades which means I can get what I want, when I want it.
Apple is really no longer interested in computers unless they offer a lifestyle image. The MacPro should offer easy access to CPU so that we can upgrade when we have the money. At the moment it's not easy.. or as easy as slotting the RAM in.
I love Apple design. However the HP blackbird 002 really does make heads turn. Really is thinking different and I get brownie points for having one on my desk. Apple must wake up.
Just reading these boards is proof enough that there is a lot of pent up demand. Just we're uncool, we want to do more than sit in Starbucks...
By the way HP workstations lated to deliver in Jan are some mean beasts..
The more OSX looks like Vista looking like OSX the easier it is to switch back and forth.
If I were a Mac Freak I'd be sending petitions.. I know Apple does not care and that is dangerous for the platform.
Intel has been on a roll, recently updating by increments that make life a lot easier for Dell and HP but Apple has decided to mostly do nothing. I bought an HP Blackbird 002... yes it's a gaming rig. It sports two Nvidia 8800 Ultra graphics cards in SLi format, with a third part software nHancer you can get 1.5GB of GPU power to work for QT, After Effects and Premiere and more. The motherboard is capable of taking the new Penrynn ... indeed the whole unit owes a lot to Apple, clean inside, thumb/push install hard drives and cards. Special liquid cooler that does not remotely leak or require attention.
I'm trying to point out that Apple is silly to think that all its MacPro customers will wait. I wanted a gaming rig that's true but the iMac does not fit that and I also wanted four cores to use when rendering video...
Leopard is so close to Vista Ultimate that it's no problem to switch. Even if Vista ultimate is a rip off of OSX it's not bad and in some places even better. For example what it can do with 3.5GB of usable RAM puts to shame OSX's ram hogishness.
I will buy another MacPro but instead of having 2, I will only buy one.
The HP Blackbird 002 is the highest ranked/graded desktop on CNET and some PC mags. It's great to use for work and play. Apple should not dismiss gaming. The resources needed for gaming can be used for high end video and DVD production. I can slide an extra Blue Ray burner in... there's no need for me to wait for HP as the whole point of a blackbird is to offer off the shelf upgrades which means I can get what I want, when I want it.
Apple is really no longer interested in computers unless they offer a lifestyle image. The MacPro should offer easy access to CPU so that we can upgrade when we have the money. At the moment it's not easy.. or as easy as slotting the RAM in.
I love Apple design. However the HP blackbird 002 really does make heads turn. Really is thinking different and I get brownie points for having one on my desk. Apple must wake up.
Just reading these boards is proof enough that there is a lot of pent up demand. Just we're uncool, we want to do more than sit in Starbucks...
By the way HP workstations lated to deliver in Jan are some mean beasts..
The more OSX looks like Vista looking like OSX the easier it is to switch back and forth.
If I were a Mac Freak I'd be sending petitions.. I know Apple does not care and that is dangerous for the platform.
A fund to launch the new Mac Pro
I am starting a fund to get the new Mac Pro launched!.
Here's how it works.
You will donate money to me now. And I will go out and buy the Mac Pro Next Monday!
By me Buying the Mac Pro Monday, it is sure to launch the new model by Friday.
It has always happened this way and I don't see any change in my apple luck.
So what say!
Lets get that new Mac Pro Model out on the sales floor next week shall we?
Donate generously and with the knowledge you are helping to advance our tech society and help me maintain a modicum of faith in my techie friends.
I am starting a fund to get the new Mac Pro launched!.
Here's how it works.
You will donate money to me now. And I will go out and buy the Mac Pro Next Monday!
By me Buying the Mac Pro Monday, it is sure to launch the new model by Friday.
It has always happened this way and I don't see any change in my apple luck.
So what say!
Lets get that new Mac Pro Model out on the sales floor next week shall we?
Donate generously and with the knowledge you are helping to advance our tech society and help me maintain a modicum of faith in my techie friends.
Yea, same here, I feel like I'm holding up the launch too. Every time I buy, the next day or a week later the newer one comes out. Happens all the time. So maybe the MacPro will never come as long as I don't buy. haha.
But once you do buy, then the rest of us will be getting a heck of an upgraded new Mac Pro. Go ahead and buy. I don't won't to do the same thing again, just like you did.
Apple does what it has always done: upgrades its hardware in such a way people can predict what is coming and when. Apple has upgraded its pro workstations "once a year" for longer than I can remember, and it's great. People can expect that the top-of-the-line workstation they buy just after the release will be "current" for 12ish months. What's wrong with that picture?
And with Apple it's not so necessary to have the latest and greatest all the time. OSX won't slow you down like Windows so you're more than happy with your Apple workstation for at least 3-5 years, most people for much longer times.
So, why would you be planning to buy one in the first place? Why even wait for something that is not worth the money for you? I don't get it. I buy whatever I need whenever I must and by following that guideline everything is always worth it.
I worded that wrong... MacPro isn't worth the price right now. A "4-Core Woodcrest" is pretty much the same price as an "8-Core Harpertown" when available. I will definitely be buying a MacPro. I'm a video editor / graphic designer and in the need of a new computer. Just don't want to get an old MacPro with a really out dated graphics card.
From what I hear on Mac forums, you might be getting a new Mac Pro with an outdated graphics card!
Just like an old western shootout! The difference is the first one to "Pull The Trigger*" loses!
*Multimedia's favorite expression.
I sure hope the next update fixes AEBSn disk sharing. It's really frustrating how it dies sporadically.
Changing video cards could be considered an upgrade, but I said perhaps a new Mac Pro but with an outdated video cards. You don't expect Apple to give you the latest and greatest video card? Do you? A years old card would be considered an upgrade.
It may be getting "old" but it's still top notch. Thats just how the market goes. Also it's going to be very "me too" as long as Apple is an Intel customer since Intel doesn't play favorites and doesn't make "special" items for their customers unless it's something easy to do, like a low end part with a higher multiplier (compensate for a low fsb) to be used in leftover laptops, a la the 400MHz FSB Core Duos and Core Solos used by Dell.
Thats not how it works. The "concept of '16x' bus speed" isn't a concept. That just means that the card uses 16 PCI Express lanes. Thats what the GPU itself is designed to interface with. You can't just throw a GPU into a socket and say "Go". ATI and NVdia would have to redesign their products to work with Intel's bus interface. Also GPUs have access to the memory on board. Thats where textures go after the on board memory is used up. The GPU doesn't care what the addressing mode of the software is since it is unto itself. The only place where 64bit might make a difference is if the video card needs to put textures into main system memory (hypermemory, etc). This is very very slow, and it's not ideal to share memory in the first place since you loose available memory for the rest of the system.
There isn't such a thing as "infinite X" either. The speed would still be limited by the FSB speed (since thats what a socketed GPU would be connected to). On that note, you'd also end up throwing out a whole batch of GPUs every time you change FSB speeds dramatically. An 800MHz FSB GPU may work on a 1066MHz FSB, but there's no way it's going to work on the new 1600 and unlikely to work with the present 1333MHz FSB. This is why new chipsets have to be designed to work with higher FSB speeds. Only rare exceptions (like the 945 chipset which started as a 667/800MHz FSB speed chip but can run 1333MHz FSB with out of spec PCIe) can do something this drastic.
This is why PCI, AGP, PCIe, etc is used. It's easy to change cards and even swap cards between platforms. A GPU for the Intel FSB isn't going to work on an AMD setup or vice versa unless a bridge chip is used... what have you gained in that case?
ATI and NVidia have been working very well for a while now. The entire reason why we have games that look as good as they do is owed to those two as well as 3DFX for pushing the envelope of what can be done. This of course is with some help from the OpenGL group and Microsoft, who defines the DirectX standard. Intel Integrated graphics have never been considered for serious 3D graphics applications.
Also why would they want to reinvent the wheel for a limited user base? The engineering required to make something "unique" like a GPU tagged onto the FSB would be possible cost effective if it was released for the entire market.
PCIe is not 20 years old. PCI itself isn't 20 years old. PCIe is also not compatible with PCI, it is a serial bus while PCI is a parallel bus. Also what does power draw have to do with a bus itself? The power draw of a GPU is because of the GPU, not how it connects to something. An 8800GTX is going to draw nearly 200W at load if it's on ISA, PCI, AGP, PCIe, or on a FSB.
Silicon Graphics also sells very specialized hardware at very high prices. No one is going to buy a Mac for the price SG charges.
That "Motherboard GPU" is just integrated graphics. It's built into the northbridge/southbridge (depending on platform - ATI or AMD). The GPU die is usually just tethered to an internal PCIe link and is a low end product meant to be a step up/competitor to Intel Integrated Graphics.
but that is why these newer video cards have a molex connector on them cause the PCI , AGP or PCIe bus cant supply the proper amount of power to the card. if you look at a Geforce2 card it uses the power from the bus, you look at a Geforce 6800 you are going to see 1 or 2 molex connectors on the card so it will function, without them it would not run, cause the bus cant supply that much power to the card.
if you overtax this bus (PCI, AGP, PCIe,) you will run into allot of issues, there is a limit to how much power the said bus can supply to the card in the slot (or combined slots). that is why they put a molex connector on the card so it can get the power directly from the PSU to avoid the bus power overtaxing issue. there is only so much power you can pull from a PCI, AGP, or PCIe
Yeah Apple need to mature their wireless a bit more. You should have some more control over how long an attached HDD spins and disconnects etc
PCI bus is more than 20 years old
Ok - the interfaces, names, etc have changed dramatically, but cards in slots have been around for a long time. Some things are a natural for them, but high power circuits that are included on every (high end) computer are not. Slots should be used for things that most computers DONT need, like fancy RAID controllers, four port ethernet connections, etc.
You can tell when a technology is on its last legs by looking at how complicated it is. These new cards get power from two places, have massive cooling problems, are expensive, etc.
The bus speed of a card is a statement about how fast the graphics card can copy memory from the system memory to the graphics card. If there is no transfer then the speed is infinite. 'Infinite x' IS really the only way to go forward with design. Here is the present path for graphics: Disk -> system memory -> PCI bus -> card memory -> PCI bus -> back to system memory when card is done with the texture -> back to card memory. With a true GPU on the main bus, you have disk->system memory. That's it.
Memory sharing is a GOOD thing. Just buy memory once, and put it all in one place so that anyone can use it. A 2 Gig graphics card is really expensive right now, but with a motherboard GPU, the graphics card can use 2 Gigs of RAM if it needs it, but usually it will not. Heck, on a 16 GB machine, a 3D walkthrough could easily have a 10 GB set of textures. With a 500MB graphics card on a bus, the textures have to unload and load from the card many hundreds of times, using memory bandwidth, etc. It just does not make sense. GPUs have access to the main memory on the computer, but it still chokes up the memory controller, plus you have 2 copies of textures sitting around..
As an example of this, look at OS X. Each window is put onto the screen by the GPU, which allows for all sorts of nice effects like the shadows, etc to happen with 'no' CPU load. But - each of the many windows on your machine has to pass through each system on the computer for each change that happens to it. For each keystroke you make, the window has to be drawn by the CPU, and then reloaded across the PCI bus to the GPU as a new texture. So even though the CPU load is low, there is lots of action on the memory - with at least 2 copies going on, etc. With the right driver and system software, all of this evaporates - the GPU can just show the window.
If the bus on the computer is too slow, then widen it and speed it up. That way the whole machine benefits from having blazing fast memory. The money spent on making that fast bus on the graphics card can be used on the system bus.
Yes they will have to redesign chips to make this work. The 'redesign' is only on the mundane i/o parts of the design. The benefits are numerous: Power problems go away, the computer maker knows where the heat is coming from, there are no cards to rattle around. Want more memory? Add more and the GPU, CPU all get it. GPUs have impressive performance on some non graphics problems, like simulations (eg weather), but now, with the GPU tucked away on some exceeding slow bus, you kinda have to either do your whole job on the card or on the CPU. With everything handy, you can program using both. The list goes on. Plus its cheaper!
A radical redesign would not JUST be for Apple, but rather eventually for everyone. It is just that Apple can ship computers with a new technology easier than the windows world. They have done that before more than a few times.
NVidia's Motheroard GPU is really a step towards this goal. The reason that intel integrated graphics are so bad has everything to do with Intel's graphics team. All the good people there work on CPUs. 'Everyone knows' that integrated graphics are bad. So NVidia needed a new marketing name.
--
Ok - the interfaces, names, etc have changed dramatically, but cards in slots have been around for a long time. Some things are a natural for them, but high power circuits that are included on every (high end) computer are not. Slots should be used for things that most computers DONT need, like fancy RAID controllers, four port ethernet connections, etc.
You can tell when a technology is on its last legs by looking at how complicated it is. These new cards get power from two places, have massive cooling problems, are expensive, etc.
The bus speed of a card is a statement about how fast the graphics card can copy memory from the system memory to the graphics card. If there is no transfer then the speed is infinite. 'Infinite x' IS really the only way to go forward with design. Here is the present path for graphics: Disk -> system memory -> PCI bus -> card memory -> PCI bus -> back to system memory when card is done with the texture -> back to card memory. With a true GPU on the main bus, you have disk->system memory. That's it.
Memory sharing is a GOOD thing. Just buy memory once, and put it all in one place so that anyone can use it. A 2 Gig graphics card is really expensive right now, but with a motherboard GPU, the graphics card can use 2 Gigs of RAM if it needs it, but usually it will not. Heck, on a 16 GB machine, a 3D walkthrough could easily have a 10 GB set of textures. With a 500MB graphics card on a bus, the textures have to unload and load from the card many hundreds of times, using memory bandwidth, etc. It just does not make sense. GPUs have access to the main memory on the computer, but it still chokes up the memory controller, plus you have 2 copies of textures sitting around..
As an example of this, look at OS X. Each window is put onto the screen by the GPU, which allows for all sorts of nice effects like the shadows, etc to happen with 'no' CPU load. But - each of the many windows on your machine has to pass through each system on the computer for each change that happens to it. For each keystroke you make, the window has to be drawn by the CPU, and then reloaded across the PCI bus to the GPU as a new texture. So even though the CPU load is low, there is lots of action on the memory - with at least 2 copies going on, etc. With the right driver and system software, all of this evaporates - the GPU can just show the window.
If the bus on the computer is too slow, then widen it and speed it up. That way the whole machine benefits from having blazing fast memory. The money spent on making that fast bus on the graphics card can be used on the system bus.
Yes they will have to redesign chips to make this work. The 'redesign' is only on the mundane i/o parts of the design. The benefits are numerous: Power problems go away, the computer maker knows where the heat is coming from, there are no cards to rattle around. Want more memory? Add more and the GPU, CPU all get it. GPUs have impressive performance on some non graphics problems, like simulations (eg weather), but now, with the GPU tucked away on some exceeding slow bus, you kinda have to either do your whole job on the card or on the CPU. With everything handy, you can program using both. The list goes on. Plus its cheaper!
A radical redesign would not JUST be for Apple, but rather eventually for everyone. It is just that Apple can ship computers with a new technology easier than the windows world. They have done that before more than a few times.
NVidia's Motheroard GPU is really a step towards this goal. The reason that intel integrated graphics are so bad has everything to do with Intel's graphics team. All the good people there work on CPUs. 'Everyone knows' that integrated graphics are bad. So NVidia needed a new marketing name.
--
That still has nothing to do with the bus itself. Power is brought through a connection to the 12v, 5v, or 3.3v rail in the power supply depending on the standard (i.e. PCIe, MXM, Mini PCIe, and Express Card use the same bus but their power delivery capabilities are different). This just has to do with how the standards are written. PCIe 2.0 for example doubles the maximum watts provided but that change did nothing to change the bus (architectural changes were all for the purposes of increasing speed like 33MHz PCI -> 66MHz PCI). All that entails is increasing the width of the traces going to the power pins. They did the same with with AGP Pro. The bus itself is agnostic to power input.
PCI isn't 20 years old. It came out in the mid 90s. The type of slot it's on is similar to VLB and MCA yes, but not identical, and that has nothing to do with PCI or how much power it can transfer. Also it really isn't that complicated to plug a molex connector into a card. At that logic, socketed motherboards are on their last legs since they need not only an extended ATX connector, but an additional 12v line to power just the CPU.
The massive cooling problems and need for all of that power would still exist if you socketed the GPUs. Most of the power is consumed by the cores.
Also if you want to talk about complicated, the latest PMU designs for Intel CPUs are 4-8 phase completely digital designs. Graphics cards still use simple MOSFETs and power transistors. Power regulation itself is not very difficult. The hard part is simply keeping tolerances in check and thats universal as IC speeds increase. The latest ATX spec will tighten the restrictions on voltage ripple for that reason. There is simply no way to avoid increasingly complex power circuitry no matter how you package it.
It would be nice if it could work that way. The problem with your "infinite X" idea is that without having memory for the card, every execution the GPU makes will require going to the memory. The entire point of having on board memory is to avoid this. It's the same reason why they put caches on CPUs. You're not saving anything. You also don't need two copies of the texture because it's cached in the memory of the card.
The card's memory is massively faster than the connection between FSB and main memory. On average, a P35 (the current fastest chipset as the X38 lags behind by a few percent) can gross ~7GB of bandwidth with quality memory at the 1333MHz FSB at maximum. The 8800 Ultra is rated at 103GB/s bandwidth due to a faster wider bus. The maximum throughput of a 1600MHz FSB is 12.8GB/s. Mind you, the FSB is considered a bottleneck which is why Intel is moving to an integrated memory controller soon.
In addition, when you cut the width of the memory bus (halving the bandwidth generally) performance is simply destroyed (not as critical with the G92 since they're still on a fast 256 bit bus but there is a noticeable difference in some games). This has been noted since the early Radeon days when companies would save money by marketing 64 and 128 bit versions of 128 and 256 bit (respectively) graphics cards.
Also why would you further bottleneck the CPU by using the memory of the system? This is the half the reason why Intel Integrated graphics have always been lamented.
Also you're assuming that you get a huge increase in performance from massive amounts of ram. While its true that textures are getting larger, you're not seeing a 1:1 relationship between memory size and performance. Unfortunately the improvement is never that good.
http://www.pcstats.com/articleview.cfm?articleid=2159&page=1
Petty much sums it up.
In fact, until you get to an utterly expansive game like Oblivion and turn every graphics option on, HDR, and full AA & AF at 1600x1200 and above, 256MB video cards are still a viable option.
http://www.firingsquad.com/hardware/xfx_geforce_8800_gt_256mb_xxx_review/page9.asp
Also the memory controller still comes into play no matter how you send data to the memory, PCIe bus or otherwise. There is no second interface to the memory that a socketed GPU could go through, not to mention that you simply would not want to do this.
[quote[As an example of this, look at OS X. Each window is put onto the screen by the GPU, which allows for all sorts of nice effects like the shadows, etc to happen with 'no' CPU load. But - each of the many windows on your machine has to pass through each system on the computer for each change that happens to it. For each keystroke you make, the window has to be drawn by the CPU, and then reloaded across the PCI bus to the GPU as a new texture. So even though the CPU load is low, there is lots of action on the memory - with at least 2 copies going on, etc. With the right driver and system software, all of this evaporates - the GPU can just show the window.
[/quote]
You're confusing the two display modes OS X will implement. While under Quartz Extreme, the GPU will handle the processing of the UI. This is exclusive from standard Quartz where the CPU tells the GPU what to draw. Thats why Quartz Extreme systems feel zippier than their disabled counterparts. Also the windows are always processed, not just when things are changing. Once again there will not be two copies under Quartz Extreme. Date is sent directly to the video card. And yes, you're right, with the right software, this does all disappear, it's called Quartz Extreme.
Easier said than done and not quite. If Intel could just snap their fingers and give us a 512bit bus, I'm sure they would, but the cost and complexity of the components would be ridiculous. Current video card designs treat the memory chip(s) as several memory busses and patch them together a la Dual Channel memory. This is very expensive to do in terms of transistor count. Also memory bandwidth is of most importance to GPUs since it's always dealing with memory since it receives a constant stream of data from it's own fast memory. CPUs on the other hand spend a considerable amount of time running logic and doing other tasks that don't necessarily require large amounts of bandwidth. How much bandwidth can MS word possibly require? Applications that require massive amounts of bandwidth are going the exact opposite way of what you're suggesting, and are being moved onto graphics cards. GPGPUs are being thought of as the next wave in technology because they offer large amounts of floating point power. Sharing a memory bus with a CPU would simply cripple a GPU.
When you widen the bus to an extreme degree and make it very fast, you can share memory, but the poor performance of the PS3 (which uses such an architecture) in some cases has been documented. Granted thats not the entire issue (the complexity of programing for 8 SPEs is another), the "technically inferior" Xbox 360 (sharing the same basic instruction set as the Cell's PPE), as well as PCs of comparable spec, which use a traditional architecture, have no such problems.
Yes they will have to redesign chips to make this work. The 'redesign' is only on the mundane i/o parts of the design. The benefits are numerous: Power problems go away, the computer maker knows where the heat is coming from, there are no cards to rattle around.[/url]
I assume the last part is just a dig. Your system may have be missing some screws if you're being literal here. The mundane IO parts you're talking about happens to be the FSB, not something thats easy to design. Also the power problems won't go away just by moving a very hot chip from a card into a socket. Thats like moving a 7L V8 engine from the front of the car to the back. You still need to cool it just the same. Also you still need power circuitry. Nothing goes away except the dedicated memory which doesn't get all that hot to begin with. Actually if you decide to increase your system memory so more is allocated to the GPU, you may end up with more memory chips and more heat.
You're right, one advantage would be more memory for the GPU, but the bottlenecks, as explained above, simply makes it not worth it. Also (as mentioned before) in this situation, the memory controller and FSB will have to deal with increased overhead from having another device to feed directly.
Something big in terms of performance would definitely never be an Apple exclusive. Apple is a customer of Intel, NVidia, and ATI, just like everyone else, and mind you they're a small one at that. Also big gaming performance is definitely something better shown off through companies like Voodoo PC and Alienware. Apple's game selection is a bit slim to be considered a serious gaming platform.
A new name won't save it from the problems inherent with integrated graphics. Honestly, they already know how to use the system memory for graphics. If it was the best way forward they would have been there already. Integrated graphics is not anything new. Nvidia has been doing it since the NForce2 IGP. Intel's solution isn't bad because they put "bad workers" on it. It's designed for business and home uses that exclude high performance 3D. You don't enter an economy car in the Daytona 500, and likewise, an 800HP stock car might make for a fun ride to the store but 4MPG car that costs hundreds of thousands of dollars probably isn't the best choice.
--------
The real future for GPUs and CPUs sharing the same turf is integration onto the die. AMD and Intel are both planning this, but it's not going to be for the highest end GPUs, just the lowest end ones. Both Intel and AMD are planning to make CPUs that feature integrated GPUs but right now the markets are fuzzy and no one knows if they'll be for joe 6 pack and his dell (to, as you said, decrease costs) or if they'll make their way into things like PDA, Cell Phones, and multimedia devices like the iPod Touch/iPhone. It would definitely improve the BOM to replace the three chips they're using now with one that featured the GPU, CPU, IMC, and the local interconnect (Intel and AMD have both shown slides showing roadmaps with CPUs integrating PCIe, GPU, CPU, and IMC).
Really this isn't new stuff besides the integrated GPU. Freescale had this planned since the Motorola days (the e800) and I believe their integrated IMC, CPU, PCI, IO chips have already been in production for a while now. A lot of your enthusiasm for these concepts is good, just placed in the wrong direction and wrong markets