It will still be the net. Because I can be like paul you ever see sea side heights new jersey. And you will be like no. And i will be like lets log on and i will show it to you . Or a store where I can go and see the product and turn it around and inspect it. But i think even past shockwave we will have things that will tax the connections we will have in the next 5 years or so. What we have mabye wont but what we had when 56k modems were first coming out they didn't even have stuff that could tax it. IT will just take longer with cable connections since they can scale some what upwards.Paul said:
How would you react when the PSX3 spec reads 240~480 GFLOPS?
- Thread starter DeadmeatGA
- Start date
Panajev2001a
Veteran
Josiah said:
PC gamers play different games in general and some gamers only play PC games ( for the mst part ) hence the need of computers to be faster and faster at rendering 3D graphics and CELL would fit the bill very well.
What? DO you mean Cell is heading to a PC? 
How will it work? Is it going to run MS stuffs? Hell, do you think MS will allow that? What type of board do you need? The other components like RAM/3Dcards/HDD/etc?
Or do you mean IBM/Sony are going to sell their own Cell powered *PCs*??? Do you think people will really buy that?
How will it work? Is it going to run MS stuffs? Hell, do you think MS will allow that? What type of board do you need? The other components like RAM/3Dcards/HDD/etc?
Or do you mean IBM/Sony are going to sell their own Cell powered *PCs*??? Do you think people will really buy that?
randycat99
Veteran
[creepy, screechy violin string sounds ensue in Chapworld...]
One of the interviews mentioned IBM would aim Cell at some of their lower end servers (legacy is too important for servers for it to jump into the high end merely because of architectural superiority for parallel workloads). They might try PCs too, and if not them open source enthousiasts might start using the servers on their desktops instead
I dont think Sony would want to compete with the PS3 though.
I dont think Sony would want to compete with the PS3 though.
chaphack said:What? DO you mean Cell is heading to a PC?
How will it work? Is it going to run MS stuffs? Hell, do you think MS will allow that? What type of board do you need? The other components like RAM/3Dcards/HDD/etc?
Hell, the PSX (and just so's to not confuse folks, I mean this thing) could perfectly well offer all the usual amenities of a computer, it's just that right now we don't know what Sony's aspirations are.
If the suckers are going to run Linux, what's to STOP them from doing exactly what PC's do for the vast bulk of the population? Since they can already attach keyboard and mouse, the only thing missing is broadly supported software, and as long as that exists and is simplistic for someone to use, it could certainly fit right into their scheme of things. (Would just take longer for people to decide to break old habits.) The only other major difference is the gaming, but that's no big deal--one offers PC gaming and the other offers PS gaming.
IBM obviously has its own plans for CELL, which likely involves a lot of big, bad-ass clusters, and will be providing all the software necessary for them itself. Obviously no one things Microsoft is "necessary" for adoption here. Why would it stand in the way of intrusion elsewhere? Is the PS3 going to step into the business world and replace file servers and employee desktops? Of course not. But the home is the crucial environment that MANY are fighting over, and the Playstations (and Sony products in general) are already well-established here.
If all the simple software is made easily available and is simple to use and has proper support, why WOULDN'T people be apt to adopt it? "Gaming" certainly wasn't an a core design of the PC, and it took long years for it to be intrinsically connected as it is now on many levels, so why should it seem surprising if we see PC functionality creep in from the console end? (Especially since we ALL recognize that "convergence is coming" in many, many arenas.) If Sony is smart, they'll make all the typical things people do with their PC's available and even through the PS3 (and maybe things like the PSX if it's embraced) because... well... why not? Obviously Microsoft thinks there's plenty there to make a grab at the living room, but Sony is already a bigger name and presence there, and offering everything people need should be a no-brainer.
The Linux community is already out there, and they will likely have all the same components as a PC, connected online... I imagine they will follow up on this even with the PS3--just likely not at launch. (They have enough to worry about.) I'd estimate 1-2 years later we see a more solid push, though there could be interesting things even before then depending how well CELL appeals and if Sony remains open to the Linux community and 3rd parties thinking along PC-application lines as well.
And all things considered, they're a more accepted part of the living room landscape, and won't be the expensive beast the Media Center PC's have been up until now.
All depends how each party handles each situation, but all have the ability to drive in one segment or another, and leakage will happen regardless. I'm looking forward to seeing what turns up, as the more players in each arena, the more interesting it all gets.
Panajev2001a
Veteran
chaphack said:What? DO you mean Cell is heading to a PC?
How will it work? Is it going to run MS stuffs? Hell, do you think MS will allow that? What type of board do you need? The other components like RAM/3Dcards/HDD/etc?
Or do you mean IBM/Sony are going to sell their own Cell powered *PCs*??? Do you think people will really buy that?
Chap, go back and read what I have been posting so far: code morphing/JIT compilation would help to take care of Windows.
It can be done efficienciently: go loop FX!32 up ( used for EVxx processors by Compaq/DEC/Alpha ).
Yes i did, but i am curious what type of parts a Cell PC is going to have? You know, whats inside a Cell PC? A normal PC consists of Intel\AMD cpu, a GPU, some RAM, some HDD, some Creative soundcard, some Intel/Nvidia/Via mobo etc.
So for instance, say you are PanaDELL or a DIY enthusiaists(?), whats going inside you PC?
So for instance, say you are PanaDELL
or a DIY enthusiaists(?), whats going inside you PC?DeadmeatGA
Banned
...
CELL PC? Never gonna happen. MS tried to move Windows to RISC platform before; the market chose X86. I mean, those Alpha PCs had it all, blazing speed, Windows and Office. Yet those never took off.
Now you are telling me that CELL will take off without Windows and Office? Forget it.
CELL PC? Never gonna happen. MS tried to move Windows to RISC platform before; the market chose X86. I mean, those Alpha PCs had it all, blazing speed, Windows and Office. Yet those never took off.
Now you are telling me that CELL will take off without Windows and Office? Forget it.
Panajev2001a
Veteran
What about Sony VAIOs ( laptops ALREADY use the non-x86 Transmeta Cruose VLIW processor ) and Toshiba's computers ( both laptops and desktops ) ?
Get an AGP bus license ( you could also have an embedded GPU by SCE+Toshiba ), a PCI one, put some USB and PS/2 connectors, put your custom ASIC for I/O and get some drivers ready and a CELL OS ( that would be loaded if you do not want to run Windows [dual boot] ), etc...
Nothing out of the realm of what Sony + Toshiba can do.
Again, Sony already uses the Crusoe CPU in several Sony Vaio laptops: the CPU is not a x86 CPU.
Get an AGP bus license ( you could also have an embedded GPU by SCE+Toshiba ), a PCI one, put some USB and PS/2 connectors, put your custom ASIC for I/O and get some drivers ready and a CELL OS ( that would be loaded if you do not want to run Windows [dual boot] ), etc...
Nothing out of the realm of what Sony + Toshiba can do.
Again, Sony already uses the Crusoe CPU in several Sony Vaio laptops: the CPU is not a x86 CPU.
Panajev2001a
Veteran
Re: ...
Alpha came in when games were not starting to push sales of new PCs and it had no price advantage ( their CPUs were not buolt to be cheap ) and the ACE consortium ( MIPS and EVxx were the two architectures proposed as successors to x86 ) was WAY too divided to be cohesive around a successor to x86.
Alpha was faster, but produced in limited quantities and expensive and the speed of Word Processing + e-mails was already reaching the point in which the speed differential between a Pentium MMX/Pentium Pro and an Alpha could not be felt that much.
Compaq/DEC would have not able to meet production and with the conflicts internals to the ACE consortium thanks to Intel F.U.D. tactics, the idea of replacing x86 lost a lot of steam.
Alpha also did not have prospects of a good presence in a big market such as the consumer electronics market ( which is now a major money making opportunity of r industries ) and could not have used that strenght to push itself in Desktop PCs.
Had the ACE consotrium stuck together, had Apha pushed to the ACE members technologies like FX!32 that would have allowed people not to worry about HAVING to recompile all their applications, had Compaq/DEC had money to increase Alpha chips volume production, maybe things could have been different.
Aside multimedia and games or running TONS of parallel processes at the same time ( which most users do not do ) there is NO incentive to have something faster than a 800 MHz Pentium III/ Athlon.
Finally this reality is starting to show in the PC market and wikl be more obvious as time goes by.
CELL can take off in the 'NIX environment with OpenOffice and a 'NIX dervative as its OS... custom drivers, code morphing/JIT compilation will take care of Windows AND Office.
We need multi-media ( 3D gaming especially ) designed processors, we need processors good at doing lots of parallel tasks ( people are starting to use the computer in such ways ), etc... and x86 does not fit the bill.
IPF does though, but again we have the problem of replacing x86 as you say
CELL can sneak on the side while Intel moves IPF down to the Desktop level.
DeadmeatGA said:
Alpha came in when games were not starting to push sales of new PCs and it had no price advantage ( their CPUs were not buolt to be cheap ) and the ACE consortium ( MIPS and EVxx were the two architectures proposed as successors to x86 ) was WAY too divided to be cohesive around a successor to x86.
Alpha was faster, but produced in limited quantities and expensive and the speed of Word Processing + e-mails was already reaching the point in which the speed differential between a Pentium MMX/Pentium Pro and an Alpha could not be felt that much.
Compaq/DEC would have not able to meet production and with the conflicts internals to the ACE consortium thanks to Intel F.U.D. tactics, the idea of replacing x86 lost a lot of steam.
Alpha also did not have prospects of a good presence in a big market such as the consumer electronics market ( which is now a major money making opportunity of r industries ) and could not have used that strenght to push itself in Desktop PCs.
Had the ACE consotrium stuck together, had Apha pushed to the ACE members technologies like FX!32 that would have allowed people not to worry about HAVING to recompile all their applications, had Compaq/DEC had money to increase Alpha chips volume production, maybe things could have been different.
Aside multimedia and games or running TONS of parallel processes at the same time ( which most users do not do ) there is NO incentive to have something faster than a 800 MHz Pentium III/ Athlon.
Finally this reality is starting to show in the PC market and wikl be more obvious as time goes by.
CELL can take off in the 'NIX environment with OpenOffice and a 'NIX dervative as its OS... custom drivers, code morphing/JIT compilation will take care of Windows AND Office.
We need multi-media ( 3D gaming especially ) designed processors, we need processors good at doing lots of parallel tasks ( people are starting to use the computer in such ways ), etc... and x86 does not fit the bill.
IPF does though, but again we have the problem of replacing x86 as you say
CELL can sneak on the side while Intel moves IPF down to the Desktop level.
Panajev2001a
Veteran
Call it interim solution chap: I also said that a embedded GPU ( which could be made of CELL technology ) would be a preferrable option.
The AGP slot would be there, on Desktop PCs, to allow users to try out combinations of CELL CPU ( the CELL embedded GPU would still help the CPU in CPU calculations ) and PC GPUs.
Currently CPUs still do a lot of physics, A.I. and do some help in processing geometry ( dynamic geometry, vertex creation and deletion [Vertex Shaders only work on Vertices, they cannot create or delete them yet] ): all areas that a 256 GFLOPS and 256 GOPS CELL CPU could do nicely.
A.I. and Physics would enjoy a CELL-like architecture quite a bit
You could expose even more of CELL in OpenGL ( which is not under Microsoft control, unlike DirectX ) which would make people like John Carmack happy.
On Beyond3D a certain poster called Nick, IIRC, showed his software DirectX project which optimized for x86 was 100x faster than the reference rasterizer MS provided.
Now imagine a custom CELL back-end ( his project provides the use of custom back-ends and various optimizations )
PC gaming is still strong chap and if you could sell PC-for-gaming and multimedia based on CELL + PlayStation 3s to customers, why wouldn't you ? This way you would also offer the bonus of having more CELL enabled devices in your house
The AGP slot would be there, on Desktop PCs, to allow users to try out combinations of CELL CPU ( the CELL embedded GPU would still help the CPU in CPU calculations
) and PC GPUs.Currently CPUs still do a lot of physics, A.I. and do some help in processing geometry ( dynamic geometry, vertex creation and deletion [Vertex Shaders only work on Vertices, they cannot create or delete them yet] ): all areas that a 256 GFLOPS and 256 GOPS CELL CPU could do nicely.
A.I. and Physics would enjoy a CELL-like architecture quite a bit
You could expose even more of CELL in OpenGL ( which is not under Microsoft control, unlike DirectX ) which would make people like John Carmack happy.
On Beyond3D a certain poster called Nick, IIRC, showed his software DirectX project which optimized for x86 was 100x faster than the reference rasterizer MS provided.
Now imagine a custom CELL back-end ( his project provides the use of custom back-ends and various optimizations )
PC gaming is still strong chap and if you could sell PC-for-gaming and multimedia based on CELL + PlayStation 3s to customers, why wouldn't you ? This way you would also offer the bonus of having more CELL enabled devices in your house
Panajev2001a
Veteran
chaphack said:
Cruose did not offer any advantage except supposely good power consumption, but even that was debatable.
DeadmeatGA
Banned
..
Crusoe was specifically engineered with X86 compatibility in mind(80 bit floating point registers, how many RISC CPUS do you know have these), if the code morphing software was truely hardware independent then you would have seen Transmeta dump Crusoe and sell a code morphing software for Xscale or some other RISC processor.
How many DX GPUs has Sony/Toshiba built to date? Zero.
Linux is the CELL OS.
Sure, if SCEI has the cash to burn....
Panajev2001a
Veteran
Re: ..
Ok, let's see... if you have 128 bits GRPs that support both FP and Integer data ( and DP FP and Extended FP by emulation, in the case of the APUs maybe...
http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=/netahtml/search-bool.html&r=31&f=G&l=50&co1=AND&d=ptxt&s1='Kahle+James+Allan'&OS=%22Kahle+James+Allan%22&RS=%22Kahle+James+Allan%22
) what is the problem supporting 80 bits FP values in registers ?
Did Alpha have many problems with FX!32 because of not supporting 80 bits floats natively like x86 does ? No, of course.
The reason they had to design some features especially for compatibility with x86 is because they knew the speed of their code morphing software and the target frequency and performance of their Cruose core and they did not want to make code morphing even slower on thier CPU by abstracting even more x86 features.
Also, PC CELL could differentiate itself from other CELL implementations ( without touching the CELL ISA ) thanks to these few tips and tricks:
One, in part, the Emeotion Engine ( > VS 3.0 , so ?
I already told you, for the PC GPUs afficionados you can have your AGP or PCI Express slot and you can put a two PEs Visualizer in a separate chip and call it GPU.
You can then go and expose CELL more through OpenGL ( DOOM III "engine"/Quake IV "engine" codepath if they approached John Carmack nicely ? ) or emulate DirectX using CELL's APUs and Rasterizing logic: again the Shaders can be JIT re-compiled and optimized by the driver for CELL ( Nick showed that even a custom x86 emulation of Difrect X was able to run at 100x the speed of the reference rasterizer MS provided and he was not using Pentium 4 Extreme Edition CPUs either ).
Red Hat 9/10 ? Knoppix ? Gen2 ?
Kernel <2.4.x or >2.4.x ?
I think it won't be exactly stock Linux
Hence, CELL OS != Linux.
You sure seem overly optimistic today Deadmeat.
DeadmeatGA said:
Ok, let's see... if you have 128 bits GRPs that support both FP and Integer data ( and DP FP and Extended FP by emulation, in the case of the APUs maybe...
http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=/netahtml/search-bool.html&r=31&f=G&l=50&co1=AND&d=ptxt&s1='Kahle+James+Allan'&OS=%22Kahle+James+Allan%22&RS=%22Kahle+James+Allan%22
) what is the problem supporting 80 bits FP values in registers ?
Did Alpha have many problems with FX!32 because of not supporting 80 bits floats natively like x86 does ? No, of course.
The reason they had to design some features especially for compatibility with x86 is because they knew the speed of their code morphing software and the target frequency and performance of their Cruose core and they did not want to make code morphing even slower on thier CPU by abstracting even more x86 features.
Also, PC CELL could differentiate itself from other CELL implementations ( without touching the CELL ISA
) thanks to these few tips and tricks:One, in part, the Emeotion Engine ( > VS 3.0 , so ?
I already told you, for the PC GPUs afficionados you can have your AGP or PCI Express slot and you can put a two PEs Visualizer in a separate chip and call it GPU.
You can then go and expose CELL more through OpenGL ( DOOM III "engine"/Quake IV "engine" codepath if they approached John Carmack nicely ? ) or emulate DirectX using CELL's APUs and Rasterizing logic: again the Shaders can be JIT re-compiled and optimized by the driver for CELL ( Nick showed that even a custom x86 emulation of Difrect X was able to run at 100x the speed of the reference rasterizer MS provided and he was not using Pentium 4 Extreme Edition CPUs either ).
Red Hat 9/10 ? Knoppix ? Gen2 ?
Kernel <2.4.x or >2.4.x ?
I think it won't be exactly stock Linux
Hence, CELL OS != Linux.
You sure seem overly optimistic today Deadmeat.