Wait! You can still buy music *and* screw the recording industry (according to their definition, anyway). Just visit your local used CD store. Typically a nice selection and great prices (~50% of retail), especially if you're looking for older stuff as you said you were.
This kernel patch (or is it a module? I'm not sure, haven't played with it yet) is already out, released within the past week. I've been doing most of my development on the RTE via redboot so as to avoid the restrictions placed by the Linux kernel (DMA chain packet rebuilding, primarily), but this I may have to look into.
What I'd really like to see is a slimmed-down version of the real PS2 kernel, including a network stack, file I/O, and basic threading, but without things like virtual memory and filesystem caching to get in the way.
"The neat thing is that this $199 (new, and will always get cheaper) device should, in theory, stay up to date with Apache."...
"Personally, I'd rather set one of those things up once and reboot once in a while rather than having to maintain a machine with failing hard-drives and so on."
Actually, to run this, you also need the PS2 Linux Kit (another $200), which comes with a hard drive and ethernet adapter. So in the end, you end up with a $400 machine with potentially failing hard drives, exactly what you were trying to avoid. Not to mention a slow processor and a severely limited amount of RAM. Very cool little setup to play with and learn from, though...
I agree, what was this guy talking about with "16 threads in Emotion"? And it's been a while since I've sat down next to a devkit, but if I remember, writing multithreaded code on a PS2 didn't seem to make much sense to me (no preemptive multithreading, though this was almost two years ago).
But anyway, I didn't quite get your comment about getting twice the fillrate running at 60fps interlaced as opposed to 30fps progressive. Sure, you are only rendering half of a frame, but you have to do it in half the time too. Fill rate stays the same, and if you've got a lot of motion then you get nasty tearing artifacts.
Amen to that, brother. Maybe I'm outgrew video games when I was a teen, or maybe they just aren't as good as they used to be. But to me, Square reached their peak right around SoM and FF3. Man, those were the days...
Speaking of which, do any of the modern versions of "classic" games (Mario, Zelda, FF) suck compared to their predecessors? It's seems that after they went 3D, they just lost something...
You are correct, higher resolution models is an answer, but keep in mind that this also increases your memory and bandwidth requirements by a huge margin (local memory on the vector units was severely lacking on the PS2). You must also transform all of those vertices in your mesh, which is a rather expensive process (4 vector multiply-adds and a divide on a PS2, just for the model->screen transformation), plus lighting, and then you must cull and clip these (a very expensive process). And then you still have to interpolate when it's all done! Whereas per-pixel shading requires far fewer vertex transformations, requires less memory, and at the same time generally offers smoother results.
Cheesewhiz,
I think you fail to realize the fundamentals of the architecture here. Based on my experience with the PS2, the machine this thing is eventually put in will have an incredibly fast bus. PC's are severely limited by the fact that any data that the graphics card needs must be sent over the AGP/PCI bus to the card. This is slow. However, the PS2 has very fast bus from RAM to the GS (DMA, no need to interrupt the CPU), and the bandwidth from the GS local memory to the GS is INSANELY fast.
So while a PC must have a lot of local storage in the video card (which isn't even on die, which makes it much slower) because texture transfers are very expensive, a PS2 (and logically its successor) doesn't need as much, simply because texture transfers are so cheap. And having the memory on die makes writing to/reading from texture buffers or the frame buffer also incredibly fast, thus increasing fill rate.
Per-Pixel shading. This is where we'll see a drastic improvement in the quality of interactive graphics. See those nice 3D renders from MAX? Those are likely Phong/Blinn shading models, which are per-pixel. As opposed to Gouraud shading, which calculates color values at each vertex and interpolates across the polygon. This latter method is what is used today in interactive 3D. This is why you can see the nasty aliasing across a low resolution mesh when using realtime lighting.
Unfortunately, this technique doesn't rely on enormous fill rates that this new Sony chip probably offers, but rather it requires an incredibly fast, integrated hardware lighting engine. Nothing in the article mentions this, and in the current PS2, rasterization and hardware T&L, while they work together, are completely seperate entities. Could be a while...
I personally have faith in the chip (if it ever comes out). I know everyone out there hates Intel, but don't forget, this thing was designed by HP too, and HP knows how to make a decent high-end processor.
Just curious (and I'm not at all comdemning the use of modern medicine or the potential of this new technology), but has anyone considered the fact that the use of these basically spell the end of human evolution? Sure, as a race we gain knowledge, however we stop getting physically stronger and more intelligent (for the most part), but the process of natural selection is essentially halted.
If a child is born with a vulnerability to a certain disease, we may be able to cure it. No one would let a child die if he/she didnt' have to. However, while his weak genes are spread on, the bacteria continues to evolve (this has actually already seen to be taking place) and get stronger. In nature, this individual would have died, thus not spreading his genes, and the species as a whole would have become stronger. The human race, however, has to carry the burden of this gene for as long as medicine can keep up with the evolving bacteria.
Unfortunately, I have to agree with most of your points. Also, I have some experience with this last summer, where I took an internship at a very respectable company as a Playstation 2 software engineer. As many of you probably know, the PS2 development tools from Sony run under Linux. So, as I'm sitting there setting the thing up to work, I turn to the boss and ask, "What about a Linux port" (PC/Xbox/PS2 was currently being developed at the time). He basically said "Hell no," siting the "fact" that most Linux users are hackers and would find a way around buying the game anyway.
Unfortunately, while you and I know these are not true, he does, and there was no convincing him otherwise. As hard as it may be to admit, I'm guessing that most professional game developers share this opinion. One must consider that to be senior software engineer for a game development studio, one most likely is one hell of a badass Windows programmer. The average Windows programming guru probably doesn't know squat about Linux compared to the average Linux-using Slashdotter like you and I. Therefore, he will have several ill-concieved biases against Linux, and likely will have no desire to have anything to do with developing for Linux. Unfortunately, I'm guessing it will take a long time before we can expect the major game studios to recognize Linux as a viable source of income.
Actually, your right, I don't use Photoshop much. I was referring to 3D modelling mostly, and while I mentioned that I wasn't sure where video editing, I forgot to add image editing. Speaking of which, anyone know a rough ratio of how all workstations sold are used across the various fields (3D, DV, Imaging)? I assume that most workstations sold are for 3D, since every workstation I see advertised is for 3D.
I think a lot of people here are confused about who Intel is marketing this processor towards. No SMP? Doesn't matter. It seems to me that this processor is not meant for use in servers (Xeon is their server processor money-maker). Rather, it is meant mainly for high end gaming/enthusiasts and rich computer illiterate people who buy computers from Best Buy.
This processor would also be ideal for workstation use, which contrary to popular belief, isn't necessarily enhanced by SMP. Here's my theory on that. I've worked in the game industry (Verant) for a while as a college intern, and visited a number of studios such as Volition, and also have done some 3D modelling at home on the side. Programs such as 3DS MAX and the like are optimized for SMP, but ONLY FOR FINAL RENDERING. When actually modelling in realtime, having more than one processor contributes little. The machines we were working on were dual PIII's, and it was a waste of money for that second processor. So most of what goes on in 3D graphics doesn't benefit from multiple processors, but rather a single hella-fast processor and a sweet video card. This applies to modelling a 500 poly airplane, a architectural model consisting of hundreds of thousands of polygons, and everywhere in between. (Yeah, a very few vendors advertise their drivers to take advantage of multiprocessors, but it's nowhere near perfect). I'm not sure where video editing fits in to this, though.
So everyone worried about Intel losing marketshare in the server and workstation market because of this announcement, I believe you are overlooking this point. If this were intended as a server chip, I certainly doubt they would release it without SMP capabilities, and it is in fact a worthy contender in the workstation arena.
I think the original point of the post was that Crusoe seems to be the bastard child of a PC processor and an embedded systems processor. Unfortunately, in this case, it seems to me to have inherited few of the virtues and most of the evils of its parents. In the PC market, it lacks both the raw power and market acceptance of Intel and AMD, and in the embedded systems market, it lacks both the acceptance (again) and the maturity of ARM. From what I understand, about the only thing the chip has going for itself is its low power consumption. And I agree: one must wonder where this chip (and its respective company) would be without Linus. Just my thoughts.
Okay, I'll try to do this again, without breaking the NDA:) The PS2 has 2 vector processing units. One is directly linked to the EE, designed for use with relatively simple transformations (ie collision detection) and other vector math. The second vector processor is directly linked to the GS, is more advanced than the first, and the preparation needed to take advantage of it are such that it's only use is for T&L for the rasterizer right next door.
"It does not have a processor specifically designed ONLY for performing transformation and lighting calculations."
BZZZZT. Wrong again. There's really no way for me to argue against this without going into detail on the instruction set of the vector units (which would be violating the NDA). Why else would you stick a vector processing unit, perfectly fitted for processing 128-bit (4x32-bit, the size of a vertex) floating point data, directly next to the GS, with a number of instructions dedicated to communicating with the GS? If that's not custom design and integration, I'm really not sure what is.
"A spare high-precision FPU certainly helps, but it hardly counts as a part custom optimized for T&L."
As far as it not being a "part custom optimized for T&L," its sole intended purpose is in fact for transformation and lighting. This CANNOT be argued, as it is plainly stated in the VU users manual.
Hmmm... I've heard this multiple times before, so I guess it's time to clarify.
How can you everyone the PS2 doesn't have hardware T&L? Does it not have 2 vector processors (operating on 128bit floating point data), with one linked directly to the graphics chipset (one could argue it's a part of the graphics chipset), for the sole purpose of transforming and lighting vertices? Seriously, there is NO WAY the >300MHz MIPS processor can transform the claimed 60 million polygons/second by itself. So, the PS2 *does* in fact have hardware accelerated transformation and lighting.
The only difference between hardware T&L on the PS2 and on a GeForce is the way it's used. On a GeForce, the D3D/OGL driver takes care of everything for you and hides all the ugliness. You just hand it some vertices and say 'Go.' It's a bit more complicated on the PS2, where the programmer has to write his own T&L engine in assembly.
By the way, I'd be willing to bet the X-box does support multithreading, as generally sound and I/O runs in it's own thread on both PC's and consoles.
p.s. To my credit, I am a PS2 software engineer (writing low level graphics routines) at a respectable game studio, with several months experience with the hardware.
Well, Origin layed off like 30 people, and the only titles still in development are UO2 and something else, so those 30 people had to be working on something.
Well, word on the street (or at least at the studio I work at, which shall remain nameless), Garriot got fired. Apparently, Origin has always sucked money from EA, due to there always being several projects in the works that never come to fruition. Apparently, some internal emails leaked out, though I haven't personally seen them yet. Garriot got to expensive, so they let him go. Hell, he probably doesn't care, though. He lives in a castle for Christ's sake, and will probably just go and start another company. Origin will never be the same, though.
NO NO NO!! For the love of god, do not compare this machine to a Saturn!! Being one of the lucky few to have actually worked with the dev kit (took a much needed semester off of school), the VPU's serve different purposes. If my memory serves me correctly, the Saturn had dual general purpose processors (the second added at the last moment after shots of the PSX were available), but still had a basic graphics subsystem. It was hard to take advantage of both processors in the Saturn just as it is tricky to write multithreaded programs for a dual Pentium machine (syncronization, etc.) That was like 5 years ago, though, so I'm not sure if I'm right about the Saturn.
Anyways, I admit I haven't read the article yet (I doubt it will tell me anything the 1000 pages of docs havent' already) the two vector processors serve different purposes. VU0 is a weak version of the VU1, having a quarter as much onboard memory, and lacking some of the advanced mathematical operations of the VPU1. It seems best used as a coprocessor to the EE, for stuff like a few general transformations in collision detection, for example. The VU1 has much more local storage (16k), and can only run as a 'standalone processor,' and not as a coprocessor. Send this thing your matrices and a block of vertices, and let it crunch through them and send them directly to the Graphics Synthesizer right next door. Not quite as easy to program for, though, everything MUST be done in assembly (libraries exist for using the VU0). It's not too difficult, though. That was next on my list, but our Codewarrior trial license ran out, and we have yet to purchase a seat.
Being the first console I have ever programmed for, having minimal knowledge of OpenGL, and little knowledge of matrix algebra (was supposed to take that class this semester), I was able to get a hair overd 1m polygons/sec, using just the EE and VU0, in the first 3 weeks. That should say something about the ease of development for this piece of equipment. Granted, it's the only console I've worked with, but it really didn't seem that hard to program for.
Basically, IMHO, the two vector processors will not be an issue, and in the hands of a reasonably competant programmer, the PSX will absolutely scream.
Isn't it now "Discreet 3D Studio MAX?" I believe Discreet Logic bought Kinetix or something. Somehow, Kinetix changed hands between Autodesk and Discreet, and the latter finally got rid of that dumb name (sounds like a toy company). I know my copy of 3.1 says Discreet on it... --Terrence
Slashdot Mirror
Very insightful.
Do you always run around plagiarizing posts from dup'ed articles?
Jackass.
None of that pussy Skole shit.
"overloads are much slower. Usually firing off all sorts of copy constructors."
Curious, what exactly do you mean by this statement?
Wait! You can still buy music *and* screw the recording industry (according to their definition, anyway). Just visit your local used CD store. Typically a nice selection and great prices (~50% of retail), especially if you're looking for older stuff as you said you were.
This kernel patch (or is it a module? I'm not sure, haven't played with it yet) is already out, released within the past week. I've been doing most of my development on the RTE via redboot so as to avoid the restrictions placed by the Linux kernel (DMA chain packet rebuilding, primarily), but this I may have to look into.
What I'd really like to see is a slimmed-down version of the real PS2 kernel, including a network stack, file I/O, and basic threading, but without things like virtual memory and filesystem caching to get in the way.
"The neat thing is that this $199 (new, and will always get cheaper) device should, in theory, stay up to date with Apache." ...
"Personally, I'd rather set one of those things up once and reboot once in a while rather than having to maintain a machine with failing hard-drives and so on."
Actually, to run this, you also need the PS2 Linux Kit (another $200), which comes with a hard drive and ethernet adapter. So in the end, you end up with a $400 machine with potentially failing hard drives, exactly what you were trying to avoid. Not to mention a slow processor and a severely limited amount of RAM. Very cool little setup to play with and learn from, though...
I agree, what was this guy talking about with "16 threads in Emotion"? And it's been a while since I've sat down next to a devkit, but if I remember, writing multithreaded code on a PS2 didn't seem to make much sense to me (no preemptive multithreading, though this was almost two years ago).
But anyway, I didn't quite get your comment about getting twice the fillrate running at 60fps interlaced as opposed to 30fps progressive. Sure, you are only rendering half of a frame, but you have to do it in half the time too. Fill rate stays the same, and if you've got a lot of motion then you get nasty tearing artifacts.
--Terry Caton
Amen to that, brother. Maybe I'm outgrew video games when I was a teen, or maybe they just aren't as good as they used to be. But to me, Square reached their peak right around SoM and FF3. Man, those were the days...
Speaking of which, do any of the modern versions of "classic" games (Mario, Zelda, FF) suck compared to their predecessors? It's seems that after they went 3D, they just lost something...
--Terry
You are correct, higher resolution models is an answer, but keep in mind that this also increases your memory and bandwidth requirements by a huge margin (local memory on the vector units was severely lacking on the PS2). You must also transform all of those vertices in your mesh, which is a rather expensive process (4 vector multiply-adds and a divide on a PS2, just for the model->screen transformation), plus lighting, and then you must cull and clip these (a very expensive process). And then you still have to interpolate when it's all done! Whereas per-pixel shading requires far fewer vertex transformations, requires less memory, and at the same time generally offers smoother results.
--Terrence
Cheesewhiz,
I think you fail to realize the fundamentals of the architecture here. Based on my experience with the PS2, the machine this thing is eventually put in will have an incredibly fast bus. PC's are severely limited by the fact that any data that the graphics card needs must be sent over the AGP/PCI bus to the card. This is slow. However, the PS2 has very fast bus from RAM to the GS (DMA, no need to interrupt the CPU), and the bandwidth from the GS local memory to the GS is INSANELY fast.
So while a PC must have a lot of local storage in the video card (which isn't even on die, which makes it much slower) because texture transfers are very expensive, a PS2 (and logically its successor) doesn't need as much, simply because texture transfers are so cheap. And having the memory on die makes writing to/reading from texture buffers or the frame buffer also incredibly fast, thus increasing fill rate.
--Terrence
Per-Pixel shading. This is where we'll see a drastic improvement in the quality of interactive graphics. See those nice 3D renders from MAX? Those are likely Phong/Blinn shading models, which are per-pixel. As opposed to Gouraud shading, which calculates color values at each vertex and interpolates across the polygon. This latter method is what is used today in interactive 3D. This is why you can see the nasty aliasing across a low resolution mesh when using realtime lighting.
Unfortunately, this technique doesn't rely on enormous fill rates that this new Sony chip probably offers, but rather it requires an incredibly fast, integrated hardware lighting engine. Nothing in the article mentions this, and in the current PS2, rasterization and hardware T&L, while they work together, are completely seperate entities. Could be a while...
--Terrence
I personally have faith in the chip (if it ever comes out). I know everyone out there hates Intel, but don't forget, this thing was designed by HP too, and HP knows how to make a decent high-end processor.
-----------
Terrence
www.umr.edu/~tcaton
Just curious (and I'm not at all comdemning the use of modern medicine or the potential of this new technology), but has anyone considered the fact that the use of these basically spell the end of human evolution? Sure, as a race we gain knowledge, however we stop getting physically stronger and more intelligent (for the most part), but the process of natural selection is essentially halted.
If a child is born with a vulnerability to a certain disease, we may be able to cure it. No one would let a child die if he/she didnt' have to. However, while his weak genes are spread on, the bacteria continues to evolve (this has actually already seen to be taking place) and get stronger. In nature, this individual would have died, thus not spreading his genes, and the species as a whole would have become stronger. The human race, however, has to carry the burden of this gene for as long as medicine can keep up with the evolving bacteria.
Just a thought...
Terrence - Programmer - Student - www.umr.edu/~tcaton
Unfortunately, I have to agree with most of your points. Also, I have some experience with this last summer, where I took an internship at a very respectable company as a Playstation 2 software engineer. As many of you probably know, the PS2 development tools from Sony run under Linux. So, as I'm sitting there setting the thing up to work, I turn to the boss and ask, "What about a Linux port" (PC/Xbox/PS2 was currently being developed at the time). He basically said "Hell no," siting the "fact" that most Linux users are hackers and would find a way around buying the game anyway.
Unfortunately, while you and I know these are not true, he does, and there was no convincing him otherwise. As hard as it may be to admit, I'm guessing that most professional game developers share this opinion. One must consider that to be senior software engineer for a game development studio, one most likely is one hell of a badass Windows programmer. The average Windows programming guru probably doesn't know squat about Linux compared to the average Linux-using Slashdotter like you and I. Therefore, he will have several ill-concieved biases against Linux, and likely will have no desire to have anything to do with developing for Linux. Unfortunately, I'm guessing it will take a long time before we can expect the major game studios to recognize Linux as a viable source of income.
Terrence
http://www.umr.edu/~tcaton
Actually, your right, I don't use Photoshop much. I was referring to 3D modelling mostly, and while I mentioned that I wasn't sure where video editing, I forgot to add image editing. Speaking of which, anyone know a rough ratio of how all workstations sold are used across the various fields (3D, DV, Imaging)? I assume that most workstations sold are for 3D, since every workstation I see advertised is for 3D.
--Terry
I think a lot of people here are confused about who Intel is marketing this processor towards. No SMP? Doesn't matter. It seems to me that this processor is not meant for use in servers (Xeon is their server processor money-maker). Rather, it is meant mainly for high end gaming/enthusiasts and rich computer illiterate people who buy computers from Best Buy.
This processor would also be ideal for workstation use, which contrary to popular belief, isn't necessarily enhanced by SMP. Here's my theory on that. I've worked in the game industry (Verant) for a while as a college intern, and visited a number of studios such as Volition, and also have done some 3D modelling at home on the side. Programs such as 3DS MAX and the like are optimized for SMP, but ONLY FOR FINAL RENDERING. When actually modelling in realtime, having more than one processor contributes little. The machines we were working on were dual PIII's, and it was a waste of money for that second processor. So most of what goes on in 3D graphics doesn't benefit from multiple processors, but rather a single hella-fast processor and a sweet video card. This applies to modelling a 500 poly airplane, a architectural model consisting of hundreds of thousands of polygons, and everywhere in between. (Yeah, a very few vendors advertise their drivers to take advantage of multiprocessors, but it's nowhere near perfect). I'm not sure where video editing fits in to this, though.
So everyone worried about Intel losing marketshare in the server and workstation market because of this announcement, I believe you are overlooking this point. If this were intended as a server chip, I certainly doubt they would release it without SMP capabilities, and it is in fact a worthy contender in the workstation arena.
--Terrence
I think the original point of the post was that Crusoe seems to be the bastard child of a PC processor and an embedded systems processor. Unfortunately, in this case, it seems to me to have inherited few of the virtues and most of the evils of its parents. In the PC market, it lacks both the raw power and market acceptance of Intel and AMD, and in the embedded systems market, it lacks both the acceptance (again) and the maturity of ARM. From what I understand, about the only thing the chip has going for itself is its low power consumption. And I agree: one must wonder where this chip (and its respective company) would be without Linus. Just my thoughts.
--Terrence
www.umr.edu/~tcaton
"No, the PS2 does not have onboard T&L."
:) The PS2 has 2 vector processing units. One is directly linked to the EE, designed for use with relatively simple transformations (ie collision detection) and other vector math. The second vector processor is directly linked to the GS, is more advanced than the first, and the preparation needed to take advantage of it are such that it's only use is for T&L for the rasterizer right next door.
Okay, I'll try to do this again, without breaking the NDA
"It does not have a processor specifically designed ONLY for performing transformation and lighting calculations."
BZZZZT. Wrong again. There's really no way for me to argue against this without going into detail on the instruction set of the vector units (which would be violating the NDA). Why else would you stick a vector processing unit, perfectly fitted for processing 128-bit (4x32-bit, the size of a vertex) floating point data, directly next to the GS, with a number of instructions dedicated to communicating with the GS? If that's not custom design and integration, I'm really not sure what is.
"A spare high-precision FPU certainly helps, but it hardly counts as a part custom optimized for T&L."
As far as it not being a "part custom optimized for T&L," its sole intended purpose is in fact for transformation and lighting. This CANNOT be argued, as it is plainly stated in the VU users manual.
Regards,
Terrence
www.umr.edu/~tcaton
err, make that " 300MHZ" (as in less than 300MHz), rather than greater than.
Hmmm... I've heard this multiple times before, so I guess it's time to clarify.
How can you everyone the PS2 doesn't have hardware T&L? Does it not have 2 vector processors (operating on 128bit floating point data), with one linked directly to the graphics chipset (one could argue it's a part of the graphics chipset), for the sole purpose of transforming and lighting vertices? Seriously, there is NO WAY the >300MHz MIPS processor can transform the claimed 60 million polygons/second by itself. So, the PS2 *does* in fact have hardware accelerated transformation and lighting.
The only difference between hardware T&L on the PS2 and on a GeForce is the way it's used. On a GeForce, the D3D/OGL driver takes care of everything for you and hides all the ugliness. You just hand it some vertices and say 'Go.' It's a bit more complicated on the PS2, where the programmer has to write his own T&L engine in assembly.
By the way, I'd be willing to bet the X-box does support multithreading, as generally sound and I/O runs in it's own thread on both PC's and consoles.
p.s. To my credit, I am a PS2 software engineer (writing low level graphics routines) at a respectable game studio, with several months experience with the hardware.
" ...and they're basically just USING ALL OF US to get free publicity... "
I'll have to disagree. I doubt they are doing this for publicity, considering all the publicity they are getting is negative.
--terrence
Well, Origin layed off like 30 people, and the only titles still in development are UO2 and something else, so those 30 people had to be working on something.
--Terrence
Well, word on the street (or at least at the studio I work at, which shall remain nameless), Garriot got fired. Apparently, Origin has always sucked money from EA, due to there always being several projects in the works that never come to fruition. Apparently, some internal emails leaked out, though I haven't personally seen them yet. Garriot got to expensive, so they let him go. Hell, he probably doesn't care, though. He lives in a castle for Christ's sake, and will probably just go and start another company. Origin will never be the same, though.
--Terrence
NO NO NO!! For the love of god, do not compare this machine to a Saturn!! Being one of the lucky few to have actually worked with the dev kit (took a much needed semester off of school), the VPU's serve different purposes. If my memory serves me correctly, the Saturn had dual general purpose processors (the second added at the last moment after shots of the PSX were available), but still had a basic graphics subsystem. It was hard to take advantage of both processors in the Saturn just as it is tricky to write multithreaded programs for a dual Pentium machine (syncronization, etc.) That was like 5 years ago, though, so I'm not sure if I'm right about the Saturn.
Anyways, I admit I haven't read the article yet (I doubt it will tell me anything the 1000 pages of docs havent' already) the two vector processors serve different purposes. VU0 is a weak version of the VU1, having a quarter as much onboard memory, and lacking some of the advanced mathematical operations of the VPU1. It seems best used as a coprocessor to the EE, for stuff like a few general transformations in collision detection, for example. The VU1 has much more local storage (16k), and can only run as a 'standalone processor,' and not as a coprocessor. Send this thing your matrices and a block of vertices, and let it crunch through them and send them directly to the Graphics Synthesizer right next door. Not quite as easy to program for, though, everything MUST be done in assembly (libraries exist for using the VU0). It's not too difficult, though. That was next on my list, but our Codewarrior trial license ran out, and we have yet to purchase a seat.
Being the first console I have ever programmed for, having minimal knowledge of OpenGL, and little knowledge of matrix algebra (was supposed to take that class this semester), I was able to get a hair overd 1m polygons/sec, using just the EE and VU0, in the first 3 weeks. That should say something about the ease of development for this piece of equipment. Granted, it's the only console I've worked with, but it really didn't seem that hard to program for.
Basically, IMHO, the two vector processors will not be an issue, and in the hands of a reasonably competant programmer, the PSX will absolutely scream.
--Terrence
Isn't it now "Discreet 3D Studio MAX?" I believe Discreet Logic bought Kinetix or something. Somehow, Kinetix changed hands between Autodesk and Discreet, and the latter finally got rid of that dumb name (sounds like a toy company). I know my copy of 3.1 says Discreet on it... --Terrence