What about lacquering the circuit boards? I imagine you would still have to apply silicon liberally around all the sockets and perhaps the big capacitor bases. Presumably you wouldn't need a fan, and you'd still have to keep the drives (and power supply) outside.
Then all you'd have to worry about is temperature regulation to keep the fish from boiling.
Yes, but that's going in the opposite direction (take a computer case, make something else out of it). Here the direction is (take something, make a computer out of it).
I've longed to build a mobile computer. "Mobile", as in that type of art object connected by strings where the pieces rotate in the breeze. Anyone seen one of these yet?
For airflow, there's a plethora of holes under the drives. For expansion cards, the part of the case that holds the motherboard is taller than the part that holds the drives, and should be tall enough for most expansion cards that don't tower too high above the metal bracket (the case may be a bit too short for FULL-height cards).
I didn't study the structural issues you mentioned, however.
I've also had my eye on a 1U rack case with the idea of mounting it under a desk. All that's missing is a 90 degree AGP card slot (on-board graphics still sucks).
And what do you do when the program is so obscure that you can't just "find a different one"? Write a new one from scratch?
Seriously now, the problem is not just programs that run a month. Even interrupting a program that runs for a few hours can be a critical deal when you're working on a big deadline.
Actually, when game consoles are released intially, they usually do cost more to make than their selling price. It requires continuous cost reductions over time (process shrinks, renegotiations, taking advantage of technological advances, etc.) to finally drive the real product cost below the actual selling cost.
I agree with you on this. Also, one of the nice things about developing on a console is that you have total control of the hardware, and therefore predictable response time. If you try to add additional functions to a console that sap away some of the resources, this would take away this advantage from the software developer. To maintain the separation required between the game function and the other functions would seem to require one of two things:
1) the other functions are diminished (ie, unavailable) while you're playing a game
2) or, you'd have to add more hardware to provide the other functions in parallel with gaming. This would lead to an expensive, complicated product.
The problem with many all-in-one type devices is that they often do each task more poorly than having dedicated devices for each function. For instance, a combined TV-VCR often has only one tuner, so you can't watch one program while taping another. When the all-in-one device lacks enough of the convenience of separates, it's better to get the separates.
The other problem is that you usually can't upgrade any single function without tossing the whole device to get another.
> Sure stuff got left out. (I thought they could have added 2 more hours. But then no one else would be sitting in the theater.)
I think they definitely could have adhered better to the story in certain parts without adding any additional length, or perhaps only a few seconds here and there. I imagine they didn't do these things simply due to the overwhelming amount of production work.
For instance, a couple more sightings of Gollum would only add a few seconds. Also, the battle sequence in Moria could have shown more of Gandalf's use of magic during the first fight.
Ah well. It wasn't bad. I suppose I may have enjoyed it more if I hadn't read the book so recently (with all the details still in my head).
I totally agree. The "physics" problems in many games are often not really due to things being unrealistic, but rather to things being inconsistent. Game players don't have any problems with weird rules, just as long as they understand the rules.
I think the biggest problem in most games has nothing to do with physics. It's usually because the "game" aspects of the game aren't so well thought out or developed. I think talking about physics in games is like talking about realism in movies. Making the movies more realistic doesn't automatically make them more interesting. Making them have better plot and better characters usually does.
Of course there are many industry reasons why some games are bound to be boring and repetitive. Publishers prefer establishing a "brand" and making lots of sequels (can you say "Tomb Raider"?) rather than trying out new concepts (ie, risks). Developers usually have an excessively tight schedule, and don't have time to refine. They're forced to get something working and ship it as soon as possible.
Wow. It sure looks a lot more elegant on the outside than on the inside. Unlike the nice simplicity of a common GameCube's guts, this thing has circuit boards and ribbon cables galore. I'm surprised it's only ~$100 more. They must expect to cost-reduce it in the future.
Many sites use images containing text instead of actual text. And the stupid ones don't have ALT tags, making them impossible to navigate in text-only mode. (Okay, it's been years since I used text-only mode, but I still think it should be supported).
The Xbox may have a lot of "power", but much of that power is spent spinnings its wheels and getting nowhere. The GameCube may have less "power", but its power is used more efficiently, always moving things along.
I say this because of the memory systems of the different systems. Xbox is crippled by UMA DRAM. DRAM has long latencies for page misses, and with UMA, you're guaranteed to have lots of page misses. GameCube has dedicated memories for framebuffer, texture buffer, and audio. So its graphics processor is never stalled waiting for memory access. And the 1T-SRAM means low low latency with no page-miss costs.
Peak performance figures may sound nice, but without efficiency and practical considerations, they're meaningless.
Peak figures are only achievable in extremely unrealistic (unuseful) situations. Hitting the peak polygon rate requires rendering no (or very few) pixels. How useful is that? Hitting the peak pixel fill rate typically requires drawing only one screen-size polygon.
Drawing any useful types of polygons will dramatically lower the performance figures (especially on the Xbox with UMA memory contention and DRAM latency).
What makes you think the Xbox is made in the US? I've heard the factories are in Hungary and Mexico (or was it Malaysia?).
Besides, the GameCube is also designed in the US. Nintendo has a hardware design team in Redmond, and with ArtX/ATI in California and IBM (here and there in the US), it's as American as the Xbox is.
The old design for the Koolance case was supposed to be extremely quiet. I would have bought one for that reason. The new design seems to be aimed at the overclocker, and noise appears to be a secondary consideration. For my needs, I think the new design is a failure.
Something else that seems stupid about the new design is that there is no apparent air cooling for the main part of the case. The plastic divider prevents the fans from helping to cool the motherboard, hard drives, etc.
> API's can't be faster or slower. They're just interfaces.
True, but API's can make a large difference in how well they lend themselves towards hardware acceleration. API's that require lots of data copying will leave you with slow-running applications.
One of the biggest problems with the current OpenGL API is that there are too many ways to try to accelerate geometry rendering, and they don't work equally well on different hardware.
In a way, it's kind of "nice" that DirectX totally rewrites itself every couple of versions. Now they have a single way of specifying primitives, and it's relatively simple to understand, and it lends itself well towards hardware acceleration.
The 1T-SRAM is indeed implemented using DRAM at the base level. However, the DRAM is broken down into a large number of small banks operating independently. The actual implementation details are complicated (there was an article about it at EETimes). The only result that matters is that it, for most practical purposes, behaves just like SRAM. Of course, the 1T-SRAM is more expensive than DRAM. It's also not as dense. I'm sure the reason for only 24MB (for the external RAM) on GameCube is due to this.
>I was also surprised to find that there was no shielding on the power supply unit, and no active fan on the CPU.
Regarding the CPU fan: it's got a big one. It just happens to mounted to the chassis right behind the heatsink. Of course, it also happens to be the system fan.
Regarding the power supply shielding: there's no need for it since the box is not intended to be opened by the consumer.
First off, the PS2 and GameCube don't use UMA. Both have embedded memories for framebuffer and textures. GameCube also has a separate audio memory (I don't know about PS2). For framebuffer and textures, UMA creates a big bottleneck. You have lots of high-bandwidth demands on memory, and only one memory to talk to. Having seperate framebuffer and texture memories reduces the bottleneck on main memory.
Also, on the topic of latency, only the GameCube has truly low latency access to memory. Their "1T-SRAM" allows fast random access, low-latency memory access, whereas any system based upon DRAM only has low-latency for accesses within the same memory page. Page misses are relatively high latency.
Nice pics. Could you put up a nice hires pic of the motherboard? What are the chip numbers on the memory chips? Just curious. It's strange that they have places for 4 memory chips but only have 2 on there.
Regarding the HD; if you've got an extra >8GB drive lying around, you might try sector-copying the Xbox drive to the larger drive and see if the Xbox will recognize the larger drive and its larger space.
> GameCube has a 162mhz ATI "Flipper" chip (which is similar to the ATI Radeon).
Flipper is only similar to the Radeon in that they are both integrated circuits with graphics capabilities. Architecture-wise, they are completely different.
Flipper offers more useable power than any PC-type graphics chips due to the dedicated bandwidth available from its internal framebuffer and texture memories, both of which are made of high-speed 1T-SRAM. It's all about high-speed, low-latency, random-access bandwidth.
Thus while GameCube's peak performance figures may not match those of other systems, you'll find that its processors are always running near peak, whereas on other systems the processors are sitting on their hands most of the time waiting on memory (slow DRAM with poor random access capability).
Of course, I still fully agree with your conclusion. Specs don't mean jack. Buy games because they're fun!
Slashdot Mirror
What about lacquering the circuit boards? I imagine you would still have to apply silicon liberally around all the sockets and perhaps the big capacitor bases. Presumably you wouldn't need a fan, and you'd still have to keep the drives (and power supply) outside.
Then all you'd have to worry about is temperature regulation to keep the fish from boiling.
That's a good step. Now they just need to upgrade from the WebTV to a real PC.
Yes, but that's going in the opposite direction (take a computer case, make something else out of it). Here the direction is (take something, make a computer out of it).
I've longed to build a mobile computer. "Mobile", as in that type of art object connected by strings where the pieces rotate in the breeze. Anyone seen one of these yet?
You haven't studied the pictures closely enough.
For airflow, there's a plethora of holes under the drives. For expansion cards, the part of the case that holds the motherboard is taller than the part that holds the drives, and should be tall enough for most expansion cards that don't tower too high above the metal bracket (the case may be a bit too short for FULL-height cards).
I didn't study the structural issues you mentioned, however.
I've also had my eye on a 1U rack case with the idea of mounting it under a desk. All that's missing is a 90 degree AGP card slot (on-board graphics still sucks).
And what do you do when the program is so obscure that you can't just "find a different one"? Write a new one from scratch?
Seriously now, the problem is not just programs that run a month. Even interrupting a program that runs for a few hours can be a critical deal when you're working on a big deadline.
Actually, when game consoles are released intially, they usually do cost more to make than their selling price. It requires continuous cost reductions over time (process shrinks, renegotiations, taking advantage of technological advances, etc.) to finally drive the real product cost below the actual selling cost.
I agree with you on this. Also, one of the nice things about developing on a console is that you have total control of the hardware, and therefore predictable response time. If you try to add additional functions to a console that sap away some of the resources, this would take away this advantage from the software developer. To maintain the separation required between the game function and the other functions would seem to require one of two things:
1) the other functions are diminished (ie, unavailable) while you're playing a game
2) or, you'd have to add more hardware to provide the other functions in parallel with gaming. This would lead to an expensive, complicated product.
The problem with many all-in-one type devices is that they often do each task more poorly than having dedicated devices for each function. For instance, a combined TV-VCR often has only one tuner, so you can't watch one program while taping another. When the all-in-one device lacks enough of the convenience of separates, it's better to get the separates.
The other problem is that you usually can't upgrade any single function without tossing the whole device to get another.
> Sure stuff got left out. (I thought they could have added 2 more hours. But then no one else would be sitting in the theater.)
I think they definitely could have adhered better to the story in certain parts without adding any additional length, or perhaps only a few seconds here and there. I imagine they didn't do these things simply due to the overwhelming amount of production work.
For instance, a couple more sightings of Gollum would only add a few seconds. Also, the battle sequence in Moria could have shown more of Gandalf's use of magic during the first fight.
Ah well. It wasn't bad. I suppose I may have enjoyed it more if I hadn't read the book so recently (with all the details still in my head).
I totally agree. The "physics" problems in many games are often not really due to things being unrealistic, but rather to things being inconsistent. Game players don't have any problems with weird rules, just as long as they understand the rules.
I think the biggest problem in most games has nothing to do with physics. It's usually because the "game" aspects of the game aren't so well thought out or developed. I think talking about physics in games is like talking about realism in movies. Making the movies more realistic doesn't automatically make them more interesting. Making them have better plot and better characters usually does.
Of course there are many industry reasons why some games are bound to be boring and repetitive. Publishers prefer establishing a "brand" and making lots of sequels (can you say "Tomb Raider"?) rather than trying out new concepts (ie, risks). Developers usually have an excessively tight schedule, and don't have time to refine. They're forced to get something working and ship it as soon as possible.
So YOU'RE the idiot who keeps turning my lights on while I'm sleeping!
It's quite odd when you're dreaming, and all of a sudden your dream gets very bright, and then you realize it's not a dream.
Anyway, please CUT IT OUT!
Wow. It sure looks a lot more elegant on the outside than on the inside. Unlike the nice simplicity of a common GameCube's guts, this thing has circuit boards and ribbon cables galore. I'm surprised it's only ~$100 more. They must expect to cost-reduce it in the future.
Many sites use images containing text instead of actual text. And the stupid ones don't have ALT tags, making them impossible to navigate in text-only mode. (Okay, it's been years since I used text-only mode, but I still think it should be supported).
I think it is indeed a digital port. If you, um, "study" the component video cable enough, you'll notice a chip inside of it which must be a DAC.
The Xbox may have a lot of "power", but much of that power is spent spinnings its wheels and getting nowhere. The GameCube may have less "power", but its power is used more efficiently, always moving things along.
I say this because of the memory systems of the different systems. Xbox is crippled by UMA DRAM. DRAM has long latencies for page misses, and with UMA, you're guaranteed to have lots of page misses. GameCube has dedicated memories for framebuffer, texture buffer, and audio. So its graphics processor is never stalled waiting for memory access. And the 1T-SRAM means low low latency with no page-miss costs.
Peak performance figures may sound nice, but without efficiency and practical considerations, they're meaningless.
Peak figures are only achievable in extremely unrealistic (unuseful) situations. Hitting the peak polygon rate requires rendering no (or very few) pixels. How useful is that? Hitting the peak pixel fill rate typically requires drawing only one screen-size polygon.
Drawing any useful types of polygons will dramatically lower the performance figures (especially on the Xbox with UMA memory contention and DRAM latency).
I think Nintendo was aiming at 700K for launch.
What makes you think the Xbox is made in the US? I've heard the factories are in Hungary and Mexico (or was it Malaysia?).
Besides, the GameCube is also designed in the US. Nintendo has a hardware design team in Redmond, and with ArtX/ATI in California and IBM (here and there in the US), it's as American as the Xbox is.
The old design for the Koolance case was supposed to be extremely quiet. I would have bought one for that reason. The new design seems to be aimed at the overclocker, and noise appears to be a secondary consideration. For my needs, I think the new design is a failure.
Something else that seems stupid about the new design is that there is no apparent air cooling for the main part of the case. The plastic divider prevents the fans from helping to cool the motherboard, hard drives, etc.
> API's can't be faster or slower. They're just interfaces.
True, but API's can make a large difference in how well they lend themselves towards hardware acceleration. API's that require lots of data copying will leave you with slow-running applications.
One of the biggest problems with the current OpenGL API is that there are too many ways to try to accelerate geometry rendering, and they don't work equally well on different hardware.
In a way, it's kind of "nice" that DirectX totally rewrites itself every couple of versions. Now they have a single way of specifying primitives, and it's relatively simple to understand, and it lends itself well towards hardware acceleration.
The 1T-SRAM is indeed implemented using DRAM at the base level. However, the DRAM is broken down into a large number of small banks operating independently. The actual implementation details are complicated (there was an article about it at EETimes). The only result that matters is that it, for most practical purposes, behaves just like SRAM. Of course, the 1T-SRAM is more expensive than DRAM. It's also not as dense. I'm sure the reason for only 24MB (for the external RAM) on GameCube is due to this.
>I was also surprised to find that there was no shielding on the power supply unit, and no active fan on the CPU.
Regarding the CPU fan: it's got a big one. It just happens to mounted to the chassis right behind the heatsink. Of course, it also happens to be the system fan.
Regarding the power supply shielding: there's no need for it since the box is not intended to be opened by the consumer.
First off, the PS2 and GameCube don't use UMA. Both have embedded memories for framebuffer and textures. GameCube also has a separate audio memory (I don't know about PS2). For framebuffer and textures, UMA creates a big bottleneck. You have lots of high-bandwidth demands on memory, and only one memory to talk to. Having seperate framebuffer and texture memories reduces the bottleneck on main memory.
Also, on the topic of latency, only the GameCube has truly low latency access to memory. Their "1T-SRAM" allows fast random access, low-latency memory access, whereas any system based upon DRAM only has low-latency for accesses within the same memory page. Page misses are relatively high latency.
Nice pics. Could you put up a nice hires pic of the motherboard? What are the chip numbers on the memory chips? Just curious. It's strange that they have places for 4 memory chips but only have 2 on there.
Regarding the HD; if you've got an extra >8GB drive lying around, you might try sector-copying the Xbox drive to the larger drive and see if the Xbox will recognize the larger drive and its larger space.
> GameCube has a 162mhz ATI "Flipper" chip (which is similar to the ATI Radeon).
Flipper is only similar to the Radeon in that they are both integrated circuits with graphics capabilities. Architecture-wise, they are completely different.
Flipper offers more useable power than any PC-type graphics chips due to the dedicated bandwidth available from its internal framebuffer and texture memories, both of which are made of high-speed 1T-SRAM. It's all about high-speed, low-latency, random-access bandwidth.
Thus while GameCube's peak performance figures may not match those of other systems, you'll find that its processors are always running near peak, whereas on other systems the processors are sitting on their hands most of the time waiting on memory (slow DRAM with poor random access capability).
Of course, I still fully agree with your conclusion. Specs don't mean jack. Buy games because they're fun!
> However, I am wondering why fibre channel and SCSI aren't more popular for the desktop?
Because SCSI and fibre channel controller's don't come standard on every motherboard sold.