This topic is locked from further discussion.
[QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.Nagidar
:lol: :lol: You know absolutely nothing about computer hardware.
owned right there...the Cell is nothing specialRemember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.NickLikesWii
Yes and because Cell was proven CRAP for games in the end, IT IS ONLY FOR DECODING MOVIES, they put in RSX. Plus Cell can't do physics as xbox 360 with branching prediciton can too, so cell is a piece of garbage for games
Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.NickLikesWiiWell, take it from someone who loves the PS3 and knows the hardware intimatly - the Cell is not as fast as a GPU at rendering graphics. That was a bit of delusion some Sony designers had before :). It's not that the Cell isn't very fast; it could do more impressive software rendering in real time than any other console or PC these days. However the Cell is very general purpose hardware compared to a GPU. On a GPU every single transistor, every FPU, is designed for graphics with huge pipelines and caches designed specifically for vertices and pixels. What some people don't get however is that the PS3 Cell + GPU can be used together to produce much better graphics than the 360's GPU (which by itself is faster than the PS3 GPU). Yee shall see in good time.
[QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.Nagidar
 :lol: :lol: You know absolutely nothing about computer hardware.
Andyou know nothing about the Cell.
if the cell was so awesome, how come no one uses them for pc's?spacebyrdBecause the Cell is not x86 architecture and because the Cell is not that good at running PC apps. It is best used for scientific use, for render farms, and you guessed it, games. Sony is working on Cell based workstations for professional use and there are already huge clusters of Cell based computers at a research center based in Spain.
[QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.HarakoMeshiWell, take it from someone who loves the PS3 and knows the hardware intimatly - the Cell is not as fast as a GPU at rendering graphics. That was a bit of delusion some Sony designers had before :). Edit: Take it from someone who has written software renderers on the Cell (yours truly). It's not that the Cell isn't very fast; it could do more impressive software rendering in real time than any other console or PC these days. However the Cell is very general purpose hardware compared to a GPU. On a GPU every single transistor, every FPU, is designed for graphics with huge pipelines and caches designed specifically for vertices and pixels. What some people don't get however is that the PS3 Cell + GPU can be used together to produce much better graphics than the 360's GPU (which by itself is faster than the PS3 GPU). Yee shall see in good time.
[QUOTE="HarakoMeshi"][QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.HarakoMeshiWell, take it from someone who loves the PS3 and knows the hardware intimatly - the Cell is not as fast as a GPU at rendering graphics. That was a bit of delusion some Sony designers had before :). Edit: Take it from someone who has written software renderers on the Cell (yours truly). It's not that the Cell isn't very fast; it could do more impressive software rendering in real time than any other console or PC these days. However the Cell is very general purpose hardware compared to a GPU. On a GPU every single transistor, every FPU, is designed for graphics with huge pipelines and caches designed specifically for vertices and pixels. What some people don't get however is that the PS3 Cell + GPU can be used together to produce much better graphics than the 360's GPU (which by itself is faster than the PS3 GPU). Yee shall see in good time.
Iv been trying to tell people this, and have been looking for some info to show them, do you know where we can get a link that talks about this?
Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.NickLikesWiiYou would have an arguement if the Cell could directly read/write from the GDDR3. When I first started reading into the PS3's architecture, I thought that would be the case, and the Cell would be able to just dump the SPEs to help with the graphics computations. Unfortunately, the Cell has to go through the RSX's bus to read/write from the memory, and that means less resources for the RSX. Sigh, could've been so shiny.
Well, take it from someone who loves the PS3 and knows the hardware intimatly - the Cell is not as fast as a GPU at rendering graphics. That was a bit of delusion some Sony designers had before :). Edit: Take it from someone who has written software renderers on the Cell (yours truly). It's not that the Cell isn't very fast; it could do more impressive software rendering in real time than any other console or PC these days. However the Cell is very general purpose hardware compared to a GPU. On a GPU every single transistor, every FPU, is designed for graphics with huge pipelines and caches designed specifically for vertices and pixels. What some people don't get however is that the PS3 Cell + GPU can be used together to produce much better graphics than the 360's GPU (which by itself is faster than the PS3 GPU). Yee shall see in good time.[QUOTE="HarakoMeshi"][QUOTE="HarakoMeshi"][QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.marvelfan
Iv been trying to tell people this, and have been looking for some info to show them, do you know where we can get a link that talks about this?
This is one example showing that the Cell can perform T&L 5 - 6x faster than a nVidia 7800 GT OC. http://gametomorrow.com/blog/index.php/2005/11/30/gpus-vs-cell/
 As someone who has done a lot of Cell programming I can tell you first hand that the Cell SPUs are every bit as fast as the published figures. The reason the SPUs can come close to their theoretical maximums is because they have no caches. They have 256KB of memory running at full 3.2GHz which holds the code and data which the SPU calculates on and there is essentially no high latency to accessing this memory. Furter to aid in the process the Cell has extremely fast DMAs that transfer data between SPU local memory and the rest of the system. If a process works on streaming data (such as Transform & Lighting) it can easily achieve the maximum performance by double buffering the data in & out of the SPU.
So it is quite easy to see that the GPU Flops plus the Cell Flops add up to very high numbers. What makes the combination even more powerful than the 360's GPU though is that the SPUs are much more flexible than the GPU alone. They can work not only 'faster' but also 'smarter' doing calculating things like LOD and procedural geometry so that you have fewer triangles going to the GPU but it looks better. Think of it like comparable to a GeForce 8800. The 8800 introduces geometry shaders - shaders that can create vertices. That is new in the world of GPU.
Cell is essentially the 'geometry shader' of the PS3, something the 360 doesn't have. Sadly for the 360 its CPUs are nowhere near as fast at 'geometry processing' as the Cell. So instead they have a faster GPU that offsets the need for 'geometry processing' off the GPU. However as we see the trend with 8800 GPU and DirectX 10 'geometry shaders' are the way of the future.
[QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.FrYGuY101You would have an arguement if the Cell could directly read/write from the GDDR3. When I first started reading into the PS3's architecture, I thought that would be the case, and the Cell would be able to just dump the SPEs to help with the graphics computations. Unfortunately, the Cell has to go through the RSX's bus to read/write from the memory, and that means less resources for the RSX. Sigh, could've been so shiny. The SPEs don't directly read the GDDR3, it is inefficient to do that. (It's possible to do but very slow). Instead the GPU must 'fetch' the data from the Cell. So the SPEs must first dump the data to XDR memory and then the GPU reads them directly from there or transfers them to GDDR3. The transfer rate when GPU is in control is very fast.
To explain why ^^ 'geometry processing' is an important step forward, here are some famous words of wisdom in the world of 3D computer graphics (one I have been involved in since the PS1 days):
"The fastest triangles are the ones that you don't draw"
[QUOTE="FrYGuY101"][QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.HarakoMeshiYou would have an arguement if the Cell could directly read/write from the GDDR3. When I first started reading into the PS3's architecture, I thought that would be the case, and the Cell would be able to just dump the SPEs to help with the graphics computations. Unfortunately, the Cell has to go through the RSX's bus to read/write from the memory, and that means less resources for the RSX. Sigh, could've been so shiny. The SPEs don't directly read the GDDR3, it is inefficient to do that. (It's possible to do but very slow). Instead the GPU must 'fetch' the data from the Cell. So the SPEs must first dump the data to XDR memory and then the GPU reads them directly from there or transfers them to GDDR3. The transfer rate when GPU is in control is very fast. That's my point. Dumping it to XDR, then the GPU reading and writing to GDDR3 is inefficient in itself. If the SPEs could dump to GDDR3 directly without going throught the GPU bus it'd make the setup a lot more effective.
[QUOTE="HarakoMeshi"][QUOTE="FrYGuY101"][QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.FrYGuY101You would have an arguement if the Cell could directly read/write from the GDDR3. When I first started reading into the PS3's architecture, I thought that would be the case, and the Cell would be able to just dump the SPEs to help with the graphics computations. Unfortunately, the Cell has to go through the RSX's bus to read/write from the memory, and that means less resources for the RSX. Sigh, could've been so shiny. The SPEs don't directly read the GDDR3, it is inefficient to do that. (It's possible to do but very slow). Instead the GPU must 'fetch' the data from the Cell. So the SPEs must first dump the data to XDR memory and then the GPU reads them directly from there or transfers them to GDDR3. The transfer rate when GPU is in control is very fast. That's my point. Dumping it to XDR, then the GPU reading and writing to GDDR3 is inefficient in itself. If the SPEs could dump to GDDR3 directly without going throught the GPU bus it'd make the setup a lot more effective. That is true, but processing polygons on the cell is not 100% required it is only a bonus, and more than likely geometry on PS3 games would be split between XDR & GDDR3 anyway since the memory is not unified. The GPU will happily process data from both memories, in fact acieving even higher bandwidth by doing both in parallel. A developer would organise some geometry & textures in GDDR3 and some in XDR. Compared to the 360, 360 has an advantage of unified GDDR3 memory that everything can read from everywhere, and also a disadvantage that bandwidth is completely shared between the GPU and 3 CPU cores. If one hammers the memory the other will be starved. Also XDR bandwidth is much higher than GDDR3. One thing that I have experimented with on the PS3 is splitting frame buffers between GDDR3 and XDR. I put Z buffer in GDDR3 and color buffer in XDR. By doing this the combined bandwidth increased the memory bandwidth such that even the nastiest alpha blending did not slow down the GPU from its peak fillrate. Sweet huh? 2 memories = 2 sets of bandwidth to make use of.
Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.NickLikesWiiSo why does the ps3 has a gpu if the cell is that powerful?
[QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.rwbojorquezSo why does the ps3 has a gpu if the cell is that powerful? Read other posts. Cell is not enough to replace a GPU. Originally Sony had considered having 2 Cells, one of them being customized to be more like a GPU. Then later a custom GPU developed by Sony & Toshiba. Then finally settled on the nVidia GPU.
[QUOTE="FrYGuY101"][QUOTE="HarakoMeshi"][QUOTE="FrYGuY101"][QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.HarakoMeshiYou would have an arguement if the Cell could directly read/write from the GDDR3. When I first started reading into the PS3's architecture, I thought that would be the case, and the Cell would be able to just dump the SPEs to help with the graphics computations. Unfortunately, the Cell has to go through the RSX's bus to read/write from the memory, and that means less resources for the RSX. Sigh, could've been so shiny. The SPEs don't directly read the GDDR3, it is inefficient to do that. (It's possible to do but very slow). Instead the GPU must 'fetch' the data from the Cell. So the SPEs must first dump the data to XDR memory and then the GPU reads them directly from there or transfers them to GDDR3. The transfer rate when GPU is in control is very fast. That's my point. Dumping it to XDR, then the GPU reading and writing to GDDR3 is inefficient in itself. If the SPEs could dump to GDDR3 directly without going throught the GPU bus it'd make the setup a lot more effective. That is true, but processing polygons on the cell is not 100% required it is only a bonus, and more than likely geometry on PS3 games would be split between XDR & GDDR3 anyway since the memory is not unified. The GPU will happily process data from both memories, in fact acieving even higher bandwidth by doing both in parallel. A developer would organise some geometry & textures in GDDR3 and some in XDR. Compared to the 360, 360 has an advantage of unified GDDR3 memory that everything can read from everywhere, and also a disadvantage that bandwidth is completely shared between the GPU and 3 CPU cores. If one hammers the memory the other will be starved. Also XDR bandwidth is much higher than GDDR3. One thing that I have experimented with on the PS3 is splitting frame buffers between GDDR3 and XDR. I put Z buffer in GDDR3 and color buffer in XDR. By doing this the combined bandwidth increased the memory bandwidth such that even the nastiest alpha blending did not slow down the GPU from its peak fillrate. Sweet huh? 2 memories = 2 sets of bandwidth to make use of. Yeah, I've heard that trick, it does sound pretty useful. Still, Sony should've ponied up for a point-to-point hypertransport interconnect, that way both the CPU and GPU could make use of both RAMs without having to resort to work arounds like going through the GPU bus, and the Cell's SPEs would be far more useful for preprocessing graphics.
[QUOTE="HarakoMeshi"][QUOTE="FrYGuY101"][QUOTE="HarakoMeshi"][QUOTE="FrYGuY101"][QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.FrYGuY101You would have an arguement if the Cell could directly read/write from the GDDR3. When I first started reading into the PS3's architecture, I thought that would be the case, and the Cell would be able to just dump the SPEs to help with the graphics computations. Unfortunately, the Cell has to go through the RSX's bus to read/write from the memory, and that means less resources for the RSX. Sigh, could've been so shiny. The SPEs don't directly read the GDDR3, it is inefficient to do that. (It's possible to do but very slow). Instead the GPU must 'fetch' the data from the Cell. So the SPEs must first dump the data to XDR memory and then the GPU reads them directly from there or transfers them to GDDR3. The transfer rate when GPU is in control is very fast. That's my point. Dumping it to XDR, then the GPU reading and writing to GDDR3 is inefficient in itself. If the SPEs could dump to GDDR3 directly without going throught the GPU bus it'd make the setup a lot more effective. That is true, but processing polygons on the cell is not 100% required it is only a bonus, and more than likely geometry on PS3 games would be split between XDR & GDDR3 anyway since the memory is not unified. The GPU will happily process data from both memories, in fact acieving even higher bandwidth by doing both in parallel. A developer would organise some geometry & textures in GDDR3 and some in XDR. Compared to the 360, 360 has an advantage of unified GDDR3 memory that everything can read from everywhere, and also a disadvantage that bandwidth is completely shared between the GPU and 3 CPU cores. If one hammers the memory the other will be starved. Also XDR bandwidth is much higher than GDDR3. One thing that I have experimented with on the PS3 is splitting frame buffers between GDDR3 and XDR. I put Z buffer in GDDR3 and color buffer in XDR. By doing this the combined bandwidth increased the memory bandwidth such that even the nastiest alpha blending did not slow down the GPU from its peak fillrate. Sweet huh? 2 memories = 2 sets of bandwidth to make use of. Yeah, I've heard that trick, it does sound pretty useful. Still, Sony should've ponied up for a point-to-point hypertransport interconnect, that way both the CPU and GPU could make use of both RAMs without having to resort to work arounds like going through the GPU bus, and the Cell's SPEs would be far more useful for preprocessing graphics. Yeah well the flexIO was supposed to be that kind of amazing interconnect but finally got an incredibly poor 16MB/s reading from GDDR3 because of some bug in the implementation. (thanks STI) But I still think they are very useful if not as useful as if they were glued directly to the GPU :D Sony already ponied up too much. I still wish they had made their Sony+Toshiba custom GPU, that would have torn apart the 360 to shreads. It had 256GByte/s EDRAM!
[QUOTE="FrYGuY101"][QUOTE="HarakoMeshi"][QUOTE="FrYGuY101"][QUOTE="HarakoMeshi"][QUOTE="FrYGuY101"][QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.HarakoMeshiYou would have an arguement if the Cell could directly read/write from the GDDR3. When I first started reading into the PS3's architecture, I thought that would be the case, and the Cell would be able to just dump the SPEs to help with the graphics computations. Unfortunately, the Cell has to go through the RSX's bus to read/write from the memory, and that means less resources for the RSX. Sigh, could've been so shiny. The SPEs don't directly read the GDDR3, it is inefficient to do that. (It's possible to do but very slow). Instead the GPU must 'fetch' the data from the Cell. So the SPEs must first dump the data to XDR memory and then the GPU reads them directly from there or transfers them to GDDR3. The transfer rate when GPU is in control is very fast. That's my point. Dumping it to XDR, then the GPU reading and writing to GDDR3 is inefficient in itself. If the SPEs could dump to GDDR3 directly without going throught the GPU bus it'd make the setup a lot more effective. That is true, but processing polygons on the cell is not 100% required it is only a bonus, and more than likely geometry on PS3 games would be split between XDR & GDDR3 anyway since the memory is not unified. The GPU will happily process data from both memories, in fact acieving even higher bandwidth by doing both in parallel. A developer would organise some geometry & textures in GDDR3 and some in XDR. Compared to the 360, 360 has an advantage of unified GDDR3 memory that everything can read from everywhere, and also a disadvantage that bandwidth is completely shared between the GPU and 3 CPU cores. If one hammers the memory the other will be starved. Also XDR bandwidth is much higher than GDDR3. One thing that I have experimented with on the PS3 is splitting frame buffers between GDDR3 and XDR. I put Z buffer in GDDR3 and color buffer in XDR. By doing this the combined bandwidth increased the memory bandwidth such that even the nastiest alpha blending did not slow down the GPU from its peak fillrate. Sweet huh? 2 memories = 2 sets of bandwidth to make use of. Yeah, I've heard that trick, it does sound pretty useful. Still, Sony should've ponied up for a point-to-point hypertransport interconnect, that way both the CPU and GPU could make use of both RAMs without having to resort to work arounds like going through the GPU bus, and the Cell's SPEs would be far more useful for preprocessing graphics. Yeah well the flexIO was supposed to be that kind of amazing interconnect but finally got an incredibly poor 16MB/s reading from GDDR3 because of some bug in the implementation. (thanks STI) But I still think they are very useful if not as useful as if they were glued directly to the GPU :D Sony already ponied up too much. I still wish they had made their Sony+Toshiba custom GPU, that would have torn apart the 360 to shreads. It had 256GByte/s EDRAM! The 360 *HAS* 256GB/s ED-RAM, which handles z-buffering, alpha-blending, fp10 HDR and 4xAA. It's a developer's dream, man...
[QUOTE="HarakoMeshi"][QUOTE="FrYGuY101"][QUOTE="HarakoMeshi"][QUOTE="FrYGuY101"][QUOTE="HarakoMeshi"][QUOTE="FrYGuY101"][QUOTE="NickLikesWii"]Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.FrYGuY101You would have an arguement if the Cell could directly read/write from the GDDR3. When I first started reading into the PS3's architecture, I thought that would be the case, and the Cell would be able to just dump the SPEs to help with the graphics computations. Unfortunately, the Cell has to go through the RSX's bus to read/write from the memory, and that means less resources for the RSX. Sigh, could've been so shiny. The SPEs don't directly read the GDDR3, it is inefficient to do that. (It's possible to do but very slow). Instead the GPU must 'fetch' the data from the Cell. So the SPEs must first dump the data to XDR memory and then the GPU reads them directly from there or transfers them to GDDR3. The transfer rate when GPU is in control is very fast. That's my point. Dumping it to XDR, then the GPU reading and writing to GDDR3 is inefficient in itself. If the SPEs could dump to GDDR3 directly without going throught the GPU bus it'd make the setup a lot more effective. That is true, but processing polygons on the cell is not 100% required it is only a bonus, and more than likely geometry on PS3 games would be split between XDR & GDDR3 anyway since the memory is not unified. The GPU will happily process data from both memories, in fact acieving even higher bandwidth by doing both in parallel. A developer would organise some geometry & textures in GDDR3 and some in XDR. Compared to the 360, 360 has an advantage of unified GDDR3 memory that everything can read from everywhere, and also a disadvantage that bandwidth is completely shared between the GPU and 3 CPU cores. If one hammers the memory the other will be starved. Also XDR bandwidth is much higher than GDDR3. One thing that I have experimented with on the PS3 is splitting frame buffers between GDDR3 and XDR. I put Z buffer in GDDR3 and color buffer in XDR. By doing this the combined bandwidth increased the memory bandwidth such that even the nastiest alpha blending did not slow down the GPU from its peak fillrate. Sweet huh? 2 memories = 2 sets of bandwidth to make use of. Yeah, I've heard that trick, it does sound pretty useful. Still, Sony should've ponied up for a point-to-point hypertransport interconnect, that way both the CPU and GPU could make use of both RAMs without having to resort to work arounds like going through the GPU bus, and the Cell's SPEs would be far more useful for preprocessing graphics. Yeah well the flexIO was supposed to be that kind of amazing interconnect but finally got an incredibly poor 16MB/s reading from GDDR3 because of some bug in the implementation. (thanks STI) But I still think they are very useful if not as useful as if they were glued directly to the GPU :D Sony already ponied up too much. I still wish they had made their Sony+Toshiba custom GPU, that would have torn apart the 360 to shreads. It had 256GByte/s EDRAM! The 360 *HAS* 256GB/s ED-RAM, which handles z-buffering, alpha-blending, fp10 HDR and 4xAA. It's a developer's dream, man...
That is correct although the actual bus from GPU to the EDRAM daughter chip (the Smart Memory) is only 32GB/s. The daughter chip is able to utilise 256GB/s when doing 4x MSAA. Essentially 4xMSAA is free. This has the effect of smoothing polygon edges but does not work on things such as wire fences or other transparent things.  Those require SSAA which is then limited by the 32GB/s. (Still faster than the PS3 though).
The only limitation of the 360 GPU is having only 10MB EDRAM which is not enough to take advantage of free 4x MSAA at 720p resolution. That would require over 28MB of frame buffer memory (1280*720*8bytes*4) so the whole scene would have to be rendered 3 times to 3 slices of memory, thus making it 'not so free' - in fact slower than the PS3 can do it.
This is the reason PS3 can boast higher resolutions including true 1080p games. At 1080p a minimum of 16MB of video ram is required without any kind of AA. 64MB with 4X AA. The 360 would have to render between 2 and 7 passes to achieve 1080p while the PS3 can do it in one pass.
On the face of things the 360 GPU appears to have an edge over the PS3, but for 720p games with 4xAA it's actually going to be slower. "Oh oh", atomic lemming destruction. "pop, pop, pop, pop, pop, pop, pop....".
Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.NickLikesWiii cant belive after all the lies your still beliveing sonys words wow how sad lol.
I read an article written by a hardware expert a while back. It explained that the Cell is indeed extremely powerful... but it's not designed for running games. The lowdown was that most of the power and potential of the Cell is wasted due to it being poorly suited to the demands of games. Don't think that more power = better games. An aircraft engine is powerful but would you use it to power a car?GreenManSorry but actually Cell is very suited for games. Most of the time spent on most games is graphics, physics and audio. Cell excells at those tasks. It's also great at decoding 1080p movies - great for BD playback.
Would you use an aircraft engine on a car? You would if you were trying to break the land speed record.
Remember Sony talking about how the PS3 can generate excellent real-time graphics using only the power of the Cell, with no GPU at all? They showed up demonstrations of this at E3.NickLikesWii
The Cell + RSX >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> xenos + xenondirtysneekers55
[QUOTE="dirtysneekers55"]The Cell + RSX >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> xenos + xenoncakeorrdeath
[QUOTE="cakeorrdeath"][QUOTE="dirtysneekers55"]The Cell + RSX >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> xenos + xenonHarakoMeshi
[QUOTE="cakeorrdeath"][QUOTE="dirtysneekers55"]The Cell + RSX >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> xenos + xenonHarakoMeshi
Its well suited to graphics compared to a regular CPU. Compared to a GPU though it sucks.I doubt Cell adds half of what RSX adds to the graphical power of PS3 in reality. And thats if it dedicated itself to graphics which it won't under normal circumstances.
Please Log In to post.
Log in to comment