Rev |
Log message |
Author |
Age |
Path |
108 |
These are the latest changes. With this the GPU will resemble the new multi core architecture with separate buses for Out-memory (no share), Texture (cross-bar interconnect) and Geometry (hub/switch network). This will allow more cores to be connected in parallel minimizing the associated bus contention (or so I hope). |
diegovalverde |
5043d 15h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
107 |
These are the latest changes. With this the GPU will resemble the new multi core architecture with separate buses for Out-memory (no share), Texture (cross-bar interconnect) and Geometry (hub/switch network). This will allow more cores to be connected in parallel minimizing the associated bus contention (or so I hope). |
diegovalverde |
5043d 15h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
106 |
These are the latest changes. With this the GPU will resemble the new multi core architecture with separate buses for Out-memory (no share), Texture (cross-bar interconnect) and Geometry (hub/switch network). This will allow more cores to be connected in parallel minimizing the associated bus contention (or so I hope). |
diegovalverde |
5043d 15h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
105 |
These are the latest changes. With this the GPU will resemble the new multi core architecture with separate buses for Out-memory (no share), Texture (cross-bar interconnect) and Geometry (hub/switch network). This will allow more cores to be connected in parallel minimizing the associated bus contention (or so I hope). |
diegovalverde |
5043d 15h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
104 |
These are the latest changes. With this the GPU will resemble the new multi core architecture with separate buses for Out-memory (no share), Texture (cross-bar interconnect) and Geometry (hub/switch network). This will allow more cores to be connected in parallel minimizing the associated bus contention (or so I hope). |
diegovalverde |
5043d 15h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
103 |
This unit is no longer present in the new architecture |
diegovalverde |
5043d 15h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
102 |
Moving this unit under the CORES/TOP folder |
diegovalverde |
5044d 13h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
101 |
Adding new folder into the hierarchy |
diegovalverde |
5044d 13h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
100 |
Changed my mind |
diegovalverde |
5044d 13h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
99 |
Changed my mind |
diegovalverde |
5044d 13h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
98 |
Adding new folder into the hierarchy |
diegovalverde |
5044d 13h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
97 |
Adding new folder into the hierarchy |
diegovalverde |
5044d 13h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
96 |
Adding new folder into the hierarchy |
diegovalverde |
5044d 13h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
95 |
Moving this unit under the GPU folder |
diegovalverde |
5044d 13h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
94 |
Moving this unit under the CORES folder |
diegovalverde |
5044d 13h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
93 |
Moving this unit under the CORES folder |
diegovalverde |
5044d 13h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
92 |
Moving this unit under the CORES folder |
diegovalverde |
5044d 13h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
91 |
Moving this unit under the CORES folder |
diegovalverde |
5044d 13h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
90 |
Adding new folder into the hierarchy |
diegovalverde |
5044d 14h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
89 |
Adding new folder into the hierarchy |
diegovalverde |
5044d 14h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
88 |
This unit is no longer present in the new architecture |
diegovalverde |
5044d 14h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
82 |
Latest changes:
Support for 4 cores working simultaneously.
The scene is divided in 4, each core renders 1/4 of the scene.
Amdahl's law is smiling upon me. |
diegovalverde |
5149d 12h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
81 |
Latest changes:
Support for 4 cores working simultaneously.
The scene is divided in 4, each core renders 1/4 of the scene.
Amdahl's law is smiling upon me. |
diegovalverde |
5149d 12h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
76 |
Latest changes:
Support for 2 cores working simultaneously.
The scene is divided in two, each core renders half of the scene.
The two cores share a bus, wish-bone protocol, and round-robin bus arbiter. |
diegovalverde |
5163d 14h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
75 |
Latest changes to make IO work in parallel with EXE |
diegovalverde |
5273d 23h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
74 |
The main changes to make IO work in parallel with EXE are here. The memory has been divided in different segments. There is a memory tha is swapping between IO and EXE. |
diegovalverde |
5273d 23h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
73 |
Changes to make IO and EXE work in parallel |
diegovalverde |
5273d 23h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
72 |
Changes some files to facilitate EXE and IO to work in parallel |
diegovalverde |
5273d 23h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
71 |
Added a flag in control unit to set the memory to swap or not, this is, while IO is populating one copy, EXE is working on the other, then they get swapped |
diegovalverde |
5274d 00h |
/theia_gpu/branches/gpu_8_cores/rtl/ |
70 |
Fixed some of the arithmetic modules, so that we now can have EXE and IO work in parallel. |
diegovalverde |
5274d 00h |
/theia_gpu/branches/gpu_8_cores/rtl/ |