GSvit documentation

open source FDTD solver with GPU support

User Tools

Site Tools

Trace:
opt:graphics_cards

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
opt:graphics_cards [2025/02/19 14:40] – adminopt:graphics_cards [2025/02/19 14:46] (current) – admin
Line 6: Line 6:
 As a result, both the capacity of computer memory and the processing speed is a limiting factor for large and long calculations in FDTD. This fact decreases the applicability of FDTD for many purposes. An alternative approach, that is used in GSvit, is based on using a graphics processing unit (graphics card) to move the calculation from the computer processor (CPU) to the graphics processing unit (GPU) and to speed it up significantly, as shown in the following graph where the time necessary for the same calculation on different cards and processors is listed (note that the x axis is logarithmic). As a result, both the capacity of computer memory and the processing speed is a limiting factor for large and long calculations in FDTD. This fact decreases the applicability of FDTD for many purposes. An alternative approach, that is used in GSvit, is based on using a graphics processing unit (graphics card) to move the calculation from the computer processor (CPU) to the graphics processing unit (GPU) and to speed it up significantly, as shown in the following graph where the time necessary for the same calculation on different cards and processors is listed (note that the x axis is logarithmic).
  
-{{ :opt:scaling.png?400 |}}+{{ :opt:scaling.png?600 |}}
  
- The capabilities of graphics cards for numerical modeling have increased dramatically over the last few years. Two main producers NVIDIA and ATI have developed drivers for running calculations on their cards, and both memory and processing power of commonly used graphical cards have increased by a large factor. In Gsvit we focus on Nvidia CUDA products (from historical reasons when this was the only graphics card that could be coded in C). Generally, the available memory and computing power of GPUs increases much more rapidly than for personal computers themselves, which is also promising. Moreover, even special supercomputers based on graphical cards can be purchased now, like Nvidia Tesla systems. In these computers there are multiple graphics cards and we want to address also computing on multiple cards therefore.+The capabilities of graphics cards for numerical modeling have increased dramatically over the last few years. Two main producers NVIDIA and ATI have developed drivers for running calculations on their cards, and both memory and processing power of commonly used graphical cards have increased by a large factor. In GSvit we focus on NVIDIA CUDA products (from historical reasons when this was the only graphics card that could be coded in C). Generally, the available memory and computing power of GPUs increases much more rapidly than for personal computers themselves, which is also promising. Moreover, even special supercomputers based on graphical cards can be purchased now, like NVIDIA Tesla systems. In these computers there are multiple graphics cards and we want to address also computing on multiple cards therefore.
  
-To use GPU for a calculation is not straightforward, unfortunataly. We cannot simply take a conventional PC executable and run it on GPU. Both data processing and memory model is completely different for GPU and for CPU and the part of the code that should be run on GPU (called kernel) must be written to fulfill these conditions. GPU is equipped by several multiprocessors, consisting of a large number of processors. Many hundreds of threads (kernel calls) grouped in thread blocks can be processed simultaneously on GPU, which is the basis of tremendous speedup that we can achieve. Memory available on GPU can be divided into a global memory - accessible by all the multiprocessors, a shared memory - accessible by processors within one multiprocessor, and a local memory - accessible by single processor. All the memories are hardware limited (for each type of GPU differently). We refer to Nvidia CUDA developer zone for further details.+To use GPU for a calculation is not straightforward, unfortunately. We cannot simply take a conventional PC executable and run it on GPU. Both data processing and memory model is completely different for GPU and for CPU and the part of the code that should be run on GPU (called kernel) must be written to fulfill these conditions. GPU is equipped by several multiprocessors, consisting of a large number of processors. Many hundreds of threads (kernel calls) grouped in thread blocks can be processed simultaneously on GPU, which is the basis of tremendous speedup that we can achieve. Memory available on GPU can be divided into a global memory - accessible by all the multiprocessors, a shared memory - accessible by processors within one multiprocessor, and a local memory - accessible by single processor. All the memories are hardware limited (for each type of GPU differently). We refer to NVIDIA CUDA developer zone for further details.
  
 To check if the graphics card installed on your computer is suitable for GSvit calculations and if GSvit was installed with graphics card support at all, you can run the solver with parameter "test 0", e.g. on Linux system (see [[start:tests|all the tests]] for more details): To check if the graphics card installed on your computer is suitable for GSvit calculations and if GSvit was installed with graphics card support at all, you can run the solver with parameter "test 0", e.g. on Linux system (see [[start:tests|all the tests]] for more details):
  
-<html> +<code>
-<pre>+
 klapetek@pejsek:~/rungsvit/bin> ./gsvit test 0 klapetek@pejsek:~/rungsvit/bin> ./gsvit test 0
 Running GSvit tests at level 0 Running GSvit tests at level 0
Line 53: Line 52:
 Max grid size           : 2147483647x65535x65535 Max grid size           : 2147483647x65535x65535
 Max threads dim         : 1024x1024x64 Max threads dim         : 1024x1024x64
-</pre> +</code>
-</html>+
  
-This means that you can use five different GPUs on this system (which does not usually happen on a standard computer). GSvit does not support use of all of them at once, but you can still run up to four different instances on different cards. The basic settings for using the GPU are therefore whether to use it at all (GPU directive in the parameter file) and eventually which one to use (UGPU directive in the parameter file). GPUs are numbered from 0, exactly the same way as what <tt>gsvit test 0</tt> outputs, so these settings in the parameter file:+This means that you can use five different GPUs on this system (which does not usually happen on a standard computer). GSvit does not support use of all of them at once, but you can still run up to four different instances on different cards. The basic settings for using the GPU are therefore whether to use it at all (GPU directive in the parameter file) and eventually which one to use (UGPU directive in the parameter file). GPUs are numbered from 0, exactly the same way as what ''gsvit test 0'' outputs, so these settings in the parameter file:
  
-<html> +<code> 
-<tt> +GPU 
-GPU<br> +1
-1<br>+
  
-<br> +UGPU 
-UGPU<br> +0 
-0<br> +</code>
-</tt> +
-</html>+
  
 will run the calculation on the Tesla K40m card on our system. will run the calculation on the Tesla K40m card on our system.
opt/graphics_cards.1739972458.txt.gz · Last modified: 2025/02/19 14:40 by admin