PNY TCSC2050-PB Scheda Tecnica

Pagina di 2
NVIDIA TESLA | DATASHEET | APR10
Technical SPeciFicaTiOnS
FORM FacTOR
9.75” Pcie x16 form factor
 
>
# OF TeSla GPUS 
1
 
>
# OF cUDa cOReS 
448
 
>
FReqUency OF cUDa cOReS 
1.15 Ghz
 
>
DOUble PReciSiOn FlOaTinG POinT  
PeRFORMance (Peak)
515 Gflops
 
>
SinGle
 PReciSiOn FlOaTinG POinT  
PeRFORMance (Peak)
1.03 Tflops
 
>
TOTal DeDicaTeD MeMORy
Tesla c2050: 3Gb GDDR5*
 
>
Tesla c2070: 6Gb GDDR5*
 
>
MeMORy SPeeD
 
1.5 Ghz
 
>
MeMORy inTeRFace
384-bit
 
>
MeMORy banDwiDTh
144 Gb/sec
 
>
POweR cOnSUMPTiOn
247w TDP
 
>
SySTeM inTeRFace
Pcie x16 Gen2
 
>
TheRMal SOlUTiOn
active Fansink
 
>
S
OFTwaRe DevelOPMenT TOOlS
cUDa c/c++/Fortran, Opencl,  
 
>
Directcompute Toolkits.  
nviDia Parallel nsight
 for visual Studio
*Note:  With ECC on, a portion of the dedicated 
memory is used for ECC bits, so the available user 
memory is reduced by 12.5%.  (e.g. 3 GB total memory 
yields 2.625 GB of user available memory.)
Based on the next-generation CUDA
 
architecture codenamed “Fermi”, 
the 20-series family of Tesla GPUs 
support many “must have” features for 
technical and enterprise computing 
including C++ support, ECC memory 
for uncompromised accuracy and 
scalability, and a 7X increase in double 
precision performance compared Tesla 
10-series GPUs.  The Tesla C2050 and 
C2070 GPUs are designed to redefine 
high performance computing and make 
supercomputing available to everyone.
Compared to the latest quad-core CPUs, 
Tesla C2050 and C2070 Computing 
Processors deliver equivalent 
supercomputing performance at 
1/10
th
 the cost and 1/20
th
 the power 
consumption.
The nviDia Tesla
 c2050 and c2070 computing Processors fuel  
the transition to parallel computing and bring the performance of  
a small cluster to the desktop.
 
TESLA
 C2050 / C2070
GPU ComPUTinG ProCESSor 
SUPErComPUTinG AT 1/10
Th
 ThE CoST
BIoSCIENCE
MEDICAL
FINANCE