TYAN NVIDIA Compatibility

TYAN NVIDIA Compatibility List

 

 

GPUs

Generation / Model SKU Supported GPUs
NVIDIA
Tesla V100/32GB
NVIDIA
Tesla V100/16GB
NVIDIA
Tesla P100/16GB
NVIDIA
Tesla P100/12GB
NVIDIA
Tesla P40/24GB
NVIDIA
Tesla P4
NVIDIA
Tesla M60
NVIDIA
Tesla M40/24GB
NVIDIA
Tesla M10
NVIDIA
Tesla K80
NVIDIA
Tesla K40M
Purley
TN76-B7102
B7102T76V8E4HR-2T-N x2 (-2T-N) x2 (-2T-N) x2 (-2T-N) x2 (-2T-N) x2 (-2T-N)
B7102T76V12HR-2T-N x2 (-2T-N) x2 (-2T-N) x2 (-2T-N) x2 (-2T-N) x2 (-2T-N)
Purley
FT48T-B7105
B7105F48TV4HR-2T-N x5 (-2T-N) x5 30C (-2T-N) x5 30C (-2T-N)
Purley
FT48B-B7100
B7100F48BV10HR-N x4 (-N) x4 (-N) x4 (-N)
Purley
FT77D-B7109
B7109F77DV14HR-2T-N(F) x833C, 6S (-2T-N/-2T-NF)
x831C, 14S (-2T-N/-2T-NF)
x86S (-2T-N/-2T-NF)
x833C, 14S (-2T-N/-2T-NF)
x86S (-2T-N/-2T-NF)
x833C, 14S (-2T-N/-2T-NF)
x8 (-2T-N/-2T-NF) x8 (-2T-N/-2T-NF) TBD
(-2T-N/-2T-NF)
Under validation
B7109F77DV10E4HR-2T-N(F) x831C (-2T-N/-2T-NF) x833C (-2T-N/-2T-NF) x833C (-2T-N/-2T-NF) x8 (-2T-N/-2T-NF) x8 (-2T-N/-2T-NF) TBD
(-2T-N/-2T-NF)
Under validation
B7109F77DV4HR-2T-NF x833C (-2T-N/-2T-NF) x8 (-2T-N/-2T-NF) x8 (-2T-N/-2T-NF) x8 (-2T-N/-2T-NF) x8 (-2T-N/-2T-NF) TBD
(-2T-N/-2T-NF)
Under validation
Purley
FA77-B7119
B7119F77V14HR-2T-N x10!,26C (-2T-N)
x10!,28C,6S (-2T-N)
x10!,26C (-2T-N)
x10!,28C,6S (-2T-N)
x10!,26C (-2T-N)
x10!,28C,6S (-2T-N)
x8 (-2T-N)
Under validation
x20$ (-2T-N)
B7119F77V10E4HR-2T-N x10!,26C (-2T-N) x10!,26C (-2T-N) x10!,26C (-2T-N) x8 (-2T-N)
Under validation
x20$ (-2T-N)
B7119F77V4HR-2T-N x10!,30C (-2T-N) x10!,30C (-2T-N) x10!,30C (-2T-N) x8 (-2T-N)
Under validation
x20$ (-2T-N)
Purley
GA88-B5631
B5631G88V2HR-2T-N x428C (-2T-N) x429C (-2T-N) x429C (-2T-N) x4 (-2T-N)
Under validation
x4 (-2T-N)
Under validation
x4 (-2T-N)
Under validation
Grantley-EX
FT76-B7922
B7922F76U8HR-4T-HE-X
x422C (-X) x430C (-X) x4 (-X)
Grantley-EP
FT77C-B7079
B7079F77CV10HR-N
B7079F77CV10HR-2T-N
x86L (-N), (-2T-N) x86L (-N), (-2T-N) x86L (-N), (-2T-N) x86L (-N), (-2T-N) x8 (-N), (-2T-N) x8 (-N), (-2T-N) x66L(-N), (-2T-N) x8 (-N), (-2T-N) x2 (-N), (-2T-N) x86L (-N), (-2T-N) x8 (-N), (-2T-N)
Grantley-EP
TA80-B7071
B7071T80V8HR-N
B7071T80V8HR-2T-N
x4 (-N), (-2T-N) x4 (-N), (-2T-N) x4 (-N), (-2T-N) x4 (-N), (-2T-N) x4 (-N), (-2T-N) x2 (-N), (-2T-N) x4 (-N), (-2T-N) x4 (-N), (-2T-N)
Grantley-EP
GA80-B7081
B7081G80V4HR-N
B7081G80V4HR-2T-N
x325C (-N), (-2T-N) x330C (-N), (-2T-N) x330C (-N), (-2T-N) x325C (-N), (-2T-N) x3 (-N), (-2T-N)
Romley-EP
FT77B-B7059
B7059F77BV10R-N
B7059F77BV10R-2T-N
x8* (-N), (-2T-N) x8 (-N), (-2T-N)
Romley-EP
FT77A-B7059
B7059F77AV6R-N /-Nxxxx
B7059F77AV6R-2T-N /-Nxxxx
x8# * (-N), (-2T-N) x8 (-N), (-2T-N)
Romley-EP
TA77-B7061
B7061T77W8HR-N x425C (-N)
Romley-EP
GA80-B7061
B7061G80W4H-N x225C (-N)

Legend:
= Not Supported
x = Maximum supported number of GPUs
33C = Max. system ambient temperature is 33-degree C when GPUs are fully populated
31C = Max. system ambient temperature is 31-degree C when GPUs are fully populated
30C = Max. system ambient temperature is 30-degree C when GPUs are fully populated
29C = Max. system ambient temperature is 29-degree C when GPUs are fully populated
28C = Max. system ambient temperature is 28-degree C when GPUs are fully populated
25C = Max. system ambient temperature is 25-degree C when GPUs are fully populated
22C = Max. system ambient temperature is 22-degree C when GPUs are fully populated
6L = 6-LFF HDDs are populated
6S = 6-SFF HDDs are populated
# = May have no power redundancy when GPUs are fully populated in the system
* = Required Tesla K80 accessory kit to be separately ordered
! = Application Note: Modern Deep Learning frameworks have been optimized to run up to 8 GPU cards on a single server. Nvidia does not support the use of 10 GPUs under a single workload. The following two configurations south of the two PLX switches are recommended: 1) 8 GPUs + 2 NICs or 2) 10 GPUs running separate workloads.
$ = Application Note: Modern Deep Learning frameworks have been optimized to run up to 16 GPU cards on a single server. Nvidia does not support the use of 20 GPUs under a single workload. The following two configurations south of the two PLX switches are recommended: 1) 16 GPUs + 2 NICs or 2) 20 GPUs running separate workloads