Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Tyan FT72-B7015 - initializing USB controllers freeze + Warning led light
19 views
Skip to first unread message
wolni...@lighthousestudio.pl
Jul 1, 2016, 6:08:02 AM7/1/16
to
Hi,
In our Tyan FT72-B7015 setup we have a problem of freezing at "initializing USB controllers" after installing 10th+11th GPU through an extender.
All runs perfectly if we heve installed:
1)
7 GPUs directly in 7 x16 PCI-e slots, power supplied from system's power suppliers &
In 8th x16 PCI-e slot one NVidia HIC that enables to connect 2 additional GPUs of NVidia Tesla S2070 system.
2)
5 GPUs directly in 5 PCI-e slots, power supplied from system's power suppliers &
In 6th and 7th x16 PCI-e slots two NVidia HIC that enables to connect 4 additional GPUs so whole NVidia Tesla S2070 system.
(Nvidia's S2070 GPU extender has 4GPUs, its own cooling, its own power supply etc. basically it is connected to the main system via 2 HIC cards where one HIC card connects 2GPUs through one PCI-e slot)
However problems occur if we try to run the below setup:
6 GPUs directly in 6 PCI-e slots, power supplied from system's PSUs &
In 7th and 8th x16 PCI-e slots two NVidia HIC cards that enable to connect 4 additional GPUs - so all of NVidia Tesla S2070 system.
The problem is that system is freezing at "initializing USB controllers" and after around 3-4 minutes the Warning led light at front panel lights up.
In systems manual we read that this led indicates one of the following: fan fail/ PSU fail / Over temperature / Over voltage
We checked fans , psus , and temperatures - all ok.
So probably system thinks that we try to connect to its PSUs more than these are capable of handling? Or can it be something completely else?
As a reminder Nvidia S2070 has its own PSUs and is completely independent system.
I would gladly check all potential solutions you could propose
If anything needs to be clarified please ask.
Thank you for helping.
In our Tyan FT72-B7015 setup we have a problem of freezing at "initializing USB controllers" after installing 10th+11th GPU through an extender.
All runs perfectly if we heve installed:
1)
7 GPUs directly in 7 x16 PCI-e slots, power supplied from system's power suppliers &
In 8th x16 PCI-e slot one NVidia HIC that enables to connect 2 additional GPUs of NVidia Tesla S2070 system.
2)
5 GPUs directly in 5 PCI-e slots, power supplied from system's power suppliers &
In 6th and 7th x16 PCI-e slots two NVidia HIC that enables to connect 4 additional GPUs so whole NVidia Tesla S2070 system.
(Nvidia's S2070 GPU extender has 4GPUs, its own cooling, its own power supply etc. basically it is connected to the main system via 2 HIC cards where one HIC card connects 2GPUs through one PCI-e slot)
However problems occur if we try to run the below setup:
6 GPUs directly in 6 PCI-e slots, power supplied from system's PSUs &
In 7th and 8th x16 PCI-e slots two NVidia HIC cards that enable to connect 4 additional GPUs - so all of NVidia Tesla S2070 system.
The problem is that system is freezing at "initializing USB controllers" and after around 3-4 minutes the Warning led light at front panel lights up.
In systems manual we read that this led indicates one of the following: fan fail/ PSU fail / Over temperature / Over voltage
We checked fans , psus , and temperatures - all ok.
So probably system thinks that we try to connect to its PSUs more than these are capable of handling? Or can it be something completely else?
As a reminder Nvidia S2070 has its own PSUs and is completely independent system.
I would gladly check all potential solutions you could propose
If anything needs to be clarified please ask.
Thank you for helping.
0 new messages