NVIDIA has updated its top of the range proposed for the graphics workstation market, bringing the onboard video memory allocation to the record value of 24 Gbytes. Nothing changes on the outside compared to the declination with 12 Gbytes of video memory, and the price remains unchanged even though it is not cheap given the peculiarity of the product: about $ 5,000 plus tax.
We talk about the model M6000 Framework, with the GM200 GPU Maxwell family and subject to a doubling in the video memory allocation that exceeds 16 Gbytes of which is equipped with the competitor AMD FirePro W9100. On several occasions we have reported as a high supply of onboard video memory does not represent significant added value for the purposes of the speed in the consumer proposals aimed at gamers, but it is not at work.
The choice to bring Framework M6000 to 24 Gbytes of video memory is useful in contexts of professional graphics, for which have a frame buffer of this size allows not to run into bottlenecks. Less useful this feature considering the use of this card for GPU computing processing, not because the video memory can not be fully exploited because of the characteristics of the GM200 GPU. Recall how in the proposals of the Maxwell family the computing power in double precision is equal to 1/32 of that in single precision, because of clear choices related to architecture.
Among other innovations of this board we report a new monitoring system and management of the GPU temperature and its clock frequency, which should allow for better management of peak loads remain within the stated TDP of 250 watts for the tab .
We talk about the model M6000 Framework, with the GM200 GPU Maxwell family and subject to a doubling in the video memory allocation that exceeds 16 Gbytes of which is equipped with the competitor AMD FirePro W9100. On several occasions we have reported as a high supply of onboard video memory does not represent significant added value for the purposes of the speed in the consumer proposals aimed at gamers, but it is not at work.
The choice to bring Framework M6000 to 24 Gbytes of video memory is useful in contexts of professional graphics, for which have a frame buffer of this size allows not to run into bottlenecks. Less useful this feature considering the use of this card for GPU computing processing, not because the video memory can not be fully exploited because of the characteristics of the GM200 GPU. Recall how in the proposals of the Maxwell family the computing power in double precision is equal to 1/32 of that in single precision, because of clear choices related to architecture.
Among other innovations of this board we report a new monitoring system and management of the GPU temperature and its clock frequency, which should allow for better management of peak loads remain within the stated TDP of 250 watts for the tab .
Comments
Post a Comment