Difference between revisions of "Blackwell (microarchitecture)"

From wikieduonline
Jump to navigation Jump to search
Tags: Mobile web edit, Mobile edit
(Nvidia claims 20 petaflops of FP4 compute with the dual GB100 die B100 accelerator)
Tags: Mobile web edit, Mobile edit
 
(One intermediate revision by the same user not shown)
Line 1: Line 1:
 
[[wikipedia:Blackwell (microarchitecture)]]
 
[[wikipedia:Blackwell (microarchitecture)]]
  
 +
* Blackwell adds support for [[FP4]] and [[FP6]] data types with eighth precision floating point processing
  
 +
Nvidia claims 20 [[petaflops]] of FP4 compute with the dual GB100 die B100 accelerator
  
 
* [[GTC]]
 
* [[GTC]]
  
 
{{NVLink}}
 
{{NVLink}}

Latest revision as of 04:07, 6 May 2024

wikipedia:Blackwell (microarchitecture)

  • Blackwell adds support for FP4 and FP6 data types with eighth precision floating point processing

Nvidia claims 20 petaflops of FP4 compute with the dual GB100 die B100 accelerator

NVLink

Advertising: