You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Intel’s official specification [1] for the Arc A380 reports the following:
Xe-Cores: 8
Render Slices: 2
XVE: 128
Which implies 16 XVE * 4 Xe-Cores * 2 Render Slices = Total 128 Xe-Cores * SIMD8 = 1024 “shader cores”. This means that there are 16 XVE per Xe-Cores which is consistent with oneAPIs GPU Optimization Guide “An Xe-core of the Xe-HPG GPU contains 16 vector and 16 matrix engines” [2]
But when using Level Zero to query the GPU configuration we get the following result:
physicalEUSimdWidth = 8
numEUsPerSubslice = 8
numSubslicesPerSlice = 8
numSlices = 2
Which reports that each Xe-Core (Subslice) has 8 XVE (EUs) and that each Render Slice has 8 Xe-Cores. Is it possible that there are two different ways of counting XVEs and Xe-Cores?
As an aside it seems that GPU-Z reports something completely different with 2048 shader cores which would result in 256 XVEs.
Intel’s official specification [1] for the Arc A380 reports the following:
Which implies 16 XVE * 4 Xe-Cores * 2 Render Slices = Total 128 Xe-Cores * SIMD8 = 1024 “shader cores”. This means that there are 16 XVE per Xe-Cores which is consistent with oneAPIs GPU Optimization Guide “An Xe-core of the Xe-HPG GPU contains 16 vector and 16 matrix engines” [2]
But when using Level Zero to query the GPU configuration we get the following result:
Which reports that each Xe-Core (Subslice) has 8 XVE (EUs) and that each Render Slice has 8 Xe-Cores. Is it possible that there are two different ways of counting XVEs and Xe-Cores?
As an aside it seems that GPU-Z reports something completely different with 2048 shader cores which would result in 256 XVEs.
[1] https://www.intel.com/content/www/us/en/products/sku/227959/intel-arc-a380-graphics/specifications.html
[2] https://www.intel.com/content/www/us/en/docs/oneapi/optimization-guide-gpu/2025-0/intel-xe-gpu-architecture.html
The text was updated successfully, but these errors were encountered: