Closed shimmybalsam closed 4 months ago
dpu-rock (AMD) 8 nodes: <html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:x="urn:schemas-microsoft-com:office:excel" xmlns="http://www.w3.org/TR/REC-html40">
| hcoll | ucc oob(radix=4) | ucc radix = 2 | ucc radix = 8 -- | -- | -- | -- | -- ppn=1 | 3.57 | 4.06 | 4.98 | 3.27 ppn=64 | 12.43 | 15.12 | 14.41 | 13.58
What
Change default radix for TL/UCP barrier to 8
Why ?
Better perf for OOB