Skip to content

Commit 97f253b

Browse files
jganganiJatin Gangani
andauthored
enable c=128 for gptoss trt (#233)
Co-authored-by: Jatin Gangani <jgangani@dc2-container-xterm-014.prd.it.nvidia.com>
1 parent d2f1254 commit 97f253b

1 file changed

Lines changed: 10 additions & 10 deletions

File tree

.github/configs/nvidia-master.yaml

Lines changed: 10 additions & 10 deletions
Original file line numberDiff line numberDiff line change
@@ -178,23 +178,23 @@ gptoss-fp4-b200-trt:
178178
- isl: 1024
179179
osl: 1024
180180
search-space:
181-
- { tp: 1, conc-start: 64, conc-end: 64 }
182-
- { tp: 2, conc-start: 4, conc-end: 64 }
183-
- { tp: 4, conc-start: 4, conc-end: 64 }
181+
- { tp: 1, conc-start: 64, conc-end: 128 }
182+
- { tp: 2, conc-start: 4, conc-end: 128 }
183+
- { tp: 4, conc-start: 4, conc-end: 128 }
184184
- { tp: 8, conc-start: 4, conc-end: 8 }
185185
- isl: 1024
186186
osl: 8192
187187
search-space:
188-
- { tp: 1, conc-start: 64, conc-end: 64 }
189-
- { tp: 2, conc-start: 4, conc-end: 64 }
190-
- { tp: 4, conc-start: 4, conc-end: 64 }
191-
- { tp: 8, conc-start: 4, conc-end: 64 }
188+
- { tp: 1, conc-start: 64, conc-end: 128 }
189+
- { tp: 2, conc-start: 4, conc-end: 128 }
190+
- { tp: 4, conc-start: 4, conc-end: 128 }
191+
- { tp: 8, conc-start: 4, conc-end: 16 }
192192
- isl: 8192
193193
osl: 1024
194194
search-space:
195-
- { tp: 1, conc-start: 64, conc-end: 64 }
196-
- { tp: 2, conc-start: 4, conc-end: 64 }
197-
- { tp: 4, conc-start: 4, conc-end: 64 }
195+
- { tp: 1, conc-start: 64, conc-end: 128 }
196+
- { tp: 2, conc-start: 4, conc-end: 128 }
197+
- { tp: 4, conc-start: 4, conc-end: 128 }
198198
- { tp: 8, conc-start: 4, conc-end: 8 }
199199

200200
gptoss-fp4-b200-vllm:

0 commit comments

Comments
 (0)