Skip to content

Commit e2d0d09

Browse files
klueskaklueska
committed
WIP: Add a POC of an alternate partitioaing scheme
Signed-off-by: Kevin Klues <kklues@nvidia.com>
1 parent fdc5827 commit e2d0d09

File tree

1 file changed

+289
-0
lines changed

1 file changed

+289
-0
lines changed

‎dra-evolution/pkg/api/poc.yaml‎

Lines changed: 289 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,289 @@
1+
sharedAttributeGroups:
2+
- name: common-attributes
3+
attributes:
4+
- name: product-name
5+
string: Mock NVIDIA A100-SXM4-40GB
6+
- name: brand
7+
string: Nvidia
8+
- name: architecture
9+
string: Ampere
10+
- name: cuda-compute-capability
11+
version: 8.0.0
12+
- name: driver-version
13+
version: 550.54.15
14+
- name: cuda-driver-version
15+
version: 12.4.0
16+
17+
sharedCapacityTemplates:
18+
- name: gpu-shared-resources
19+
capacities:
20+
- name: memory
21+
quantity: 40Gi
22+
- name: multiprocessors
23+
quantity: "98"
24+
- name: copy-engines
25+
quantity: "7"
26+
- name: decoders
27+
quantity: "5"
28+
- name: encoders
29+
quantity: "0"
30+
- name: jpeg-engines
31+
quantity: "1"
32+
- name: ofa-engines
33+
quantity: "1"
34+
- name: memory-slices
35+
intRange: 0-7
36+
37+
deviceTemplates:
38+
- name: gpu
39+
sharedAttributeGroups:
40+
- common-attributes
41+
attributes:
42+
- name: memory
43+
quantity: 40Gi
44+
- name: mig-capable
45+
bool: true
46+
sharedCapacitiesConsumed:
47+
- sharedCapacityTemplateName: gpu-shared-resources
48+
capacities:
49+
- name: multiprocessors
50+
quantity: "98"
51+
- name: copy-engines
52+
quantity: "7"
53+
- name: decoders
54+
quantity: "5"
55+
- name: encoders
56+
quantity: "0"
57+
- name: jpeg-engines
58+
quantity: "1"
59+
- name: ofa-engines
60+
quantity: "1"
61+
- name: memory-slices
62+
intRange: 0-7
63+
- name: mig-1g.5gb
64+
sharedAttributeGroups:
65+
- common-attributes
66+
attributes:
67+
- name: mig-profile
68+
string: 1g.5gb
69+
- name: memory
70+
quantity: 4864Mi
71+
sharedCapacitiesConsumed:
72+
- sharedCapacityTemplateName: gpu-shared-resources
73+
capacities:
74+
- name: multiprocessors
75+
quantity: "14"
76+
- name: copy-engines
77+
quantity: "1"
78+
- name: decoders
79+
quantity: "0"
80+
- name: encoders
81+
quantity: "0"
82+
- name: jpeg-engines
83+
quantity: "0"
84+
- name: ofa-engines
85+
quantity: "0"
86+
87+
devices:
88+
- name: gpu
89+
namespace: gpu-0
90+
deviceTemplateName: gpu
91+
attributes:
92+
- name: index
93+
string: "0"
94+
- name: minor
95+
string: "0"
96+
- name: uuid
97+
string: GPU-0eaad900-5263-4fd6-b020-f03d30efac31
98+
- name: mig-1g.5gb-0
99+
namespace: gpu-0
100+
deviceTemplateName: mig-1g.5gb
101+
attributes:
102+
- name: index
103+
string: "0:0"
104+
- name: parentUUID
105+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
106+
sharedCapacitiesConsumed:
107+
- sharedCapacityTemplateName: gpu-shared-resources
108+
capacities:
109+
- name: memory-slices
110+
intRange: "0"
111+
- name: mig-1g.5gb-1
112+
namespace: gpu-0
113+
deviceTemplateName: mig-1g.5gb
114+
attributes:
115+
- name: index
116+
string: "0:1"
117+
- name: parentUUID
118+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
119+
sharedCapacitiesConsumed:
120+
- sharedCapacityTemplateName: gpu-shared-resources
121+
capacities:
122+
- name: memory-slices
123+
intRange: "1"
124+
- name: mig-1g.5gb-2
125+
namespace: gpu-0
126+
deviceTemplateName: mig-1g.5gb
127+
attributes:
128+
- name: index
129+
string: "0:2"
130+
- name: parentUUID
131+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
132+
sharedCapacitiesConsumed:
133+
- sharedCapacityTemplateName: gpu-shared-resources
134+
capacities:
135+
- name: memory-slices
136+
intRange: "2"
137+
- name: mig-1g.5gb-3
138+
namespace: gpu-0
139+
deviceTemplateName: mig-1g.5gb
140+
attributes:
141+
- name: index
142+
string: "0:3"
143+
- name: parentUUID
144+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
145+
sharedCapacitiesConsumed:
146+
- sharedCapacityTemplateName: gpu-shared-resources
147+
capacities:
148+
- name: memory-slices
149+
intRange: "3"
150+
- name: mig-1g.5gb-4
151+
namespace: gpu-0
152+
deviceTemplateName: mig-1g.5gb
153+
attributes:
154+
- name: index
155+
string: "0:4"
156+
- name: parentUUID
157+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
158+
sharedCapacitiesConsumed:
159+
- sharedCapacityTemplateName: gpu-shared-resources
160+
capacities:
161+
- name: memory-slices
162+
intRange: "4"
163+
- name: mig-1g.5gb-5
164+
namespace: gpu-0
165+
deviceTemplateName: mig-1g.5gb
166+
attributes:
167+
- name: index
168+
string: "0:5"
169+
- name: parentUUID
170+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
171+
sharedCapacitiesConsumed:
172+
- sharedCapacityTemplateName: gpu-shared-resources
173+
capacities:
174+
- name: memory-slices
175+
intRange: "5"
176+
- name: mig-1g.5gb-6
177+
namespace: gpu-0
178+
deviceTemplateName: mig-1g.5gb
179+
attributes:
180+
- name: index
181+
string: "0:6"
182+
- name: parentUUID
183+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
184+
sharedCapacitiesConsumed:
185+
- sharedCapacityTemplateName: gpu-shared-resources
186+
capacities:
187+
- name: memory-slices
188+
intRange: "6"
189+
- name: gpu
190+
namespace: gpu-1
191+
deviceTemplateName: gpu
192+
attributes:
193+
- name: index
194+
string: "1"
195+
- name: minor
196+
string: "1"
197+
- name: uuid
198+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
199+
- name: mig-1g.5gb-0
200+
namespace: gpu-1
201+
deviceTemplateName: mig-1g.5gb
202+
attributes:
203+
- name: index
204+
string: "1:0"
205+
- name: parentUUID
206+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
207+
sharedCapacitiesConsumed:
208+
- sharedCapacityTemplateName: gpu-shared-resources
209+
capacities:
210+
- name: memory-slices
211+
intRange: "0"
212+
- name: mig-1g.5gb-1
213+
namespace: gpu-1
214+
deviceTemplateName: mig-1g.5gb
215+
attributes:
216+
- name: index
217+
string: "1:1"
218+
- name: parentUUID
219+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
220+
sharedCapacitiesConsumed:
221+
- sharedCapacityTemplateName: gpu-shared-resources
222+
capacities:
223+
- name: memory-slices
224+
intRange: "1"
225+
- name: mig-1g.5gb-2
226+
namespace: gpu-1
227+
deviceTemplateName: mig-1g.5gb
228+
attributes:
229+
- name: index
230+
string: "1:2"
231+
- name: parentUUID
232+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
233+
sharedCapacitiesConsumed:
234+
- sharedCapacityTemplateName: gpu-shared-resources
235+
capacities:
236+
- name: memory-slices
237+
intRange: "2"
238+
- name: mig-1g.5gb-3
239+
namespace: gpu-1
240+
deviceTemplateName: mig-1g.5gb
241+
attributes:
242+
- name: index
243+
string: "1:3"
244+
- name: parentUUID
245+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
246+
sharedCapacitiesConsumed:
247+
- sharedCapacityTemplateName: gpu-shared-resources
248+
capacities:
249+
- name: memory-slices
250+
intRange: "3"
251+
- name: mig-1g.5gb-4
252+
namespace: gpu-1
253+
deviceTemplateName: mig-1g.5gb
254+
attributes:
255+
- name: index
256+
string: "1:4"
257+
- name: parentUUID
258+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
259+
sharedCapacitiesConsumed:
260+
- sharedCapacityTemplateName: gpu-shared-resources
261+
capacities:
262+
- name: memory-slices
263+
intRange: "4"
264+
- name: mig-1g.5gb-5
265+
namespace: gpu-1
266+
deviceTemplateName: mig-1g.5gb
267+
attributes:
268+
- name: index
269+
string: "1:5"
270+
- name: parentUUID
271+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
272+
sharedCapacitiesConsumed:
273+
- sharedCapacityTemplateName: gpu-shared-resources
274+
capacities:
275+
- name: memory-slices
276+
intRange: "5"
277+
- name: mig-1g.5gb-6
278+
namespace: gpu-1
279+
deviceTemplateName: mig-1g.5gb
280+
attributes:
281+
- name: index
282+
string: "1:6"
283+
- name: parentUUID
284+
string: GPU-4404041a-04cf-1ccf-9e70-f139a9b1e23c
285+
sharedCapacitiesConsumed:
286+
- sharedCapacityTemplateName: gpu-shared-resources
287+
capacities:
288+
- name: memory-slices
289+
intRange: "5"

0 commit comments

Comments
 (0)