Change snmg index to use updated multi gpu resource API

[jinsolp]

Description

These changes are dependent on this breaking PR in raft.
This PR itself doesn't introduce any new features, but makes changes to the existing code to work with the breaking changes in the PR in raft.

[jinsolp]

This needs to be reverted

[viclafargue]

Thanks @jinsolp, LGTM apart from the benchmark files. Could you switch the raft::device_resources_snmg to raft::device_resources_snmg_nccl?

[jinsolp]

@viclafargue Thanks for pointing out Victor, I also had to change function calls in snmg.cuh so that it works with the change function names in raft!

[cjnolet]

@jinsolp don't forget to revert the change in get_raft.cmake before this is merged.

[achirkin]

Hi Jinsol, did you have a chance to run the ann_bench with the new changes?
I've just tried this branch on a system with two GPUs, but limited to only one via the env variable CUDA_VISIBLE_DEVICES=0

(cuml_dev) CUDA_VISIBLE_DEVICES=0 ./cpp/build/bench/ann/ANN_BENCH \
  --search \
  --mode=latency \
  --benchmark_filter=cuvs_mg_cagra.dim32_vpq \
  --benchmark_min_time=3s \
  --benchmark_min_warmup_time=0.01 \
  --benchmark_counters_tabular \
  --override_kv=dataset_memory_type:\"host\" \
  --override_kv=query_memory_type:\"pinned\" \
  --override_kv=graph_memory_type:\"host_pinned\" \
  --override_kv=persistent:false \
  --data_prefix=/tmp/data \
  --index_prefix=/ann/index \
  /ann/conf/deep-100M-cagra-persistent.json
[I] [12:00:04.876774] Using the query file '/tmp/data/deep-1B/query.public.10K.fbin'
[I] [12:00:04.876805] Using the ground truth file '/tmp/data/deep-100M/groundtruth.neighbors.ibin'
2025-05-13T12:00:04+00:00
Running ./cpp/build/bench/ann/ANN_BENCH
Run on (32 X 3729.49 MHz CPU s)
CPU Caches:
  L1 Data 32 KiB (x16)
  L1 Instruction 32 KiB (x16)
  L2 Unified 512 KiB (x16)
  L3 Unified 32768 KiB (x4)
Load Average: 0.21, 4.55, 12.52
command_line: ./cpp/build/bench/ann/ANN_BENCH --search --mode=latency --benchmark_filter=cuvs_mg_cagra.dim32_vpq --benchmark_min_time=3s --benchmark_min_warmup_time=0.01 --benchmark_counters_tabular --override_kv=dataset_memory_type:"host" --override_kv=query_memory_type:"pinned" --override_kv=graph_memory_type:"host_pinned" --override_kv=persistent:false --data_prefix=/tmp/data --index_prefix=/ann/index /ann/conf/deep-100M-cagra-persistent.json
dataset: deep-100M
dim: 96
distance: euclidean
gpu_driver_version: 13.0
gpu_gpuDirectRDMASupported: 1758070544
gpu_hostNativeAtomicSupported: 0
gpu_mem_bus_width: 6016
gpu_mem_freq: 2619000000.000000
gpu_mem_global_size: 99951181824
gpu_mem_shared_size: 233472
gpu_name: NVIDIA H800 NVL
gpu_pageableMemoryAccess: 1
gpu_pageableMemoryAccessUsesHostPageTables: 0
gpu_runtime_version: 12.8
gpu_sm_count: 132
gpu_sm_freq: 1785000000.000000
host_cores_used: 16
host_cpu_freq_max: 3000000000
host_cpu_freq_min: 1500000000
host_pagesize: 4096
host_processors_sysconf: 32
host_processors_used: 32
host_total_ram_size: 270107455488
host_total_swap_size: 0
max_k: 100
max_n_queries: 10000
***WARNING*** CPU scaling is enabled, the benchmark real time measurements may be noisy and will incur extra overhead.
terminate called after throwing an instance of 'raft::cuda_error'
  what():  CUDA error encountered at: file=~/workspace/raft/cpp/include/raft/core/device_resources_snmg.hpp line=102: call='cudaSetDevice(main_gpu_id_)', Reason=cudaErrorInvalidDevice:invalid device ordinal
Obtained 18 stack frames
#1 in /tmp/build/bench/ann/libcuvs_mg_ann_bench.so: raft::cuda_error::cuda_error(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) +0x5a [0x7f0a3c323c6a]
#2 in /tmp/build/bench/ann/libcuvs_mg_ann_bench.so: raft::device_resources_snmg::has_resource_factory(raft::resource::resource_type) const +0x1fe [0x7f0a3c32530e]
#3 in /tmp/build/bench/ann/libcuvs_mg_ann_bench.so: raft::resource::get_cuda_stream(raft::resources const&) +0x14 [0x7f0a3c32abd4]
#4 in /tmp/build/bench/ann/libcuvs_mg_ann_bench.so: non-virtual thunk to cuvs::bench::cuvs_mg_cagra<float, unsigned int>::get_sync_stream() const +0xd [0x7f0a3c32af7d]
#5 in ./cpp/build/bench/ann/ANN_BENCH: void cuvs::bench::bench_search<float>(benchmark::State&, cuvs::bench::configuration::index const&, unsigned long, std::shared_ptr<cuvs::bench::dataset<float, int> const>, bool) +0x662 [0x5557863a4342]
#6 in ./cpp/build/bench/ann/ANN_BENCH: benchmark::internal::LambdaBenchmark<benchmark::RegisterBenchmark<void (&)(benchmark::State&, cuvs::bench::configuration::index const&, unsigned long, std::shared_ptr<cuvs::bench::dataset<float, int> const>, bool), cuvs::bench::configuration::index&, unsigned long&, std::shared_ptr<cuvs::bench::dataset<float, int> const>&, bool&>(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, void (&)(benchmark::State&, cuvs::bench::configuration::index const&, unsigned long, std::shared_ptr<cuvs::bench::dataset<float, int> const>, bool), cuvs::bench::configuration::index&, unsigned long&, std::shared_ptr<cuvs::bench::dataset<float, int> const>&, bool&)::{lambda(benchmark::State&)#1}>::Run(benchmark::State&) +0x54 [0x555786364cc4]
#7 in ./cpp/build/bench/ann/ANN_BENCH(+0xc0721) [0x5557863e7721]
#8 in ./cpp/build/bench/ann/ANN_BENCH(+0xa66b0) [0x5557863cd6b0]
#9 in ./cpp/build/bench/ann/ANN_BENCH(+0xa703d) [0x5557863ce03d]
#10 in ./cpp/build/bench/ann/ANN_BENCH(+0xa7626) [0x5557863ce626]
#11 in ./cpp/build/bench/ann/ANN_BENCH(+0xa886e) [0x5557863cf86e]
#12 in ./cpp/build/bench/ann/ANN_BENCH(+0x87d95) [0x5557863aed95]
#13 in ./cpp/build/bench/ann/ANN_BENCH(+0x8992d) [0x5557863b092d]
#14 in ./cpp/build/bench/ann/ANN_BENCH(+0x89a97) [0x5557863b0a97]
#15 in ./cpp/build/bench/ann/ANN_BENCH: cuvs::bench::run_main(int, char**) +0xfa1 [0x5557863ab411]
#16 in /usr/lib/x86_64-linux-gnu/libc.so.6(+0x2a1ca) [0x7f0a42d4d1ca]
#17 in /usr/lib/x86_64-linux-gnu/libc.so.6: __libc_start_main +0x8b [0x7f0a42d4d28b]
#18 in ./cpp/build/bench/ann/ANN_BENCH: _start +0x2e [0x5557863584de]

Aborted (core dumped)

[jinsolp]

@achirkin I only verified the gtests, let me try running the ANN BENCH thanks for pointing out : )

[jinsolp]

Should be fixed now

[achirkin]

/merge