eBay — 103.66 €

YouTube video by 情報の灯台

Hi, I am testing a miniapp that tests multi-gpu Fortran that is used in our production code. The code is located at:  multigpu-test-code The specific version being tested here is:   psi_multigpu_test_code_stdpar_ompdata.f90 I am testing on the Stampede3 system at TACC on the MAX 1550 GPU. When I try to compile the code with: **mpif90 -f90=ifx -fiopenmp -fopenmp-target-do-concurrent -fopenmp-targets=spir64 -fopenmp-do-concurrent-maptype-modifier=present psi_multigpu_test_code_stdpar_ompdata.f90 -o psi_multigpu_test_code_stdpar_ompdata** I get the error: **psi_multigpu_test_code_stdpar_ompdata.f90(669): error #9036: A list item that appears in a use_device_ptr or use_device_addr clause must not be a component of a derived type. [R]** **!$omp target data use_device_ptr(v%r,v%t,v%p)** **-----------------------------------^** Is this error due to the OpenMP Target specification, or is it a temporary limitation of the compiler? The code works correctly on NVIDIA GPUs with the nvfortran compiler, implying the latter (although nvfortran is known to support the unsupported sometimes...) In either case, is there a current workaround to allow passing derived type arrays into GPU-aware MPI/library calls? Thanks!  - Ron

はじめに こんにちは、きまま / Easygoing です。 今回は ComfyUI 上級編で、カスタムノードを使った高速化について見ていきます。 高速化するカスタムノード 今回ご紹介するカスタムノードのは次の２つです ComfyUI-MultiGPU（VRAM 管理を最適化） ComfyUI-TeaCache（キャッシュを利用して高速化） １．ComfyUI-Mul...

I got OOM error when finetuning phi4-mini-reasoning with a batch size of 1 and length of 1024 tokens with 3 v100 16gb, tried both DDP and FSDP, both got OOM, but when I use single gpu, it pass with a batch size of 2 and peak memory at <15GB, another know what’s work? the following is my training script #!/usr/bin/env python import os, torch, argparse from datasets import load_from_disk from transformers import ( AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig, TrainingArguments, Trainer ) from peft import ( LoraConfig, prepare_model_for_kbit_training, get_peft_model ) from accelerate import DistributedDataParallelKwargs from accelerate.utils import DistributedType from functools import partial import torch.distributed as dist # Added for explicit init MODEL_ID = "microsoft/Phi-4-mini-reasoning" DATA_PATH = "./flattened_distilled_dataset-win1k-phi" OUT_DIR = "./lora_phi4" def main(): # ───────────────────────────── Accelerator info ── local_rank = int(os.environ.get("LOCAL_RANK", 0)) world_size = int(os.environ.get("WORLD_SIZE", 1)) device = torch.device(f"cuda:{local_rank}") torch.cuda.set_device(local_rank) # Explicitly initialize distributed process group (safer) dist.init_process_group( backend="nccl", init_method="env://", world_size=world_size, rank=local_rank ) # ───────────────────────────── Tokenizer ───────── tok = AutoTokenizer.from_pretrained(MODEL_ID, use_fast=True) # ───────────────────────────── 4-bit model ─────── bnb_cfg = BitsAndBytesConfig( load_in_4bit = True, bnb_4bit_quant_type = "nf4", bnb_4bit_compute_dtype = torch.float16, bnb_4bit_use_double_quant= True, ) # each process loads on *its* GPU base_model = AutoModelForCausalLM.from_pretrained( MODEL_ID, quantization_config = bnb_cfg, device_map = {"": device}, # crucial trust_remote_code = True, use_cache = False, ) base_model = prepare_model_for_kbit_training(base_model) # ───────────────────────────── LoRA ────────────── lora_cfg = LoraConfig( r=32, # Reduced from 64 to save memory lora_alpha=128, lora_dropout=0.08, bias="none", task_type="CAUSAL_LM", target_modules=["q_proj", "k_proj", "v_proj", "o_proj", "up_proj", "down_proj", "gate_proj", "fc1", "fc2", "dense"], ) model = get_peft_model(base_model, lora_cfg) try: import xformers, xformers.ops model._set_memory_efficient_attention_xformers(True) except Exception: pass model.print_trainable_parameters() # Print GPU memory usage after model load (for debugging) if dist.get_rank() == local_rank: print(f"Rank {local_rank}: GPU memory allocated after model load: {torch.cuda.memory_allocated(device) / 1e9:.2f} GB") print(f"Rank {local_rank}: GPU memory reserved after model load: {torch.cuda.memory_reserved(device) / 1e9:.2f} GB") # ───────────────────────────── Dataset ─────────── # Load full dataset, then shard per process to save RAM full_ds = load_from_disk(DATA_PATH) train_ds = full_ds.shard(num_shards=world_size, index=local_rank) if local_rank == 0: print(f"Sharded dataset: each process has {len(train_ds)} examples (total: {len(full_ds)})") def collate_fn(batch): ids = [torch.tensor(x["input_ids"]) for x in batch] mask = [torch.tensor(x["attention_mask"]) for x in batch] return { "input_ids" : torch.stack(ids), "attention_mask": torch.stack(mask), "labels" : torch.stack(ids), } # ───────────────────────────── TrainingArguments ─ args = TrainingArguments( output_dir = OUT_DIR, num_train_epochs = 2, per_device_train_batch_size= 1, # per-GPU! gradient_accumulation_steps= 2, # effective * world_size gradient_checkpointing = True, learning_rate = 3e-4, warmup_ratio = 0.03, logging_steps = 10, save_steps = 500, save_strategy = "steps", fp16 = True, optim = "paged_adamw_8bit", # Changed to paged BNB optimizer for memory efficiency report_to = "none", remove_unused_columns = False, ddp_find_unused_parameters = False, # speeds up DDP ddp_bucket_cap_mb = 25, # Balanced value ) trainer = Trainer( model = model, args = args, train_dataset = train_ds, tokenizer = tok, data_collator = collate_fn, ) # Print GPU memory after Trainer init (before train) if dist.get_rank() == local_rank: print(f"Rank {local_rank}: GPU memory allocated after Trainer init: {torch.cuda.memory_allocated(device) / 1e9:.2f} GB") print(f"Rank {local_rank}: GPU memory reserved after Trainer init: {torch.cuda.memory_reserved(device) / 1e9:.2f} GB") trainer.train() # Only rank-0 writes to disk if local_rank == 0: trainer.save_model(OUT_DIR) tok.save_pretrained(OUT_DIR) if __name__ == "__main__": main()

With IterableDataset like ConstnatLengthDataset model will train longer than with predefined tokenized dataset. Moreover, with IterableDataset multigpu training is more complex. It needs something like DistributedSample, DistributedLoader and so on.