Intel Gaudi 3 128GB vs Intel Gaudi 3 HL-338
The Gaudi 3 128GB and the Gaudi 3 HL-338 deliver near-identical FP8 throughput (1,678 vs 1,678 TFLOPS dense).
Intel Gaudi 3 128GB
Full specs →Intel Gaudi 3 HL-338
Full specs →Specifications
Performance (TFLOPS)
FLOPS by Precision
What actually differs
Dense throughput for the Gaudi 3 128GB against the Gaudi 3 HL-338: FP32 229 vs 229 TFLOPS, FP16 459 vs 459 TFLOPS, FP8 1,678 vs 1,678 TFLOPS. Sparse figures, where the vendor publishes them, appear under each dense number in the performance table.
Memory is 128 GB against 128 GB, fed at 3.7 TB/s versus 3.7 TB/s. More memory per GPU means larger models fit before you have to shard across cards, which often matters more than raw TFLOPS for inference.
Power budgets are 900 W for the Gaudi 3 128GB and 600 W for the Gaudi 3 HL-338. At FP32 that works out to 0.25 against 0.38 TFLOPS per watt.
For LLM training and inference, weight the FP8 and FP16 rows and memory capacity most heavily. For scientific and HPC workloads, the FP64 row is the one to read.
Frequently asked questions
Is the Gaudi 3 128GB faster than the Gaudi 3 HL-338?
They are close on paper: both deliver about 1,678 TFLOPS of dense FP8 throughput. Memory, bandwidth, and power are the deciding differences.
Which has more memory, the Gaudi 3 128GB or the Gaudi 3 HL-338?
Both GPUs carry 128 GB of VRAM. Memory bandwidth is 3.7 TB/s for the Gaudi 3 128GB and 3.7 TB/s for the Gaudi 3 HL-338.
How much power do the Gaudi 3 128GB and the Gaudi 3 HL-338 draw?
The Gaudi 3 128GB is rated at 900 W TDP and the Gaudi 3 HL-338 at 600 W. On FP32 throughput per watt, the Gaudi 3 HL-338 is the more efficient part.
Get Comparison Updates
New GPUs added weekly. Be the first to see how they compare.