AMD Instinct MI300X 192GB vs AMD Instinct MI300A 128GB

The Instinct MI300X delivers 33% more FP8 throughput than the Instinct MI300A (2,615 vs 1,961 TFLOPS dense). The Instinct MI300X also carries 64 GB more memory (192 GB vs 128 GB).

Instinct MI300X: CDNA 3, 2023 Instinct MI300A: CDNA 3, 2023
FP8 dense lead
+33%
Instinct MI300X: 2,615 vs 1,961 TFLOPS
VRAM
192 vs 128 GB
64 GB more for the Instinct MI300X
Bandwidth
5.3 TB/s vs 5.3 TB/s
Even
TDP
750 W vs 760 W
Instinct MI300X draws 10 W less

Specifications

Instinct MI300X
Instinct MI300A
Architecture
CDNA 3CDNA 3
Launch Year
20232023
Form Factor
OAMAPU SH5 socket
VRAM
192 GB128 GB
Memory Bandwidth
5.3 TB/s5.3 TB/s
TDP
750 W760 W
Process Node
5nm FinFET5nm FinFET

Performance (TFLOPS)

Instinct MI300X
Instinct MI300A
FP64
81.7 TFlops 61.3 TFlops
FP32
163.4 TFlops 122.6 TFlops
FP16
1307.4 TFlops
2614.9 TFLOPS sparse
980.6 TFlops
1961.2 TFLOPS sparse
BF16
1307.4 TFlops
2614.9 TFLOPS sparse
980.6 TFlops
1961.2 TFLOPS sparse
FP8
2614.9 TFlops
5229.8 TFLOPS sparse
1961.2 TFlops
3922.3 TFLOPS sparse
INT8
2614.9 TFlops
5229.8 TOPS sparse
1961.2 TFlops
3922.3 TOPS sparse

FLOPS by Precision

What actually differs

Both GPUs launched in 2023: the Instinct MI300X on CDNA 3 and the Instinct MI300A on CDNA 3.

Dense throughput for the Instinct MI300X against the Instinct MI300A: FP64 82 vs 61 TFLOPS, FP32 163 vs 123 TFLOPS, FP16 1,307 vs 981 TFLOPS, FP8 2,615 vs 1,961 TFLOPS. Sparse figures, where the vendor publishes them, appear under each dense number in the performance table.

Memory is 192 GB against 128 GB, fed at 5.3 TB/s versus 5.3 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 750 W for the Instinct MI300X and 760 W for the Instinct MI300A. At FP32 that works out to 0.22 against 0.16 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 Instinct MI300X faster than the Instinct MI300A?

At FP8 precision the Instinct MI300X reaches 2,615 TFLOPS dense against 1,961 TFLOPS for the Instinct MI300A. The performance table on this page lists every published precision for both GPUs.

Which has more memory, the Instinct MI300X or the Instinct MI300A?

The Instinct MI300X carries 192 GB of VRAM versus 128 GB for the Instinct MI300A. Memory bandwidth is 5.3 TB/s for the Instinct MI300X and 5.3 TB/s for the Instinct MI300A.

How much power do the Instinct MI300X and the Instinct MI300A draw?

The Instinct MI300X is rated at 750 W TDP and the Instinct MI300A at 760 W. On FP32 throughput per watt, the Instinct MI300X is the more efficient part.

Can I rent the Instinct MI300X or the Instinct MI300A in the cloud?

Yes. Live cloud listings tracked by Flopper start at $0.50 per GPU hour. The rental pricing section on this page lists current providers and rates for both GPUs.

Where to Rent

AMD Instinct MI300X 192GB

ProviderConfigurationPrice/GPU-hrChecked
RunPod
1× Instinct MI300X 192GB Community $0.50 just now View →
RunPod
1× Instinct MI300X 192GB $2.39 just now View →
DigitalOcean
1× Instinct MI300X 192GB $2.59 just now View →
DigitalOcean
8× Instinct MI300X 192GB $2.59 just now View →
Crusoe
1× Instinct MI300X 192GB $3.45 just now View →
Oracle Cloud (OCI)
1× Instinct MI300X 192GB $6.00 just now View →
Microsoft Azure
8× Instinct MI300X 192GB $6.00 just now View →
Microsoft Azure
8× Instinct MI300X 192GB $6.00 just now View →
All Instinct MI300X listings and price history →

Get Comparison Updates

New GPUs added weekly. Be the first to see how they compare.