Jake Tech Lab

AMD MI300X vs Intel Gaudi 3

AMD MI300X comes out ahead of Intel Gaudi 3 on 5 of the 7 things that decide this, going by what each maker publishes. The one that matters for you may be one the leader loses.

AMD MI300XWinner
AMD MI300X
6.8
Fair
5 of 7 rounds
vs
Intel Gaudi 3
Intel Gaudi 3
6.2
Fair
2 of 7 rounds

The verdict

AMD MI300X takes it, 5–2 on rounds.

On the published figures, AMD MI300X takes 5 of 7 rounds. Treat that as a starting point and check the current figures on both pages, because prices and plans move.

Choose AMD MI300X if…

  • • You want AMD MI300X's edge on how much memory it has and how fast its memory is

Choose Intel Gaudi 3 if…

  • • You want Intel Gaudi 3's edge on how fast it runs fp16 ai math and how new it is

Round by round

  1. Round 1How much memory it hasAMD MI300X wins
    AMD MI300X8/10

    192 GB HBM3

    dedicated memory size

    Source: amd.com ↗

    Intel Gaudi 35/10

    128 GB HBM2E

    hbm capacity

    • 128 GB HBM2E, against 192 GB HBM3

    Source: cdrdv2-public.intel.com ↗

    AMD MI300X wins this round on what the company itself publishes.

  2. Round 2How fast its memory isAMD MI300X wins
    AMD MI300X8/10

    5.3 TB/s

    peak memory bandwidth

    Source: amd.com ↗

    Intel Gaudi 35/10

    3.7 TB/s

    hbm bandwidth

    • 3.7 TB/s, against 5.3 TB/s

    Source: cdrdv2-public.intel.com ↗

    AMD MI300X wins this round on what the company itself publishes.

  3. Round 3How fast it runs FP8 AI mathAMD MI300X wins
    AMD MI300X8/10

    2.61 PFLOPs

    peak fp8 performance

    Source: amd.com ↗

    Intel Gaudi 35/10

    1,678 TFLOPS

    fp8 mme (matrix) tflops

    • 1,678 TFLOPS, against 2.61 PFLOPs

    Source: cdrdv2-public.intel.com ↗

    AMD MI300X wins this round on what the company itself publishes.

  4. Round 4How fast it runs FP16 AI mathIntel Gaudi 3 wins
    AMD MI300X5/10

    1.3 PFLOPs

    peak fp16 performance

    • 1.3 PFLOPs, against 1,678 TFLOPS

    Source: amd.com ↗

    Intel Gaudi 38/10

    459 TFLOPS

    FP16 (signed) matrix (MME) peak, Gaudi 3 technical paper Table 5; 1,678 is the BF16/FP8 figure

    • bf16 mme (matrix) tflops

    Source: cdrdv2-public.intel.com ↗

    Intel Gaudi 3 wins this round on what the company itself publishes.

  5. Round 5How much power it drawsAMD MI300X wins
    AMD MI300X8/10

    750W

    typical board power (peak)

    Source: amd.com ↗

    Intel Gaudi 35/10

    900W

    tdp

    • 900W, against 750W

    Source: cdrdv2-public.intel.com ↗

    AMD MI300X wins this round on what the company itself publishes.

  6. Round 6Which memory generation it usesAMD MI300X wins
    AMD MI300X8/10

    HBM3

    dedicated memory type

    Source: amd.com ↗

    Intel Gaudi 35/10

    HBM2E

    memory type

    • HBM2E, against HBM3

    Source: cdrdv2-public.intel.com ↗

    AMD MI300X wins this round on what the company itself publishes.

  7. Round 7How new it isIntel Gaudi 3 wins
    AMD MI300X5/10

    12/06/2023

    launch date

    • 12/06/2023, against September 24, 2024

    Source: amd.com ↗

    Intel Gaudi 38/10

    September 24, 2024

    launch press release date

    Source: intel.com ↗

    Intel Gaudi 3 wins this round on what the company itself publishes.

AMD MI300XAMD MI300X
Intel Gaudi 3Intel Gaudi 3

Side by side

Overall score
6.8
Best
6.2
Price
See price
See price
How fast its memory is
7.0
6.2
How much memory it has
7.0
6.2
How much power it draws
6.5
5.8
How fast it runs FP16 AI math
6.5
7.4
How new it is
6.6
7.3
Head-to-head findings
How fast its memory is
5.3 TB/s
3.7 TB/s
How much memory it has
192 GB HBM3
128 GB HBM2E
How much power it draws
750W
900W
How fast it runs FP16 AI math
1.3 PFLOPs
459 TFLOPS
How new it is
12/06/2023
September 24, 2024
Which memory generation it uses
HBM3
HBM2E
How fast it runs FP8 AI math
2.61 PFLOPs
1,678 TFLOPS
How it handles scientific FP64 math
163.4 TFLOPs
—
Record
Head-to-head record
4–2
2–3
Head-to-heads won
67 %
40 %
Rounds won
56 %
38 %
Pros
    • bf16 mme (matrix) tflops
    Cons
    • 1.3 PFLOPs, against 1,678 TFLOPS
    • 12/06/2023, against September 24, 2024
    • 192 GB HBM3, against 288 GB HBM3E
    • 5.3 TB/s, against 8 TB/s
    • 128 GB HBM2E, against 192 GB HBM
    • 3.7 TB/s, against 7.37 TB/s
    • 1,678 TFLOPS, against 4,614 TFLOPs
    • 1,678 TFLOPS, against 2,307 TFLOPs
    Add a third option to this comparison →

    More AI accelerators match-ups