Intel Axes Data Center GPU Max 1350 Preps Max 1450 for “Different Markets”
Intel has successfully rebuilt the corporate’s Information Heart GPU Max sequence of computing GPUs. Chip maker accepted Tom’s {Hardware} Information Heart Max has determined to take away the GPU 1350 from the product stack. Nevertheless, it’s going to launch the newly manufactured Information Heart Max GPU 1450 later this yr with diminished I/O bandwidth to cater to ‘totally different markets’. The transfer comes simply after Intel’s determination to cancel the Rialto Bridge GPUs following the restructuring of the AXG graphics division.
The primary Ponte Vecchio GPU lineup consisted of Information Heart Max GPU 1550, Information Heart Max GPU 1350 and Information Heart Max GPU 1100. Intel launched the 1550 mannequin within the first quarter of this yr. The upcoming Q2 launch for the 1100 mannequin continues to be unaffected by the Information Heart Max GPU 1450, which was lately added to the product stack. Nevertheless, Intel has not but offered a particular launch date for the Information Heart Max GPU 1450, however we all know it’s going to come this yr.
“We launched the Intel Information Heart Max GPU 1550 (600W), which was initially focused for liquid-cooled options solely. We have now since expanded our help to incorporate the Intel Information Heart Max GPU 1550 (600W) to incorporate air-cooled options.
“Because of this, we’re modernizing our product choices by eradicating the Intel Information Heart Max GPU 1350 (450W) designed for air-cooled options. Will probably be in a position to make use of IO bandwidth and air and liquid cooling options for various markets,” an Intel spokesperson stated. Tom’s {Hardware}.
Swipe to scroll horizontally
| Header Cell – Column 0 | Information Heart GPU Max 1100 | Information Heart GPU Max 1350 | Information Heart GPU Max 1450 | Information Heart GPU Max 1550 |
|---|---|---|---|---|
| kind issue | PCIe | OAM | ? | OAM |
| Tiles + Reminiscence | ? | ? | ? | 39 + 8 |
| transistors | ? | ? | ? | 100 billion |
| Xe HPC Cores | Calculation Items | 56 | 112 | ? | 128 |
| RT Cores | 56 | 112 | ? | 128 |
| 512-bit Vector Engines | 448 | 896 | ? | 1,024 |
| 4096-bit Matrix Engines | 448 | 896 | ? | 1,024 |
| Base Clock (MHz) | 1,000 | 750 | ? | 900 |
| Most Dynamic Clock (MHz) | 1,550 | 1,550 | ? | 1,600 |
| L1 Cache | 12MB | 48MB | ? | 64MB at 105 TB/s |
| L2 Rambo Cache | 108MB | 216MB | ? | 408MB at 13TB/s |
| Reminiscence | 48GB | 96GB | ? | 128GB at 3.2TB/s |
| Reminiscence Interface | 1024bit | ? | ? | 1024bit |
| Reminiscence Bandwidth (GB/s) | 1,228.8 | ? | ? | 3276.8 |
| Multi GPU IO | 8 | 16 | ? | 16 |
| TDP | 300W | 450W | ? | 600W |
Intel has not but shared the technical specs of the Information Heart GPU Max 1450. Nevertheless, logic implies that this will probably be a scaled down model of the Information Heart GPU Max 1550, however we’ve not confirmed it but. The Information Heart GPU Max 1450 will help air and liquid cooling options and we suspect it might carry the identical 450W TDP ranking because the canceled 1350 mannequin. We have reached out to Intel for additional affirmation.
A key element from Intel’s announcement is that the chipmaker has tailor-made the Information Heart GPU Max 1450 for “totally different markets” that we suspect are Chinese language. In keeping with Intel, the Information Heart GPU Max 1450 will include diminished I/O bandwidth ranges; This can be a transfer almost definitely meant to adjust to US rules on GPU exports to China.
Keep in mind, US export guidelines now mandate that chip-to-chip I/O bandwidth for GPUs related to China should be lower than 600 GB/s. For instance, Nvidia has changed the H100 (Hopper) GPU and renamed it H800 to adapt it to the now restricted Chinese language market. It’s affordable to count on the Information Heart GPU Max 1450 to take the same method.
Along with changing the Information Heart Max GPU 1350, Intel has additionally expanded the cooling choices for its flagship Information Heart Max GPU 1550. The 600W GPU was initially solely out there with liquid-cooled options. Nevertheless, Intel will now supply air cooling on the Information Heart Max GPU 1550.
Intel’s preliminary plan was to launch the Rialto Bridge this yr because the successor to Ponte Vecchio and launch Falcon Shores in 2024 to interchange the Rialto Bridge. Nevertheless, the chipmaker in the end determined to ditch the Rialto Bridge and push Falcon Shores to 2025. Because of this, Intel must make do with the Ponte Vecchio for not less than two extra years, competing in opposition to Nvidia’s H100 and AMD’s Intuition MI300.
#Intel #Axes #Information #Heart #GPU #Max #Preps #Max #Markets
