Bolt Graphics presents the GPU Zeus, aimed at representation, HPC and Games Claims include a faster 10x representation, 6x FP64, 300x simulation gains Zeus admits a memory of up to 2.25 TB, 6x 800gbe, 8x pcie gen5
Bolt Graphics has announced that what he says is the fastest graphic processor in history, which is a great statement.
Sunnyvale, California, the semiconductor startup says that Zeus has a “completely new GPU design for high -performance workloads, including representation, HPC and games” and that “addresses performance, efficiency and limitations of functionalities with legal GPUs.”
Bolt Graphics adds Zeus offers up to 10 times faster, 6 times higher FP64 HPC and up to 300 times faster electromagnetic wave simulation speeds compared to inherited GPUs.
It supports the expanded memory through Bolt’s architecture, allowing up to 384 GB per PCIe card and up to 2.25TB per unit Zeus on a 2U server, with frame settings that reach 180TB. Zeus also integrates 400 GBE Ethernet interfaces and 800 GBE directly in the GPU, reduces energy use despite the highest performance and will be available in PCIE, server and cloud factors.
“Zeus increases performance while simultaneously reducing energy consumption,” Darwesh Singh, founder and CEO of Bolt Graphics, said. “I am proud of the dedicated effort of the Bolt team to create a solution that addresses the client’s key pain, allowing them to be more productive and give life to their ideas.”
Slides published by Servethehome Show Zeus is based on an RVA23 scalar nucleus out of order RVA23 and there are a series of different architectures.
The Zeus 1C26-032 is a single-chipple design and admits Displayport 2.1a and HDMI 2.1B, along with an RJ45 BMC port, typically seen in DPU, servers and other infrastructure components. There is also a QSFP-DD port for 400GBE and two PCIE GEN5 X16 slots. The 1C26-032 includes 32 GB of LPDDR5X and up to 128 GB of DDR5.
The Zeus 2C26-064/128 is a two chipple module, while the 4C26-256 is a four chipplen configuration. In the case of the latter, which is 500W of chip power, each chiplet is combined with 64 GB of LPDDR5X and up to 512 GB of DDR5, which carries the total memory capacity to 256 GB of LPDDR5X and up to 2TB of DDR5. Combined, the GPU admits up to 2.25 TB of total memory.
It also includes six 800gbe (OSFP) ports that Servethehome Notes is “a lot of connectivity that comes out of a GPU. Putting that in perspective, that is, ~ 12x pcie gen5 x16 lanes of bandwidth.”
The slides show the Zeus 2C26-064/128 exceed Nvidia GeForce RTX 4090 and RTX 5090 in route tracking loads in approximately 9x and 4.8x, delivering 154 Gigarays compared to 17 Gigarays (RTX 4090) and 32 Gigarays (RTX 5090). In FP64 calculation yield, Zeus achieves 10 Tflop, significantly ahead of 1.4 Tflops (RTX 4090) and 1.6 Tflops (RTX 5090) – Approximately 7.1xy 6.25x higher, respectively.
Comparing the Zeus 4C26-256 bolt with the NVIDIA consumption GPUs, the route tracking performance is approximately 18x higher than the RTX 4090 and almost 10x higher than the RTX 5090, while the FP64 Computa performance is more than 12 times higher.
All this must be taken with a pinch of salt, of course. As Servethehome Note: “This feels like an advertisement in which we really want to see the product. The fact that we are still a few months after early developer kits makes this feel a little early. On the other hand, if Bolt Graphics has Zeus competing with a combination of NIC and GPU of consumption, while offering more memory, then it could be a really orderly combination. I am always a bit skeptical with the products until we see them live. “
You may also like
