News
Intel 13th Gen CPUs and Arc GPU! Sapphire Rapids and Ponte Vecchio News
September 28, 2022
6 min read
Intel Innovation 2022 was another event that took place this month with announcements of new components of their own. Let's get with the highlights:
Intel Raptor Lake Overview
Intel Raptor Lake 13th Generation Core CPUs are finally here set to release on October 20, 2022. Packed with a whole bunch of new specs and tech, this next generation of Intel delivers competent performance at a reasonable value.
| Cores/Threads | Base Clock | Boost Clock | Cache L2/L3 | Base/Max Power | Price K/KF Spec |
|---|---|---|---|---|---|---|
Core i9 13900K/KF | 24/32 (8P+16E) | 3.0GHz | 5.8GHz | 68MB | 125W/253W | $589/$564 |
Core i7 13700K/KF | 16/24 (8P+8E) | 3.4GHz | 5.4GHz | 54MB | 125W/253W | $409/$384 |
Core i5 13600K/KF | 14/20 (6P+8E) | 3.5GHz | 5.1GHz | 44MB | 125W/181W | $319/$294 |
Differences between the K and the KF model is integrated graphics and price. K will include Intel UHD Graphics 770 for those wanting flexibility with integrated or discrete graphics. The KF variant holds all the capabilities as the K, does not have integrated graphics, but is slightly cheaper in price.
This generation of Intel CPUs includes a performance and efficiency core increase speeds of up to 24% with 15% single-threaded performance gains and a claimed huge 41% multi-threaded performance gain! Intel decided to double the number of Efficiency Cores of every model over the last generation since those cores are solely responsible for the large performance increase of multi-threaded workloads.
Higher IPC will enable them to compete with competitors in the market. The 13th Generation Intel CPUs come equipped with PCIe Gen 4.0 and support both DDR4 and DDR5 so when spec-ing out your new system, be sure to pick compatible parts (or let our engineers at Exxact spec out a solution for you).
The new 13th Gen Intel hits a peak of 5.8GHz but Intel doesn't stop there. They announced later this year they will release a CPU with a 6.0GHz boost clock speed right out of the box! Of course, they’re talking about the future i9 13900KS. We shall wait and see…
Intel ARC
The long-awaited Intel GPU is finally here after a few delays. Intel plans to bring mainstream GPUs to the market at easy and affordable pricing and availability with their top-of-the-line Intel ARC A770. They will be available in 8GB and 16GB variantsIntel Arc 770 GPUs.
The Intel Arc 770 will be available Oct 12 starting at $329 with its less powerful brothers soon to follow.
Intel Data Center GPU Flex Series
Intel Data Center GPU Flex Series offers flexible and robust GPU solutions for the intelligent visual cloud. The Flex Series GPUs support a diverse range of workloads in the industry starting with media streaming and cloud gaming, support for AI visual inference, and virtual desktop infrastructure workloads.
With an open, standards-based software stack optimized for density and quality with critical server capabilities for high reliability, availability, and scalability. The Flex Series Data Center GPUs include an AV1 Encoder for lower operational expense and will deliver extremely high fidelity video quality.
This GPU is only supported on Intel Xeon Scalable 3rd Gen Processors at this time. They come in both full-width single slots with 16GBs and half-width single slot variants with 12GBs.
Sapphire Rapids: Intel Xeon Scalable 4th Generation
There was some talk about Intel Xeon Scalable 4th Generation but very limited. They had a demo showcasing its speeds and capabilities in matrix multiplication but had no major details on the release or availability.
We are excited to see this CPU but it has yet to see the sockets of our servers and solutions. The chiplet design with a modular chiplet EMIB interconnecting the 4 individual dies is very interesting. We will continue to keep a lookout for an official release date for Sapphire Rapids.
Ponte Vecchio
Ponte Vecchio is Intel’s other Deep Learning AI accelerator GPU for data centers. Intel sees Habana GPUs as a complementary technology to Ponte Vecchio GPUs. There was a small showcase of the GPU alongside the aforementioned Sapphire Rapids HBM CPU.
Ponte Vecchio is a tile-based SOC with 47 dies manufactured on TSMC 5nm, TSMC 7nm, and Intel 7nm nodes. It houses 408MB of L2 and L3 cache, 128GB of HBM2e memory, and 3.2TB/s bandwidth is a monstrous accelerator. Ponte Vecchio has been delayed a couple of times and we hope intel can deliver because this looks to be a very interesting GPU accelerator to compete in the extreme HPC OAM/SXM accelerator socketed GPUs.
Ponte Vecchio GPUs are shipping to Argonne National Laboratory alongside the Xeon Scalable 4th Gen to power the Aurora Super Computer.
Gaudi2 Deep Learning GPU Processor
Habana Labs, its data center team focused on AI deep learning processor technologies, launched its second-generation deep learning processors for training and inference: Habana Gaudi2 and Habana Greco.
At Intel Innovation, they showed off the Gaudi2, a beautiful OAM socketed accelerator. Though not featured during the keynote, Habana Greco is a single-slot, half-width PCIe GPU. Habana Labs is a subsection of Intel that specializes in developing AI accelerators.
The Gaudi2 comes in 7nm equipped with new FP8 data types, higher tensor cores counts, on-chip media processing, 96GB of HBM2E memory, and 2.45 TB/s bandwidth enable peak performance for AI workloads to accelerate things like Computer Vision, Natural Language Processing AI.
Thoughts
Lots of exciting new announcements for new products coming out to both the consumer and enterprise markets. When it comes to performance; we won't know for sure until they are tested and benchmarked by 3rd party sources. We are extremely excited to see how these new CPUs perform as well as excited to see Intel finally deliver Sapphire Rapids and Ponte Vecchio.
Don't miss out on these new releases!
Contact Exxact Today!
News
Intel 13th Gen CPUs and Arc GPU! Sapphire Rapids and Ponte Vecchio News
September 28, 20226 min read
Intel Innovation 2022 was another event that took place this month with announcements of new components of their own. Let's get with the highlights:
Intel Raptor Lake Overview
Intel Raptor Lake 13th Generation Core CPUs are finally here set to release on October 20, 2022. Packed with a whole bunch of new specs and tech, this next generation of Intel delivers competent performance at a reasonable value.
| Cores/Threads | Base Clock | Boost Clock | Cache L2/L3 | Base/Max Power | Price K/KF Spec |
|---|---|---|---|---|---|---|
Core i9 13900K/KF | 24/32 (8P+16E) | 3.0GHz | 5.8GHz | 68MB | 125W/253W | $589/$564 |
Core i7 13700K/KF | 16/24 (8P+8E) | 3.4GHz | 5.4GHz | 54MB | 125W/253W | $409/$384 |
Core i5 13600K/KF | 14/20 (6P+8E) | 3.5GHz | 5.1GHz | 44MB | 125W/181W | $319/$294 |
Differences between the K and the KF model is integrated graphics and price. K will include Intel UHD Graphics 770 for those wanting flexibility with integrated or discrete graphics. The KF variant holds all the capabilities as the K, does not have integrated graphics, but is slightly cheaper in price.
This generation of Intel CPUs includes a performance and efficiency core increase speeds of up to 24% with 15% single-threaded performance gains and a claimed huge 41% multi-threaded performance gain! Intel decided to double the number of Efficiency Cores of every model over the last generation since those cores are solely responsible for the large performance increase of multi-threaded workloads.
Higher IPC will enable them to compete with competitors in the market. The 13th Generation Intel CPUs come equipped with PCIe Gen 4.0 and support both DDR4 and DDR5 so when spec-ing out your new system, be sure to pick compatible parts (or let our engineers at Exxact spec out a solution for you).
The new 13th Gen Intel hits a peak of 5.8GHz but Intel doesn't stop there. They announced later this year they will release a CPU with a 6.0GHz boost clock speed right out of the box! Of course, they’re talking about the future i9 13900KS. We shall wait and see…
Intel ARC
The long-awaited Intel GPU is finally here after a few delays. Intel plans to bring mainstream GPUs to the market at easy and affordable pricing and availability with their top-of-the-line Intel ARC A770. They will be available in 8GB and 16GB variantsIntel Arc 770 GPUs.
The Intel Arc 770 will be available Oct 12 starting at $329 with its less powerful brothers soon to follow.
Intel Data Center GPU Flex Series
Intel Data Center GPU Flex Series offers flexible and robust GPU solutions for the intelligent visual cloud. The Flex Series GPUs support a diverse range of workloads in the industry starting with media streaming and cloud gaming, support for AI visual inference, and virtual desktop infrastructure workloads.
With an open, standards-based software stack optimized for density and quality with critical server capabilities for high reliability, availability, and scalability. The Flex Series Data Center GPUs include an AV1 Encoder for lower operational expense and will deliver extremely high fidelity video quality.
This GPU is only supported on Intel Xeon Scalable 3rd Gen Processors at this time. They come in both full-width single slots with 16GBs and half-width single slot variants with 12GBs.
Sapphire Rapids: Intel Xeon Scalable 4th Generation
There was some talk about Intel Xeon Scalable 4th Generation but very limited. They had a demo showcasing its speeds and capabilities in matrix multiplication but had no major details on the release or availability.
We are excited to see this CPU but it has yet to see the sockets of our servers and solutions. The chiplet design with a modular chiplet EMIB interconnecting the 4 individual dies is very interesting. We will continue to keep a lookout for an official release date for Sapphire Rapids.
Ponte Vecchio
Ponte Vecchio is Intel’s other Deep Learning AI accelerator GPU for data centers. Intel sees Habana GPUs as a complementary technology to Ponte Vecchio GPUs. There was a small showcase of the GPU alongside the aforementioned Sapphire Rapids HBM CPU.
Ponte Vecchio is a tile-based SOC with 47 dies manufactured on TSMC 5nm, TSMC 7nm, and Intel 7nm nodes. It houses 408MB of L2 and L3 cache, 128GB of HBM2e memory, and 3.2TB/s bandwidth is a monstrous accelerator. Ponte Vecchio has been delayed a couple of times and we hope intel can deliver because this looks to be a very interesting GPU accelerator to compete in the extreme HPC OAM/SXM accelerator socketed GPUs.
Ponte Vecchio GPUs are shipping to Argonne National Laboratory alongside the Xeon Scalable 4th Gen to power the Aurora Super Computer.
Gaudi2 Deep Learning GPU Processor
Habana Labs, its data center team focused on AI deep learning processor technologies, launched its second-generation deep learning processors for training and inference: Habana Gaudi2 and Habana Greco.
At Intel Innovation, they showed off the Gaudi2, a beautiful OAM socketed accelerator. Though not featured during the keynote, Habana Greco is a single-slot, half-width PCIe GPU. Habana Labs is a subsection of Intel that specializes in developing AI accelerators.
The Gaudi2 comes in 7nm equipped with new FP8 data types, higher tensor cores counts, on-chip media processing, 96GB of HBM2E memory, and 2.45 TB/s bandwidth enable peak performance for AI workloads to accelerate things like Computer Vision, Natural Language Processing AI.
Thoughts
Lots of exciting new announcements for new products coming out to both the consumer and enterprise markets. When it comes to performance; we won't know for sure until they are tested and benchmarked by 3rd party sources. We are extremely excited to see how these new CPUs perform as well as excited to see Intel finally deliver Sapphire Rapids and Ponte Vecchio.
Don't miss out on these new releases!
Contact Exxact Today!
