thelastword
Banned
"AMD has introduced two graphics cards for the Apple Mac Pro lineup. Both are using their latest Vega 20 GPU which was currently available on the Radeon VII graphics cards and now aiming the Radeon Pro market. It utilizes the Vega 20 GPU architecture which is a refinement of Vega itself on a 7nm process node. The two products in the 7nm Radeon Pro lineup for Mac Pro include the dual chip powered Radeon Pro Vega II Duo and the single chip powered Radeon Pro Vega II. Some key capabilities and features of AMD Radeon Pro Vega II GPUs include:
When it comes to specifications, we will start with the single chip variant which is based on the full 64 CU enabled Vega 20 GPU and houses 4096 stream processors. The chip has a max rated clock of 1.7 GHz and can deliver a peak throughput of 14.2 TFLOPs (FP32). The chip makes use of an infinity fabric with transfer speeds of 84 GB/s. The chip also comes with HBM2 stacked DRAM with up to 32 GB in capacity, operating at 1 TB/s alongside a 4096-bit wide bus interface."
--------------------------------------
A single Vega II can give you up to 32GB in HBM memory at 14 Tflops and Dual Vega PRO Duo's can max up to 128GB, 56.8TF.....
Now I always told folk that AMD has a plan, not just for CPU's but for GPU's.......You may say, "hey"....this product is for the professional market, but AMD is also testbedding their M-GPU technology for future Navi/Desktop products with it...…Everybody knows what the limitations of M-GPU's or even SLI was in the past, but that is being solved here with Infinity fabric and higher bandwidth between several GPU's on a board......
I remember that I made a post recently on how AMD could bypass the 64CU limit on Vega, same with stream processors, rops etc......There is just no way we can continue to just depend on node shrinks and cranking Mhz on GPU's, we have to go the M-GPU route now and make it viable and work now, or else there's no way we are going to have enough compute power to deliver 4k-8k raytraced 60fps games in the future.....Things would just rush to a standstill if not, there would not be great advancements in graphics and fidelity if not.......So AMD is on the ball here, whilst Nvidia feels AMD is just on this 7nm craze, NO, they're are thinking way past that......And hence the reason they said, their investment in NAVI is one that transcends this current generation......
Even recently, when AMD showed Navi and it's bandwidth capabilities under PCIe 4.0 against a 2080ti, some youtube creators could not see the forest for the trees and claimed AMD was being dicey, Navi does not outperform a 2080ti in games, it's only bandwidth they said.....Well, AMD was clear that the Navi which launches this year is only going for a bit over RTX 2070 performance....And youtube creators went on... no GPU is going to use all that bandwidth, yet just how they are so shortsighted that any game will use more than the 8Gb's of the RTX 2080, now or in the future....Just think about it, PCIE4.0 is going to do wonders for dual GPU's and as we speak PCIE5.0's specs has already been finalized......Multiple-GPU's is the future....Look at the CPU, where did anyone see threadripper coming in and making a dent last gen.....Before the the last 3 years, did we ever forsee that desktop CPU's would be sporting 12-16cores? Had intel been left to its wiles, we would still be rocking 4 core I7's in 2019......Nvidia is only giving us slightly better pascal performance for exorbitant prices, leeching off subpar RTX performance and support in a first to market campaign, "which they deem gamers can't do without", yet no one wants to kill their framerates and rez for a hybrid solution...…..In essence, so much stagnation the last few years till AMD came to the party.......I think, they are going to revolutionize the GPU market too, they're all poised...….Their infinity fabric, high speed PCIE lanes, leveraging their compute capabilities in RDNA, even Samsung has seen the potential....There's a lot to look forward to in the GPU space.....It's been a long time, but thank the heavens that upcoming PC technology is exciting again...
- Leading-edge compute performance – The AMD Radeon Pro Vega II GPU delivers up to 14 TFLOPS of single-precision FP32 performance and up to 28 TFLOPS of half-precision FP16 performance.
- Support for Infinity Fabric Link GPU interconnect technology – With up to 84GB/s per direction low-latency peer-to-peer memory access, the scalable GPU interconnect technology enables GPU-to-GPU communications up to 5X faster than PCIe Gen 3 interconnect speeds.
- Ultra-fast HBM2 memory – 32GB of high-speed HBM2 memory delivers 1TB/s memory bandwidth, providing the memory capacity and data transfer speeds required by today’s high-resolution, multi-display setups, 8K video, and other demanding content creation workloads.
When it comes to specifications, we will start with the single chip variant which is based on the full 64 CU enabled Vega 20 GPU and houses 4096 stream processors. The chip has a max rated clock of 1.7 GHz and can deliver a peak throughput of 14.2 TFLOPs (FP32). The chip makes use of an infinity fabric with transfer speeds of 84 GB/s. The chip also comes with HBM2 stacked DRAM with up to 32 GB in capacity, operating at 1 TB/s alongside a 4096-bit wide bus interface."
--------------------------------------
A single Vega II can give you up to 32GB in HBM memory at 14 Tflops and Dual Vega PRO Duo's can max up to 128GB, 56.8TF.....
Now I always told folk that AMD has a plan, not just for CPU's but for GPU's.......You may say, "hey"....this product is for the professional market, but AMD is also testbedding their M-GPU technology for future Navi/Desktop products with it...…Everybody knows what the limitations of M-GPU's or even SLI was in the past, but that is being solved here with Infinity fabric and higher bandwidth between several GPU's on a board......
I remember that I made a post recently on how AMD could bypass the 64CU limit on Vega, same with stream processors, rops etc......There is just no way we can continue to just depend on node shrinks and cranking Mhz on GPU's, we have to go the M-GPU route now and make it viable and work now, or else there's no way we are going to have enough compute power to deliver 4k-8k raytraced 60fps games in the future.....Things would just rush to a standstill if not, there would not be great advancements in graphics and fidelity if not.......So AMD is on the ball here, whilst Nvidia feels AMD is just on this 7nm craze, NO, they're are thinking way past that......And hence the reason they said, their investment in NAVI is one that transcends this current generation......
Even recently, when AMD showed Navi and it's bandwidth capabilities under PCIe 4.0 against a 2080ti, some youtube creators could not see the forest for the trees and claimed AMD was being dicey, Navi does not outperform a 2080ti in games, it's only bandwidth they said.....Well, AMD was clear that the Navi which launches this year is only going for a bit over RTX 2070 performance....And youtube creators went on... no GPU is going to use all that bandwidth, yet just how they are so shortsighted that any game will use more than the 8Gb's of the RTX 2080, now or in the future....Just think about it, PCIE4.0 is going to do wonders for dual GPU's and as we speak PCIE5.0's specs has already been finalized......Multiple-GPU's is the future....Look at the CPU, where did anyone see threadripper coming in and making a dent last gen.....Before the the last 3 years, did we ever forsee that desktop CPU's would be sporting 12-16cores? Had intel been left to its wiles, we would still be rocking 4 core I7's in 2019......Nvidia is only giving us slightly better pascal performance for exorbitant prices, leeching off subpar RTX performance and support in a first to market campaign, "which they deem gamers can't do without", yet no one wants to kill their framerates and rez for a hybrid solution...…..In essence, so much stagnation the last few years till AMD came to the party.......I think, they are going to revolutionize the GPU market too, they're all poised...….Their infinity fabric, high speed PCIE lanes, leveraging their compute capabilities in RDNA, even Samsung has seen the potential....There's a lot to look forward to in the GPU space.....It's been a long time, but thank the heavens that upcoming PC technology is exciting again...
