AMD Brings Vega To Workstations With Radeon Pro WX9100, Radeon Pro SSG
During AMD’s recent Vega / Threadripper tech day at Siggraph 2017, the company also announced new Radeon Pro WX 9100 and SSG cards. The Radeon Pro branding, announced last year amid a reorganization, is still fairly new, and the family has lacked a true flagship for the workstation market. Radeon Pro WX 9100 is that product.
AMD Radeon Pro WX 9100
AMD claims the Radeon Pro WX 9100 will double the performance of its previous-generation W9100, as well as fold in new functionality associated with the Vega architecture. For example, HBCC allows Vega 10 to fall back to main memory when on-package HBM2 is exhausted, enabling much larger data sets than were previously possible (even if this means incurring a performance hit from essentially paging data in and out of main memory).
AMD demonstrated loading a city model from Baahubali: The Beginning consisting of billions of polygons and rendering it using AMD’s ProRender GPU renderer. Other GPUs would simply get an out-of-memory error when even attempting to load this, according to company representatives, but the WX 9100 managed to do so with a modicum of interactivity.
AMD Radeon Pro WX9100 | |
|---|---|
Memory | 16GB HBM2 w/ ECC |
Memory Interface | 2,048-bit |
Memory Bandwidth | 483.84GBps |
Compute Units | 64 nCU |
Stream Processors | 4,096 |
Engine Clock | 1,500MHz |
API Support | OpenCL 2.0, OpenGL 4.5, DirectX 12, Vulkan 1.0 |
Display Support | Up to 6 DisplayPort @ 4K |
Power | 250W |
AMD Radeon Pro SSG
Building on the SSG (Solid State Graphics) technology announced at Siggraph last year, AMD will soon introduce the Radeon Pro SSG, a card that unites the SSG and Vega architecture. Performance-wise, the Radeon Pro SSG is said to be identical to the WX 9100. The difference is its on-board non-volatile storage.
The Vega architecture's HBCC means the SSG memory can function more seamlessly, allowing the solid state storage to be seen and used as local memory for loading extremely large data sets. The version of the card that AMD announced has 2TB of SSG memory on it, and that extra space is capable of up to 8 GBps reads and 6 GBps writes. While that's much slower than the on-package HBM2 the Vega 10 uses as cache, it's still faster than going across PCI Express to system memory and back.
AMD demonstrated the Radeon Pro SSG being used for real-time playback of uncompressed 8K footage in Adobe Premiere Pro. The 8K playback and scrubbing through the timeline was smooth; AMD also demonstrated doing so without SSG, and it stuttered to the point of being unusable.
Adobe has already taken advantage of the SSG technology for Premiere, and other application developers are expected to follow.
The AMD Radeon Pro WX 9100 is expected to have an MSRP of $2,200, and the Radeon Pro SSG is expected to sell for $7,000. Both cards should be available on September 13.
I will bet they did evaluate it. I'm sure working with studios they came up with this size and RAM was out of the question for the 2TB range. Interesting stuff anyhow.
But new xpoint memory could be even better than ssd in here though. It is expensive as hell, but faster than ssd and Also more durable. I am guite sure that big renders like industrial light and Magic and similar companies put a lot of money if They can speed upp the rendering time in their films!
"Performance-wise, the Radeon Pro SSG is said to be identical to the WX 9100. The difference is its on-board non-volatile storage."
and
"The AMD Radeon Pro WX 9100 is expected to have an MSRP of $2,200, and the Radeon Pro SSG is expected to sell for $7,000.".
The price for the " Samsung 960 PRO Series - 2TB PCIe NVMe - M.2 Internal SSD (MZ-V6P2T0BW)" is listed as U$1,109.99 on one well known Site.
There's a U$4800 price difference for a U$1100 SSD ? - There is something that doesn't add up.
Even if they used 8 512GB SSDs (U$280*8) and a RAID setup there's still some money on the table.
I would pay an extra hundred for 64GB iPhone over a 32GB one, it's not like a cheaper OTG Solution exists or would be better.
With a U$7K Budget I could by 2 Radeon Vega Frontiers for U$2200, and 2 Samsung 960 PRO Series for U$2200 and spend the remainder on the Computer to house them.
Not only is that U$4400 spent less than the price difference but I'd get 52.4 TFlops Half Precision and 26.2 Single and 32 Gbps access speed using DirectGMA.
Easy to add more Graphics Cards or SSDs (assuming I can find an Epyc CPU and a Motherboard with enough PCIe Slots using the remaining U$2600).
Vega Frontier Edition
Samsung 960
I understand paying a bit more to get what you choose or paying a premium for a premium product - I always pay a bit more and never use cheapness as a purchasing decision (not even for Garbage Bags which I intend to throw away).
The point is that DirectGMA already provides the same solution and spending U$4800 (or more) for less (except the all-in-one convenience) seems pointless - especially when it's going into a Chassis and will be "all-in-one" enough.
It's a ridiculous difference in price, as TECH2020 also pointed out:
http://www.tomshardware.com/news/amd-vega-workstations-wx9100-pro-ssg,35128.html#comment-rsp_en_20015861
There's a U$4800 price difference for a U$1100 SSD ? - There is something that doesn't add up.
Even if they used 8 512GB SSDs (U$280*8) and a RAID setup there's still some money on the table."
For the record, its definitely a raid array - a/ cos amd said so on the prev ssg model, & b/ cos of the cited data transfer rates of 8GBps - its pretty likely its a striped pair.
""The AMD Radeon Pro WX 9100 is expected to have an MSRP of $2,200, and the Radeon Pro SSG is expected to sell for $7,000.".
The price for the " Samsung 960 PRO Series - 2TB PCIe NVMe - M.2 Internal SSD (MZ-V6P2T0BW)" is listed as U$1,109.99 on one well known Site.
There's a U$4800 price difference for a U$1100 SSD ? - There is something that doesn't add up.
Even if they used 8 512GB SSDs (U$280*8) and a RAID setup there's still some money on the table."
Whatever, but we can extrapolate a bit from single drive speeds.
it cant be that fast (8GB read & 6GB/s write), & be a single nvme ssd (singele samsung 960 pro 500GB = ~3.2GB/s).
In theory, 2 drives could almost be that fast, but not really given benchmarks. They may gain advantage using NAND in their proprietary environment?
So maybe its 4x 500GB nvmeS?, but I would expect them to be faster.