Malogeek
Golden Member
Just FYI it's RTG (Radeon Technologies Group)Then TRG (the radeon group)
Just FYI it's RTG (Radeon Technologies Group)Then TRG (the radeon group)
Link to Si interposer latency paper?By arrive, I mean another strategy. I never meant they have finished the journey.
Yes, they appear to be delayed.
Ryzen appears more energy efficient than Intel so GloFlo/Samsung process can't be that bad. Porting to TSMC is a non starter long term. Nvidia will have 1st call on wafers.
Executing large ASICs don't matter with the small-die approach. That is one of it's great value. Early good yield access to new nodes. Do you really think AMD stating quite clearly their early bold [foolhardy?] move to 7nm has nothing to do with yields? They must be fabbing small die for both CPUs and GPUs.
There is no true reticle limit to composite ASIC size. Interposer limits become the barriers to max size. What about a 1000mm^2 + stitched together top model clocked for great energy efficiency.
IF can scale to 512 bits at least, according to AMD. As I said, Naples and Threadripper will tell a lot more of it's capabilities.
Last year in the multi-die thread, I linked a Xilink paper showing signal latency using an SI interposer is the same as on die. I am assuming the multi-die are all on a shared interposer, not through the PCB.
Why do you think multi-die means 2? I'm thinking 1 to 8 ratio, top to bottom. Huge product stack, small development cost.
Small die are lot cheaper/mm^2. We might very well have a much larger AMD die competing with a smaller Nvidia one, while costing equal or even less to fab.
Certainly development cost has large savings, similarly fabbing one product must be amazing for inventory control and product flexibility. [Points to Ryzen]
The amazing thing is we that we have a stated goal as scalability in Navi, before the comments were removed [Raja spilled the beans too early?]. We have a very early use of 7nm. New nodes being traditional only used for small mobile ASICs due to associated costs and dfficulties with larger ones. We have the CPU side following a similar strategy. Is it really such a leap?
Just FYI it's RTG (Radeon Technologies Group)
Its about the Vertex-7 FPGALink to Si interposer latency paper?
you don't have to let them go, just keep them mining and they'll pay for their own upkeep.Looking forward to retiring my dual 290s!! Bought in 2013 and they've been rock solid whether gaming or mining. First time I'll be sad to let cards go lol.
I guess I could, but I only have 3 pci-e slots in my main comp and don't want an extra comp for mining. I was going to put 2 Vegas into my main comp and watercool them (as my 290s are). Having 3 cards in there might be too cramped, but it's an option.you don't have to let them go, just keep them mining and they'll pay for their own upkeep.
As much as I am looking forward to replacing my 480 with a Vega based card, I dread trying to actually find one. Miners have ruined this whole market for actual gamers. eBay is flooded with cards that have been hammered on, and new ones are difficult to find. Even a year after launch 480/580's are hard to find. If anything is screwing AMD out of market share, its this. I know two people that have gone nVidia because it was literally not possible to find a 480 or a 580. Bots sweep in and buy the the moment they are available.
Sure AMD is getting sales, but it hurts them in gaming share badly, and they make no money off the used cards being dumped afterwards.
Its truly infuriating.
I mostly agree, but there are good things about this:
It is good for AMDs revenue numbers, which not only puts money into their back pocket for more R&D but it also looks good to investors.
A lot of people who would never have considered AMD cards have now purchased them and have discovered that not only do they make money, but they can be run 24/7 without dying. When you ask most people about AMD hardware, they don't have a good opinion of it. They think it runs hot, or it dies easily, or Nvidia has better drivers etc.
Now many, many people are being exposed to AMD's hardware and software (for mining purposes) and discovering that they aren't really any worse than Nvidia cards.
Hopefully this means that the next time they choose a GPU for gaming, they won't be so biased towards Nvidia.
Also even if they don't buy AMD cards for mining, everyone is now getting the message that AMD cards are the best for mining. Sure, it doesn't necessarily mean that they are the best for gaming, but the point is, they are desirable. I think previously a lot of gamers thought nobody would ever want AMD hardware unless it was free. Now they see this is incorrect.
Well.. it remains to be seen how scarce HBM2 supplies are. Thankfully due to the Wafer Service Amendment, AMD is able to fab at Samsung on the identical 14nm process.. their capacity for making the actual chips shouldn't hold back Vega production.. not to mention the way they are launching FE first it means production will have a full month additional time to be also creating RX Vega dies.As much as I am looking forward to replacing my 480 with a Vega based card, I dread trying to actually find one. Miners have ruined this whole market for actual gamers. eBay is flooded with cards that have been hammered on, and new ones are difficult to find. Even a year after launch 480/580's are hard to find. If anything is screwing AMD out of market share, its this. I know two people that have gone nVidia because it was literally not possible to find a 480 or a 580. Bots sweep in and buy the the moment they are available.
Sure AMD is getting sales, but it hurts them in gaming share badly, and they make no money off the used cards being dumped afterwards.
Its truly infuriating.
If your post was remotely true you would tie it into the multiple times in the past amd cards were in high demand....
? Buying GTX 1060's and 1070's and mining Zcash instead to trade to ether. There really isn't any positive spin you can make for AMD's situation right now other than they're selling the cards they make. In 6 months when this bubble crashes or ether forks to proof of stake AMD will be unable to sell any of their RX 500 series cards at MSRP because you will be able to find them for 50% of that on Ebay.
start releasing mining only variants of the cards, and finding a way to lock the RX cards to only be for gaming. These mining variants would have no video outputs, be traditional reference style designs for expelling heat out of chasis, and be marked up a little bit to increase margins.
So doesn't the exact same issue affect Nvidia if people are buying 1060 / 1070 for mining??
Why would a miner buy a more expensive less functional GPU with no resell value?
That's the beauty of GPU mining, you can resell the card for about what you paid (heck right now probably even more due to demand) because its still a great gaming card.
No miner would pay more for a Mining GPU with less features.
They would if AMD disabled the ability to hash on RX gaming series cards, because they would have no choice.
So AMD should gimp their compute abilities? Yeah not going to happen...
Sounds about right for the market?https://www.scan.co.uk/shop/compute...d-radeon-vega-frontier-edition-graphics-cards
Air cooled version for 999$? AIO version 1499$?
Many factors could contribute to this. If Charlie's words are true and Vega was delayed, it would mean AMD has more time to tune the process and get higher clocks. So while AMD's original release estimate would put them at around 12.5 TFLOPS, the extra time afforded them to get to 13.1 TFLOPS.It appears that even AMD had no idea, how high they can push the GPUs in terms of clock speeds.
They still refer to it as 12.5 TFLOPs GPU, despite it running 1.6 GHz core clock, and giving 13.1 TFLOPs, actually.
And it also confirms 1800+ MHz (~1880) for HBM2Also confirmation on ~1600mhz (13.1 TFLOPs).
I'm curious about that DX12.1 support. Will it fully supports DX12.1 or will it supports it in "nvidia style"?Features• “Vega” GPU Architecture.
• 64 Next-Gen Compute Units (nCUs) (4096 Stream Processors).
• 16GB High Bandwidth Cache (HBC) Memory.
• 483 GB/s Memory Bandwidth.
• 90 Gpixels/s Fillrate.
• 13.1 TFLOPS Peak FP32 Compute Performance.
• 26.2 TFLOPS Peak FP16 Compute Performance.
• 3x DisplayPort™ 1.4 HBR3/HDR Ready, & 1x HDMI™ 4K60 Display Outputs.
• DirectX® 12.1, OpenGL® 4.5, OpenCL™ 2.0, & Vulkan® 1.0 API Support.
• Dual-slot, full length (10.5”) Form Factor.
• Liquid Cooling Solution. Air-cooled versions also available.
And it also confirms 1800+ MHz (~1880) for HBM2I'm curious about that DX12.1 support. Will it fully supports DX12.1 or will it supports it in "nvidia style"?
Many factors could contribute to this. If Charlie's words are true and Vega was delayed, it would mean AMD has more time to tune the process and get higher clocks. So while AMD's original release estimate would put them at around 12.5 TFLOPS, the extra time afforded them to get to 13.1 TFLOPS.