Discussion in 'Videocards - AMD Radeon' started by WhiteLightning, Sep 28, 2018.
*PUSH PUSH PUSH* I hope that helps.
hahaha thanks! I am definitely going team red, just waiting for the right time. Last time I was with team red was in 2004... a while ago. I am excited.
looks like vega 20 can drive up to 6 4k displays,
that would be quite the eyefinity setup
Rumor: AMD Navi 10 to Launch in Mid 2019 & Compete With RTX 2080
AMD is allegedly readying an RX 3000 Series 7nm based graphics card featuring its upcoming Navi 10 GPU that’s going to compete with NVIDIA’s RTX by summer of 2019.
The past week has been a little crazy with two massive bombshell leaks dropping virtually out of nowhere overnight detailing 10 different 7nm Ryzen 3000 series processors, including a 16 core, 5.1 GHz monster chip,
and 3 different Navi based 7nm RX 3000 series graphics cards that are set to be announced at CES a month from now, the fastest of which is said to deliver RTX 2070 class performance at half the price.
Unlike last week’s leak, which came via one of youtubes’ more solid leakers AdoredTV, this latest rumor comes via overclock3d who spotted it on the infamous Chinese Chiphell forums.
So before we proceed I feel compelled to urge you to Sprinkle a bunch of salt on this one, because it’s definitely going to be hard to swallow.With that established, let’s take a look at the details that this rumor alleges.
The source alleges that AMD’s Navi 10 GPU will deliver high-end performance at “surprising” power efficiency. The alleged leaker didn’t provide any more detail regarding power, other than that it will be very frugal.
They also allege that AMD is planning to use Navi 10 to “compete” with NVIDIA’s RTX 2080.
Whether that means generally offering a high-end Radeon alternative to NVDIA’s big guns or actually deliver comparable performance to the 2080 is to be seen.
It’s important to point out that this claim goes somewhat contrary to the previous leak, which we regard as more credible, pegging the RX 3080, powered by Navi 10, as an RTX 2070 competitor rather than a rival to the RTX 2080.
Whilst i love that rumor. I'm not believing it. This sounds more hype than good sense. I mean 2080 performance for $250. I mean really, c'mon.
Maby it's too good to be true.
But, looking into our GPU history it's not the first time, such a Tech Jump occured
Im waiting for actuall GPUs and Tests (in Guru3D and in other sites).
Besides good Hype is always a Hype in the end of't'day
We can Jump or restrain -> Im happy with my Vega for now..... Arcturus is for me ! (i need min. 2x Vega XTX performance so 144Hz solid w/HDR)
I love the idea of performance matching the 2080. But if it does, no way in Hell it's going to be $250. You can almost double that price.
.... Summarized HBCC Tech in our Vega thread Now with 32GB LLT
I would guess the 3080 to be $399 or $449.
If its out q2 of 2019, a 2080 level perf even at 400-500€ its a gamer changer. 2080 its around 700-900€ right now, a price drop should be in order if AMD can hit that mark.
So my Leak was True?
We will have Vega 2 7nm or VII nm
Now we need to do the math, how fast will it be?
@JonasBeckman ? Bring the Crystal ball
Faster HBM2 4x4GB, Better GPU clocks & maby some enchancements.
Vega XTX at ~1757MHz/1200HBM2 gives >28k GPU pts. in 3Dmark
Drv. 188.8.131.52 (Our Bratan' from overclock.net)
So if Vega2 7nm will be 4x4 HBM2 (~1TB/s) and >1800MHz then Yes 29k-30k is possible
Im talking 7nm "shrink" of My Vega, but if they do some magic on silicon level then that score can be even at 32k ballpark.
Spoiler: New Golden Logo
Spoiler: B/W Older Logo
SAPPHIRE To Release Black Shroud Nitro+ Radeon RX 590
I love the minimalistic approach to Vega 2 logo.
AMD/ATI Radeon Software Adrenalin 2019 Edition
As has become tradition, AMD is yet again on the cusp of launching its next big driver package as a holiday gift to all of its Radeon fans.
Adrenalin 2019 Edition is the fifth consecutive installment in AMD’s big yearly GPU driver package releases since it all began back in 2014 with the Omega Driver package.
The only slide with meaningful performance figures compares the new driver to 17.12.1 driver.
Most of the games in this comparison were not even out when 17.12.1 was available, performance increased (up to) 15% for Radeon RX 570.
Game Advisor – “provides game settings guidance for a personalized & improved experience”
This tool works in conjunction with Radeon Overlay. What it does is rather simple: it suggests how to improve performance and frame rating by adjusting game settings.
Settings Advisor – this is designed for new users who never used Radeon Settings before. It recommends which features should be enabled (VSR, FreeSync etc.) based on measured performance.
Upgrade Advisor – system analyzer for “minimum and recommended game compatibility”
Auto-GPU & Memory Overclocking
Temperature-dependent fan curves
Unlocked DPM states for RX Vega
FreeSync 2 HDR – improved auto-tone mapping for a more detailed experience
Virtual Super Resolution – support for 21:9 displays
Display Settings – in-game controls: enhanced sync, FreeSync, per-game settings
Wattman – in-game adjustment for: gpu frequency, gpu voltage, gpu temperature, memory timing, memory frequency, load/save profiles
Refined performance metrics – framerate and frametimes – adjust colors, columns, position, size and transparency
QR Code linking
Voice Control – streaming, recording, screenshots, instant replay, min/avg/max fps, gpu temp, gpu clocks, mem clocks, fan speed
Wattman – same settings as in Radeon Overlay
Enhanced Performance Metrics — detailed analytics (basically turning a phone into a benchmarking tool).
ReLive – View, Edit, Stream recordings/screenshots from a mobile device
AMD ReLive streaming
Wireless streaming up to 4K60FPS to mobile devices
Free on Android and IOS
70ms (AMD) vs 125ms (competitor) responsiveness
Works with AMD Link (through Game Explorer tab)
AMD ReLive for VR: support for HTC Vive Focus, Oculus GO, Samsung Gear VR, Google Daydream
— features: 3rd party Bluetooth compatibility, up to 1440x1440p
Image Quality comparison IQ
JEDEC Updates HBM Standard With Up To 24 GB Memory & 307 GB/s Bandwidth Support Per Stack
JEDEC has published an update for the high-bandwidth memory standard which allows higher capacities and faster pin speeds.
With the update, system manufacturers adopting HBM can leverage from higher memory capacities and various device configurations as HBM memory serves several applications include Graphics, High-Performance Computing, Server, Networking and Client applications.
JEDEC Published New Update For High-Bandwidth Memory Standard – Higher Memory Capacities of Up To 24 GB, Faster Pin Speeds
The existing HBM2 memory allows for up to 8 GB memory in 4-Hi or 8-Hi stacks.
The products that make the most use of the HBM2 standard currently in the graphics industry are the Tesla V100 that is produced by NVIDIA and the upcoming Instinct MI60 from AMD.
Both cards would feature 32 GB of HBM2 memory and feature a Terabyte of bandwidth (900 GB/s for the Tesla V100).
The key here is that these cards make use of 4 HBM2 stacks, each with 8 GB of memory per stack.
These can be either 4-Hi or 8-Hi, depending on the memory configuration (Tesla V100 is also available in 16 GB HBM2 variant).
JEDEC standard JESD235B for HBM leverages Wide I/O and TSV technologies to support densities up to 24 GB per device at speeds up to 307 GB/s.
This bandwidth is delivered across a 1024-bit wide device interface that is divided into 8 independent channels on each DRAM stack.
The standard can support 2-high, 4-high, 8-high, and 12-high TSV stacks of DRAM at full bandwidth to allow systems flexibility on capacity requirements from 1 GB – 24 GB per stack.
This update extends the per pin bandwidth to 2.4 Gbps, adds a new footprint option to accommodate the 16 Gb-layer and 12-high configurations for higher density components, and updates the MISR polynomial options for these new configurations.
Additional clarifications are provided throughout the document to address test features and compatibility across generations of HBM components.
With the updated JEDEC standard, HBM will be updated to support 24 GB of memory per stack. The memory would be made up of 16 Gb dies and can scale all the way from 1 GB to 24 GB per stack depending on the height of the stacks themselves.
The standard can support 2-Hi, 4-Hi, 8-Hi and 12-Hi TSV stacks and each would feature a 1024-bit wide bus interface.
A single stack thus will offer speeds of 307 GB/s. If you look at the topmost configuration that one can get out of the new standard, you get 96 GB of memory and 1.2 TB/s bandwidth along a 4096-bit wide interface.
This is an unprecedented amount of memory and bandwidth you can get from a single chip!
THX to WCCFtech
With how fast GDDR6 memory is, it will be interesting to see if AMD sticks with HBM moving forward.
Have there been any more leaks or rumors about the performance (and release date) of AMD's next gen series? I'm going to likely be in the market for one, but I need to decide what to do for the near future if they aren't releasing until Summer 2019.
Nothing more unfortunately. Hopefully AMD will spill the beans at CES in January.
AMD Will Be Launching The Ryzen 3000 Series CPUs, APUs And A Radeon GPU At CES
I have received confirmation that AMD will be launching 3 classes of product at their CES keynote: the Ryzen 3000 Series CPUs, Ryzen 3000 Series (APUs?) with Vega Graphics and a consumer Radeon graphics card – which is probably the Radeon Vega II.
Lisa will also be talking in depth about AMD’s 7nm progress at the CES keynote.
Update: AMD to launch a Radeon (consumer) GPU, Ryzen 3000 series and discuss 7nm progress at CES 2019 Keynote
Now the Ryzen 3000 series is fairly straightforward. We have been hearing about the upcoming Picasso series for quite a while now and there is even talk on the grapevine that this will be featured in an upcoming XBOX console.
That said, AMD will be launching both the CPU and APU SKUs at CES 2019 at their Keynote. The APUs will have vega graphics and will be competing against Intel/NVIDIA combos in the mobility segment.
Perhaps the more interesting news is the confirmation that AMD will be launching a Radeon, consumer graphics cards. This much I have confirmed. As for what it is, based on my past talks with some industry insiders,
I believe this is going to be the Radeon Vega II – the 7nm consumer variant of the Vega 20 GPU. This is something that I was told was one of the last projects of Mike Rayfield and should be the one AMD launches at CES 2019.
(Update 4:39 AM 12/25/2018 PKT) Availability will be in 1H 2019 after the CES launch.
That’s it for the update if you want to stick to our source and verified data. If you want to read on some more juicy (but unvetted) deets, here they are from past rumors and leaks:
Rumored specifications and details based on past leaks:
The Ryzen 3000U series has already had a couple of leaks and will likely be one of the lineups that AMD reveals at the show in January.
The U stands for Ultra Low Voltage variants and will be used in netbooks and laptops with battery in mind (ie not gaming laptops).
Here’s the thing though, past rumors indicated that the 3000U series will be based on the 12nm process and not the 7nm one while as some others have indicated that the Ryzen 3000 CPU series will be based on the 7nm node so it is unclear which process the 3000 series will hail from.
The 7nm 3000 series is supposed to feature Zen 2 and the new X570 PCH (while staying on the AM4) socket and will be a considerable performance update over the last generation.