Discussion in 'Videocards - AMD Radeon' started by WhiteLightning, Sep 28, 2018.
"With respect to accelerator features, 7nm Vega and the resulting MI60 & MI50 cards differentiates itself from the previous Vega 10-powered MI25 in a few key areas.
7nm Vega brings support for half-rate double precision – up from 1/16th rate – and AMD is supporting new low precision data types as well.
These INT8 and INT4 instructions are especially useful for machine learning inferencing, where high precision isn’t necessary, with AMD able to get up to 4x the perf of an FP16/INT16 data type when using the smallest INT4 data type.
However it’s not clear from AMD’s presentation how flexible these new data types are – and with what instructions they can be used – which will be important for understanding the full capabilities of the new GPU.
All told, AMD is claiming a peak throughput of 7.4 TFLOPS FP64, 14.7 TFLOPS FP32, and 118 TOPS for INT4."
In another words Vega 7nm can have +16->25% more Performance and
Vega 14nm at 1750MHz have 264tW
Vega 7nm at 1750MHz have 175tW
or at 1700MHz 145tW -> Well done ATI, not great but still GJ
Also this one have 4xHBM2 stacks (2nd Gen) with Top >1.2TB/s (at 1100MHz)
My Vega at 1200 HBM2 can have ~650GB/s
I think than Dr.Lisa can give Vega2 7nm Green light
Vega 2 7nm 4x4GB HBM2 -> and we can have up to 1800-1900MHz Monster.
AMD Polaris 30 GPU Based Radeon RX 590 To Clock As High As 1680 MHz, 3DMark Firestrike Extreme Score Leaks Out
Recapping the specifications for those who missed out our previous article, the AMD Polaris 30 GPU will be based on the 12nm process node.
It’s not known whether AMD will be relying on the Global Foundries 12nm process or the much advanced TSMC 12nm node which is being used to make the NVIDIA Turing GPUs. Considering that Polaris GPUs have been featuring a FinFET process design from Glofo, it won’t be a surprise if AMD ends up using them for 12nm too.
In terms of specs, the AMD Polaris 30 GPU would feature 2304 Stream processors, 144 TMUs and 32 ROPs.
The VRAM includes the same 8 GB GDDR5 design, clocked at 8000 MHz across a 256-bit bus. Most of the performance benefits would come from the architecture tuning that is achieved with the 12nm die shrink but doesn’t expect it to be a lot.
The card is expected to hit the pricing around $199 to $250 US ??.
Now coming to the models, both PowerColor and XFX would be launching their custom models. PowerColor’s Radeon RX 590 Red Devil was already leaked a while ago by Videocardz.
The model would be clocked at 1576 MHz out of the box while the XFX Radeon RX 590 custom model would be clocked slightly higher at 1580 MHz.
At the same time, two differently clocked cards from the same brands were reported which ran at 1645 MHz and 1680 MHz, respectively.
PowerColor Radeon RX 590: 1,576 MHz (Default), 1,645 MHz (OC)
XFX Radeon RX 590: 1,580 MHz (Default), 1,680 MHz (OC)
PowerColor 1,576 MHz, 1,645 MHz
XFX 1,580 MHz, 1,680 MHz
Fire Strike Extreme 7350 pic.twitter.com/umq3KPxEK9
— APISAK (@TUM_APISAK) November 8, 2018
It says finfet 14nm?
Well, my FuryX is about to die, so this could be a nice replacement until Cyberpunk comes out. I dont want to spend more than 200-250€.
Exclusive: First AMD Navi GPU Will Have 40 CUs And Is Codenamed Navi 12
So I have an update on the AMD Navi GPU front. I have been informed that AMD has finalized the first Navi design and codenamed it Navi 12. We also know that the GPU will have 40 CUs.
If the ratio of stream processors in the CU is the same as that of GCN, then you are looking at 64 CUs each for a grand total of 2560 stream processors.
This is going to be posited as the upgrade path from Polaris 30 and will probably feature in the Radeon 600 series.
AMD’s Navi 12 GPU will have 40 CUs
Now off the bat, I will tell you that the Navi 12 is an odd codename. We expected AMD to launch with the “Navi 10” designation but that does not appear to be the case.
AMD’s nomenclature is based on the timeline at which the dies were designed and what this essentially means is that AMD is not ready to roll out Navi 10 just yet –
which is good news since Navi 12 appears to be a continuation of the mainstream upgrade path that the Polaris series has become famous for.
Now you might remember my old exclusive in which I revealed that we can’t expect a flagship GPU from AMD until 2020.
I also talked about how Vega was designed specifically for Apple and Navi has been designed specifically for the Sony PS5 and the question that I immediately thought was whether this is the GPU that we can expect to see featured – the answer – nope.
The Navi 12 is not going to be the GPU that gets featured in PS5, its a derivative of the actual Navi die and has been created specifically so AMD can get it to market for the PC audience primarily.
I have also been told that Navi will be a new microarchitecture (in other words the first AMD Radeon uArch to not be based on GCN). This is going to be the same IP that powers the succeeding Navi dies (such as the Navi 10 and Navi 20).
Now I do know that the Navi 12 GPU will have 40 CUs, what I do not know is if they are going to have the same number of stream processors as GCN does – in which case the core count is going to be 2560 sps.
That does seem likely to be the case because from what I have heard the card slots in somewhere around the Vega 56 depending on the clock rate but much cheaper!
I don’t have a concrete timeline for the part but early estimates put it in H1 2019. AMD can also choose to accelerate this part while they work on Navi 10/20.
As I have previously mentioned the true TITAN-killer from Radeon will not be landing anytime soon –
early estimates put the beefy Navi 20 somewhere around 2020 and at this early in the lifecycle, these timelines can vary widely.
AMD Roadmap TLDR as I have heard:
Vega 7nm will not be coming to gamers.
Navi 12 will be the first Navi part to arrive and will be landing sometime in 1H 2019. Navi 10 has either been scrapped or will follow later sometime in late 2019 or early 2020, depending on a couple of factors.
The performance level of this part will be equivalent to Vega and it will be a small GPU based on 7nm.
Navi 20 is going to be the true high-end GPU built on the 7nm node and as things stand right now, you are tentatively looking at it landing sometime around late 2020 – 2021.
Navi will also be the first architecture to transition away from GCN (and along with it, the 4096 SP / 64 CU limit that is inherent to the uArch implementation).
‘Next-Gen’ architecture is the uArch formerly codenamed KUMA internally before AMD decided it didn’t like that name too much (oops) and will be based on the same brand new major architecture that AMD rolls out with Navi.
It is probably going to be more than that. I'd say €350. RX580 is currently at around €250.
A man can dream.... If that one is not cheap im screwed with a 1440p monitor and a broken FuryX
There is always 1280x720 to have 2:1 scaling... J/K.
But I can dream too... about chiplets with 40CU Navi. Well, maybe gen. or two after Navi.
Thx for the idea, maybe ill try that on the 65" 4k tv, so the ratio is better i wont even see those damn pixels
Meh, not cool how the shitty furyx is dying Maybe if the pump stops working completely is worth keeping it... maybe it stays cool enough and dead silent lol
Or DIY Goiur Cooling Solutions !
Just look for some good Vega Deals, and sell this one cheaper (100-150€ as "to repair" on e-Bay)
DYI-ing it will end up with me breaking the card lol
Im looking for deals on vega, best ive seen is in Geizhals BUT the dont send card to spain so ill have to wait to amazon deals
CaseKing? or CaseKing OC UK
Delivery to other countries on request
"AMD will definitely respond to DirectX Raytracing" - David Wang
Japenese gaming site 4Gamer has managed to get a scoop from AMD’s David Wang (Senior Vice President of Engineering for Radeon Technologies Group) about the company’s plans to support DXR or DirectX RayTracing in games.
According to Wang, the company does not have plans to support DXR at the point in time until it becomes available across their entire product line – from the low end to the high end.
The Japanese Gaming Website 4Gamer has managed to snag an interview with the Radeon Technologies Group's Senior Vice President of Engineering, David Wang, to discuss the company's latest Radeon Instinct graphics cards as well as their future products.
One thing is clear with this interview, with David Wang stating that, in his personal opinion, "AMD will definitely respond to DirectX Raytracing", though for now, the company plans to focus on Radeon ProRender, which is a free tool that is designed to accelerate the production of CG renders.
This is where things get a little tricky, as machine translations of foreign language websites are notoriously unreliable. Some translators read as if AMD has no plans to back Ray Tracing until they can apply it to their whole product stack (from low to high-end), while others say that ray tracing itself will not be utilised widely in games until hardware support is available on both low-end and high-end products.
To get to the bottom of this issue, I contacted @KOMANCHI_ENSAKA on Twitter, a native Japanese speaker who frequently posts Radeon-related information leaks and other tech-related content,
and asked if we could get an accurate translation of the quotes related to DirectX Raytracing.
Thankfully, Komachi obliged and delivered us an accurate English translation of the text, which reads as follows.
Mr. Wang said that "AMD will definitely respond to DXR," after preposing that
"This is a personal view", but
"For the time being, AMD is providing it free of charge"
We will focus on improving the offline CG production environment centered on Radeon ProRender ".
"The spread of Ray-Tracing's game will not go unless the GPU will be able to use Ray-Tracing in all ranges from low end to high end," he said.
As you can see, this translation is a far cry from what some machine translations have generated, so much so that at least one website is claiming that AMD won't implement DirectX Raytracing DXR "until its offered in all product ranges", which is not what David Wang is trying to say here. Beyond that, what David Wang expressed was an opinion, not a statement that should be considered as being from AMD as a company.
Regardless, the supposition that AMD will not offer support for DXR until they can support it across their entire product stack is false, as it has no relevance to David Wang's actual statements to 4Gamers. All David Wang said what that he thought that DXR wouldn't become mainstream until the feature is available on products that range from low-end to high-end, which has nothing to do with AMD's hardware roadmap.
So far, up till 12_1 feature levels have been officially announced and most of AMD’s GPUs are on 12_0. NVIDIAs Pascal is 12_1 and Turing is probably even higher on an as of yet undeclared feature level.
Now there is some debate whether Turing’s hypothetical 12_2 feature level (or whatever higher number is allocated, could be 13_0 as well for eg) is required to support DXR or it can run on 12_1 as well just fine. In any case, it does not look like running it on 12_1 without dedicated raytracing cores will be feasible from a performance POV and it is doubtful the IHVs would allow that.What this essentially means is that trying to implement DXR on anything other than 12_2/13_0 would result in significant performance costs.
On the other hand, NAVI is due in less than a year and that is brand new IP. Depending on whether AMD managed to make that 12_1 (or greater) the company could easily support DXR in the future but with a big performance cost – unless they managed to squeeze in raytracing cores (which is unlikely).
That said, from what we have been told, the first Navi GPU to arrive will be a mainstream part and will have Vega 56 like performance, which is barely enough to sustain the heavy performance requirements of DXR.
So AMD might actually make the strategic decision not to support it till heavy hitters like the Navi 20 arrive much later (and even then they might not because of the performance impact associated with enabling DXR without dedicated RT cores).
The biggest implication I can think of is that future consoles like the PS5, which are always based on AMD tech, will probably not be supporting DXR either.
In fact, your best bet for DXR in a Radeon card is what AMD has planned as a follow up to NAVI: the architecture formerly codenamed KUMA,
now dubbed simply as “Next-Gen” and will almost certainly be a feature level 12_2 or higher architecture.
The moral of the story here is that you shouldn't trust machine language translations, especially when it involves technical matters, as small changes can easily result in the spread of incorrect information.
Otherwise known as the December Omega Driver Release
That's disappointing to hear. I was hoping that Navi would at least support it.
The feature level shouldn't matter unless there's some specific criteria DXR makes use of although I'm not sure if there's documentation on how it's implemented via D3D12 yet.
AMD also has their own AMD Rays via GPU Open though going via D3D and DXR with D3D12 would make adaption a bit more standardized I suppose though it could also be that AMD and NVIDIA both use their own versions even if it's going through DXR as well.
Similar to Vulkan then and vendor specific extensions which should have landed now for NVIDIA's RTX functionality in the newest SDK version.
Although Vulkan also has the advantage of not being bound to Windows 10 with adaption still being pretty low and Microsoft's recent messes with 1809 isn't exactly helping either.
It's google Translator miss Translate
"AMD will definitely respond to DirectX Raytracing" - David Wang
Please re-read news again.
wccftech fault -> lol
AMD's rumoured to release 7nm Navi 12 GPU in mid-2019
AMD's Navi graphics architecture is expected to succeed Vega in 2019, bringing Radeon's consumer graphics cards to 7nm while transitioning away from the company's longlived GCN microarchitecture.
WCCFTECH has reported that AMD's NAVI 12 GPU will feature a total of 40 CUs (compute units), giving it a total of 2560 stream processors, assuming that Navi has 64 stream processors per CU like Vega.
Navi 12 is expected to act as a mainstream product, effectively working as a replacement for Polaris, while offering performance levels that are similar to the RX Vega series.
Rumour has it that NAVI was designed with Sony's PS5 console in mind, bringing with it architectural enhancements that were defined by the console manufacturer.
This isn't necessarily a bad thing for PC gamers, as these architectural tweaks should also improve the architecture's gaming performance in PC titles.
Right now, AMD's 7nm Vega graphics hardware isn't expected to ever make it to PC gamers, making it likely that the architecture will be exclusive to the machine learning market.
AMD isn't expected to release a high-end NAVI graphics card until a later date, making the launch of NAVI similar to Polaris.
At this time, little is known about AMD's NAVI architecture, or if it will add any new features to the Radeon graphics lineup, like support for AI acceleration or Ray Tracing Acceleration for Microsoft's DXR (DirectX Raytracing) or DirectML (DirectX Machine Learning) APIs.
With Sony's focus on checkerboard rendering on the PS4 Pro, it would be a smart move from Sony to invest in an AI compute solution that is similar to Nvidia's Tensor cores
to offer Machine Learning image enhancement functions like DLSS, a feature set that could easily transfer to DirectML and PC gaming.