Tuesday, June 19th 2018

AMD Radeon Vega 12 and Vega 20 Listed in Ashes Of The Singularity Database

Back at Computex, AMD showed a demo of their Vega 20 graphics processor, which is produced using a refined 7 nanometer process. We also reported that the chip has a twice-as-wide memory interface, effectively doubling memory bandwidth, and alsomaximum memory capacity. The smaller process promises improvements to power efficiency, which could let AMD run the chip at higher frequencies for more performance compared to the 14 nanometer process of existing Vega.

As indicated by AMD during Computex, the 7 nanometer Vega is a product targeted at High Performance Compute (HPC) applications, with no plans to release it for gaming. As they clarified later, the promise of "7 nanometer for gamers" is for Navi, which follows the Vega architecture. It's even more surprising to see AOTS results for a non-gaming card - my guess is that someone was curious how well it would do in gaming.
The device ID 66A0 has been never seen before and is sufficiently far away from Vega 10 (687F) to indicate that AMD treats Vega 20 as a really new product and not as a minor revision. We expect Vega 20 to be used on a Radeon Instinct card first, with 32 GB of HBM2 memory and 4096 shaders, a professional workstation card called "Radeon Pro Vega 20" could also be possible, as indicated by other leaks, but I'm skeptic whether AMD really wants to put the "20" from the GPU codename into the official marketing name, going against what they used for "Vega 64".
The second card listed is Vega 12, for which very little is known. Its device ID 69A0 is also from a new, unused value range. It is rumored that Vega 12 is targeted at lower-end, mobile markets, possible to offer an alternative to Polaris. We also wouldn't be surprised to see improved power management on Vega 12, to cater better for mobile devices, which require higher energy efficiency. Please note that Vega 12 does not indicate the number of CUs (like on Vega 64 for instance), but rather represents the internal GPU codename, like Vega 10.
Sources: Vega 12, Vega 20
Add your own comment

51 Comments on AMD Radeon Vega 12 and Vega 20 Listed in Ashes Of The Singularity Database

#26
TheoneandonlyMrK
T4C FantasyVega 12 has all of the features as 10 but smaller like polaris 12 is to 10
So 1028 shaders or less? Or closer to Polaris at 2560.

I suppose it's lowends turn tbh.
Posted on Reply
#27
T4C Fantasy
CPU & GPU DB Maintainer
theoneandonlymrkSo 1028 shaders or less? Or closer to Polaris at 2560.

I suppose it's lowends turn tbh.
Right now i dont know the cu count but one is 4gb hbm2 i think
Posted on Reply
#28
PowerPC
lynx29I think he was referring to AMD gpu's in a general sense mate. lol
When? Where did I say this?
Posted on Reply
#29
jabbadap
T4C FantasyRight now i dont know the cu count but one is 4gb hbm2 i think
Do we have the name for that RX Vega Mobile chip, which should be ~28cus. Maybe it's the Vega 12. This amd mobile aots score is quite interesting, what is that other gpu on it?
Posted on Reply
#30
T4C Fantasy
CPU & GPU DB Maintainer
jabbadapDo we have the name for that RX Vega Mobile chip, which should be ~28cus. Maybe it's the Vega 12. This amd mobile aots score is quite interesting, what is that other gpu on it?
This is where it gets confusing for most, no vega 12 is not a successor to vega 11 apu graphics, vega 12 will be like rx vega 18 mobile or rx vega 24 etc
Posted on Reply
#31
efikkan
While we all want AMD to be competing in the high end, we know it's not going to happen anytime in the next few years. Vega 20 will arrive ~Q1 2019 but will be a professional product. Navi is expected mid to late 2019, but is targeting a performance level close to Vega 10. By that time, it will be competing with GTX 1060's replacement "GTX 1160", so only competing in the lower mid-range and downward.

We've seen the hype ahead of Fiji, Polaris and Vega, and we know how they turned out. Now once again the forums heat up, this time about Navi and shrinks of Vega, and the usual Youtubers pour fuel on the fire. It's astounding how much faith some put in TSMC's 7nm node to solve all of AMD's problems. Even if we assume TSMC 7nm is free of the problems 40nm/28nm/20nm/16nm all suffered from, we might hope for something like ~40% improved efficiency or ~30% more performance(through higher clocks), but that will be towards the end of the 7nm lifecycle, not Q1 2019. And AMD will have to choose between better efficiency or higher clocks, or a mix, but they can't get the full potential of both.

People should remember that a GTX 1080 performs about 80% better per watt than RX 480 or RX Vega 64, so just a die shrink will not be enough for AMD to keep up. "Consumer Volta" would easily beat a die shrink of Vega, and would probably do well against Navi too, despite the advantage of a smaller node. Also remember that 7 nm will not ship in massive volumes in Q1 2019, so AMD will have too choose which chips to prioritize.
Posted on Reply
#32
TheoneandonlyMrK
efikkanWhile we all want AMD to be competing in the high end, we know it's not going to happen anytime in the next few years. Vega 20 will arrive ~Q1 2019 but will be a professional product. Navi is expected mid to late 2019, but is targeting a performance level close to Vega 10. By that time, it will be competing with GTX 1060's replacement "GTX 1160", so only competing in the lower mid-range and downward.

We've seen the hype ahead of Fiji, Polaris and Vega, and we know how they turned out. Now once again the forums heat up, this time about Navi and shrinks of Vega, and the usual Youtubers pour fuel on the fire. It's astounding how much faith some put in TSMC's 7nm node to solve all of AMD's problems. Even if we assume TSMC 7nm is free of the problems 40nm/28nm/20nm/16nm all suffered from, we might hope for something like ~40% improved efficiency or ~30% more performance(through higher clocks), but that will be towards the end of the 7nm lifecycle, not Q1 2019. And AMD will have to choose between better efficiency or higher clocks, or a mix, but they can't get the full potential of both.

People should remember that a GTX 1080 performs about 80% better per watt than RX 480 or RX Vega 64, so just a die shrink will not be enough for AMD to keep up. "Consumer Volta" would easily beat a die shrink of Vega, and would probably do well against Navi too, despite the advantage of a smaller node. Also remember that 7 nm will not ship in massive volumes in Q1 2019, so AMD will have too choose which chips to prioritize.
Their are other nodes then 7nm and mainstream isn't likely to get to 7nm until 2020 ,i think they have a very scalable core with Vega, mine works anywhere between 120-400+ watts and various performance levels obviously, flat out it equals a 1080ti in modern Apis but at a higher wattage on older APIs it does lag a bit ,i think they could make something competitive for midrange easily, top end for better or worse is sorted and selling well, mud range and below needs a new GPU asap.

And your post is the first I've Seen sensationaling the dedate here.
@jabbadap I heard Vega 14 is for mobile uses.
Posted on Reply
#33
T4C Fantasy
CPU & GPU DB Maintainer
theoneandonlymrkTheir are other nodes then 7nm and mainstream isn't likely to get to 7nm until 2020 ,i think they have a very scalable core with Vega, mine works anywhere between 120-400+ watts and various performance levels obviously, flat out it equals a 1080ti in modern Apis but at a higher wattage on older APIs it does lag a bit ,i think they could make something competitive for midrange easily, top end for better or worse is sorted and selling well, mud range and below needs a new GPU asap.

And your post is the first I've Seen sensationaling the dedate here.
@jabbadap I heard Vega 14 is for mobile uses.
Vega 14 doesnt exist afaik. as a core only 12 and 20 this year, mass production of 7nm is planned for 2019 possibly late of course but it will be there before 2020, 2020 will have navi 20 or w/e next gen is. As a gpu database maintainer my job isnt the results of the specs but the specs themselves. I dont care if vega 20 beats 1080ti or is worse than a 1070, my passion is graphics cards as a whole.
Posted on Reply
#34
efikkan
theoneandonlymrkTheir are other nodes then 7nm and mainstream isn't likely to get to 7nm until 2020 ,i think they have a very scalable core with Vega, mine works anywhere between 120-400+ watts and various performance levels obviously, flat out it equals a 1080ti in modern Apis but at a higher wattage on older APIs it does lag a bit ,i think they could make something competitive for midrange easily, top end for better or worse is sorted and selling well, mud range and below needs a new GPU asap.
AMD have said Navi 10 will replace Vega 10, and there is no node shrink of Vega 10 planned. They might have a backup plan, but right now the next target is Navi.

How is Vega going to "scale" out of this problem? Pascal offers ~80% more performance per watt, consumer "volta" will push that even further. Not even the best node shrink imaginable will come close to compensating for that.

And you're one of those who still spin the "AMD is better at newer APIs"-story. Well, there is nothing inherently benefiting Vega in either Direct3D 12 nor Vulkan, and the games people like to use as examples doesn't even use these APIs natively. We've been hearing this since the early days of GCN, and we're still waiting for AMD's "future proof" hardware to pay off…
Posted on Reply
#35
jabbadap
T4C FantasyThis is where it gets confusing for most, no vega 12 is not a successor to vega 11 apu graphics, vega 12 will be like rx vega 18 mobile or rx vega 24 etc
Well that run has two gpus, GPU #1 is that raven ridge vega 11 and GPU #2 Radeon RX Vega... Mobile, ⁵⁶ or what. I'm not really familiar with those scores or fps:s on that bench, if I remember correctly it's horrible bench anyway to compare.
Posted on Reply
#36
T4C Fantasy
CPU & GPU DB Maintainer
jabbadapWell that run has two gpus, GPU #1 is that raven ridge vega 11 and GPU #2 Radeon RX Vega... Mobile, ⁵⁶ or what. I'm not really familiar with those scores or fps:s on that bench, if I remember correctly it's horrible bench anyway to compare.
well yes i understand every naming convention, but ive run into so many conversations where people get confused, i didnt do a good job clearing it up but

Raven = Vega 3, 6, 8, RX Vega 10, 11
Vega 12 = RX Vega 18~24 Desktop and or Mobile (naming is unknown)
Vega 11 = RX Vega 28~36 Desktop and or Mobile (naming is unknown) <-- may have been canceled
Vega 10 = RX Vega 56, 64
Posted on Reply
#37
FordGT90Concept
"I go fast!1!11!1!"
If Vega 11 has been cancelled, they can't seriously be thinking about pushing Polaris for another generation, can they?
Posted on Reply
#38
TheoneandonlyMrK
efikkanAMD have said Navi 10 will replace Vega 10, and there is no node shrink of Vega 10 planned. They might have a backup plan, but right now the next target is Navi.

How is Vega going to "scale" out of this problem? Pascal offers ~80% more performance per watt, consumer "volta" will push that even further. Not even the best node shrink imaginable will come close to compensating for that.

And you're one of those who still spin the "AMD is better at newer APIs"-story. Well, there is nothing inherently benefiting Vega in either Direct3D 12 nor Vulkan, and the games people like to use as examples doesn't even use these APIs natively. We've been hearing this since the early days of GCN, and we're still waiting for AMD's "future proof" hardware to pay off…
Again with sensational nonesense where did I say it's better?? I have used both btw have you.
Posted on Reply
#39
T4C Fantasy
CPU & GPU DB Maintainer
FordGT90ConceptIf Vega 11 has been cancelled, they can't seriously be thinking about pushing Polaris for another generation, can they?
I think the Vega 28 sku will be igp only like raven and wont carry the vega 11 core name
The placement of the id is in igp range
Vega 12 and 20 are the only new cores right now
Posted on Reply
#40
JoniISkandar
Radeon Group has new leader right ?? he said Navi is not MCM, but he mention than Gaming and Workstation GPU will be separated., unlike Raja want put Vega under same roof Workstation Hybrid Gaming card
Posted on Reply
#41
medi01
lynx29I hope the Vega 20 can handily beat an OC'd 1080 ti, otherwise Nvidia is going to release performance increases at a snails pace like Intel did for so long.
I hope AMD lags in GPU department for a while, for exactly this reason.
Oh, and price, increase the price, Huang, the market is ready.
Posted on Reply
#42
msroadkill612
I find it odd, that to ensure one is part of the latest and most advanced AMD cpu & gpu ecosystem, is a choice between selling a kidney for a vega dgpu, or buying their very cheapest product - the apu.

usually cheap means superannuated old crap.
Posted on Reply
#43
Casecutter
FordGT90ConceptIf Vega 11 has been cancelled, they can't seriously be thinking about pushing Polaris for another generation, can they?
IDK but from all this talk above and knowing Polaris was only once re-named, that might be their cheapest option and see if Nvidia holds back more.

Just nothing more than Ryzen+ to Polaris ... that's entails few tweaks to perhaps the memory controller or whatever cheap and quick "corrections" and just say it's going to the 12nm process. That could tide over many (ie 1080p) who haven't had a chance to upgrade since before Polaris and I think there's a bunch of them. And, as with Ryzen+ get some price reductions, sure GDDR is supposedly hard to get and pricey, but see if they can hold the same MSRP that they had in June 2016. I don't see Nvidia looking to release some *1160* at anything south of $329, while first Founder Editions at $400.
Posted on Reply
#44
jabbadap
T4C FantasyI think the Vega 28 sku will be igp only like raven and wont carry the vega 11 core name
The placement of the id is in igp range
Vega 12 and 20 are the only new cores right now
IGP? So some Kaby Lake G like BGA mobile apu perhaps(that fenghuang raven thingy), I don't think that would fit to AM4 socket(cpu+28cu+hbm2).

But I agree there's only Vega 12 and Vega 20 chips are pretty much only certain new vegas.
Posted on Reply
#45
Readlight
The most annoying game ever created. Loads in 3,30 minutes, whit not working DX12 whit every setting change needs to restart and works on 20fps.
Posted on Reply
#46
Fluffmeister
ReadlightThe most annoying game ever created. Loads in 3,30 minutes, whit not working DX12 whit every setting change needs to restart and works on 20fps.
Lies! It's the poster child of DX12 and gets rolled out as required.
Posted on Reply
#47
Readlight
FluffmeisterLies! It's the poster child of DX12 and gets rolled out as required.
DX12exe. always crashes.
Posted on Reply
#48
msroadkill612
jabbadapIGP? So some Kaby Lake G like BGA mobile apu perhaps(that fenghuang raven thingy), I don't think that would fit to AM4 socket(cpu+28cu+hbm2).

But I agree there's only Vega 12 and Vega 20 chips are pretty much only certain new vegas.
I am not so sure on die HBMis very restricted by die space. They seem to talk of it as like adding a mezzanine to the die.
Posted on Reply
#49
Tom_
High End Navis will be released after the mid Range Ones.
Vega is slower than anticipated, but not so bad.
I just bought a Vega 64 Liquid Edition and a NITRO+ Vega 64.
Posted on Reply
#50
John Naylor
T4C Fantasyboth amd and nvidia always made the largest chip first then released cut down versions,
Not since the 600 series .... there's enough industrial espoinage going on that each side knows whats coming out months before the actual release. With the 700 series, many pundits noted that the early leaked specs for the 780 didn't quite live up to the pre-release specs but almost exactly matched the early 770 specs. When AMD launched their marketing campaign for th next generation cards, boom out of nowhere came the 780 Ti to rain on AMDs parade... which oddly enough lined up perfectly with the early leaked 780 specs. Not seeing anything to challenge what became the 980, they kept the 980 Ti on the shelf again. I'm quite sure that when the 1180 / 280 whatever is released, it will not be their top card in the series.
Posted on Reply
Add your own comment
Nov 12th, 2026 13:36 PST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts