MSI Spatium M480 2TB NVMe SSD review
AMD Ryzen 5 5600G and Ryzen 7 5700G review
Announcement AMD Radeon RX 6600 XT
Arctic Liquid Freezer II 360 A-RGB review
Cooler Master Masterair MA624 Stealth/MA612 Stealth ARGB review
EVGA GeForce RTX 3080 Ti FTW3 Ultra Review
F1 2021: PC graphics performance benchmark review
Asustor Drivestor 4 Pro (AS3304T) NAS Review
Colorful iGame GeForce RTX 3090 KUDAN review
Corsair Virtuoso RGB Wireless XT headset review
AMD Vega 20 3DMark11 benchmark surfaces
We've heard about the GPU called VEGA20 a number of times already. Vega 20 is to be a die shrunk version of Vega, with more memory. So it is expected to have the same 64CUs as in Vega 10 but would be fabbed at 7nm. A few days ago the card was already mentioned as 'spotted' in the AMD Labs, today however a result surfaced in the 3Dmark database.
We mentioned Vega20 a few times in the past already, and that would be a die shrunk Vega, much like what Ryzen 2000 will be towards the original Ryzen procs. Whatever the production node is, we don't know 100% yet though as that can be anything from 7 to 12 nm.
Earlier on a Linux driver entry was listing no less than six new device IDs referring towards "Vega 20". Vega 20 was mentioned to be released the second half of 2018 which makes 7nm a no-go. the card will have the same 64 CUs / 4096 shader processors as in Vega 10. VEGA20 BTW is suggested to also feature PCIe 4.0. compatibility.
We can now add to that a Futuremark 3Dmark 11 (UL) entry, which isn't going to say much in terms of performance with such an old benchmark. However, the card indeed was spotted, and with 32GB graphics memory it seems. That makes it an AI/Deep learning reference model being tested, likely something Radeon instict. Here is the Vega Frontier vs Vega 20 3DMark11 score as spotted by videocardz, the device ID of 66A0 belongs to Vega 20:
AMD Vega 20 and Ryzen 5 Pro Mobile Say Hello - 09/26/2017 08:49 AM
On a newly leaked roadmap slash set of media slides some new info on AMD Vega 20 and Ryzen 5 Pro Mobile have exposed itself. Let's have a peek at them....
AMD Vega RX 56 With Triple-Fans Surfaces - 08/14/2017 08:35 AM
Interesting, if you have had a peek at some of the photos in our unboxing of Vega, you should have noticed that the last photo showed Vega 56. That product has a blower single fan style cooler. New ph...
AMD Vega Frontier Edition Unboxed and Benchmark (updated) - 06/28/2017 03:42 PM
With AMDs Vega Frontier Edition now available it was onlt a matter of time before a consumer would get his hands on it. User #define @ Disqus purchased one, took unboxing photos and ran a 3DMark Fir...
AMD VEGA Die-shot Photo - 06/06/2017 07:19 AM
Well, GPU shot reallt. Check it out, I checked with AMD if this is a real photo and not a fakes one. It is the real thing though. The VEGA shot shows a thing or two, you can count the 8 blocks that go...
AMD Vega To get 20K Units Released at launch and new Zen+ chatter - 05/09/2017 08:48 AM
The ongoing rumors about Radeon RX Vega are just, on-going. In the latest news it is reported that there will be 20K Units for Vega available state some media. There might be news about Zen+ inbound a...
Ssateneth
Member
Posts: 99
Joined: 2007-02-20
Member
Posts: 99
Joined: 2007-02-20
#5542419 Posted on: 04/30/2018 09:04 PM
It's an engineering sample. The Vega20 part is the scores on the RIGHT, not the left. The clock speeds could be read incorrectly, or set low on purpose due to being an engineering sample, or there may be some frequency boost that readings aren't picking up on, so take those numbers with a grain of salt. I'm excited for 4096bit bus width on top of the higher clock speed on HBM. The core could just be more of the same though, just with lower power usage (AMD projects 150W)
It's an engineering sample. The Vega20 part is the scores on the RIGHT, not the left. The clock speeds could be read incorrectly, or set low on purpose due to being an engineering sample, or there may be some frequency boost that readings aren't picking up on, so take those numbers with a grain of salt. I'm excited for 4096bit bus width on top of the higher clock speed on HBM. The core could just be more of the same though, just with lower power usage (AMD projects 150W)
Kaarme
Senior Member
Posts: 2439
Joined: 2013-03-10
Senior Member
Posts: 2439
Joined: 2013-03-10
#5542423 Posted on: 04/30/2018 09:09 PM
It seems kind of doubtful they would get results like that at 1000MHz and then up the clocks to, say, 1700MHz for the released products, creating a huge difference. But if that was really the case, it would be quite a card.
It seems kind of doubtful they would get results like that at 1000MHz and then up the clocks to, say, 1700MHz for the released products, creating a huge difference. But if that was really the case, it would be quite a card.
AMDfan
Member
Posts: 48
Joined: 2017-12-21
Member
Posts: 48
Joined: 2017-12-21
#5542447 Posted on: 04/30/2018 10:26 PM
I believe this is the first MCM GPU, with 2 Vega 64 on a single die. That would be this kind of performance, hence that this is an early sample running just 1Ghz and 2 modules on 7Nm.
If true that it also supports PCIe4 than it is a little low in performance..... but it's as i said earlyer an early sample. We have to wait and see what the real performance out of the box will be.
Oh, they also have a lot of Vega 20 ID's.
My guess is they wil also create consumers versions with up to 16 GB HBM2
I believe this is the first MCM GPU, with 2 Vega 64 on a single die. That would be this kind of performance, hence that this is an early sample running just 1Ghz and 2 modules on 7Nm.
If true that it also supports PCIe4 than it is a little low in performance..... but it's as i said earlyer an early sample. We have to wait and see what the real performance out of the box will be.
Oh, they also have a lot of Vega 20 ID's.
My guess is they wil also create consumers versions with up to 16 GB HBM2
Imglidinhere
Senior Member
Posts: 248
Joined: 2010-10-02
Senior Member
Posts: 248
Joined: 2010-10-02
#5542472 Posted on: 05/01/2018 12:05 AM
Check the clock speed differences. It's 3% difference but with the new chip running at 1GHz vs the current Vega running at 1.6GHz. Sorry, but... that's pretty important.
Hmm 3% combined increase. Is it really worth making this product? Isn't an overclocked Vega 10 card in this area of performance already? I'm confused why AMD is putting all this effort in to an blah Gpu instead of using their resources for a new series? Is this their attempt at a Vega TI?
Check the clock speed differences. It's 3% difference but with the new chip running at 1GHz vs the current Vega running at 1.6GHz. Sorry, but... that's pretty important.
Click here to post a comment for this news story on the message forum.
Senior Member
Posts: 1353
Joined: 2017-08-18
I don't think AMD needs to pump out a brand new architecture. There is plenty good about Vega, it just needs refinement. But, if this is really just nothing more than a die shrink with better memory, this is looking very bleak. AMD had plenty of time and money to fine-tune Vega. The miners were a saving grace to them, and so far it looks like they're not taking advantage of that. They can't afford making another product that's mediocre to the gaming community (I personally think Vega is great, for non-gaming purposes).
The way I see it, whatever Vega 20 comes out to be would still be less than what people were expecting Vega 10 should've been. I sincerely hope Vega 20 is more interesting than we think.
i think your reasoning is generally spot-on. however, this is such an early stage that the Process may be ready, but not the new architecture. so this just might be a smokescreen.