Radeon VII

[aoaaron]

Personally I think Vega 7 is pointless already.

It's priced too high.

Again, I could buy a 2080 for probably the same price (or maybe cheaper) than what the V7 will be.

Why, oh why (unless I was a rabid fanboy) would I buy this over a 2080?

Wouldn't the 16GB of RAM swing you towards AMD?

(I have a 2080).

I think sadly everyone is right in that AMD have released a 1080ti competitor (released in 2017) in 2019 for the same retail price the 1080ti was released.

They're just lucky Nvidia pulled a fast one so they can draw comparisons with the 2080 instead.

 

[Kelt]

Wouldn't the 16GB of RAM swing you towards AMD?

(I have a 2080).

I think sadly everyone is right in that AMD have released a 1080ti competitor (released in 2017) in 2019 for the same retail price the 1080ti was released.

They're just lucky Nvidia pulled a fast one so they can draw comparisons with the 2080 instead.

16GB would be nice no doubt, but not necessary for me at the moment.

I'm running a 1070 @ 1440P and though I've gotten close to 8GB usage, nothing has quite exceeded it.

This card should have had 8-12GB of GDDR6 memory and priced at around £500.

 

[Gerard]

This card should have had 8-12GB of GDDR6 memory and priced at around £500.

No real way of them doing that, they would have had to make a ton of changes both to the core and the pcb which would have cost more in the long run.

 

[CAT-THE-FIFTH]

Some more details:
https://www.anandtech.com/show/13832/amd-radeon-vii-high-end-7nm-february-7th-for-699
https://www.youtube.com/watch?v=IcL4XFwptHo

So it has 128 ROPs,a base clock of 1450MHZ and a maximum boost clock of 1850MHZ. So clockspeeds seem to be only around 20% higher,and the transistor count increase is quite small over Vega MK1,ie,700 million transistors.

 

[IT Troll]

With Nvidia now supporting VRR on FreeSync monitors they have taken this USP away from AMD. Why wouldn’t you just go ahead and buy an Nvidia card now, if you have that kind of budget and want that level of performance?

I don’t like the high pricing of the RTX cards but the R7 isn’t significantly cheaper and it would seem lacks any form of DXR. It doesn’t seem to offer performance, new features, or value. Very underwhelming.

 

[LePhuronn]

To be honest this feels like a PR product, nothing more.

The RX 590 felt like something to fulfil GloFo's wafer agreement and to give AiB partners something new to sell over Christmas and has a 9 month shelf life. Now with the raster performance of RTX not being much higher than GTX 10, and ray tracing performance utterly woeful, the Radeon VII cannibalises MI50 sales just to say "look, we can match Nvidia at the top end, honest". And for a price point that doesn't really offer much incentive to actually buy Radeon VII over the RTX 2080.

I would've much preferred AMD stick their chest out and defiantly say "no, we have a plan and we're sticking to it. We're taking back the midrange with Navi. We're not chasing Nvidia's halo products right now. We're getting our GPU house in order and then we'll talk Arcturus in 2020".

Very disappointed in a $700 PR stunt.

It also brings into question that hint one of AdoredTV's sources dropped when they said they'd seen a "Big Navi" running and it was impressive. Given the RX 3080 as leaked is still a chunk of performance away from Radeon VII, could it be the "Big Navi" was actually just Vega 20 after all?

Roll on Computex when I think we'll see some Navi talk. I'm still excited at the prospect of Vega 64 + 15% for £250.

 

[Scougar]

Was looking at some of the specs, and it does really seem like another RX 590. Apart from rop count and memory doubling, most changes seem to come from drop to 7nm.

I want a shiny GPU, and my birthday in Feb... Was very much hoping navi would at least be announced.

 

[Aretak]

Not really he is doing his job by defending "his" products from the competition shareholders wouldn't be to happy if he bigged the Vega 7 card would they

That would be like tim cook saying the galaxy range of phones are brilliant

It would breed confidence in the shareholders and so on

Company executives are generally very respectful of their competition if they comment on them at all. You know, because they have some professionalism. It's more than possible to talk up your product as being the best without resorting to petty attacks. LJM lashing out, calling AMD's product garbage and making absurd claims about how FreeSync "doesn't work" (which will be surprising news to those that have been using it for the past few years) just makes him seem like a slighty pathetic, classless, angry little man with a lot of problems on his mind (not least Nvidia's tanking share price I'm sure).

 

[Legend]

I would've much preferred AMD stick their chest out and defiantly say "We're not chasing Nvidia's halo products right now.".

I'd hardly call it a chase, and this is no halo product contest. AMD are putting out a card with the performance levels we had access to nearly 2 years ago with the 1080Ti! That's all the 2080 is, with some extra tensor cores that aren't being used... which the Radeon VII obviously doesn't have. It's basically a 7nm 1080Ti, 2 years late. At the same price a 1080Ti was 2 years ago. Way to innovate AMD!

 

[DONOHUE07]

Long time AMD user here. Though I currently own a 1070ti.

This is a terrible, again. This card if it matches the 2080, is on par with the now much older 1080ti...

Both AMD and NVIDIA are offering tiny increments in performance, but for a hugely inflated fee.

One of the best desicions I made regarding PC gaming was to sell off a 4K monitor, I’m staying at 1080p 60hz, the high end market of PC gaming is just a bottomless money pit, and it’s likely to get even worse

 

[Calin Banc]

And now 6gb is borderline. So soon the 1060s will be crippled or stuck in low detail settings while 8gb 480/580/590s cruise on.

I'm betting Doom Eternal will eat more than 6, seeing as the 2016 one wanted 5 for highest texture settings. In which case bad times for anyone who ordered a 2060.

You make it sound worse than it is. Is really not a comparison of highest settings vs low. You turn down a setting or two (when needed) which probably will go unnoticed in the long run for most. With the R290 4GB, in Eyefinity, the limiting factor kicks in as the GPU power, not vRAM size.

 

[easyrider]

I'd hardly call it a chase, and this is no halo product contest. AMD are putting out a card with the performance levels we had access to nearly 2 years ago with the 1080Ti! That's all the 2080 is, with some extra tensor cores that aren't being used... which the Radeon VII obviously doesn't have. It's basically a 7nm 1080Ti, 2 years late. At the same price a 1080Ti was 2 years ago. Way to innovate AMD!

Its a shame as I was really hoping that AMD would shake up the Industry. Its a decent enough card but basically its a 1080ti or 2080 with more ram. There is nothing compelling about it. I think that the 1080 ti will become a classic and dare I say it I may keep mine for a couple of years maybe more. There is nothing in the GFX card space that compels me to upgrade.

This isn't an Easyrider going on about his 1080ti again post. Its a genuine reflection on the Market today.

People defending the Radeon 7 two years after the release of the 1080ti when its the same speed and the only justification is a few few games need to have a serious chat with themselves.

Its a mediocre release....The only saving grace is its worth buying over the 2080... its just a shame its the same performance just with more ram.

I would have preferred less ram say 12GB and a retail of £499 this would then make the 2080 obsolete.

 

[D.P.]

16GB would be nice no doubt, but not necessary for me at the moment.

I'm running a 1070 @ 1440P and though I've gotten close to 8GB usage, nothing has quite exceeded it.

This card should have had 8-12GB of GDDR6 memory and priced at around £500.

They can't do that from a salvaged part though. Using GDDR6 would require spinning out a new chip with a redesigned memory controller. It then likely would end up more expensive due to 7nm yields at the moment.

 

[Raviolay]

60CUs. It is a salvage part of the Vega20 Instinct. 300w with performance between 1080 and 1080ti.

Expect limited supply. Likely street price higher than 2080.

There won't be a 8GB model since it would have half the bandwidth.


The fact that Navi wasn't mentioned is a bad sign IMO. I expect the 7nm process is not mature enough yet. Would explain a salvage part from AMD and Nvidia going with 12nm

Or maybe wafers are needed for the CPU chiplets, and this card is just a way of offloading left over 7nn Vega. Why announce Navi when it's going to be announced latter in the year. As part of the new PlayStation and kill any chance shifting old stock. Navi for PC will be announced and launched this time next year my guess.

 

[drunkenmaster]

People need to get a grip, it's a card that was designed in every way to simply replace professional versions of their Hawaii cards, nothing about it was designed to improve gaming performance, every single addition was to support more compute functions and better improve efficiency for gaming. A by product of making this professional compute card is it's on 7nm, it runs faster and is more efficient because of that so is superior to Vega 64. So they either don't make a gaming version, or they do. If they don't people say why not just release it for consumers, if they do, people compare it to a pure gaming card from Nvidia and wonder why it's not trashing it. A it's not in the same class, they didn't go from a Vega 64 with 480mm2 die to a 480mm^2 7nm card, they shrunk it, added some compute and released it.

There is nothing about it that was designed for gaming efficiency, it wasn't designed to be a die size or transistor count that wasn't achieveable on 14/12nm, it's a professional card repurposed because the higher clocks make it a little faster than Vega 64, nothing more or less.

Midrange Navi is likely to be around 30% smaller die, completely optimised for gaming alone, no compute that can't be actively used for gaming, it fp16, but nothing fp64, it will be significantly more efficient both in die size, power, etc. I'm guessing a ~240-250mm^2 core, something along the lines of 2080 performance and at ~£300 mark. gddr6 and significantly smaller die size as well as not being the first 7nm product and on a process they've had time to learn from and tape out a second chip all mean it can be way cheaper.

In terms of the card itself, we'll have to see real world performance, clock speeds and maybe most importantly overclocks but if it's $699 comes with 3 decent games and ends up being 30-40% faster real world it's not particularly bad. It's better than not having the option at all. It's not a great card for gaming but they have it and can make it available so they did. if people want it, good, but this was never about making an efficient gaming card.

 

[D.P.]

Some more details:
https://www.anandtech.com/show/13832/amd-radeon-vii-high-end-7nm-february-7th-for-699
https://www.youtube.com/watch?v=IcL4XFwptHo

So it has 128 ROPs,a base clock of 1450MHZ and a maximum boost clock of 1850MHZ. So clockspeeds seem to be only around 20% higher,and the transistor count increase is quite small over Vega MK1,ie,700 million transistors.

The extra transistors are for 1:2 FP64 support used in the Instinct MI60, it has no value for gaming. AMD did a direct die shrink and added back in the FP64 support for HPC uses. This isn't a new GPU designed for gaming, it is failed MI60 cores. IPC (performance per clock tick) will be identical to Vega64, the only real difference is far more bandwidth.

AMD's performance numbers are liekly very optimistic.

EDIT:
Actually, from Anantech AMd are reporting 20-42% faster than Vega 64, with an average of 29%.
Using TPU: https://www.techpowerup.com/reviews/Zotac/GeForce_RTX_2080_AMP_Extreme/31.html

A 29% increase over Vega64 at 1440p givs 1.29*65% = 84% of the OC'd 2080, and would be 84/95 = 12% slower than a 2080 stock/FE.


We will see when the reviews comes out but I think 10% slower than a 2080 is a very realistic exception.

 

[drunkenmaster]

Or maybe wafers are needed for the CPU chiplets, and this card is just a way of offloading left over 7nn Vega. Why announce Navi when it's going to be announced latter in the year. As part of the new PlayStation and kill any chance shifting old stock. Navi for PC will be announced and launched this time next year my guess.

CPUs get talked about much earlier than GPUs, it's always been that way, always will. CPUs need an entire ecosystem of support to work, OS integration and motherboard/bios support, etc. GPUs is like a usb hard drive, as long as a hard drive supports usb specs, it will plug into any usb port fine. If a gpu supports pci-e properly, it will fit in anywhere and their own drivers aren't as tightly integrated as OS scheduling on a CPU so there is far less need there.

CPUs take 3 times the validation timeframe.

If Ryzen 3 comes in June then talking more about it now and then again in a couple months is normal. If Navi comes in June then they'd start leaking minor details like 3 months from now and they'd only show it like this within weeks of launch if not at the launch itself.

Navi was always scheduled for more like Q3 this year, Vega 7nm was always the first product on 7nm and end of 2018, Navi was always months later, Ryzen 3 was presumed to be late Q2, but because of some leaks and another massive round of hyping because AMD had a speech everyone convinced themselves Navi, Ryzen, that 7nm EUV GPU called Next Gen and Ryzen 4 were all coming out in February.

 

[DragonQ]

Some more details:
https://www.anandtech.com/show/13832/amd-radeon-vii-high-end-7nm-february-7th-for-699
https://www.youtube.com/watch?v=IcL4XFwptHo

So it has 128 ROPs,a base clock of 1450MHZ and a maximum boost clock of 1850MHZ. So clockspeeds seem to be only around 20% higher,and the transistor count increase is quite small over Vega MK1,ie,700 million transistors.

Makes sense, since their claim is 25% more performance for the same power. Vega was stupidly clocked to begin with (hence the crazy stock voltages) so only increasing clocks by 20% should make it a bit less power hungry out of the box compared to Vega 64.

To be honest this feels like a PR product, nothing more.

It is. This is essentially a result of AMD still working on a new architecture but needing to compete with the nVidia RTX series somehow. They've taken an originally designed-for-compute card and made it into a gaming product. They're pricing it high presumably because of the high build costs and the fact they know they aren't going to win any big market share with it anyway (due to nVidia now supporting VRR and other features on their cards like DXR and DLSS). Like the RX 590, it's stopgap largely for shareholders and to keep AMD in the GPU news cycle.

I'm sure the card will be competitive but most gamers will choose the RTX 2080 at this price point. Double the VRAM probably won't entice people more than DXR/DLSS.

 

[D.P.]

Makes sense, since their claim is 25% more performance for the same power. Vega was stupidly clocked to begin with (hence the crazy stock voltages) so only increasing clocks by 20% should make it a bit less power hungry out of the box compared to Vega 64.


It is. This is essentially a result of AMD still working on a new architecture but needing to compete with the nVidia RTX series somehow. They've taken an originally designed-for-compute card and made it into a gaming product. They're pricing it high presumably because of the high build costs and the fact they know they aren't going to win any big market share with it anyway (due to nVidia now supporting VRR and other features on their cards like DXR and DLSS). Like the RX 590, it's stopgap largely for shareholders and to keep AMD in the GPU news cycle.

I'm sure the card will be competitive but most gamers will choose the RTX 2080 at this price point. Double the VRAM probably won't entice people more than DXR/DLSS.

Going by AMD's own numbers, power requirements have probably gone up for Vega7. If they Claim Vega7 is 25% faster at same power and Vega 7 is 29% faster than Vega64, then we might be looking at 320w or soemthign outrageous. However, I suspect the reality is sticking to a 300w ceiling and performance is closer to the 22-25% improvement, so at least 10% behind 2080.

 

[TNA]

I'd hardly call it a chase, and this is no halo product contest. AMD are putting out a card with the performance levels we had access to nearly 2 years ago with the 1080Ti! That's all the 2080 is, with some extra tensor cores that aren't being used... which the Radeon VII obviously doesn't have. It's basically a 7nm 1080Ti, 2 years late. At the same price a 1080Ti was 2 years ago. Way to innovate AMD!

I don’t expect to see any innovation from AMD until Arcturus which seems like it will be closer 2021 now.

Navi might be bringing better price for performance to mid range, at least that is what people are saying, but I doubt it will be as good as what some are expecting, like say better than Vega 64 performance for £250.