Introducing AMD Radeon RX Vega

[aeliusg]

That graphic is not confusing at all: it's a simple chart. Imagine if it had columns and check marks instead of points. Too bad Vega is a flop. I'm disappointed.

 

[Syntax_Error]

Sometimes I get the feeling that the entire point behind AMD is just letting some rich corporate folks play the stock market.

1. Buy big after a product launch.

2. Announce next "future-proof" product that truly "shines", and comes with some gimmicky crap that will "revolutionize" gaming forever.

3. Release vague info full of weasel words in the official forums and social media.

4. Unleash the army of fanboy drones in the tech forums to bolster hype with ridiculous promises of performance, and attack technically proficient members who can see the BS and dare to speak against it.

5. Then, sell right at the apex of the hype for the next one.

6. Rinse and repeat.

 

[Delicieuxz]

That graphic is not confusing at all: it's a simple chart. Imagine if it had columns and check marks instead of points. Too bad Vega is a flop. I'm disappointed.

I'm running a GTX 1070 with a 1440p FreeSync monitor. Getting a Vega 64 card will bump me up to GTX 1080 performance territory in DX 11, while giving me adaptive-sync, while also freeing up the additional GPU power that is currently going towards v-sync. Vega 64 is about perfect for me, except for the power usage, which, to me, is its one downside. But, I checked what the extra electricity that I'll be paying for with Vega 64 if I game 3 hours a day for a year (which is probably way more than I actually game) will cost me, and it worked out to being around only an extra $14 CND each year to power Vega 64 instead of my GTX 1070.

So, other than the power thing, Vega 64 looks like it's the sweet spot for me, getting me FreeSync with a bump up to GTX 1080 performance in DX 11, and presumably more in async compute situations.

 

[aeliusg]

I'm running a GTX 1070 with a 1440p FreeSync monitor. Getting a Vega 64 card will bump me up to GTX 1080 performance territory in DX 11, while giving me adaptive-sync, while also freeing up the additional GPU power that is currently going towards v-sync. Vega 64 is about perfect for me, except for the power usage, which, to me, is its one downside. But, I checked what the extra electricity that I'll be paying for with Vega 64 if I game 3 hours a day for a year (which is probably way more than I actually game) will cost me, and it worked out to being around only an extra $14 CND each year to power Vega 64 instead of my GTX 1070.

So, other than the power thing, Vega 64 looks like it's the sweet spot for me, getting me FreeSync with a bump up to GTX 1080 performance in DX 11, and presumably more in async compute situations.

I can see that working out if you can get it for MSRP and are able to sell your 1070 at miner prices. That might be easier said than done, though.

 

[defaultluser]

Here is another performance indicator for RX Vega 64:

Of course, this will be subject to the actual reviewers recreating these results, much like the Fury X had tough numbers for reviewers to recreate (or settings that made no sense, purposefully selected to highlight the strengths of architecture). And in this very short list of games, you may not get a realistic feel for performance.

All this time nobody will call the FE results "final" because "it's not a gaming card." So tired of this bullshit wait - just release the cards already with their current broken drivers and watch the blood fly.

First June, then July, and now mid-August for actual card availability. Supposedly.

It's not a part availability problem. You can buy an FE right now on Newegg. Just read the reviews about the poor drivers to get a feel for why this isn't sold-out. Despite this being the "best-value" pure compute card on the market.

https://www.newegg.com/Product/Product.aspx?Item=N82E16814105073

Only a value if your time spent troubleshooting is worth nothing.

 

[FrgMstr]

Broke your foot? I want to see pictures of the computer which took the kick.

You got a telescope?

 

[Quartz-1]

I don't believe the higher power requirement of Vega will be an issue. Or rather, it will only be an issue if it forces the purchase of a new PSU.

 

[jbc029]

Here is another performance indicator for RX Vega 64:

No. Just no. A Fury X is not 20-30% faster than a 980ti. The numbers thrown up here are definitely not minimums, those are averages for 1440p UW gaming, and regarding a Fury X, they're being generous. WTF is with the Fury X *matching* a Vega 64 in BF1 and beating a 1080? I *know* that isn't true for a 1080 and I hope it isn't for the Vega. So much fail in this slide.

 

[Delicieuxz]

No. Just no. A Fury X is not 20-30% faster than a 980ti. The numbers thrown up here are definitely not minimums, those are averages for 1440p UW gaming, and regarding a Fury X, they're being generous. WTF is with the Fury X *matching* a Vega 64 in BF1 and beating a 1080? I *know* that isn't true for a 1080 and I hope it isn't for the Vega. So much fail in this slide.

The slide says the numbers show the 99% minimum FPS. This image shows that the difference between DX 11 and DX 12 for a Fury X running Battlefield 1 can indeed be 20%+.

Ashes of the Singularity, Battlefield 1, Deus Ex MD, Forza, are all async-compute titles, so the AMD cards have a big advantage in them. If those were average FPS and Vega 64 was basically matching the GTX 1080 in async-compute heavy games, then something would be very wrong and Vega 64 would be no better than a GTX 1070, at best.


Edit: I just found out that what I though minimum % means is incorrect. 99% low would mean the average FPS for the lowest 99% of all FPS encountered during gaming. So, it's basically an average of all the FPS except for the occasional maximum peaks. That looks to me like it means that Vega doesn't have a big advantage over the GTX 1080 in async-compute games... and I wonder what that says about Vega 64's performance.

 

[bbf]

I'm running a GTX 1070 with a 1440p FreeSync monitor. Getting a Vega 64 card will bump me up to GTX 1080 performance territory in DX 11, while giving me adaptive-sync, while also freeing up the additional GPU power that is currently going towards v-sync. Vega 64 is about perfect for me, except for the power usage, which, to me, is its one downside. But, I checked what the extra electricity that I'll be paying for with Vega 64 if I game 3 hours a day for a year (which is probably way more than I actually game) will cost me, and it worked out to being around only an extra $14 CND each year to power Vega 64 instead of my GTX 1070.

So, other than the power thing, Vega 64 looks like it's the sweet spot for me, getting me FreeSync with a bump up to GTX 1080 performance in DX 11, and presumably more in async compute situations.

Definitely since you already have a Freesync capable monitor, the Vega 64 upgrade option looks good.

For me, I already have a real 10-bit 4K 60Hz monitor, so will not be replacing that anytime soon, my only option was to get a card that could do 60fps consistently, and that meant a GTX 1080ti. This is from a guy that has been with ATI since at least the HD 4850's... it's been a while... my memory's fuzzy on what cards were in my system prior to that... but my first ATI card was the EGA Wonder, then I went for some other brands, then nVidia and then back to ATi for a decade or so and just recently back to nVidia. So I am definitely *not* an nVidia fanboi.

All nVidia has to do to put the final nails into ATi's coffin is to lower licensing costs on its G-Sync technology so G-Sync monitors will be within $25 of freesync ones. Or just support freesync on GTX 1070 and lower GPUs.

 

[Sith'ari]

The graph isn't ranking FPS, it's just showing how many titles each card manages to get above 48 FPS minimum (the right-side graph), or for which the minimum FPS is less than 48 FPS. The ordering of the GPU marks isn't by FPS, but like this in each of the graphs: GTX 980 Ti, GTX 1080, Fury X, RX Vega 64.

Each of the cards has a placement for each of the games in one of the two graphs. If a card isn't listed for a game on one graph, it's because it's listed on the other graph. If the card has a minimum FPS of less than 48, it will be on the left-side graph. If its minimum FPS is more than 48, it will be shown on the right-side graph.

Some thoughts about this AMD chart you posted:
These charts are misleading:
1. Check pcper's comment: ""AMD’s testing ignores the fact that NVIDIA G-Sync users don’t worry about whether or not their display integrates LFC (low frame rate compensation). All G-Sync displays have effective 0 FPS minimum frame rate (meaning they don’t regress to tearing/stutter below some panel threshold). So, the GTX 1080, even when it runs below the 40 FPS mark on the created situation AMD shows above, will not stutter, meaning they still have a “good” gaming experience.""" ( https://www.pcper.com/reviews/Graph...ga-64-and-Vega-56-Specs-Prices-Power-Detailed )
2. Also, equally important: EVEN IF the minimum numbers that AMD shows at this graph are correct, this doesn't mean anything, and can be misleading info.
Check Gamer's Nexus video about 1% low & 0,1% low, in which, it explains why there is a need to measure the average-from the lowest framerates, instead from only 1 minimum value (*which may occur only 1 time in a game and never happen again)