AMD reveals their CAPSAICIN SIGGRAPH event

[WYP]

AMD has revealed their CAPSAICIN SIGGRAPH event, where the Radeon Technologies Group will be showcasing their Radeon Vega family of products and more.



Read more on AMD's Capsaicin SIGGRAPH event.

 

[Radioman]

What should be just as interesting if not more so is the Quarterly Investor Conference call which should take place one week before this event. At that time we will be able to see what AMD's Financial Numbers look like with a full quarter of Ryzen Sales under their belts. It will also be interesting to see what AMD Senior Management has to say about the new Cryptocurrency Mining Cards. How will they organize their Production Lines.

It appears the RX Vega will slot in between the 1080 and the 1080ti in terms of performance. If that is the case it seems to me the logical thing to do is to price Vega as cheap as possible say $500. That way AMD can work 2 cohorts. The traditional AMD GPU Client base which has been buy Polaris Cards and possibly glean off some of the bottom end NVIDIA Base who are sick and tired of being fleeced. During the summer and spring AMD can produce and market the just announced Mining specific RX-470. With Polaris Cards gone from the stores and stock rooms for months RX Vega Cards should also do very well when they hit the shelves. I would expect Polaris Cards to be restocked no sooner than the middle of the 4th Quarter in time to have they on the shelves before the Holiday Rush.

And then comes Ryzen 3. Despite the recently linked specs on the Chip. I can't see AMD disabling perfectly good cores to satisfy this market. And AMD seems to concur. Reports are out of extraordinarily high yield numbers for Ryzen wafers. AMD has announced a new line of Ryzen Pro chips. It is clear that AMD wants to test and segregate the over performing cores and charge a premium price for them. This makes perfect sense considering Wall Street is always hammering AMD for catering to Low Margin Markets. The question is where will the Ryzen 3 Chips come from ?

Then there will be what AMD Senior Suits will say about Epyc. Like Ryzen in April AMD will only have 1 month of Epyc sales data to report. But how AMD guides for the 3rd and 4th quarters should give us some clues about what their strategy will be in 2018.

 

[Korreborg]

Where have you found info on 1080+ speed?
Everywhere I have looked it's between 1070 and 1080. As I see it = total fail. Those cards are 1 year old.
But this crypto mining will probably save them. It's amd and the power companies that have started the mining trend

 

[russler]

as much as id love to see above 1080 performance theres no info suggesting it, we will just have to wait and see hopefully not too long as im graphics card-less at the mo.

 

[Avet]

Only AMD advantage will be price/performance. If they offer Vega in 1060/70/80 segment at lower prices they will take big chunk of the market.

 

[NeverBackDown]

Where have you found info on 1080+ speed?
Everywhere I have looked it's between 1070 and 1080. As I see it = total fail. Those cards are 1 year old.
But this crypto mining will probably save them. It's amd and the power companies that have started the mining trend

Where have you found a review of a RX Vega card? Because all I see are Pro cards being benchmarked in games.

 

[Radioman]

Only AMD advantage will be price/performance. If they offer Vega in 1060/70/80 segment at lower prices they will take big chunk of the market.

AMD already has a big chunk of that market. 85% of all GPU Cards sold are in the 100 to 300 dollar range and AMD has that segment of the market cornered in much the same way as NVIDIA has the segment of the market cornered that furnishes Big Ticket GPUs. The profit margin for the Big Ticket Cards is much higher which explains the reason why NVIDIA rakes in the loot the way they do.

Returning to the topic at hand benchmarking the AMD Vega GPU against the NVIDIA 1080ti is a facile analysis. It would be great if Vega could make in roads into NVIDIAs customer base but it still can be very successful without achieving that goal. There is a very large cohort of people using Polaris Cards and a fair number want to upgrade. If priced properly, say 500 to 580 USD, Vega would give those users a cheaper alternative than the 1080 or 1080ti. A better comparison is the RX Vega against the old Fury X which opened at 1100 USD when placed on the market.

 

[Radioman]

Where have you found info on 1080+ speed?
Everywhere I have looked it's between 1070 and 1080. As I see it = total fail. Those cards are 1 year old.
But this crypto mining will probably save them. It's amd and the power companies that have started the mining trend

The data I'm using is from the PCPer tests that were done last Thursday. I'm going by their data not their editorial opinions.

BTW Red Gaming Tech expects a considerable jump in performance when the new Drivers optimized for games are released. That should take place in a few weeks when the RX Vega Cards are introduced at Siggraph.

It will be interesting to see what effects the Mining Craze will have on AMD and its planning. There are many things going down at Red right now and AMD Senior management is up to its eyeballs in alligators.

The future for AMD rests on the success of Epyc and the company's position in the Server/Enterprise Biz

 

[Korreborg]

Where have you found a review of a RX Vega card? Because all I see are Pro cards being benchmarked in games.

So it does not give a huge hint towards the gaming card?

 

[Dicehunter]

Where have you found a review of a RX Vega card? Because all I see are Pro cards being benchmarked in games.

Linus gave a bit of a hint in one of his recent videos when he talked about the performance and that in reality we shouldn't be expecting RX version to magically super charge the exact same cores and architecture as only really the memory size will be different will differently tuned drivers, I'm paraphrasing but that was the gist of it.

I have my money ready either way.

 

[Korreborg]

I have to ask. What exactly are you expecting.
As a company you can't ship a card at 1000$ and then ship a card that's better and cheaper. It just don't make sense. People felt sorry for the titan guys. A move like that a month after frontier launch would be much worse.

 

[Radioman]

I have to ask. What exactly are you expecting.
As a company you can't ship a card at 1000$ and then ship a card that's better and cheaper. It just don't make sense. People felt sorry for the titan guys. A move like that a month after frontier launch would be much worse.

Your argument presupposes the market for Vega FE is entirely focused on Gamers to the exclusion of all else when exactly the opposite is the case. Vega FE is target as an Entry Level Professional Design Tool. The Gaming Drivers supplied with the card are minimalist at best. Gamers Nexus did an engineering analysis video that was published Sunday Night and the hardware development of the card justifies the price being charged.

It's time for Gamers to take stock of the fact they are not the only market out there. As a matter of fact they are a fraction of a very small market.

 

[Korreborg]

No I don't. The argument is whether or not the gaming Vega will show FAR better gaming results than a frontier. And it would not make sense to not use the extra possible power in their 1000$ card if the performance is there. = I don't believe the RX Vega will be a lot better.
My guess: It will be minus 8gigs of HBM2 and maybe fewer cores + higher frequency. = performance between 1070 and 1080.
Hopefully close to 1080 at a good price = pressure on Nvidia.

 

[Dicehunter]

I have to ask. What exactly are you expecting.
As a company you can't ship a card at 1000$ and then ship a card that's better and cheaper. It just don't make sense. People felt sorry for the titan guys. A move like that a month after frontier launch would be much worse.

Personally I'm guesstimating we'll be getting a card with half the HBM and maybe 100-200MHz higher core clock with drivers tuned more for gaming rather than pro apps which is pretty much what AMD have said minus the clock increase and memory but it makes sense.

 

[os2wiz]

No I don't. The argument is whether or not the gaming Vega will show FAR better gaming results than a frontier. And it would not make sense to not use the extra possible power in their 1000$ card if the performance is there. = I don't believe the RX Vega will be a lot better.
My guess: It will be minus 8gigs of HBM2 and maybe fewer cores + higher frequency. = performance between 1070 and 1080.
Hopefully close to 1080 at a good price = pressure on Nvidia.

That is a rather inane conclusion totally ignoring the fact that the Vega FE card only had Fiji level drivers that did not take advantage of any new technology in the Vega card. The video bios as well is totally different than RX Vega. I have seen here and on other web sites too many flat liners who are either ill informed or purposely pushing garbage to damage RX Vega before it even gets out of the door. Shame on you.

 

[NeverBackDown]

That is a rather inane conclusion totally ignoring the fact that the Vega FE card only had Fiji level drivers that did not take advantage of any new technology in the Vega card. The video bios as well is totally different than RX Vega. I have seen here and on other web sites too many flat liners who are either ill informed or purposely pushing garbage to damage RX Vega before it even gets out of the door. Shame on you.

It's ok to shame people but you should definitely backup all the stuff you are claiming

 

[Radioman]

No I don't. The argument is whether or not the gaming Vega will show FAR better gaming results than a frontier. And it would not make sense to not use the extra possible power in their 1000$ card if the performance is there. = I don't believe the RX Vega will be a lot better.

Depending on whose observations you trust/respect there is a wide variety opinion on the matter. there are some opinions that driver changes alone could bring RX Vega fairly close to the 1080ti.

My guess: It will be minus 8gigs of HBM2 and maybe fewer cores + higher frequency. = performance between 1070 and 1080.
Hopefully close to 1080 at a good price = pressure on Nvidia.

Removing one of the HBM dies is a distinct possibility.

AMD doesn't put pressure on NVIDIA unless it beats NVIDIA. AMD and NVIDIA each have their own clientele. They are pretty much unique and exclusive with the exception of those AMD Customers that seek to upgrade out of the medium tier products that AMD offers. In this respect pricing of the RX Vega is far more critical than performance. If Vega represents a considerable jump in performance to an RX480 user for a price of 500 USD he or she will think long and hard before spending the extra 100 dollars for the 1080 if the RX Vega is the equal. Spending an extra 300 dollars to acquire a 1080i requires a pressing need and the extra money. There are those users that will "shoot-the-works" and buy the top of the line NVIDIA puts out but that is more of a marketing question than it is having anything to do with performance or hardware.

 

[AlienALX]

Personally I'm guesstimating we'll be getting a card with half the HBM and maybe 100-200MHz higher core clock with drivers tuned more for gaming rather than pro apps which is pretty much what AMD have said minus the clock increase and memory but it makes sense.

If you read the PCPER review you would have seen that the card throttled like a pig. 1600mhz clocks were a pipe dream and it guzzled down power too.

AMD have had an issue with temps for years. 290. 290x, 390, 390x.. Do I need to go on? their stock coolers just can not cope with their kitchen sink tech.

I do not expect a retail card with an air cooler on to be able to reach 1600mhz. The fan on Vega is pretty big too, and the entire card is alu but none of that seems to help. So you will be back to Fury X territory - water or nothing + £.

Honestly Dice, if you buy one of these things then you will need to swallow some serious humble pie. If their professional card can not even run at the advertised speeds and costs a grand how do you expect the retail one to run faster?

Nvidia.. Titan - 780 - practically identical.
Titan Black - 780Ti - practically identical.
Titan X (M) - 980Ti - practically identical.
Titan XP - 1080Ti - practically identical.

So how are AMD going to go from Vega Frontier - Vega and make up 20%?

They're not, dude. And clocking it faster will not make it run cooler either. Why do you think Vega comes with a AIO for £500 more?

 

[Dicehunter]

If you read the PCPER review you would have seen that the card throttled like a pig. 1600mhz clocks were a pipe dream and it guzzled down power too.

AMD have had an issue with temps for years. 290. 290x, 390, 390x.. Do I need to go on? their stock coolers just can not cope with their kitchen sink tech.

I do not expect a retail card with an air cooler on to be able to reach 1600mhz. The fan on Vega is pretty big too, and the entire card is alu but none of that seems to help. So you will be back to Fury X territory - water or nothing + £.

Honestly Dice, if you buy one of these things then you will need to swallow some serious humble pie. If their professional card can not even run at the advertised speeds and costs a grand how do you expect the retail one to run faster?

Nvidia.. Titan - 780 - practically identical.
Titan Black - 780Ti - practically identical.
Titan X (M) - 980Ti - practically identical.
Titan XP - 1080Ti - practically identical.

So how are AMD going to go from Vega Frontier - Vega and make up 20%?

They're not, dude. And clocking it faster will not make it run cooler either. Why do you think Vega comes with a AIO for £500 more?

All good points, I just want an all AMD rig ^_^

 

[NeverBackDown]

Are people still complaining about the Pro card? Like c'mon... Let's move on and look forward to RX Vega.