Vega/Navi Rumors (Updated)

[Atreidin]

It says right in the image "Radeon RX Vega performance estimated based on architecture, shader count and clocks." This is nothing but rumors and speculation, not actual benchmark results.

Truly, Tweaktown has demonstrated the greatest level of journalistic integrity to write such an authoritative headline as "AMD Radeon RX Vega loses to GTX 1080 at 1080p" for an article based on pure speculation.

Oh, and with the smaller-font sub-heading (A.K.A. "more factual title") "AMD's next-gen Radeon RX Vega rumored to be slower than GTX 1080 at 1080p" as a kind of out for not being total liars.

 

[Malogeek]

Wait, it's not WCCFTech? But just as bad?

 

[keymaster151]

I think the formula that TechPowerUp used to estimate the performance is [RX 480 * (4096 / 2304) * (1200 / 1266)]. They've simply changed the shader count and clocks to what we've seen from leaks so far, and are assuming no architectural changes. Kind of odd that they would even have such an estimation up on their site, I don't know if they usually do this for unreleased products. Still for a site to spin a story based on this "info" with such a headline is pretty embarassing.

 

[CatMerc]

http://www.tweaktown.com/news/57313....it&utm_medium=twitter&utm_campaign=tweaktown

Every time this TPU database link is posted, a kitten gets murdered.

Seriously, it's based on the Compubench result from 2 months ago where the clockspeeds were 1GHz and 1.2GHz boost. This is likely an engineering sample and not at all representative of the final clocks. Why you ask? Because MI25 is 12.5 TFLOPS (25 half precision). This is literally the only confirmed information we have of Vega directly from AMD.

Stop spreading this, it tells us nothing, and gives traffic for websites who do zero research and regurgitate everything from less than stellar sources.

 

[Glo.]

I think some people got offended by that link.

This thread is called: Vega/Navi RUMORS.

Just from journalist duty it was worth reposting here, regardless of how clueless the article actually is.

 

[CatMerc]

I think some people got offended by that link.

This thread is called: Vega/Navi RUMORS.

Just from journalist duty it was worth reposting here, regardless of how clueless the article actually is.

I think only things that pass the most basic sniff test should be spread. Otherwise, ignore.

 

[parvadomus]

It says right in the image "Radeon RX Vega performance estimated based on architecture, shader count and clocks." This is nothing but rumors and speculation, not actual benchmark results.

Yep, and the architecture used to interpolate performance is Polaris lol
This is just a placeholder in TPU's database..

 

[Malogeek]

This thread is called: Vega/Navi RUMORS.

It's not even a rumor. It's one site badly extrapolating a potential speed of the GPU, actually allowing it to be public, and another site creating a click bait article from it. It's not journalism, and it's not anyone's duty to post it.

Would you share all the stupid clickbait stuff on Facebook everywhere?

 

[IEC]

So, even in the worst-case scenario (no architectural improvements at all and worse than Polaris clockspeeds) it would match a 1080.

I think that puts a surprisingly high ceiling for what the actual performance of Vega could be.

 

[Paul98]

I am quite excited to see Vega, it's the first major change since the introduction of GCN, it has been a long time. Major updates to almost every part of the GPU which greatly help GPU utilization, game performance, and power consumption. Many of the changes are to let the GPU get the same results with far less work and less resource use.

 

[Snarf Snarf]

The only thing in that article even worth talking about is the mention of extremely limited quantities of HBM2. If that holds any truth Vega could be the second coming of Jesus, but we won't be able to get our hands on it 🙁

 

[Glo.]

The only thing in that article even worth talking about is the mention of extremely limited quantities of HBM2. If that holds any truth Vega could be the second coming of Jesus, but we won't be able to get our hands on it 🙁

AMD has bought last year volume of some GPU parts, that they could not use at that time, because there was no tech available, for over 80 Mln USD. It has been said in one of their conference calls. At that time, the HBM2 2 Gbps chips were made.

Based on clues, people thought that that buyout was indeed about HBM2 memory stacks.

 

[Snarf Snarf]

AMD has bought last year volume of some GPU parts, that they could not use at that time, because there was no tech available, for over 80 Mln USD. It has been said in one of their conference calls. At that time, the HBM2 2 Gbps chips were made.

Based on clues, people thought that that buyout was indeed about HBM2 memory stacks.

We can hope, that with Lisa and Raja in charge, AMD actually had some foresight and decided to buy literally every HBM2 chip Hynix could pump out if they knew supplies were going to be extremely limited.

If they are going to have 4-5 SKU's featuring HBM2 it would take quite a bit of supply to get enough stock to have a good launch. I don't doubt that Vega is going to be a compelling product, I just have fears that the launch is going to be a rocky one, similar to x370 motherboards being limited supply. If partner cards take 2 months to come out with non reference designs on the air cooled Vega's that puts them dangerously close to having the wind taken out of their sails by Nvidia making a Volta announcement late summer. I have a hard enough time getting my customers to buy AMD products unless they're twice the performance at half the cost, even then they still look at me with puppy dog eyes and say "...but I love NVidia and Intel". I actually spent a solid 45 minutes to an hour talking a customer into a 1500x with an RX 470 instead of getting an i3 and a 1050ti, it's like pulling teeth with some people. /endrant

Ranting aside, even that TPU article shows that worst case scenarios aren't THAT bad for Vega. That assumption is based on 0 advancements from Polaris IPC and the clocks look terrible.

 

[Mockingbird]

I have a hard enough time getting my customers to buy AMD products unless they're twice the performance at half the cost, even then they still look at me with puppy dog eyes and say "...but I love NVidia and Intel". I actually spent a solid 45 minutes to an hour talking a customer into a 1500x with an RX 470 instead of getting an i3 and a 1050ti, it's like pulling teeth with some people. /endrant

That's bad business practice.

 

[guachi]

An article with a headline of "Vega worst case: 5% slower than a 1080" would be useful and less click bait.

If HBM2 really is in short supply then a 4GB model makes perfect sense. You can physically make more cards to sell. I wouldn't be surprised if the 8GB models were only for the top end and much more expensive.

Though if you could beat a 1080ti at 4k you could basically charge what you wanted.

 

[Glo.]

That's bad business practice.

May be bad business practice, but good consumer advice in grand scheme of things is great business practice.

 

[Snarf Snarf]

That's bad business practice.

Giving advice on how to get more performance for the same amount of money or less is bad business? By my tone of the post it might sound like I'm an AMD fan but if my customers need CUDA and AVX I'll build them an NVidia/Intel workstation. However when someone comes in with a "my friend who knows all about PC's said this is the build I want" and they have incorrect information, I give them the right information and direct them in the best way to spend their money. Most people are not like us on these forums, they have little technical knowledge, and usually have very tight budgets for PC's. I've never lost a customer before in the 5 years I've been doing this, so it has worked out well for us in the long run. Anyway this is super off topic so that's enough for this conversation.

So a question for everyone here, if Vega only excels at high resolutions and loses to a GTX 1080 at 1080p, how do you think that would effect the perception of the card to the market? I'll entertain Mockingbird's claim that its 1080 +10%, what if that is only at 1080p, if it matches 1080ti at 1440p and wins at 4k, will AMD be able to sell this card at $650?

 

[darkswordsman17]

I think some people got offended by that link.

This thread is called: Vega/Navi RUMORS.

Just from journalist duty it was worth reposting here, regardless of how clueless the article actually is.

Wait, so you're a journalist? Have you disclosed that before?

No, and that is absolutely reprehensible logic. Unfortunately there's far too many people who seem to be of the same mind, hence why we get false equivalencies because "I saw it on the news" so people think it has the weight of actual journalism behind it. Look I can even link the article, total proof!

The duty of using basic logic would dictate that you consider things before parroting them. Now if you want to debate the issues with the article, which you clearly didn't as you just slapped the link in and left it at that, and so others have to point out the issues. If you knew it was clueless then you needed to point that out when linking, or else you showed you didn't do another major "journalist duty" by vetting the information before spreading it.

In short, it is exactly stuff like this that causes this subforum to be a near constant mess, and fuels all these garbage sites spreading junk. And because people keep parroting it, it then becomes accepted by people, which then just screws everything up, because then it causes discourse to have to debate that alongside the actuality. Take Polaris where one side went substantially negative, and the other side went the other way, and then it basically ended up right in the middle of the two (which doesn't actually happen that much, hence why people never seem to learn lessons about speculating), and then we're still having discussions where people are citing the made-up stuff to prop up their argument or having to refute some accepted knowledge that was never actual knowledge but fabrication.

I'm really baffled, as you seem to generally not be someone with anti-AMD agenda (I'll stop short of saying pro-AMD, but I think you generally have a more positive take on AMD, or at least regularly refute some of the anti-AMD posts from certain members) and you generally discuss things more. Were you trying to come off "fair and balanced" or something?

 

[Jhatfie]

if it matches 1080ti at 1440p and wins at 4k, will AMD be able to sell this card at $650?

I would say no, at least not if they hope to sell in any real volume. They should be able to, but the at price and performance parity, Nvidia will still get a lions share of the market because of perception. AMD will need to offer obvious performance advantage to charge the same for most users to consider them or sell it for less if performance is similar.

I on the other hand would be in the likely minority and go AMD even at the same price, because history seems to show that over time their forward thinking architecture stands the test of time better.

 

[Mockingbird]

May be bad business practice, but good consumer advice in grand scheme of things is great business practice.

Giving suggestions is not the same as convincing to do something.

If he/she run into issues later on or don't like it, guess who he/she will blame?

 

[Mockingbird]

Giving advice on how to get more performance for the same amount of money or less is bad business? By my tone of the post it might sound like I'm an AMD fan but if my customers need CUDA and AVX I'll build them an NVidia/Intel workstation. However when someone comes in with a "my friend who knows all about PC's said this is the build I want" and they have incorrect information, I give them the right information and direct them in the best way to spend their money. Most people are not like us on these forums, they have little technical knowledge, and usually have very tight budgets for PC's. I've never lost a customer before in the 5 years I've been doing this, so it has worked out well for us in the long run. Anyway this is super off topic so that's enough for this conversation.

So a question for everyone here, if Vega only excels at high resolutions and loses to a GTX 1080 at 1080p, how do you think that would effect the perception of the card to the market? I'll entertain Mockingbird's claim that its 1080 +10%, what if that is only at 1080p, if it matches 1080ti at 1440p and wins at 4k, will AMD be able to sell this card at $650?

Suppose that I want an Intel Core and NVIDIA Geforce build and you instead insists on a AMD Ryzen and AMD Radeon build.

Three months down the road, I encounter issues with the build or just don't like it.

Are you really going to let me return it for an Intel Core and NVIDIA Geforce build?

 

[Ajay]

Man, if I were actually planning on buying Vega, vs just being interesting in the tech, I would have gone nutz by now. Stealthiest Radeon release evah!

 

[Glo.]

Wait, so you're a journalist? Have you disclosed that before?

No, and that is absolutely reprehensible logic. Unfortunately there's far too many people who seem to be of the same mind, hence why we get false equivalencies because "I saw it on the news" so people think it has the weight of actual journalism behind it. Look I can even link the article, total proof!

The duty of using basic logic would dictate that you consider things before parroting them. Now if you want to debate the issues with the article, which you clearly didn't as you just slapped the link in and left it at that, and so others have to point out the issues. If you knew it was clueless then you needed to point that out when linking, or else you showed you didn't do another major "journalist duty" by vetting the information before spreading it.

In short, it is exactly stuff like this that causes this subforum to be a near constant mess, and fuels all these garbage sites spreading junk. And because people keep parroting it, it then becomes accepted by people, which then just screws everything up, because then it causes discourse to have to debate that alongside the actuality. Take Polaris where one side went substantially negative, and the other side went the other way, and then it basically ended up right in the middle of the two (which doesn't actually happen that much, hence why people never seem to learn lessons about speculating), and then we're still having discussions where people are citing the made-up stuff to prop up their argument or having to refute some accepted knowledge that was never actual knowledge but fabrication.

I'm really baffled, as you seem to generally not be someone with anti-AMD agenda (I'll stop short of saying pro-AMD, but I think you generally have a more positive take on AMD, or at least regularly refute some of the anti-AMD posts from certain members) and you generally discuss things more. Were you trying to come off "fair and balanced" or something?

The problem people have is getting out of their minds(perception) and understanding that reposting something in a thread does not always may mean that the reposter agrees with what has been written in the reposted link.

I have open mind about Vega GPUs. So far, people who have seen the architecture tend to be positive about it, compared to Polaris. But I do not project my expectations over this architecture. I am just curious, because I consider myself a tech enthusiast and that should be absolutely clear by now.

 

[GodisanAtheist]

Suppose that I want an Intel Core and NVIDIA Geforce build and you instead insists on a AMD Ryzen and AMD Radeon build.

Three months down the road, I encounter issues with the build or just don't like it.

Are you really going to let me return it for an Intel Core and NVIDIA Geforce build?

-Not to just keep dog-piling the OP here but: I've done my time in a Microcenter Peripherals department and while its always a good idea to discover what someone actually wants and expects from their PC as well as to present multiple options, I never thought it was a good idea to "bring them to the light". If someone rejected the options I laid out for them (with my reasoning presented up front), then lets do what they want.

When that guy goes home and his PC doesn't boot, or doesn't play everything 4k/Ultra, or has a driver issue with a program (all of which happen regardless of vendor), it will be the sales associates/company's/AMD's fault this happened and if he had just gone with what his friend had told him he wouldn't be having the issue.

Not saying that's whats right, but I've seen it happen too many times.

 

[Paratus]

Wow. 5% slower than a 1080 at 1080P?

Makes no sense. 5% slower than a 1080 puts it at a bit over 60% faster than the RX 580, but with a 10% slower Boost clock.

If we assume a straight Polaris design that would put the die at 420mm^2. Which means the 500mm^2 die rumor is wrong. To keep all the rumors:

• 500mm^2
• 5% slower than 1080
• 1200mhz clock

It would have to have a xtor density 20% less than Polaris.

So basically they are saying Vega:

• Doesn't clock as high as Polaris let alone Pascal
• Is less die efficient than Polaris let alone Pascal

Seems like complete BS to me.