AMD Vega MegaThread! FAQ and Resources

[goldstone77]

Is Nvidia a threat to Intel. INTC stock $46.52 vs. NVDA stock $209.53 2:02 p.m. EST. The future of the world is A.I., and I've kind of been hitting this hard in my commentary over the last year with the projection of strong growth in AR/VR over the next 5 years. Over the next 5 years we will see an integration of AR/VR into the commercial sector. This will increase the demand for video cards, which Nvidia controls the market. I personally believe that within the next 10 years we should see a push from conventional cell phones into AR glasses. Cell phones will reach a point at which they won't offer anything really new, and people will upgrade less often. We seen this trend with desktop computing. IMO AR glasses will be the new thing. Intel can deflate Nvidia to a certain degree by using AMD instead of Nvidia for graphics. The problem is that I feel the shift is too late. Intel has already helped create the monster, and I think it's spread into sectors beyond Intel's control. Intel is playing catch up at this point.

We are starting to get off topic here, but very importantly you cannot compare two businesses via their stock price.
Using a real measure of the size of a business - Market capitalization, Intel is nearly double the size of Nvidia.

While many of your points are valid, Intel has more than enough R&D resources to find a way to make it work. They are behind, but they have the resources and money to fix it, in time.

In terms of looking at companies via the stock market the stock price is the first thing you see when comparing any company. Stock price and the history of it is the basis for all technical charting. That said I can't argue 213.31B market cap of Intel vs. 129.68B market cap of Nvidia. Intel is the larger company and has billions of dollars on hand, but it also has a somewhat checkered past when it comes to acquisitions that have led to waste of 10's of billions of dollars and nothing to show for it. There are a lot of factors that come into play when evaluating a company stock price is one of many.

 

[Rogue Leader]

Is Nvidia a threat to Intel. INTC stock $46.52 vs. NVDA stock $209.53 2:02 p.m. EST. The future of the world is A.I., and I've kind of been hitting this hard in my commentary over the last year with the projection of strong growth in AR/VR over the next 5 years. Over the next 5 years we will see an integration of AR/VR into the commercial sector. This will increase the demand for video cards, which Nvidia controls the market. I personally believe that within the next 10 years we should see a push from conventional cell phones into AR glasses. Cell phones will reach a point at which they won't offer anything really new, and people will upgrade less often. We seen this trend with desktop computing. IMO AR glasses will be the new thing. Intel can deflate Nvidia to a certain degree by using AMD instead of Nvidia for graphics. The problem is that I feel the shift is too late. Intel has already helped create the monster, and I think it's spread into sectors beyond Intel's control. Intel is playing catch up at this point.

We are starting to get off topic here, but very importantly you cannot compare two businesses via their stock price.
Using a real measure of the size of a business - Market capitalization, Intel is nearly double the size of Nvidia.

While many of your points are valid, Intel has more than enough R&D resources to find a way to make it work. They are behind, but they have the resources and money to fix it, in time.

In terms of looking at companies via the stock market the stock price is the first thing you see when comparing any company. Stock price and the history of it is the basis for all technical charting. That said I can't argue 213.31B market cap of Intel vs. 129.68B market cap of Nvidia. Intel is the larger company and has billions of dollars on hand, but it also has a somewhat checkered past when it comes to acquisitions that have led to waste of 10's of billions of dollars and nothing to show for it. There are a lot of factors that come into play when evaluating a company stock price is one of many.

First thing you see, but again you cannot compare two stock prices directly. The stock price in relation to what the price was previously is the basis for some technical charting, not all. Correct that its one of many metrics, however you specifically were calling Nvidia a threat to Intel based on their stock price, which again has nothing to do with the size of the company. Theres any number of factors, Market Cap being the best to compare size. Revenue, EBITA, Net Income, Enterprise Value, to compare performance and profitability, and a whole bunch of other things that make more sense to compare the two than stock price. Its almost an arbitrary number. In fact at my prior companies our competitor analysis barely featured stock price in any comparison. Stock price was used as a basis for some metrics using its relation to itself, as in percentage increase and decrease related to prior time periods.

 

[goldstone77]

Spoiler

Is Nvidia a threat to Intel. INTC stock $46.52 vs. NVDA stock $209.53 2:02 p.m. EST. The future of the world is A.I., and I've kind of been hitting this hard in my commentary over the last year with the projection of strong growth in AR/VR over the next 5 years. Over the next 5 years we will see an integration of AR/VR into the commercial sector. This will increase the demand for video cards, which Nvidia controls the market. I personally believe that within the next 10 years we should see a push from conventional cell phones into AR glasses. Cell phones will reach a point at which they won't offer anything really new, and people will upgrade less often. We seen this trend with desktop computing. IMO AR glasses will be the new thing. Intel can deflate Nvidia to a certain degree by using AMD instead of Nvidia for graphics. The problem is that I feel the shift is too late. Intel has already helped create the monster, and I think it's spread into sectors beyond Intel's control. Intel is playing catch up at this point.

We are starting to get off topic here, but very importantly you cannot compare two businesses via their stock price.
Using a real measure of the size of a business - Market capitalization, Intel is nearly double the size of Nvidia.

While many of your points are valid, Intel has more than enough R&D resources to find a way to make it work. They are behind, but they have the resources and money to fix it, in time.

In terms of looking at companies via the stock market the stock price is the first thing you see when comparing any company. Stock price and the history of it is the basis for all technical charting. That said I can't argue 213.31B market cap of Intel vs. 129.68B market cap of Nvidia. Intel is the larger company and has billions of dollars on hand, but it also has a somewhat checkered past when it comes to acquisitions that have led to waste of 10's of billions of dollars and nothing to show for it. There are a lot of factors that come into play when evaluating a company stock price is one of many.

First thing you see, but again you cannot compare two stock prices directly. The stock price in relation to what the price was previously is the basis for some technical charting, not all. Correct that its one of many metrics, however you specifically were calling Nvidia a threat to Intel based on their stock price, which again has nothing to do with the size of the company. Theres any number of factors, Market Cap being the best to compare size. Revenue, EBITA, Net Income, Enterprise Value, to compare performance and profitability, and a whole bunch of other things that make more sense to compare the two than stock price. Its almost an arbitrary number. In fact at my prior companies our competitor analysis barely featured stock price in any comparison. Stock price was used as a basis for some metrics using its relation to itself, as in percentage increase and decrease related to prior time periods.

Stock price that is stagnant or depreciating vs. a stock price showing tremendous growth speaks volumes about a company at a glance! Companies may or may not want to talk about their stock price for various reasons especially when comparing to it's competitors. Let me show you why I used stock price as a comparison (which I use on a daily basis for comparison of companies I might want to invest in), and it's relevance in my forward looking statements on A.I. and AR/VR. Below are 1 year and 5 year price comparisons. This tells me that there is massively better growth, and positive sentiment for Nvidia compared that of Intel. The stock price has been on a tear since 2016. Knowing both companies, and Intel's reaction to jump into the GPU market is no surprise when looking at these charts. To put is simple looking at Nvidia's stock price we can see that the future is GPU's, and everyone is betting on them.

 

[Rogue Leader]

Spoiler

Is Nvidia a threat to Intel. INTC stock $46.52 vs. NVDA stock $209.53 2:02 p.m. EST. The future of the world is A.I., and I've kind of been hitting this hard in my commentary over the last year with the projection of strong growth in AR/VR over the next 5 years. Over the next 5 years we will see an integration of AR/VR into the commercial sector. This will increase the demand for video cards, which Nvidia controls the market. I personally believe that within the next 10 years we should see a push from conventional cell phones into AR glasses. Cell phones will reach a point at which they won't offer anything really new, and people will upgrade less often. We seen this trend with desktop computing. IMO AR glasses will be the new thing. Intel can deflate Nvidia to a certain degree by using AMD instead of Nvidia for graphics. The problem is that I feel the shift is too late. Intel has already helped create the monster, and I think it's spread into sectors beyond Intel's control. Intel is playing catch up at this point.

We are starting to get off topic here, but very importantly you cannot compare two businesses via their stock price.
Using a real measure of the size of a business - Market capitalization, Intel is nearly double the size of Nvidia.

While many of your points are valid, Intel has more than enough R&D resources to find a way to make it work. They are behind, but they have the resources and money to fix it, in time.

In terms of looking at companies via the stock market the stock price is the first thing you see when comparing any company. Stock price and the history of it is the basis for all technical charting. That said I can't argue 213.31B market cap of Intel vs. 129.68B market cap of Nvidia. Intel is the larger company and has billions of dollars on hand, but it also has a somewhat checkered past when it comes to acquisitions that have led to waste of 10's of billions of dollars and nothing to show for it. There are a lot of factors that come into play when evaluating a company stock price is one of many.

First thing you see, but again you cannot compare two stock prices directly. The stock price in relation to what the price was previously is the basis for some technical charting, not all. Correct that its one of many metrics, however you specifically were calling Nvidia a threat to Intel based on their stock price, which again has nothing to do with the size of the company. Theres any number of factors, Market Cap being the best to compare size. Revenue, EBITA, Net Income, Enterprise Value, to compare performance and profitability, and a whole bunch of other things that make more sense to compare the two than stock price. Its almost an arbitrary number. In fact at my prior companies our competitor analysis barely featured stock price in any comparison. Stock price was used as a basis for some metrics using its relation to itself, as in percentage increase and decrease related to prior time periods.

Stock price that is stagnant or depreciating vs. a stock price showing tremendous growth speaks volumes about a company at a glance! Companies may or may not want to talk about their stock price for various reasons especially when comparing to it's competitors. Let me show you why I used stock price as a comparison (which I use on a daily basis for comparison of companies I might want to invest in), and it's relevance in my forward looking statements on A.I. and AR/VR. Below are 1 year and 5 year price comparisons. This tells me that there is massively better growth, and positive sentiment for Nvidia compared that of Intel. The stock price has been on a tear since 2016. Knowing both companies, and Intel's reaction to jump into the GPU market is no surprise when looking at these charts. To put is simple looking at Nvidia's stock price we can see that the future is GPU's, and everyone is betting on them.

I agree however thats not what you said. You specifically compared the days closing price of Intel vs Nvidia, without any other context, or discussion of growth or anything else. Just stock price. Growth charts? Great comparison, as I said.

Either way we are WAY off topic at this point.

 

[BurgerandChips66]

Hot on the heels of that XFX card preview the other week comes this... https://videocardz.com/73947/sapphire-radeon-rx-vega-64-nitro-pictured-and-tested

 

[renz496]

3 8 pin.....Vega really needs that. i saw Vega at 100% fan speed with load temp below 60c and still can't even maintain stable clock. the LC version have much more stable clock not because of better cooling...but because of the power ceiling is higher.

 

[BurgerandChips66]

This version is a bit excessive however!

 

[-Fran-]

This version is a bit excessive however!

Hahaha, I have to say that made me laugh really loud here at the office.

That would be a proper heater alright... What is that, like 2KW?

I think "Photoshop Edition" is a very good marketing term for it.

Cheers!

 

[goldstone77]

I love it! That's gave me quite a chuckle!

 

[TechyInAZ]

LOL that's good. Then the little 6 pin at the very end LOLz.

 

[BurgerandChips66]

Here's another custom Vega, from Power Color...

https://videocardz.com/74007/powercolor-radeon-rx-vega-64-red-devil-pictured

So far that makes Asus, Gigabyte, Sapphire, XFX and Power Color all with custom Vega cards. MSI also have their 'Air Boost' model, if you can call that a custom design. Though none are actually available yet...

 

[goldstone77]

AMD’s Adrenalin overlay makes Shadowplay look like you're capturing on a wax tablet
By
Dave James

Spoiler

The AMD Radeon Adrenalin update is here, giving the Radeon GPU software it’s now customary yearly update. Don’t worry, they haven’t abandoned the whole Radeon-means-red stuff, moving on from the Crimson and Crimson ReLive releases we’ve now got a software update named after a deep red rose. We’ve also got a mobile app and a rather sexy game overlay too.
We’re not talking about a Radeon Crimson ReLive Redux release - thankfully that was just a wee joke from the AMD Vega launch that seemed to have gotten out of hand. It would have been a terrible name, “fire that marketing guy!” as AMD’s Sasa Marinkovic, head of software marketing at AMD said when we were introduced to the new Radeon Adrenalin.

Named after the Adrenalin rose, because AMD are already starting to run dry of cool-sounding shades of red (and no, they can’t use Ruby because someone else has already bagged that one), the new Radeon Software Adrenalin update is a mixture of little changes, some bigger updates, and a whole new slab of extra functionality.
The headline-grabbing features surround the accessibility of the powerful ReLive capture and streaming feature, which they introduced last year. The main one being the impressive-looking overlay which sits on top of your game and gives you access to a whole host of settings, capture, and performance metrics with which to play.

It allows you to screenshot, record, and stream your gameplay without having to alt-tab out of the action in the same way the much more basic original Crimson ReLive overlay did.

But it’s the instant, direct access to much of AMD’s graphics enhancement technology, such as Radeon Chill and FreeSync, that makes the Adrenalin overlay so useful. It’s all saved on a per-game basis, so any games that don’t play nice with FreeSync (there are a few) or ones that you don’t want to limit with Chill, can be set up with their own profile in just a few clicks. It's far easier than remembering to dig into the Radeon Setttings desktop app before you dive into a game.

Radeon Chill, the algorithmic feature which monitors what's happening in-game and limits frame rate, and therefore power draw, when it's not needed has itself had a rather big update. It used to operate on a whitelist basis, where compatible games were added, allowing you to enable Chill. Now it's switched around so that every game is deemed compatible unless it's added to a blacklist. AMD have been working their way through a host of games and so far the blacklist remains empty.

Chill is an impressive, borderline invisible bit of software which can really save on power and also heat generation. It really ought to be on by default but, as an AMD rep explained, people simply don't like stuff enabled by default.
For us hardware nerds it also gives us access to a bunch of performance metrics too. These can be overlayed themselves during gameplay, with or without them showing up in any captured footage you record. You can also log the performance too, giving you figures for GPU, CPU, and RAM utilisation, temperature, GPU and memory clockspeeds, GPU power, fan speed, and yes, that all-important frames per second figure. You can set the sampling interval as well, from one to ten seconds.
Unfortunately the only thing which doesn’t work on a per-game basis is the colouring option. With the Radeon Adrenalin overlay you can adjust the colour (sorry, color) settings and see instantly how it will look in-game. Sadly this is a global setting, so if you happen to like a pink hue to your Wolfenstein gibbing that will stick around once you withdraw to the comparative safety of your Windows desktop.

We've tested the overlay in pre-release form and it's been remarkably stable for the most part. We've had a few issues where it wouldn't function in certain games, and only a single crash. But the capture is solid, barely seems to impact on gaming frame rates, and the performance metrics collection is going to be seriously helpful for us.

AMD have really moved forward with the functionality and stability of their software releases since Catalyst Omega, and it shows.

The new overlay also has an off-site companion in the shape of the AMD Link app. This is an Android or iOS application which ties into your PC (so long as both it and the phone are connected to the same network) and gives you both insight and control away from your PC.

The connection is super-easy and just requires you to point the app at a QR code and the rest is done for you. Then you’ll have real-time monitoring of your system (mostly on the graphics side, obvs.), whether you want to have an overview or a constant single-screen look at the FPS or GPU clockspeed.
The app will also allow you to externally control capture of video and screenshots, as well as starting a broadcast with a single tap. By default the app stops the mobile device’s screen from going to sleep, which means you can have it sat by your keyboard with a big green button waiting to be tapped to start your stream or capture.

Who wants to remember the complicated digital gymnastics needed to nail the various shortcuts?
It's most definitely a gen one version of the application, and you can see where AMD can go from here - adding in customisable full-screen buttons, and streamed playback from your library of your files, for example. But it's not without it's issues. I had to continually reconnect the app as it wouldn't automatically sync when it was opened, and honestly it's more for the streamer than anyone else.

But this all speaks to AMD’s desire to work with the huge streaming community, and not just the casual streamers either. They’ve entered a partnership with Stage TEN to use the ReLive output as a stream in their browser-based live mixer. It’s a cloud-based setup which essentially works like a mixing desk, letting you flow pre-recorded and live streaming content together.
They’ve also enabled live chat integration for your ReLive stream, so you can see the horrendous abuse you’re getting for your gaming skills overlayed onto your screen. Like all good interweb chats it’s a one-way thing so you can’t actually respond to the torrent of hate that will come your way from the Twitch, Facebook, Mixer, or YouTube chat integration.

AMD have also split the audio tracks for voice and game, a much-needed feature of ReLive for streamers, and have brought in chroma key support for transparencies and borderless region capture support too. Most importantly though, AMD have also cut down the performance overheads associated with the ReLive capture, which means streamers’ rigs aren’t going to be as likely to drop frames as a result of the stream.

The update is out now, and anyone with a Radeon graphics card should absolutely get themselves over to AMD's download site and get the Adrenalin installed. If you're an AMD gamer, and/or hardware nerd it's a joy, highlighting and making the impressive Radeon tech more accessible, and if you're a streamer it's a must.

The app, well, that's going to be more of a niche offering.

 

[BurgerandChips66]

https://videocardz.com/74474/sapphire-launches-rx-vega-nitro-series

Here at last! I almost always prefer Sapphire cards over the competition. Might give your wallet a heart-attack though...

 

[-Fran-]

Even though that it's a great piece of hardware, it's heart is still a bad purchase IMO. If only Sapphire would make a 1080 or a 1080ti with that much care, it would sell out in a single second. AMD really needs to put out something that is worthy of their OEMs like Sapphire that really put pieces of hardware like that out there.

Cheers!

 

[Martell1977]

Even though that it's a great piece of hardware, it's heart is still a bad purchase IMO. If only Sapphire would make a 1080 or a 1080ti with that much care, it would sell out in a single second. AMD really needs to put out something that is worthy of their OEMs like Sapphire that really put pieces of hardware like that out there.

Cheers!

Hard to say Vega is bad considering it has higher IPC than nVidia's offerings, it's just power hungry. I've always read that the excess power is due to some extra resources AMD bakes in, which is also why they do so well in productivity.

As for Sapphire, they really are one of the best. Whenever recommending an AMD card, I tell people to get Sapphire or ASUS. Just can't beat they quality and customer service. XFX is a third choice, but only if they are cheaper.

 

[goldstone77]

Even though that it's a great piece of hardware, it's heart is still a bad purchase IMO. If only Sapphire would make a 1080 or a 1080ti with that much care, it would sell out in a single second. AMD really needs to put out something that is worthy of their OEMs like Sapphire that really put pieces of hardware like that out there.

Cheers!

Hard to say Vega is bad considering it has higher IPC than nVidia's offerings, it's just power hungry. I've always read that the excess power is due to some extra resources AMD bakes in, which is also why they do so well in productivity.

As for Sapphire, they really are one of the best. Whenever recommending an AMD card, I tell people to get Sapphire or ASUS. Just can't beat they quality and customer service. XFX is a third choice, but only if they are cheaper.

XFX honored their lifetime warranty! Never had a problem with XFX.

 

[Rogue Leader]

Vega still has the top single card score on our superposition thread. It took TWO 1070's overclocked to beat it, and not by much

http://www.tomshardware.com/forum/id-3536560/tom-superposition-thread.html

The raw power is there. I got that score by enabling HBCC as well.

 

[goldstone77]

Vega still has the top single card score on our superposition thread. It took TWO 1070's overclocked to beat it, and not by much

http://www.tomshardware.com/forum/id-3536560/tom-superposition-thread.html

The raw power is there. I got that score by enabling HBCC as well.

5035!!!! Damn, that's smoking the 1080's with 7700s.

 

[Martell1977]

Even though that it's a great piece of hardware, it's heart is still a bad purchase IMO. If only Sapphire would make a 1080 or a 1080ti with that much care, it would sell out in a single second. AMD really needs to put out something that is worthy of their OEMs like Sapphire that really put pieces of hardware like that out there.

Cheers!

Hard to say Vega is bad considering it has higher IPC than nVidia's offerings, it's just power hungry. I've always read that the excess power is due to some extra resources AMD bakes in, which is also why they do so well in productivity.

As for Sapphire, they really are one of the best. Whenever recommending an AMD card, I tell people to get Sapphire or ASUS. Just can't beat they quality and customer service. XFX is a third choice, but only if they are cheaper.

XFX honored their lifetime warranty! Never had a problem with XFX.

I've had a few XFX cards and my only complaint has been badly applied thermal paste. My Sapphire and ASUS cards have always been rock solid.

 

[renz496]

Even though that it's a great piece of hardware, it's heart is still a bad purchase IMO. If only Sapphire would make a 1080 or a 1080ti with that much care, it would sell out in a single second. AMD really needs to put out something that is worthy of their OEMs like Sapphire that really put pieces of hardware like that out there.

Cheers!

Hard to say Vega is bad considering it has higher IPC than nVidia's offerings, it's just power hungry. I've always read that the excess power is due to some extra resources AMD bakes in, which is also why they do so well in productivity.

with different architecture it is hard to tell which one have higher IPC. people often said AMD lower clock compared to nvidia for the same performance is proof AMD have much higher IPC but people ignore everything else. sure RX480/580 have much lower clock than GTX1060 but those RX also have more SP (2304 vs 1280) and die size wise polaris also bigger than GP106. with pascal nvidia able to push the clock over 2Ghz from maxwell that usually hit 1.4-1.5Ghz. nvidia using this high clock speed advantage to reduce the amount of CUDA cores for roughly the same performance. prime example is GTX980 vs GTX1060. IPC wise maxwell and pascal should be outright identical according to anandtech. but because of much higher clock speed GTX1060 was able to reach 980 performance with much less CUDA cores (2048 vs 1280). so with pascal nvidia was able to reduce their die size not only with node shrink but also because of the high clock speed.

 

[-Fran-]

Even though that it's a great piece of hardware, it's heart is still a bad purchase IMO. If only Sapphire would make a 1080 or a 1080ti with that much care, it would sell out in a single second. AMD really needs to put out something that is worthy of their OEMs like Sapphire that really put pieces of hardware like that out there.

Cheers!

Hard to say Vega is bad considering it has higher IPC than nVidia's offerings, it's just power hungry. I've always read that the excess power is due to some extra resources AMD bakes in, which is also why they do so well in productivity.

with different architecture it is hard to tell which one have higher IPC. people often said AMD lower clock compared to nvidia for the same performance is proof AMD have much higher IPC but people ignore everything else. sure RX480/580 have much lower clock than GTX1060 but those RX also have more SP (2304 vs 1280) and die size wise polaris also bigger than GP106. with pascal nvidia able to push the clock over 2Ghz from maxwell that usually hit 1.4-1.5Ghz. nvidia using this high clock speed advantage to reduce the amount of CUDA cores for roughly the same performance. prime example is GTX980 vs GTX1060. IPC wise maxwell and pascal should be outright identical according to anandtech. but because of much higher clock speed GTX1060 was able to reach 980 performance with much less CUDA cores (2048 vs 1280). so with pascal nvidia was able to reduce their die size not only with node shrink but also because of the high clock speed.

I was going to give an argument along these lines as well.

The only thing I could add is, and it is unfortunate, that AMD has failed time and time again to make developers use their "special sauce" and optimize for it. So, even if it is there and you can get great benchmark numbers, it is moot if it is not really used in games. Or, if it is used, it doesn't add much to the overall image processing itself.

I do acknowledge saying it is a "bad" GPU is absolutely unfair, but you have to still compare them to their respective contenders at the price points being sold. In the context of the Sapphire Nitro+ Vega, it's just a bad purchase from the performance standpoint, even if it is a great piece of technology.

Cheers!

 

[Martell1977]

with different architecture it is hard to tell which one have higher IPC. people often said AMD lower clock compared to nvidia for the same performance is proof AMD have much higher IPC but people ignore everything else. sure RX480/580 have much lower clock than GTX1060 but those RX also have more SP (2304 vs 1280) and die size wise polaris also bigger than GP106. with pascal nvidia able to push the clock over 2Ghz from maxwell that usually hit 1.4-1.5Ghz. nvidia using this high clock speed advantage to reduce the amount of CUDA cores for roughly the same performance. prime example is GTX980 vs GTX1060. IPC wise maxwell and pascal should be outright identical according to anandtech. but because of much higher clock speed GTX1060 was able to reach 980 performance with much less CUDA cores (2048 vs 1280). so with pascal nvidia was able to reduce their die size not only with node shrink but also because of the high clock speed.

Actually, it's no different than the comparison between Intel and AMD, but in that case AMD is behind. People love pounding the IPC of Intel over AMD's heads. However, Intel's IPC is higher and the higher clock speed is just icing on the cake. At the same clocks, Intel is still ahead. I have a feeling that, at the same clock, nVidia would lose handily. Considering they need to use the extra clock speed to do the same work.

Yes nVidia has a smaller die, but they need that extra speed to keep up. AMD went with a lower clock speed and more resources. If they could get the power budget under better control, AMD would still get slammed, people never give them a break. If not hammering them about power usage, they dredge up years old driver issues...like nVidia have a perfect record...

I'm not an AMD fanboy, but my experiences between the AMD and nVidia cards I've had, AMD has been much better. And at this time, AMD seems to have some advantages over nVidia, just not when comparing power usage. AMD seems to be trying to get it under control, but it seems they might have to strip down some of those extra resources to do it...like nVidia did.

 

[TMTOWTSAC]

with different architecture it is hard to tell which one have higher IPC. people often said AMD lower clock compared to nvidia for the same performance is proof AMD have much higher IPC but people ignore everything else. sure RX480/580 have much lower clock than GTX1060 but those RX also have more SP (2304 vs 1280) and die size wise polaris also bigger than GP106. with pascal nvidia able to push the clock over 2Ghz from maxwell that usually hit 1.4-1.5Ghz. nvidia using this high clock speed advantage to reduce the amount of CUDA cores for roughly the same performance. prime example is GTX980 vs GTX1060. IPC wise maxwell and pascal should be outright identical according to anandtech. but because of much higher clock speed GTX1060 was able to reach 980 performance with much less CUDA cores (2048 vs 1280). so with pascal nvidia was able to reduce their die size not only with node shrink but also because of the high clock speed.

Actually, it's no different than the comparison between Intel and AMD, but in that case AMD is behind. People love pounding the IPC of Intel over AMD's heads. However, Intel's IPC is higher and the higher clock speed is just icing on the cake. At the same clocks, Intel is still ahead. I have a feeling that, at the same clock, nVidia would lose handily. Considering they need to use the extra clock speed to do the same work.

Yes nVidia has a smaller die, but they need that extra speed to keep up. AMD went with a lower clock speed and more resources. If they could get the power budget under better control, AMD would still get slammed, people never give them a break. If not hammering them about power usage, they dredge up years old driver issues...like nVidia have a perfect record...

I'm not an AMD fanboy, but my experiences between the AMD and nVidia cards I've had, AMD has been much better. And at this time, AMD seems to have some advantages over nVidia, just not when comparing power usage. AMD seems to be trying to get it under control, but it seems they might have to strip down some of those extra resources to do it...like nVidia did.

I don't think this is entirely accurate. Vega 64's main competitor is the gtx 1080, which it generally matches or surpasses depending on tuning. But, Vega 64 has 12.5 Billion transistors compared to the 1080's 7.2 Billion. That is a non-trivial increase. Nearly 74% more. This isn't a direct comparison of course. The two cards budget their transistors very differently to support various proprietary technologies, which in Vega's case almost certainly include non-gaming oriented features. But per Anandtech:

https://www.anandtech.com/show/11680/radeon-rx-vega-unveiled-amd-announecs-499-rx-vega-64-399-rx-vega-56-launching-in-august/3

Talking to AMD’s engineers, what especially surprised me is where the bulk of those transistors went; the single largest consumer of the additional 3.9B transistors was spent on designing the chip to clock much higher than Fiji. Vega 10 can reach 1.7GHz, whereas Fiji couldn’t do much more than 1.05GHz. Additional transistors are needed to add pipeline stages at various points or build in latency hiding mechanisms, as electrons can only move so far on a single clock cycle; this is something we’ve seen in NVIDIA’s Pascal, not to mention countless CPU designs. Still, what it means is that those 3.9B transistors are serving a very important performance purpose: allowing AMD to clock the card high enough to see significant performance gains over Fiji.

And you can compare Vega 64 against a similarly sized die from Nvidia with nearly identical clocks. The gtx 1080 ti. Vega 64 is 486 mm² with 12.5 Billion transistors, and the gtx 1080 ti is 471 mm² with 12 Billion transistors. Stock Vega 64 clocks at 1274 MHz / 1546 MHz but Tom's just did a review of the Sapphire RX Vega 64 Nitro, which clocks at 1423 MHz / 1630 MHz , while the 1080 ti at stock is 1480 MHz / 1582 MHz.

http://www.tomshardware.com/reviews/sapphire-rx-vega-64-nitro,5388.html

 

[renz496]

Actually, it's no different than the comparison between Intel and AMD, but in that case AMD is behind.

then try looking at RX480 vs R9 390X. stock vs stock the RX480 is clocked much higher than 390X. but most often 390X is faster than RX480. if you just look at it with face value we can say the 390X have much higher IPC than RX480.

Yes nVidia has a smaller die, but they need that extra speed to keep up

but AMD for their part will also want to have this kind of advantage that nvidia had.

If they could get the power budget under better control, AMD would still get slammed, people never give them a break. If not hammering them about power usage, they dredge up years old driver issues...like nVidia have a perfect record...

and you think people never complaining about nvidia? in fact people complain about many things more with nvidia than AMD.

AMD seems to be trying to get it under control, but it seems they might have to strip down some of those extra resources to do it...like nVidia did.

in AMD case they probably need to ditch GCN altogether if they really want to catch up to nvidia. right now AMD add a lot of things but the base architecture is still the same GCN that they use since 2012. IMO they need to do something about that first before adding more feature.

 

[Sakkura]

GCN is not the same today as it was in 2012.