By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - PC Discussion - Nvidia Gets SALTY

caffeinade said:

We've known about 7nm for Vega for quite some time now.
This was nearly a whole year ago.

The whole year is not so long if we talk big tech things. They said one picture is worth more than a thousand of words, so



Around the Network
Pemalite said:
Bofferbrauer2 said:

Seriously Perm?

If it was just a die Shrink plus increased clock speeds and more bandwith, then it wouldn't have outperformed the old Vega 64 by such a large margin. Radeon VII has some architectural improvements, because the numbers don't add up otherwise. It's certainly not the Bandwith, as many of these games listed weren't limited in that domain to begin with.

Vega 7's clockspeed increased by 16.4%. - That means the Render Output Pipelines (A big limiter on Vega), Texture Mapping Units operate that much faster. - The ROPS is a big one as AMD always seems to be ROP starved.

Compute increased by 9.5% over Vega 64.

Bandwidth increased by a whopping 112.27%.

The 25% or so increases is more or less from increases in clockrate and that bandwidth boost... Vega was never compute bound to begin with.

So yes, seriously. Vega 7 is just like RX 590, few enhancements... Bulk of the gains due to clock increases thanks to the smaller fabrication process opening up extra headroom.

Is there some secret sauce hidden somewhere? Possibly, but we don't know at this point in time... And it is best to leave such speculation until Anandtech has done a thorough analysis on the hardware when it releases.

Just to add in something I wasn't sure at that point, Vega 10 FP64 can only can reach 1/16th of FP32 while Vega 20 can do 1/2, so the chip must have changed quite a bit under the hood. Double Precision is also something that cost quite a lot of energy, hence why it got cut down  in modern GPUs both by AMD and NVidia. NVidia nowadays mostly uses 1/32th, and all RTX cards, including the Quadro, do so. The original Titan had 1/4, while the Volta-based Titan V has 1/2 like the Vega 20, making the latter the probable target if the FP64 capability of the Radeon VII hasn't been cut down.



shikamaru317 said:

That pic actually reminded me that the leakers said Xbox Scarlett Anaconda would be using AMD's "next-gen" chipset. So maybe Anaconda will actually use Navi's 2020 successor paired with Ryzen 3, instead of the rumored Ryzen 2 + Navi on PS4, with a goal to release at $500 as opposed to $400 on PS5?

My hearth tells me yes, but my brain...

Last edited by KingofTrolls - on 15 January 2019

KingofTrolls said:

My hearth tell me yes, but my brain...

Your hearth?



thismeintiel said:

As for Navi, I have a feeling MS may not have access to it.  Sony is rumored to be working pretty closely with AMD on Navi.  I kinda doubt they would want all of their input to go into their competition's machine.  I would imagine Sony is going to have console exclusivity on Navi, at least for the beginning of the gen.  I think Radeon VII, most likely customized, is the way MS is going to go.  We have over a year for them to drop prices, which, like you pointed out, aren't the prices MS pays, anyway.  It would give them a numbers win, most likely, in terms of Tflops.  Of course, it won't be that large of a win, will cause it to be ~$499 vs a ~$399 PS5, and have fewer of the new technologies found in Navi.  In the end, MS is going to make sure they can tout the most powerful system next gen, but I think it will mainly be on paper.  Sony's 1st parties are going to show just what next gen can do.

Wishful thinking I guess. AMD is not a Sony subsidiary to give them such a priority over Microsoft. Both MS and Sony are equal partners to AMD. The only way I can see AMD favouring any side is if someone pays them way more money than the other. But overall, AMD, unlike console warriors, is interested In both parties to succeed. It's their business. So, there is no way they dump Microsoft like this especially now that MS is going to use their tech for xCloud. 

shikamaru317 said:

Yeah, I honestly think this is why AMD doesn't try very hard on the high end anymore. Why work your butt off, spending a ton of R&D dollars trying to make the absolute best high end chipset, when the market for it is so tiny? 

AMD does try a bit harder on the low-mid range though, Polaris for instance. When Polaris released, it's top of the line chipset, RX 480, roughly traded blows with Nvidia's 1060 in performance, beating the 1060 in games that were optimized for AMD cards and losing to it in games that were optimized for Nvidia GPU's (RX 480 also won on Vulkan games like Doom). And the best thing was the Rx 480 could usually be found for lower prices than the 1060, sometimes as much as $50 lower.

I honestly think that AMD has the right strategy focusing primarily on the low-mid range rather than the high range. Could they still try harder on the low-mid range than they do? For sure, AMD is definitely behind Nvidia when it comes to efficiency even if they do roughly match them on performance and price, the 1060 had a 30 watt lower TDP than the RX 480 for instance. I would definitely like to see AMD step up their game when it comes to efficiency, and it certainly wouldn't hurt if they stepped up their game in terms of performance as well, as currently more games get optimized for Nvidia GPU's than AMD GPU's since Nvidia is the market leader, means that AMD needs to work harder if they want the performance crown on the low-mid range. 

GPUs like 1080, 1080Ti, 2080, 2080Ti are very important to PC gaming community though. The enthusiasts are willing to pay for such cards. And it's a shame that it looks like AMD struggles to be competitive in this segment as well. Because right now usually there is no choice for high-end PC gamers other than to go Nvidia route.

freebs2 said:
derpysquirtle64 said:

Not only because it's cheaper. I guess both Sony and Microsoft doesn't really want to work with Nvidia at this point. Nvidia screwed both companies in the past. They screwed Microsoft with GPU prices for original Xbox which led to a lawsuit. And they made the worse GPU for PS3 than what was inside Xbox360 which came out a year before. I guess it costed more for Sony than X360 GPU as well.

Don't really care of being Nvidia's defence force, but this is not very accurate.

Yes, the original XB GPU was more expensive but it was largely superior to other consoles and yes, they (Nvidia) wanted and still want a larger cut of profits (that's what I'm saying from the beginning). As for PS3, the reason why the GPU was weaker is mostly due to the fact that until the later stages of HDW development Sony planned to use the Cell to render games instead of a regular GPU. The RSX was inferior simply because it was engeneered on a short notice.

It was inexcusable for nVidia to not lower GPU prices for original Xbox when their factory costs decreased after a couple of years. nVidia was just in it to get a lot of money from Microsoft and they succeeded. This was one of the main reasons why Microsoft lost around 5 billion dollars on it.

As for PS3, while the first part that Sony initially intended to use 2 Cell CPUs is true, I don't think the other one is.



 

Around the Network
derpysquirtle64 said:

It was inexcusable for nVidia to not lower GPU prices for original Xbox when their factory costs decreased after a couple of years. nVidia was just in it to get a lot of money from Microsoft and they succeeded. This was one of the main reasons why Microsoft lost around 5 billion dollars on it.

As for PS3, while the first part that Sony initially intended to use 2 Cell CPUs is true, I don't think the other one is.

Sony Abandoned using Cell as a GPU very early on... Because it would have been rubbish at the task, the Cell is still very general purpose and lacks any of the parts of a GPU that accelerates various tasks like texture samplers.

Sony did hope that Toshiba would provide the GPU at some point, but that never materialized, so nVidia was given the task.

In short... People clinged to Sony's statements from E3 when they essentially stated that the PS3 can generate excellent real-time graphics using only the power of the Cell... Cell isn't that good I am afraid, history has more than proven that.

Bofferbrauer2 said:

Just to add in something I wasn't sure at that point, Vega 10 FP64 can only can reach 1/16th of FP32 while Vega 20 can do 1/2, so the chip must have changed quite a bit under the hood. Double Precision is also something that cost quite a lot of energy, hence why it got cut down  in modern GPUs both by AMD and NVidia. NVidia nowadays mostly uses 1/32th, and all RTX cards, including the Quadro, do so. The original Titan had 1/4, while the Volta-based Titan V has 1/2 like the Vega 20, making the latter the probable target if the FP64 capability of the Radeon VII hasn't been cut down.

Not really. Graphics Core Next is extremely modular remember, you can update part of a chip and leave the rest identical.
Besides... AMD has a Vega GPU with 1/2 FP64 on the market right now... Meaning that the design of Vega 7 isn't new anyway.

In short, Vega 7 is a simply a GPU ported to 7nm with a CU removed to increase yields... Doubling of DRAM and Bandwidth and a big increase in clockrates, it was minimal effort by AMD... And because of that, it's unlikely it will beat a Geforce 1080Ti.

KingofTrolls said:

The collaboration is not the only hint we got on the topic,thoug. Forbes' leak points simply that MS will not use Navi technology, also it points the Vega will be a disappointment because resources were moved to Navi, and kinda it was proven true. So far, the leak is accurate.

It doesn't matter if Microsoft doesn't use Navi technology, Navi isn't likely to be anything special anyway. - AMD does have next-gen being designed remember.
With that said... A leak posted by Forbes can and shall be taken with grains of salt.

HoloDust said:

2020 is most likely date, though, X360 had ATI's custom GPU that had unified shaders way before their desktop cards with same feature hit the shelves. So, something like this may happen again, AMD focusing on delivering next-gen GPUs for consoles, and only later releasing desktop cards with those next-gen featues.

That just plays into my statements that GPU's and CPU's take years to design... And because of that, when you take a semi-custom approach, you have the opportunity to implement newer features that are set for future GPU designs.
The Xbox 360 is after-all is the perfect example of that.






--::{PC Gaming Master Race}::--

HoloDust said:
Alby_da_Wolf said:
this next step in AMD roadmap should be the final evolution of current architecture, so it's quite natural that more radical innovations will start appearing in the successive step, the first versions of next gen AMD graphics architecture. This makes quite likely too that next gen Sony and MS console won't appear earlier than in late 2020, as using this tech they'd look immediately old as soon as next gen is launched.

2020 is most likely date, though, X360 had ATI's custom GPU that had unified shaders way before their desktop cards with same feature hit the shelves. So, something like this may happen again, AMD focusing on delivering next-gen GPUs for consoles, and only later releasing desktop cards with those next-gen featues.

Pemalite said:
HoloDust said:

2020 is most likely date, though, X360 had ATI's custom GPU that had unified shaders way before their desktop cards with same feature hit the shelves. So, something like this may happen again, AMD focusing on delivering next-gen GPUs for consoles, and only later releasing desktop cards with those next-gen featues.

That just plays into my statements that GPU's and CPU's take years to design... And because of that, when you take a semi-custom approach, you have the opportunity to implement newer features that are set for future GPU designs.
The Xbox 360 is after-all is the perfect example of that.

With Polyphony showing off GT Sport using ray tracing, it's quite unlikely they were doing that just for fun, and even more unlikely it's distant future tech for PS6. AMD has mentioned recently they are working on ray tracing tech for their future products. If PS5 is going to use Navi, then at the very least, even if AMD isn't planning to use ray tracing tech until their "Next Gen" gpu products, they could very well be baking it into a semi custom Navi, just like how PS4 Pro used Polaris but had some Vega tech baked in.



Bofferbrauer2 said:
Pemalite said:

Just to add in something I wasn't sure at that point, Vega 10 FP64 can only can reach 1/16th of FP32 while Vega 20 can do 1/2, so the chip must have changed quite a bit under the hood. Double Precision is also something that cost quite a lot of energy, hence why it got cut down  in modern GPUs both by AMD and NVidia. NVidia nowadays mostly uses 1/32th, and all RTX cards, including the Quadro, do so. The original Titan had 1/4, while the Volta-based Titan V has 1/2 like the Vega 20, making the latter the probable target if the FP64 capability of the Radeon VII hasn't been cut down.

Ask, and ye shall receive.

https://wccftech.com/amd-radeon-vega-vii-5000-units-64-rops-no-fp64-compute/



 

 

 

 

 

EricHiggin said:
HoloDust said:

2020 is most likely date, though, X360 had ATI's custom GPU that had unified shaders way before their desktop cards with same feature hit the shelves. So, something like this may happen again, AMD focusing on delivering next-gen GPUs for consoles, and only later releasing desktop cards with those next-gen featues.

Pemalite said:

That just plays into my statements that GPU's and CPU's take years to design... And because of that, when you take a semi-custom approach, you have the opportunity to implement newer features that are set for future GPU designs.
The Xbox 360 is after-all is the perfect example of that.

With Polyphony showing off GT Sport using ray tracing, it's quite unlikely they were doing that just for fun, and even more unlikely it's distant future tech for PS6. AMD has mentioned recently they are working on ray tracing tech for their future products. If PS5 is going to use Navi, then at the very least, even if AMD isn't planning to use ray tracing tech until their "Next Gen" gpu products, they could very well be baking it into a semi custom Navi, just like how PS4 Pro used Polaris but had some Vega tech baked in.

Well. AMD's GPU's are capable of Ray Tracing today... Ray Tracing is typically a very compute and memory bound scenario... And what is Graphics Core Next great at? Compute... Hence why miners typically preferred them.

But, current hardware just doesn't have the horsepower to brute force it, hence nVidia's fixed-function approach, plus methods to reduce the processing load by reducing the work required.

It will be interesting to see what official path AMD takes to Ray Tracing though going forward, nVidia has applied the pressure that it's a must-have technology now.

As for the Playstation 4 Pro... It's not actually using Polaris as a base, it's using the same identical GPU found in the base Playstation 4, just doubled/mirrored. - It implemented features from both Polaris and Vega... But we need to keep in mind that the Playstation 4 also had changes away from the original Graphics Core Next design anyway. (Like an increase in ACE units.)

AMD designs it's GPU's in terms of "blocks". - And they can update, mix and match those blocks to suit their needs.

My own hopes is that next-gen will not be using Graphics Core Next, but AMD's next-gen architecture... But the realist in me says that both consoles will be Graphics Core Next based with features taken from Vega and Navi... And that to me isn't next-gen console hardware.

Last edited by Pemalite - on 15 January 2019

--::{PC Gaming Master Race}::--

KingofTrolls said:
caffeinade said:

We've known about 7nm for Vega for quite some time now.
This was nearly a whole year ago.

The whole year is not so long if we talk big tech things. They said one picture is worth more than a thousand of words, so

This is a whole year from announcement to release, not development to release.
The AMD Radeon Vega VII isn't a whole die, and the full 64CU die isn't even that big.
It isn't like this is a massive die, and this isn't a new architecture.
An ancient slide doesn't change the fact that Vega 7nm has clearly been in the pipeline for ages, considering the product that it is.