• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

So just how are AMD going to shoot themselves in the foot with the NAVI launch?

Associate
Joined
21 Apr 2007
Posts
2,483
LOL, so wishful some of you.

At every release in recent years AMD have set prices for the mid and high end cards that equal Nvidia's. FuryX slower than 980ti but same RRp and more expensive on the street. Vega 64 vs 1080. Radeon 7 vs 2080.


OF course AMD are going to price these at $499 for 2070 performance. The 7nm node is very expensive.

I hate to say it but I think ur right. AMD is using free sync and consoles as a way to win support not raw performance which obviously appeals to PC gamers
 
Associate
Joined
13 Jul 2009
Posts
523
I haven't read the thread but AMD seem to have pretty much not gone anywhere in the last 3 years (in terms of gpus). Just same performance under different names. At least they are making progress with their cpus.

Nvidia pretty much no better either, also same performance under difference names with just 1 better card (excluding titan) for double as much. How nvidia are charging over Titan prices for their standard high end card is staggering.

I feel i keep beating the same drum but ill continue to moan about this because in 19 years of owning a pc we have hit the point where it doesnt seem worth it/fun anymore. I miss the days of reasonable prices for decent upgrades and crazy fun overclocking (especially with cpus, getting those old 1.6Ghz intel chips to what was it 2.8Ghz+, big AMD opteron clocks, 3dmark2001, 2003, 2005, 2006, o and counter strike of course/uni... i have nostalgia). Its not like that anymore. Hell its been over 10 years since crysis, hardly moved on from that graphically. Maybe im just getting old...
 
Last edited:
Caporegime
Joined
18 Oct 2002
Posts
32,615
If the next gen arch next year is a flop then I'll be fully in agreement. I get what AMD were trying to do when they bought ATi, but with the CPU division falling on its ass as well it just messed the company up too much to make good on their plans in any reasonable time frame.

Give Lisa Su her props though, she seems to have worked some magic and gotten momentum back, but "Next Gen" will be entirely on her watch, not leftovers from RTG still trying to be independent. If her magic touch doesn't extend to graphics as well come Q1 2021 then perhaps it might be time to cut it loose again.

Unless, of course, graphics is still sustainable and doesn't require propping up with CPU profits.

One of the issues is the benefits of combined GPU + CPU just haven't been realized and it is not really clear where this will go. AMD did score the console deals because of this, although it is unlikely to have paid off the cost of buying ATI in the first place (especially given indirect costs). And it is not like AMD would be left on the sidelines without GPUs if they had a good CPU offering. However many year later and the market still following an Intel CPU with IGPU, or intel CPU + Discrete nvidia GPU. As great as Ryzen is, it is not like people people can buy a Ryzen APU that makes buying a discrete GPU obsolete so Ryzen owners still buy Nvidia GPUs. Nvidia and Intel are as strong as ever despite only servicing CPU or GPU and not both.

The failure of AMD to push APU up to the mid-level is puzzling to me. I really expected that cards card to the 1060/580 level would just cease to exist because you could get that level in a APU for less money. If Intel rumours are true then it will be Intel that causes the death of low-end discrete GPUs and not AMD. I think this even took Nvidia by surprise, they expected low end discrete GPUs to die off and concentrated on mid and high end along with data center/HPC but low end GPUs still thrive.


AMD do need a next generation architecture, and 2021 is the earliest that will happen. I still feel that there will be limitations in R&D for such a product. Especially since AMD will really need to divide R&D between HPC, deep-learning, embedded Autonomous vehicle, as well as consumer gaming focused parts. Nvidia has been splitting up their GPU designs since Pascal.
 
Associate
Joined
17 Sep 2018
Posts
1,425
One of the issues is the benefits of combined GPU + CPU just haven't been realized and it is not really clear where this will go. AMD did score the console deals because of this, although it is unlikely to have paid off the cost of buying ATI in the first place (especially given indirect costs). And it is not like AMD would be left on the sidelines without GPUs if they had a good CPU offering. However many year later and the market still following an Intel CPU with IGPU, or intel CPU + Discrete nvidia GPU. As great as Ryzen is, it is not like people people can buy a Ryzen APU that makes buying a discrete GPU obsolete so Ryzen owners still buy Nvidia GPUs. Nvidia and Intel are as strong as ever despite only servicing CPU or GPU and not both.

The failure of AMD to push APU up to the mid-level is puzzling to me. I really expected that cards card to the 1060/580 level would just cease to exist because you could get that level in a APU for less money. If Intel rumours are true then it will be Intel that causes the death of low-end discrete GPUs and not AMD. I think this even took Nvidia by surprise, they expected low end discrete GPUs to die off and concentrated on mid and high end along with data center/HPC but low end GPUs still thrive.

AMD do need a next generation architecture, and 2021 is the earliest that will happen. I still feel that there will be limitations in R&D for such a product. Especially since AMD will really need to divide R&D between HPC, deep-learning, embedded Autonomous vehicle, as well as consumer gaming focused parts. Nvidia has been splitting up their GPU designs since Pascal.

Interesting theory about future APUs killing off the mid range GPU. Why isn't there PS4 Pro level APUs? I assume it's because everything would have to be engineered on a board, from the cpu, apu, ram rather than just the chip. Even if you could get an APU You have the RAM bottleneck, right? People have talked about HBM2 being on the actual chip to get around that problem.

The APU on Zen 2 will be quite interesting
 
Soldato
Joined
14 Aug 2009
Posts
2,695
I haven't read the thread but AMD seem to have pretty much not gone anywhere in the last 3 years (in terms of gpus). Just same performance under different names. At least they are making progress with their cpus.

Nvidia pretty much no better either, also same performance under difference names with just 1 better card (excluding titan) for double as much. How nvidia are charging over Titan prices for their standard high end card is staggering.

I feel i keep beating the same drum but ill continue to moan about this because in 19 years of owning a pc we have hit the point where it doesnt seem worth it/fun anymore. I miss the days of reasonable prices for decent upgrades and crazy fun overclocking (especially with cpus, getting those old 1.6Ghz intel chips to what was it 2.8Ghz+, big AMD opteron clocks, 3dmark2001, 2003, 2005, 2006, o and counter strike of course/uni... i have nostalgia). Its not like that anymore. Hell its been over 10 years since crysis, hardly moved on from that graphically. Maybe im just getting old...

Look at it differently: with a 2500k (and maybe a 2600k as games evolved a bit), plus a R290/X, you can still game nicely at 1080p@60fps or higher. Back in the day, people would complain about the need to upgrade often if they wanted to play newer games at decent setting, but that has come and past. Sure, maybe it hasn't been that great for enthusiasts, but enthusiast, in a way, have helped nVIDIA be where it is by buying into an ever increasing price bracket - same with Intel.

Now looking back at that system above, it lasted as long as a console generation or even more. From that point, is far more important than the lack of major jumps for enthusiast - for gaming in general.

Interesting theory about future APUs killing off the mid range GPU. Why isn't there PS4 Pro level APUs? I assume it's because everything would have to be engineered on a board, from the cpu, apu, ram rather than just the chip. Even if you could get an APU You have the RAM bottleneck, right? People have talked about HBM2 being on the actual chip to get around that problem.

The APU on Zen 2 will be quite interesting

Memory bandwidth plus power required to get that going, maybe some deal between AMD and Sony+Microsoft to not have competing products on the market in the same performance bracket? But, besides the form factor, I don't see why, as a gamer, you'd went with an APU instead of a dedicated CPU+GPU, which you can mix and mach as you want, upgrade as per your needs, etc. So perhaps the market isn't big enough just yet.
 
Soldato
Joined
26 Sep 2010
Posts
7,146
Location
Stoke-on-Trent
The failure of AMD to push APU up to the mid-level is puzzling to me. I really expected that cards card to the 1060/580 level would just cease to exist because you could get that level in a APU for less money.
Shared memory and memory bandwidth probably. Buying in 16GB RAM just to give the graphics portion sufficient leg room and still have enough left over to run the system nicely defeats the point getting a cheap APU to begin with. And even then, we always see the graphics held back because system RAM is never fast enough to keep the APU fed.

I remember when the Kaveri APUs came out and I was toying with building a tiny HTPC with an A8 or A10. I commented on an AMD Facebook post about it casually saying "but what we really want is dedicated HBM strapped to the APU, guys, to let that GPU core fly!". They actually replied with "all in good time ;)". That was 2014 I think?

I do genuinely believe there is a crazy master plan at work, but it's so ambitious it was always going to be a long term endeavour and then further compounded by AMD's financial woes. 5 years after a meaningless Facebook exchange we get an AdoredTV video speculating on a custom Zen 3 EPYC with 22 CPU and GPU chiplets, and HBM2 strapped directly to the I/O die. Give it another 5 years and that Facebook exchange might just come to pass on the desktop.
 
Associate
Joined
17 Sep 2018
Posts
1,425
Memory bandwidth plus power required to get that going, maybe some deal between AMD and Sony+Microsoft to not have competing products on the market in the same performance bracket? But, besides the form factor, I don't see why, as a gamer, you'd went with an APU instead of a dedicated CPU+GPU, which you can mix and mach as you want, upgrade as per your needs, etc. So perhaps the market isn't big enough just yet

It's not what I necessarily want, it's more that direction mainsteam gaming is expected to be going if you listen to the analysts. If you suffer from upgraditis, you may want to upgrade to 4k, need a GPU to match that 2080 ti and you'll upgrade more than the gamer who's happy at 1080p for now, has a 290 and I7 2700x and will upgrade when. The latter gamer will be very happy with a custom APU that would offer Vega 56 and Ryzen 3600 performance and let him upgrade to 1440p when he's at it. That would be a compelling mass market product but the memory bandwidth would need to be worked around.

That's why 7nm Zen 2 should be fascinating from an APU perspective. How many compute units can they fit in at 7nm and how fast can it go? The last gen APUs had 11 compute units. There was a part tested with 20 compute units that hit Vega 56 performance. Now that could well be fabricated and I highly doubt Zen 2 hits that on both the size of the APU and with the memory bottleneck
 
Associate
Joined
17 Sep 2018
Posts
1,425
Shared memory and memory bandwidth probably. Buying in 16GB RAM just to give the graphics portion sufficient leg room and still have enough left over to run the system nicely defeats the point getting a cheap APU to begin with. And even then, we always see the graphics held back because system RAM is never fast enough to keep the APU fed.

I remember when the Kaveri APUs came out and I was toying with building a tiny HTPC with an A8 or A10. I commented on an AMD Facebook post about it casually saying "but what we really want is dedicated HBM strapped to the APU, guys, to let that GPU core fly!". They actually replied with "all in good time ;)". That was 2014 I think?

I do genuinely believe there is a crazy master plan at work, but it's so ambitious it was always going to be a long term endeavour and then further compounded by AMD's financial woes. 5 years after a meaningless Facebook exchange we get an AdoredTV video speculating on a custom Zen 3 EPYC with 22 CPU and GPU chiplets, and HBM2 strapped directly to the I/O die. Give it another 5 years and that Facebook exchange might just come to pass on the desktop.

This issue is the cost of HBM2 and whether the number of Compute Units/Shaders needed to make it worthwhile. So as AMD has invested in HBM2, it gets more compute units on a chip and as the price drops maybe this could be a thing
 
Caporegime
Joined
18 Oct 2002
Posts
32,615
Shared memory and memory bandwidth probably. Buying in 16GB RAM just to give the graphics portion sufficient leg room and still have enough left over to run the system nicely defeats the point getting a cheap APU to begin with. And even then, we always see the graphics held back because system RAM is never fast enough to keep the APU fed.

I remember when the Kaveri APUs came out and I was toying with building a tiny HTPC with an A8 or A10. I commented on an AMD Facebook post about it casually saying "but what we really want is dedicated HBM strapped to the APU, guys, to let that GPU core fly!". They actually replied with "all in good time ;)". That was 2014 I think?

I do genuinely believe there is a crazy master plan at work, but it's so ambitious it was always going to be a long term endeavour and then further compounded by AMD's financial woes. 5 years after a meaningless Facebook exchange we get an AdoredTV video speculating on a custom Zen 3 EPYC with 22 CPU and GPU chiplets, and HBM2 strapped directly to the I/O die. Give it another 5 years and that Facebook exchange might just come to pass on the desktop.




Well I assumed it would be possible to have 8GB GDDR5/6 for the APU but have a separate DDR3/4 sytem RAM as a hybrid solution.


If there is a fundamental reason thios ism;t achiveable then it makes the whole purchase of ATI by AMD somewhat limited in scope.
 
Soldato
Joined
26 Sep 2010
Posts
7,146
Location
Stoke-on-Trent
Well I assumed it would be possible to have 8GB GDDR5/6 for the APU but have a separate DDR3/4 sytem RAM as a hybrid solution.


If there is a fundamental reason thios ism;t achiveable then it makes the whole purchase of ATI by AMD somewhat limited in scope.
You remember Hybrid Crossfire and SidePort memory on Phenom II boards? Bring that pupper back. If you're not going to put all the RAM into a single package, just slap the GDDR6 and DDR 4 onto the motherboard. OK, it's not going to be upgradable but it you're designing a system for pre-determined purposes then you can just spec the amount of RAM you'd need.
 
Soldato
Joined
14 Aug 2009
Posts
2,695
It's not what I necessarily want, it's more that direction mainsteam gaming is expected to be going if you listen to the analysts. If you suffer from upgraditis, you may want to upgrade to 4k, need a GPU to match that 2080 ti and you'll upgrade more than the gamer who's happy at 1080p for now, has a 290 and I7 2700x and will upgrade when. The latter gamer will be very happy with a custom APU that would offer Vega 56 and Ryzen 3600 performance and let him upgrade to 1440p when he's at it. That would be a compelling mass market product but the memory bandwidth would need to be worked around.

That's why 7nm Zen 2 should be fascinating from an APU perspective. How many compute units can they fit in at 7nm and how fast can it go? The last gen APUs had 11 compute units. There was a part tested with 20 compute units that hit Vega 56 performance. Now that could well be fabricated and I highly doubt Zen 2 hits that on both the size of the APU and with the memory bottleneck

It will be easier and cheaper to make a simpler motherboard to house the CPU only and add the dedicated card later . The gamer who doesn't mind and upgrades once every 4-5 years, can always buy a prebuild system and upgrade only the parts required as the years pass - again cheaper as you'll need to change perhaps just the GPU or CPU, add some RAM, etc.

If you're talking about an APU that will fit the regular socket AM4, then you're limited by bandwidth and power. You can't have vega 56 performance from AMD, not even on 7nm and for sure not on the same package with a 8c16t cpu. If you want a dedicated socket with a motherboard build to take a hungry APU with dedicated memory on it (btw, also the connection between the GPU and memory from the motherboard has to be just as fast), then will cost a lot and a possible replacement won't be cheap and upgrades difficult and more expansive. So, if you're not going for smaller format, there's no difference between an APU and a standard solution.

If you want a small form factor gaming rig and you make an APU with HBM directly or dedicated memory as a sort of "console with windows" for OEMs, that's different. :)
 
Soldato
Joined
25 Sep 2009
Posts
9,616
Location
Billericay, UK
Well I assumed it would be possible to have 8GB GDDR5/6 for the APU but have a separate DDR3/4 sytem RAM as a hybrid solution.


If there is a fundamental reason thios ism;t achiveable then it makes the whole purchase of ATI by AMD somewhat limited in scope.
I seem to recall the whole idea behind the acquisition was to bring in a era of fusion and heterogeneous computing where a CPU core and GPU could be used interchangeably by an application. As with most things AMD it was an approach that was ahead of it's time for a market that wasn't quite ready for such a fundamental shift in technology. IMO AMD would have been better off just licensing the technology from ATI, ATI needed the money after the disastrous launch of their DX10 cards and I can't see them saying no to such an offer (ah the beauty of hindsight).
 

bru

bru

Soldato
OP
Joined
21 Oct 2002
Posts
7,360
Location
kent
Well I just watched the Computex keynote and I must say I'm impressed so far.

I would love to say/moan that they haven't reached clock speeds with Ryzen 3 or that the core names against the TDP's were not as expected. The Navi info didn't really tell us much more than we already knew and the benchmarks shown didn't really show us anything at all.
But really, they didn't do anything wrong, it was a good teaser for what is to come and now we have a date to know when it will hit the shelves.

Of course they can still end up with their foot in their mouth, but so far it all looks good. So congratulations AMD. I look forward to the new GPU's and CPU's.
 
Soldato
Joined
26 Sep 2010
Posts
7,146
Location
Stoke-on-Trent
What's the after-the-fact answer to this thread then?

Price?
Yes. Definitely price.

I don't like Zen 2's pricing but I understand it. Intel have nothing to counter the AMD CPU juggernaut and it's entirely understandable that AMD will shed their "budget brand" image and charge accordingly.

But the same logic does not apply to the GPUs. I know full well it takes a while to get new architectures designed and taped up, especially when Navi would've been in the works during AMD's darkest financial periods and near bankruptcy. To that end I wasn't expecting performance miracles. But I've also said a few times that if AMD pitch Navi prices near to their Nvidia counterparts then they just won't be attractive. Enthusiasts know ray tracing and DLSS are in their infancy and right now not worth the premium, but it's still a marketing upsell in the eyes of the lesser-informed and if you can get Nvidia's swanky new tech for only $50 more than AMD who don't have it then just pay the premium.

I never believed the Navi leaks were accurate in pricing, but I did believe they were indicative of AMD's desire to shake up the market and take advantage of Nvidia's blatant greed. Looks like now they've just capitulated and gone for a slice of that greed themselves. You can still make profit through high volume, low margin sales, and AMD would've cleaned house if these 2 cards came in $100 under Nvidia.
 
Soldato
Joined
21 Apr 2012
Posts
4,927
Location
Bristol
Don't forget, we're going to be back with the hot and loud memes that come with the ****** reference cooler as that'll be the one that gets reviewed. Completely deserved in my opinion, AMD needs to follow NVidias route and not tout terrible stock cooling options even if its cheap
 
Back
Top Bottom