Navi discussion thread

reaper12 said:

People complained that it was priced too high.

Click to expand...

Not me. I think Polaris was priced perfectly. One of AMD's best GPU moves in recent history.

AMD will still have products in the $200/250 range. The trouble is that people think AMD should be selling a 2070 competitor at $250 and that's the problem. Getting back to your AMD losing mindshare as well as marketshare. Well, that's brilliant, because all that 10 years of lowering prices has done is to make AMD look like a budget brand. And trying to win market share with that Strategy hasn't really worked. They have made no significant gains but lost money.

Click to expand...

Navi pricing is $449-$499 for 5700XT and $379 for 5700. 2070 and 2060 competitors, respectively. Drop that by $50-$100, and they'd fly off the shelves. The truth of the matter is that AMD is the budget brand with respect to GPUs. If they wanted to compete on absolute performance, it would be different. The AMD purchase is going to be primarily price-driven. I don't necessarily like this. I like the days when both were competing for absolute performance. But it is what it is.

A company can't keep cutting into it's margins to gain market share. It's not sustainable. And AMD weren't even winning market share, they have been in a steady decline for the last 10-12 years.

Click to expand...

As harmattan said, AMD does not exist in a vacuum. If they want to increase marketshare, they must create a compelling reason to prefer them over Nvidia. Since absolute performance isn't in their wheelhouse, it must be price/perf. And it must be significant enough to move the needle - as roughly equivalent pricing will favor Nvidia as the stronger brand. The greater the price band difference, the more compelling the reason to favor AMD.

Chad's earlier point, however, was well taken. AMD may not care about increasing marketshare, and may instead be focusing GPU development on consoles, integrated solutions, etc... In other words, internally they might view the discrete GPU market as the leftovers, so to speak. If they sell some good amount of discrete cards, great. If not... well, they will sell a lot of GPUs anyway. It would explain a great deal.

Caveat: if the 3rd party benchmarks show the Navi cards as significantly faster than their respective Nvidia competitors - not just a little bit, but a lot - then their current pricing would be fine.

DuronBurgerMan said:

As harmattan said, AMD does not exist in a vacuum. If they want to increase marketshare, they must create a compelling reason to prefer them over Nvidia. Since absolute performance isn't in their wheelhouse, it must be price/perf. And it must be significant enough to move the needle - as roughly equivalent pricing will favor Nvidia as the stronger brand. The greater the price band difference, the more compelling the reason to favor AMD.

Click to expand...

But since they don't exist in a Vacuum, NVidia would almost certainly react to pricing that would shift market share.

I haven't seen a reasonable argument why NVidia would just sit by and let AMD take market share.

If NVidia adjust pricing to maintain the gap, then all that AMD drastically cutting prices would accomplish is that both AMD and NVidia get less money, and that would harm AMD more than NVidia.

IOW, it doesn't make good sense for AMD to start any kind of price war that would harm itself more than NVidia.

Well, we don't have to wait too much longer to see how these things sell. My guess: they won't be anywhere near as successful as Polaris unless/until we see a good price cut.

Snowdog said:

IOW, it doesn't make good sense for AMD to start any kind of price war that would harm itself more than NVidia.

Click to expand...

If AMD is outsold by 10 to 1 you think it matters more for AMD then for Nvidia ?

sabrewolf732 said:

For everyone saying amd pricing cards cheaper than the competition doesn't change market share:
https://www.eurogamer.net/articles/...d-fall-of-amd-intel-and-nvidia-graphics-cards

The hd4800 series was instrumental in taking back market share at least according to steam survey. It was also an insane value proposition.

Click to expand...

If you look at the actual market share at that time, you will see this isn't really true. When AMD released the 2000 series cards, some of the worst cards ever, their market share was around 32%. When they released the HD 4000 cards, their market share briefly rose to 40% but, less than 6 months later it fell to under 30%. Sure, it might have shaken up the market for a few months, but, they didn't make any money and they didn't keep that market share.

They had some small success with the 5xxx cards, but only because Nvidia messed up their 4xx cards and were 6 months late. The cheap pricing didn't work at all with the 6xxx cards, no market share gain.

Pieter3dnow said:

If AMD is outsold by 10 to 1 you think it matters more for AMD then for Nvidia ?

Click to expand...

But it isn't.

AMD Discrete GPU share is something like 30% vs 70% for NVidia. Thats 7:3, not 10:1, it's not even 3:1, and AMD has much less company profit to work with before it goes in the Red, so yeah it matters more to AMD.

Snowdog said:

But it isn't.

AMD Discrete GPU share is something like 30% vs 70% for NVidia. Thats 7:3, not 10:1, it's not even 3:1, and AMD has much less company profit to work with before it goes in the Red, so yeah it matters more to AMD.

Click to expand...

Not sure you have been following the markets that well. NV and AMD are showing very close to identical quarterly revenue these days. With both showing mid 40% profit margins.

NV just reported quarterly revenue of 2.2 billion which is down from 3.2 billion the year prior. AMD just reported 1.8 billion which is up almost 50% from the year prior. Profit margins are very close... NV might have a slight edge in margin right now. I guess there is more profit on GPUs vs CPUs. (or one cost a lot more to market... I didn't break down the numbers that hard)

What you are saying WAS true a few years ago.... and no doubt it takes a few years for R&D to pay off. So navi may have been designed on a much tighter budget. Going forward however unless something changes in a big way with Ryzen. By the end of the year AMD is on track to be pulling more revenue at = or very close to = profit margins as NV.

There is a reason AMD stock is trending up quickly. They have grown their market quickly. The GPU side is still smaller of course. Having said that the GPU side of their business is also trending up sharply on the back of cloud server GPU wins. Perhaps the world isn't buying 10s of thousands of Radeon VIIs... the MI60 and MI70s however are selling extremely well.

ChadD said:

Not sure you have been following the markets that well. NV and AMD are showing very close to identical quarterly revenue these days. With both showing mid 40% profit margins.

NV just reported quarterly revenue of 2.2 billion which is down from 3.2 billion the year prior. AMD just reported 1.8 billion which is up almost 50% from the year prior. Profit margins are very close... NV might have a slight edge in margin right now. I guess there is more profit on GPUs vs CPUs. (or one cost a lot more to market... I didn't break down the numbers that hard)

Click to expand...

Yes I have been following it. Recent Q1 for both:

https://www.anandtech.com/show/14360/nvidia-q1-fy-2020-earnings-report-postcrypto-reset
https://www.anandtech.com/show/14288/amd-q1-fy-2019-earnings-report-revenue-down-but-margins-up


Net Income:

NVidia: 394 Million
AMD: 16 Million

NVidia made nearly 25 Times as much Income. That is why any price war would hurt AMD more. They are not that far above break even.

Also starting a price war with NVidia, and Intel might see this as a good time to start a CPU price war, and AMD is getting financially crushed by price wars on two fronts.

Really this just rehashing the arguments we had before Navi prices were released. People made the same arguments in support of Low Navi prices that never materialized.

I wonder how many times AMD has to demonstrate they are not interested in price wars, before some people start to accept it.

Snowdog said:

Yes I have been following it. Recent Q1 for both:

https://www.anandtech.com/show/14360/nvidia-q1-fy-2020-earnings-report-postcrypto-reset
https://www.anandtech.com/show/14288/amd-q1-fy-2019-earnings-report-revenue-down-but-margins-up


Net Income:

NVidia: 394 Million
AMD: 16 Million

NVidia made nearly 25 Times as much Income. That is why any price war would hurt AMD more. They are not that far above break even.

Also starting a price war with NVidia, and Intel might see this as a good time to start a CPU price war, and AMD is getting financially crushed by price wars on two fronts.

Really this just rehashing the arguments we had before Navi prices were released. People made the same arguments in support of Low Navi prices that never materialized.

I wonder how many times AMD has to demonstrate they are not interested in price wars, before some people start to accept it.

Click to expand...

I am not arguing AMD wants a price war on Consumer GPUs. They don't that is clear. Your right anyone thinking AMD was going to release a 200 dollar geforce killer where smoking something stronger then the legal stuff.

I don't believe if there was a war it would hurt AMD any more then NV however. NV can not afford any dip in margins AT ALL. They have been showing weaker then previous year revenue for over a year now. Their margins haven't slipped, they have went up... because if you read past the first paragraph you will see its their high margin stuff that isn't shrinking. Automotive has seen a small tick up while everything else is contracting. That leads to higher margins.

I am not suggesting NV is in a terrible place either. Their financials seem steady even if their revenue has been slipping.... investors are locked in on their automotive and AI plays.

Having said that a price war... that saw NV move a lot of gaming product at lower margins would drop their 60% margin investors love down into the 50s or 40% which they do not. AMD gets away with it because investors are expecting them to battle on price against Intel.

In any event we agree for the most part... NO Amd is not going to do something silly and release a 200 dollar card that destroys all other mid range competition. At best they will price 50 bucks or so lower based on performance. AMD and NV both are padding their revenue from other sources. Form the anandtech articles you posted... NV is padding their margins with automotive and AI, and AMD is padding theirs with semi custom work (bottom of the article 68 million in end of the day profit there)

I would think that AMD's revenue heading north has more to do with there CPU sales more than the GPU side. And that is only likely to increase with the upcoming launch of the next RyZen chips.

Maddness said:

I would think that AMD's revenue heading north has more to do with there CPU sales more than the GPU side. And that is only likely to increase with the upcoming launch of the next RyZen chips.

Click to expand...

Talking sales volume and revenue I'd agree. Margins probably shift toward the GPU side, but net profit back to CPUs for the balance sheet.

Still, while their CPUs are currently the star of the show, their architecture is still barely competitive with one Intel put out half a decade ago, and we know that Intel has had at least one architecture update waiting for their fabs to catch up.

Actually making long-term inroads into the CPU market is going to be far harder for AMD than making inroads into the GPU market where there's more room for innovation and growth. And at this time, AMD is still uniquely positioned with competitive IP in both spaces, something neither Intel nor Nvidia can yet quite claim.

I think this is the ideal time for AMD to try to shrug-off the budget image. They have have so much positive hype going for their CPU line-up as well as their name being mentioned in connection with Googles service and the consoles of course. They are being mentioned a lot by the press.
It looks like they will be competitive in the RTX2070 and under bracket (80%+ of the market?) where RT is a non-factor. It will be expected that AMD will keep a slight price advantage and will have some big stickers on the box such as PCIe 4.0, and Lowest Latency etc.. There may even be some sort of incentive if you go all AMD with your new fancy Ryzen 3000 rig!

funkydmunky said:

I think this is the ideal time for AMD to try to shrug-off the budget image. They have have so much positive hype going for their CPU line-up as well as their name being mentioned in connection with Googles service and the consoles of course. They are being mentioned a lot by the press.
It looks like they will be competitive in the RTX2070 and under bracket (80%+ of the market?) where RT is a non-factor. It will be expected that AMD will keep a slight price advantage and will have some big stickers on the box such as PCIe 4.0, and Lowest Latency etc.. There may even be some sort of incentive if you go all AMD with your new fancy Ryzen 3000 rig!

Click to expand...

I don't like how this is turned around, Nvidia sells overpriced cards, AMD is not a budget brand in any way shape or form.
It is not an image it is not how the brand is developed it is an additive forced upon by people that have little to no concept how they are over paying for hardware.

Both Nvidia and Intel do the same thing that does not make AMD the budget brand.

Pieter3dnow said:

AMD is not a budget brand in any way shape or form.

Click to expand...

With respect to graphics, they have been as long as they've had a graphics division. ATi was not, but AMD is.

This is due to the inferior quality and performance their products have exhibited over the course of their ownership of ATi. A decade of shoddy drivers, poor OEM coolers, and heat, noise, and even power issues will do that.

And we should be specific to this being a perception of AMD. The reality is, over the last decade if they hadn't priced lower than their competition, they would have lost even more market share. They didn't have the product catalog to compete toe to toe, and this is the result.

IdiotInCharge said:

blah blah blah

Click to expand...

There plenty of cards that were faster and cheaper in the history of AMD and nothing that screams budget , my argument is not about budget it is about overpriced hardware from other companies then AMD.
AdoredTV has made some excellent videos about this explaining how Nvidia mid range cards with Nvidia mid range silicon ended up in a higher price bracket.

The off topic issues you mention Nvidia has had exploding cards on fire from EVGA brand, drivers that over clocked and cooked the gpu, bump gate and the last batch of Nvidia cards had fucking space invaders as a feature , not the kind you would be playing but the memory hardware failure kind where the memory would not work for some reason. If I need to remind you you can check the [H] review I think they got a bunch of cards which failed. Your budget assessment with your reasons attached to it are rather stupid since you know yourself that Nvidia did not do much better but worse.

But like I said before overpriced hardware for people who don't know better...

Pieter3dnow said:

There plenty of cards that were faster and cheaper in the history of AMD

Click to expand...

Very, very few, and they were always (and I'm not exaggerating) hotter and louder.

Pieter3dnow said:

AdoredTV has made some excellent videos

Click to expand...

Which I will never watch. Get a reputable source.

Pieter3dnow said:

The off topic issues you mention

Click to expand...

Are on topic in the scope of comparing Navi to the rest of the market.

Pieter3dnow said:

But like I said before overpriced hardware for people who don't know better...

Click to expand...

I pay for performance. I haven't bought an AMD card for that in a decade. I certainly have bought AMD cards, just not for that.

IdiotInCharge said:

Talking sales volume and revenue I'd agree. Margins probably shift toward the GPU side, but net profit back to CPUs for the balance sheet.

Still, while their CPUs are currently the star of the show, their architecture is still barely competitive with one Intel put out half a decade ago, and we know that Intel has had at least one architecture update waiting for their fabs to catch up.

Actually making long-term inroads into the CPU market is going to be far harder for AMD than making inroads into the GPU market where there's more room for innovation and growth. And at this time, AMD is still uniquely positioned with competitive IP in both spaces, something neither Intel nor Nvidia can yet quite claim.

Click to expand...

Kinda-sorta agree. They are uniquely positioned with IP in both spaces. It's great for the console market, where AMD can position themselves as a one-stop shop. It's great for integrated, too. Intel may be catching up here with Ice Lake, but I expect that a Navi-based Zen 2 APU will drop early next year, and that'll bury Intel again. AMD is usually a step (or at least half of one) ahead of Intel in this respect. I remain very skeptical of Intel discrete GPU projects.

But I think the 'barely competitive' architecture is a stretch, with regard to Zen. AMD did change the game with Ryzen. They made some tradeoffs to do it with latency, but they gained the ability to rapidly scale core count, and that was probably a good trade to make. Steadily increasing core counts from Intel in response are an indication of that. And by the time we have Sunny Cove on the desktop, AMD will have another revision or two of Zen ready. AMD is well-positioned for recovery in the CPU space, long term. But I do agree with you that making inroads into CPUs is harder. That's entirely long game. AMD has to execute consistently, again and again, with no Bulldozer-like missteps. So far, they have. AMD once approached almost 50% of the CPU market. This was several years after introducing Athlon, which marked them taking the performance market seriously. It took years to push that high. And it will take years to do it again. They can't afford another Bulldozer!

GPUs, I agree, are more forgiving in this respect. Release a killer GPU, and brand loyalty (at least at the high end) will fizzle like a fart in a hurricane. Much faster turnaround. But this tends to be a function of the high end market more than the middle. In the mid-range, if performance is about the same, and price is about the same, Nvidia will win more sales, because they are perceived to be the stronger, higher-performing brand. This is why AMD must either compete on price (to adjust the price/perf calculus significantly in their favor), or must return to competing at the top, to be perceived as superior to Nvidia on straight-up performance. Either would work.

Polaris was an example of the former, done well. And the latter... we all remember when AMD could steal the performance crown from Nvidia on occasion. My old 7970 still powers my HTPC. Good stuff!

DuronBurgerMan said:

But I think the 'barely competitive' architecture is a stretch, with regard to Zen.

Click to expand...

Where I'm coming from: with respect to enthusiasts that can use more cores, AMD has a winning product. However, for the average business user / desktop user / mobile user? More cores just increase cost, power draw, noise, and heat. Nothing they do gets actually faster. That's where the volume is, and that's the market that AMD has to win. Further, for stuff that is CPU dependent, faster cores matter more often than more cores.

And I'll point it out again: AMDs refusal to put graphics on their higher core-count parts is hurting them here significantly. The volume markets don't care about discrete graphics. They just need something. Intel gives them that, and it works very very well for their purposes.

So when I say 'barely competitive', I do mean toward the whole market. You'll see me recommending Ryzen CPUs for enthusiasts regularly, but for your average desktop or laptop, AMD is neither targeting the space fully nor are they getting the OEM attention necessary to be a volume producer.

DuronBurgerMan said:

AMD has to execute consistently, again and again, with no Bulldozer-like missteps.

Click to expand...

Even if they don't make a misstep- they have to get competitive with Ice Lake-level IPC and soon. That's another jump above Zen 2 / Ryzen 3000, and as it stands, getting more IPC out of the base Zen core is going to get harder and harder.

DuronBurgerMan said:

In the mid-range, if performance is about the same, and price is about the same, Nvidia will win more sales, because they are perceived to be the stronger, higher-performing brand.

Click to expand...

While it's not really true today that Nvidia is the 'stronger, higher-performing brand' in the mid-range, though that does depend significantly on price, AMD has a long reputation of less than stellar graphics drivers and OEM coolers. Nvidia has a reputation largely the opposite, today.

Now, AMDs drivers are quite good today, and third-party coolers are really the same between AMD and Nvidia, so again there's not a lot of truth to the common perceptions of AMD when it comes to the products that most people would buy, but as you note these perceptions will take some time to 'right'.

IdiotInCharge said:

Where I'm coming from: with respect to enthusiasts that can use more cores, AMD has a winning product. However, for the average business user / desktop user / mobile user? More cores just increase cost, power draw, noise, and heat. Nothing they do gets actually faster. That's where the volume is, and that's the market that AMD has to win. Further, for stuff that is CPU dependent, faster cores matter more often than more cores.

Click to expand...

To a point. There is a countervailing force at work, too. Hardware Unboxed did a recomparison of the R5 1600 and the 7600k. At launch, the R5 was accounted a bit slower at gaming than the 7600k when you average out a bunch of games. Today, it is somewhat faster when you do the same, though the 7700k still eats the 1800X's lunch here - at least in average frametimes. But underlying point is, 4c/4t is no longer enough. Developers are maxing out a 7700k at 4c/8t. With core counts increasing, there is a trend - albeit a delayed trend - of developers using those extra available cores. And many business use tasks can benefit from core scaling. Financial uses, developers (compiling, multitasking between heavy applications), content creation, etc...

There is a delay in game developers using these resources, and an even longer delay, in some cases, for business use. But it is happening. So if AMD can stay in the game long enough, they create a good justification for preferring their products to Intel: "we give you more horsepower for future software." Business purchase cycles generally last a long time. Getting an extra year or two out of the hardware because more CPU resources are available could be a huge money saver. Of course, the counter argument is that Intel is scaling core count too, just a step or two behind AMD.

And I'll point it out again: AMDs refusal to put graphics on their higher core-count parts is hurting them here significantly. The volume markets don't care about discrete graphics. They just need something. Intel gives them that, and it works very very well for their purposes.

Click to expand...

I strongly agree here. With that I/O die, I'm sure they could have packed in a super-basic Vega or Navi derivative - not even at the level of their APUs, just something to get the job done. I'm very surprised they didn't. I think this was a misstep for them. But with the dual-die option on tap, maybe we see an 8-core die + GPU setup in the near future to cover this market. Some folks have been saying they know AMD won't do that (for some reason), but why wouldn't they? Seems obvious to me. Maybe they just needed some extra development time for it - that, and their APU products are usually half a generation behind these days.

So when I say 'barely competitive', I do mean toward the whole market. You'll see me recommending Ryzen CPUs for enthusiasts regularly, but for your average desktop or laptop, AMD is neither targeting the space fully nor are they getting the OEM attention necessary to be a volume producer.

Click to expand...

For the desktop, I think they are positioned okay. For el-cheapo office rigs, they offer the APUs, and AFAIK, those sell pretty decently (considering AMD is barely returning to the market at this point). But laptop... the main reason they can sell any of them is more because of stupidity on the part of system builders, offering ridiculous high-end Intel mobile CPUs (and charging accordingly), then pairing off with mediocre GPUs, or just using the iGPU. From what I've seen, low-to-mid range Ryzen builds are much more sensible. But this is an accident, and AMD shouldn't count on it continuing.

Even if they don't make a misstep- they have to get competitive with Ice Lake-level IPC and soon. That's another jump above Zen 2 / Ryzen 3000, and as it stands, getting more IPC out of the base Zen core is going to get harder and harder.

Click to expand...

If they do it by ~2021, they'll be fine. If the next refresh, in 2020, gives AMD another +3% IPC over previous, and a minor frequency bump, as Zen+ did, and AMD is offering extra cores/threads compared to Ice Lake, then they buy themselves another year even if desktop Ice Lake drops next year and matches (or slightly exceeds) Zen 2 clockspeeds. But they will need a +10%-15% or more IPC bump by 2021. Or, of course, a comparable frequency bump (but that seems unlikely). They executed on this well twice. A tock and a tick and then another tock, in the Intel parlance. I'm inclined to give them the benefit of the doubt, now, and suggest they can do it.

While it's not really true today that Nvidia is the 'stronger, higher-performing brand' in the mid-range, though that does depend significantly on price, AMD has a long reputation of less than stellar graphics drivers and OEM coolers. Nvidia has a reputation largely the opposite, today.

Now, AMDs drivers are quite good today, and third-party coolers are really the same between AMD and Nvidia, so again there's not a lot of truth to the common perceptions of AMD when it comes to the products that most people would buy, but as you note these perceptions will take some time to 'right'.

Click to expand...

I remember even in the ATi days, the driver reputation was rather poor. If AMD can change that perception someday, it would be excellent for them.

DuronBurgerMan said:

Financial uses, developers (compiling, multitasking between heavy applications), content creation, etc...

Click to expand...

Bigger point is that these uses exist yet are a tiny fraction of the overall PC market. I'd almost call these users 'low-end workstation users', but they'd still be mostly well-served by high-end consumer socket gear.

DuronBurgerMan said:

But underlying point is, 4c/4t is no longer enough. Developers are maxing out a 7700k at 4c/8t. With core counts increasing, there is a trend - albeit a delayed trend - of developers using those extra available cores.

Click to expand...

We're seeing that, but we've also been beyond four cores for a while now. Further, buying a 4C 4T CPU for high-end gaming was silly post-Sandy Bridge. That frametime difference has existed for quite some time, especially on everyday systems that people use vs. stripped-down gaming rigs. At the time, if your only choices were 7600K and R5 1600, recommending the 7600k would have only happened for a very specific subset of applications and not in any way forward looking, for someone say that planned to upgrade sooner rather than later. For stuff that depends heavily on single-core performance, it wouldn't have been a bad choice over Ryzen, it's still faster when overclocked, but man would it be limited.

DuronBurgerMan said:

Getting an extra year or two out of the hardware because more CPU resources are available could be a huge money saver.

Click to expand...

Depending on the business, sure. General reality is that businesses are either on a depreciation cycle so stuff gets tossed regardless, or they're on a 'run it till it breaks' cycle where there are many things that would likely fail before the CPU. As above up top, there's a relatively narrow market for real performance out of consumer desktops before you want workstations that are dedicated for your specific workload.

DuronBurgerMan said:

But with the dual-die option on tap, maybe we see an 8-core die + GPU setup in the near future to cover this market.

Click to expand...

This is what I see as most plausible, and yet, the cost involved as well as the power draw are likely to dissuade AMD from going down this route, i.e., where they replace the second CPU die with a GPU + HBM 'chiplet'. If they could get say RX560 performance out of it perhaps it would be a decent solution.

But that problem flows into the next...

DuronBurgerMan said:

But laptop... the main reason they can sell any of them is more because of stupidity on the part of system builders, offering ridiculous high-end Intel mobile CPUs (and charging accordingly), then pairing off with mediocre GPUs, or just using the iGPU. From what I've seen, low-to-mid range Ryzen builds are much more sensible. But this is an accident, and AMD shouldn't count on it continuing.

Click to expand...

AMDs challenge in mobile is efficiency. Flat-out, if they can't drop an APU into an XPS13 or a Ryzen 7 / 9 into an XPS15 and get similar battery life to the current Intel parts, then they don't compete. And while using 7nm might get them close, given how far away they are with Zen+ vs. Intel's current 14nm CPUs, and given that Intel's 10nm CPUs are looking really good in the mobile space, it seems that AMD needs a bit of a 'come to Jesus' moment to really justify their mobile product line.

Now, don't get me wrong, AMD is bringing performance- but as with your average consumer and business desktop user, that takes a back seat to other factors.

Short version of all the above:

AMD is winning the DIY enthusiast desktop market. They were already doing this with Ryzen 2000 series, that should only increase with Ryzen 3000.
Intel is winning the Laptop market and OEM desktop market. This is most of the PC market.

We have to wait for AMD 7nm to see what they can do to Laptop/OEM.

It will likely be an even LONGER wait until Intel starts winning back desktop DIY.

IdiotInCharge said:

AMDs challenge in mobile is efficiency. Flat-out, if they can't drop an APU into an XPS13 or a Ryzen 7 / 9 into an XPS15 and get similar battery life to the current Intel parts, then they don't compete. And while using 7nm might get them close, given how far away they are with Zen+ vs. Intel's current 14nm CPUs, and given that Intel's 10nm CPUs are looking really good in the mobile space, it seems that AMD needs a bit of a 'come to Jesus' moment to really justify their mobile product line.

Now, don't get me wrong, AMD is bringing performance- but as with your average consumer and business desktop user, that takes a back seat to other factors.

Click to expand...

On the CPU side, this is definitely their weakest link. However, I think Zen 2's chiplet design brings forward some opportunities for them. Original Zen and Zen+ weren't great on laptops. They had some bright spots (the decent integrated GPUs brought forward some possibilities), but AMD had to push Zen and Zen+ past the efficiency sweet spot to get competitive performance without a core count advantage. And even then, Intel (paradoxically) is leading on mobile core count offerings. My recent laptop buy was Intel for this reason.

Zen 2 may provide opportunity to change this. Efficiency appears to have improved significantly. And I think there's something they can do with this dual die setup. But they haven't done it yet, and we'll probably have to wait until early next year to see what they can bring to the mobile table here. AMD has a lot of IP they can leverage with regard to providing a fully integrated solution.

Although I continue to maintain that Ryzen on mobile is a pretty decent buy for low-end and low-to-mid range laptops, for the reasons I already stated - plus AMD has been smart on mobile pricing (and that goes into the business purchasing calculations as well). Mid-range and above on mobile, no. Go Intel. I did.

DuronBurgerMan said:

Although I continue to maintain that Ryzen on mobile is a pretty decent buy for low-end and low-to-mid range laptops, for the reasons I already stated - plus AMD has been smart on mobile pricing (and that goes into the business purchasing calculations as well). Mid-range and above on mobile, no. Go Intel. I did.

Click to expand...

My ultrabooks are all Intel, and I mostly don't mind the compromise, but I think what I'm looking for is a 14"-class folder with real connectivity (the dongles, they multiply...), better battery life, and a strong APU with VRR and 120Hz support on the display. And right now, this is something that only AMD can do- if you run a discrete part, you can either have dynamic GPU switching between integrated and discrete, or you can have VRR- not both at the same time.

I want both

DuronBurgerMan said:

Not me. I think Polaris was priced perfectly. One of AMD's best GPU moves in recent history.




Navi pricing is $449-$499 for 5700XT and $379 for 5700. 2070 and 2060 competitors, respectively. Drop that by $50-$100, and they'd fly off the shelves. The truth of the matter is that AMD is the budget brand with respect to GPUs. If they wanted to compete on absolute performance, it would be different. The AMD purchase is going to be primarily price-driven. I don't necessarily like this. I like the days when both were competing for absolute performance. But it is what it is.



As harmattan said, AMD does not exist in a vacuum. If they want to increase marketshare, they must create a compelling reason to prefer them over Nvidia. Since absolute performance isn't in their wheelhouse, it must be price/perf. And it must be significant enough to move the needle - as roughly equivalent pricing will favor Nvidia as the stronger brand. The greater the price band difference, the more compelling the reason to favor AMD.

Chad's earlier point, however, was well taken. AMD may not care about increasing marketshare, and may instead be focusing GPU development on consoles, integrated solutions, etc... In other words, internally they might view the discrete GPU market as the leftovers, so to speak. If they sell some good amount of discrete cards, great. If not... well, they will sell a lot of GPUs anyway. It would explain a great deal.

Caveat: if the 3rd party benchmarks show the Navi cards as significantly faster than their respective Nvidia competitors - not just a little bit, but a lot - then their current pricing would be fine.

Click to expand...

Let's say that AMD launched the Radeon RX 5700 for $299 and Radeon RX 5700 XT for $349.

Is NVIDIA going to do nothing and let AMD take its market share?

Heck, no.

NVIDIA would cut the prices of the GeForce RTX 2060 to $299 and GeForce RTX 2070 to $349.

So, cutting the prices won't improve AMD's position.

Furthermore, it worsen AMD's position (relative to NVIDIA's) because AMD don't have the economies of scale that NVIDIA does.

Saw this over on the AMD subreddit. YMMV. The system specs are in the Reddit thread, but my eyes weren't able to read them.

cageymaru said:

Click to expand...

You are so evil cageymaru . How can I resist buying this now?

3DMark Time Spy Result shows Radeon RX 5700 XT close to GeForce RTX 2070

https://www.guru3d.com/news-story/3...deon-rx-5700-xt-passing-geforce-rtx-2070.html

trash leak is trash

https://wccftech.com/amd-radeon-rx-5700-and-nvidia-geforce-rtx-2070-super-gpu-benchmarks-leak/


interesting that the timespy leak score (8700pts) matches radeon vii (~8900).

Auer said:

3DMark Time Spy Result shows Radeon RX 5700 XT close to GeForce RTX 2070

https://www.guru3d.com/news-story/3...deon-rx-5700-xt-passing-geforce-rtx-2070.html

Click to expand...

wow just posted on this too, nvidia has a fair bit of a lead in timespy vs competitive products (i.e 2080vs radeon vii) so that's actually pretty impressive.

DuronBurgerMan said:

If they want to increase marketshare, they must create a compelling reason to prefer them over Nvidia.

Click to expand...

I think they could do better by just releasing new products in line with when NV releases new products. Say if Navi came out Last year, or even 6 months ago, much closer to RTX launch, then people would have options. However, 8+ months after RTX, Navi is just coming out... but meanwhile people already have their NV cards. Even if performance is similar to RTX 2060/2070 etc, many people have already bought the 2060/2070.

sabrewolf732 said:

wow just posted on this too, nvidia has a fair bit of a lead in timespy vs competitive products (i.e 2080vs 2070) so that's actually pretty impressive.

Click to expand...

I think 5700xt might be getting up to VII performance if it can be OCed to 1900 and sustain those clocks. it looks like they really fixed the issues they had with GCN in some games where it just couldn't perform on par with nvidia. Like Assasins creed performance AMD showed, which really led me to believe you won't be seeing AMD slower in games where GCN lacked. I think you won't see RDNA lack in dx11 games either from the looks of it.

TheRookie said:

Let's say that AMD launched the Radeon RX 5700 for $299 and Radeon RX 5700 XT for $349.

Is NVIDIA going to do nothing and let AMD take its market share?

Heck, no.

NVIDIA would cut the prices of the GeForce RTX 2060 to $299 and GeForce RTX 2070 to $349.

So, cutting the prices won't improve AMD's position.

Furthermore, it worsen AMD's position (relative to NVIDIA's) because AMD don't have the economies of scale that NVIDIA does.

Click to expand...

Yep! People don't think that way, they think AMD would gain market share if they sold these for cheap. Look what happened with Polaris. It just doesn't work that way. I think the price is what it is because of Nvidia's positioning. I won't be surprised if AMD has another 50 or so price cut priced a few months down the road.

NKD said:

I think 5700xt might be getting up to VII performance if it can be OCed to 1900 and sustain those clocks. it looks like they really fixed the issues they had with GCN in some games where it just couldn't perform on par with nvidia. Like Assasins creed performance AMD showed, which really led me to believe you won't be seeing AMD slower in games where GCN lacked. I think you won't see RDNA lack in dx11 games either from the looks of it.

Click to expand...

What's the stock Graphics score for the VII in timespy? A quick look at my results gives my stock PNY RTX2070 OC a 9043 graphics score vs the 5700XT's 8719. I expect the VII to do a whole lot better than my 2070 no matter how much I OC it.

NKD said:

they think AMD would gain market share if they sold these for cheap.

Click to expand...

The only way AMD gains market share by lowering prices is if they lower their prices below the point that Nvidia is willing to lower prices on Nvidia GPUs. That means that AMD would have to have SKUs that are cheap to produce that AMD is willing to make essentially no money on because Nvidia can likely match any price decrease AMD attempts.

IdiotInCharge said:

The only way AMD gains market share by lowering prices is if they lower their prices below the point that Nvidia is willing to lower prices on Nvidia GPUs. That means that AMD would have to have SKUs that are cheap to produce that AMD is willing to make essentially no money on because Nvidia can likely match any price decrease AMD attempts.

Click to expand...

Hasn't that been the case the entire time though? Polaris could have been had for 50 below pascal but people still bought pascal. AMD knows this hence they are not really going to start a price war. They will at the top end when they have true high end with RDNA 2 or navi 20 whatever its called. Nvidia will happily sell their card 50 above AMD either case. At then end AMD loses.

Auer said:

What's the stock Graphics score for the VII in timespy? A quick look at my results gives my stock PNY RTX2070 OC a 9043 graphics score vs the 5700XT's 8719. I expect the VII to do a whole lot better than my 2070 no matter how much I OC it.

Click to expand...

about 8800.

sabrewolf732 said:

about 8800.

Click to expand...

Really?

I would have expected more.

Auer said:

Really?

I would have expected more.

Click to expand...

Nvidia does significantly better in timespy, that's why parity in it bodes well for amd