Do you think AMD will be able to compete with Nvidia again during the next few years?

[Biscuit]

If we have learned anything in recent times, it's that we should never count AMD out.

Intel stagnated for years with the core architecture, giving tiny performance bumps, having a tonne of artificially hobbled SKUs, and blocking upgrade paths by changing their underlying platform year after year. All of this without even the smallest indication of trickling HEDT technology to mainstream.

In servers they had (... or still have, tbh) a complete monopoly and were charging whatever they wanted for high core count parts. (although this is also to do with their monolithic manufacturing process...).

Laptops and mobile, AMD's parts just did not compare. Sure AMD had some interesting APU parts over the years but they weren't appealing enough to get into any mass produced, high end devices.

Intel left the entire PC market to stagnate as there was simply no reason for anyone who had a second generation(or newer) quad to change to upgrade. They left the door wide open for AMD, who did an incredible job in creating a profitable and scalable platform to get back into contention for just about every market.

Although it's easy to make the comparison, I don't however, see the GPU market being the same as the CPU one.

Nvidia and AMD, over the years between the 4*** series to the R9 3** series were pretty equal with Nvidia; trading price/performance/watts metrics from generation to generation, but they haven't seen a relative hold on the market to represent that. People buy Nvidia because, "it's the way it's meant to be played".

We're now in a situation where Nvidia doesn't have to release their parts into the mainstream at all, but due to AI and the various sub-markets that relate to it, they are still being pushed to develop them.

Look at Volta. They claim that it was never designed for consumer products which is why it only ended up in an uber expensive titan or quaddro sku, but you better believe it would have found its way into them if there was any competition...

We still dont really know what Turing is going to bring to the table, I suspect we never will know what Turing COULD bring to the table, as Nvidia has nothing pushing them to drive their higher end parts down the market segments. They will do just enough to make people buy them, but not enough to really shake things up.

I think we all thought intel had parts sat waiting to drop on AMD to keep control of the market, so we were a bit surprised when the best intel could do wasn't all that impressive. With Nvidia though, they haven't left the door open, there isnt really anyway for AMD to slide in without a swift Nvidia response that I can see.
If Navi really is just a 580 replacement, then I suspect the mainstream GPU market will continue as it is, with Nvidia dominating the high end and AMD doing an OK job at providing a good value option for the larger middle section of the market.

If I was to offer a suggestion about what they should do, it would be to target the frustration that gamers have with Nvidias 20XX series, and that is the lack of focus towards CURRENT games in 4K. If they can get a part out that pushes the framerate of current titles with current technology beyond what Turing can deliver @4K, they could undermine Nvidia somewhat. I just don't know how they could do that... but like I said, don't count them out.

[scaryjim]

… AMD had some interesting APU parts over the years but they weren't appealing enough to get into any mass produced, high end devices. ...

... People buy Nvidia because, "it's the way it's meant to be played". ...

Two different statements, same key point. AMD have always struggled with developer/OEM relations.

To be fair, there was a period, around the DX10/DX11 switch-over, where AMD's developer relations where actually doing quite well - I remember at least one Hexus graphics review commenting that they were using a majority of AMD-sponsored titles for the first time - but it rather seems the damage was already done, and I guess the lack of immediate results, combined with tight purse strings, led to the devrel programme getting chopped (it certainly dropped off again).

But for OEM relations AMD have always struggled. Even when they had the technically better processors, they struggled to get through Intel's (sometimes questionable, if not outright illegal) hegemony. Their APUs really should have sold by the bucket load in mainstream laptops, but they mostly ended up in cheap low-end machines hobbled by bad configurations, and even the ones in good laptops didn't really sell, because there was no effort (maybe just because there was no budget) to make them seem desirable. Perhaps there is simply no getting past the marketing juggernaut Intel has thrown behind making people think that you need an Intel processor to have a good Windows PC....

[Biscuit]

Two different statements, same key point. AMD have always struggled with developer/OEM relations.

Well I think they also struggle with the tribalism of people who are easily dazzled by Nvidia's superior marketing. As much as I like pointing the finger at OEMs, I think the every day "gamer" has a fair bit of blame in AMD's GPU circumstances. No point trying to have a halo product if everyone is going to buy Nvidia anyway.
I'm sure someone else posted something along these line, perhaps more elequently, elsewhere on the forums recently... might have even been you

[kalniel]

Re: nVidia's marketing, I really think it's time that sites and forums like Hexus started to operate a mandatory transparency policy and request that all staff have to include a disclaimer that states if they have received (and kept) any free products from a company. It'd be nice if the same thing was also part of the terms of use in the forums too, but I guess you can't stop people ignoring it even when it's pretty obvious when a new sign-up is a shill.

[Biscuit]

There was a big controversy recently around PCper, their "side" business dealing and not disclaiming some of the cross over benefits. Once again it was Jim from Adored TV on YT who highlighted it...
I would say the saga is well worth a watch if you want a bit of tech drama.

[peterb]

Re: nVidia's marketing, I really think it's time that sites and forums like Hexus started to operate a mandatory transparency policy and request that all staff have to include a disclaimer that states if they have received (and kept) any free products from a company. It'd be nice if the same thing was also part of the terms of use in the forums too, but I guess you can't stop people ignoring it even when it's pretty obvious when a new sign-up is a shill.

Thats a fair point - however I don't know whether HEXUS receives freebie items for review from mfrs (which of course would give the mfr the opportunity to 'tinker' with the sample).

But I don't I think for one mS that a HEXUS review would be swayed by the method that the item was obtained.

[kalniel]

Thats a fair point - however I don't know whether HEXUS receives freebie items for review from mfrs (which of course would give the mfr the opportunity to 'tinker' with the sample).

But I don't I think for one mS that a HEXUS review would be swayed by the method that the item was obtained.

Oh I'm certainly not questioning the objectivity of Hexus, but it would help a consumer get a better picture of the review landscape if (say) 90% of all the writers they were reading from were continuing to use free products beyond a review period.

[Hoonigan]

Re: nVidia's marketing, I really think it's time that sites and forums like Hexus started to operate a mandatory transparency policy and request that all staff have to include a disclaimer that states if they have received (and kept) any free products from a company. It'd be nice if the same thing was also part of the terms of use in the forums too, but I guess you can't stop people ignoring it even when it's pretty obvious when a new sign-up is a shill.

I'm a review editor for a website (won't mention which as we're a competitor to Hexus). We keep 99% of the stuff we receive, either for later testing or for personal use. The odd thing requires us to rotate it around, especially expensive laptops and the likes, but it's often seen as a perk of the job.

I have a couple of racks of Screwfix shelving here that's stacked to the ceiling with bits and pieces from all over the Parish.

I would never say it sways my opinion of the product, because we get to keep just about everything.

EDIT: To add a little further. I've spent this week going over our fan reviews, re-testing and ensuring all of the results are correct, while adding in something we get asked for a lot. So it's handy to keep hold of products for reasons like this.
Also, something like testing a keyboard, I have at least one of every common mechanical keyswitch here, useful for testing if you want to go in-depth.

[Millennium]

There was a big controversy recently around PCper, their "side" business dealing and not disclaiming some of the cross over benefits. Once again it was Jim from Adored TV on YT who highlighted it...
I would say the saga is well worth a watch if you want a bit of tech drama.

Yeah I caught this at the time. I think Jim was being overly harsh, but with good subconscious intent, on this occasion I side with PCPER, who do some of the best technical content out there (so good in fact I don't bother following it sometimes on the podcast). Now they have a lovely table with every review showing how it might or might not be biased bi their peoples.

[kalniel]

I'm a review editor for a website (won't mention which as we're a competitor to Hexus). We keep 99% of the stuff we receive, either for later testing or for personal use. The odd thing requires us to rotate it around, especially expensive laptops and the likes, but it's often seen as a perk of the job.

I have a couple of racks of Screwfix shelving here that's stacked to the ceiling with bits and pieces from all over the Parish.

I would never say it sways my opinion of the product, because we get to keep just about everything.

EDIT: To add a little further. I've spent this week going over our fan reviews, re-testing and ensuring all of the results are correct, while adding in something we get asked for a lot. So it's handy to keep hold of products for reasons like this.
Also, something like testing a keyboard, I have at least one of every common mechanical keyswitch here, useful for testing if you want to go in-depth.

Keeping stuff around for retests etc. is one thing. But keeping a review product for personal use is another. The way biases work mean that even if you are adamant that it's a perk of the job that won't affect you, it probably is affecting you on an unconscious level.

In an ideal world, you should be renumerated enough that there's no need for perks of the job, or at least that you make sure never to review something in the same field/by the same company - so for instance if you can possibly have several people in your team you might be able to segment the reviews so that people who review GPUs never get the free GPUs, but might get free keyboards or whatever that they never review themselves.

Failing that, being 100% transparent is a good starting point because it makes the writer (and reader) aware of the possible perception of bias.

[Millennium]

Going back to AMD, I think they are very likely to get the next gen consoles (at least the important western one) what's baffling is that this is probably getting really annoying for game devs because they have to develop on something like Vulkan for console OS though because nVidia are not very friendly they may end up hobbling all their PC releases (over and above not having a compiler they would use optimally for every CPU - if that's correct still)


_

[emilsobo]

Currently i have don't have high hopes of amd making a comeback since their vega series was kind of lacking and pricy I really want them to make better gpus since nvidia has a market monopoly since they have no competion and they can realease what ever they want eg the 1080ti is bacically the 2080 but cheaper im fairly sure if amd had some better gpus the 2070 would outpreform the 1080ti

[watercooled]

Correct me if I'm wrong, by that sounds like the same old 'I wish AMD had better GPUs so I could buy something better from Nvidia' statement that is probably a part of the reason AMD has pulled back from the top-end consumer market for the time being. Why bother, when marketing wins out, people split hairs over nonsense, and flock to Nvidia anyway?

Having said that, IMO Nvidia are right now in a weaker position, in terms of products and pricing, than they have been for some time. It don't imagine it would take much persuading to sell a decent alternative to the 2000 series right now and it would be difficult for Nvidia to undercut a competitor with their currently huge die size for a given performance. Not that any competitor can just up and release something as a knee-jerk reaction to market conditions. I've said it before but IMO the release of Turing as it was, was a pretty daring move by Nvidia and they must have predicted (correctly) the market wouldn't have moved much since Pascal else it would be a very hard sell. Or maybe they would have actually made an effort to make the ray tracing gimmicks look remotely interesting in reviews if they had competition? I still can't shake how poor Turing looked in reviews given how well Nvidia usually market things! Heck, even when there's nothing in it performance-wise, they still cobble together some storm in a teacup and have crowds singing from the marketing hymn sheet about 20W extra power consumption or something! Or how SLI is better on Nvidia, though when AMD actually fix it to the point of being objectively better, Nvidia basically destroy that feature altogether. Funny how something goes from being a major selling point to impossible in one generation and next to no-one pipes up about it despite flaming the competition for much the same.

WRT the 2070 performance - I can't personally see why the 2070 would comfortably beat the 1080Ti, not least because they're on the same node so yields aside Nvidia themselves aren't really seeing more transistors for the same cost. On the contrary, they've spent shedloads on new masks etc so from a financial standpoint it would have been cheaper to just refresh Pascal and shift the tiers for example.

[philehidiot]

Remember AMD's strategy slide where they showed they were targetting 3 peaks on the sales graph. The low end had a peak (maybe this is their integrated graphics section), the mid range had a peak which is where we see the RX580ish and then the high end had a peak at Vega which kinda matched the 1070Ti/1080 performance. The sales really dropped off towards 1080Ti performance and so they decided not to waste R&D competing there and just accept the "AMD can't compete in the high end" accusations which may be true or they may be a result of simply not wanting to and pitching their products slightly lower down.

Assuming they're going to continue with that and looking at Turing, I think that again they won't even try to outdo the 2080Ti, they'll rejig Vega to make it cheaper to build and slightly faster and cooler and that'll probably see it match the 2080. There are musings that the new RX models will have 1070 performance but again it looks like it's going to be last generation performance and too late to make a real impact on the market.

[watercooled]

The whole 'x can't compete with y because they don't have a top-end product' is a fallacy, and it wouldn't surprise me if halo products serve more as a marketing tool than a profit maker themselves!

I agree, and IMO it doesn't make much sense for AMD to try to push to the top-end market at the moment; there's not much room for competition in a fairly low-volume market what with fab costs nowadays, and if the e-peen crowd want to make a rod for their own backs by continuing to pay more and more for less and less then so be it! It's not like AMD ever see much benefit from dethroning Nvidia. Check what happened with the 290X which was a great GPU in many ways, but because of an admittedly rubbish stock cooler, and something like 20-30W more power draw it was almost universally panned with such idiotic comments.

[philehidiot]

If you look at the performance (or might have been price, but they're reflective of each other) Vs sales graph, there's very little at the high end but just back from that there's a massive spike which is pretty much where Vega 64 sat before the RTX series came out. I'm sure if AMD really wanted to throw money at the R&D they could provide a card that would compete at that ultra high end but I doubt that's where their R&D was focussed looking at that strategy slide. As you say, it's probably a marketing thing but the thing about marketing things is that they're marketing things rather than directly profitable things. What I mean by that is that, whilst the 1080Ti / 2080Ti might not be a profit makers in themselves, the extra custom they bring in to the rest of their range, plus the R&D bonus they provide (little tips and tricks to up performance in the ultra high end may filter down to the lower down cards to improve efficiency, etc) might be seen valuable enough by the company to carry on with it.

The idea that "Nvidia make the best card therefore they make the best cards" works for loads of brands with halo products and people just seem to forget that all companies have a range and that they will all cut corners to meet a price point at different levels in that range. It depends only on where those corners are cut and how good their engineers are at cutting costs whilst retaining performance / longevity.

I can build you a bridge across a river. It'll take decades, it'll cost you a fortune and on top of that it might not last very long but I can do it. The real skill comes in building a bridge on a budget, within a timescale and that lasts. In many ways therefore, the halo product requires a totally different set of engineering skills to the lower down products as the price is often the secondary objective with e-peen being primary. And RGB being absolute.