QOTW: Does AMD Ryzen need to be faster than Intel Core?

[Percy1983]

I think this is a really interesting point, actually. One of the issues for Bulldozer/Piledriver was that in some DX11 games it had low enough single-thread performance for it to become the bottleneck. So if Ryzen can improve performance to the point where it's no longer the bottleneck, it doesn't need to outright beat Intel in every benchmark.

The bigger issue is that for most day to day tasks even a mid-range AMD APU is already perfectly adequate, yet many people still equate Intel with "good PC" and AMD with "cheap and second rate". That's a stigma that's not going to go away any time soon, and is only likely to be shifted by AMD not only beating Intel with a new CPU release, but staying clearly ahead of them for several years. That's going to be a pretty demanding task...

This is so true, my second PC, HTPC netbook and laptop all have AMD APUs as they are epic for the money.

I will say I only bought the laptop this week and when looking in PC worth it was hard to even find one with an APU, likewise on the website if you filter by processor type you have to click see more to list anything AMD.

Interestingly if Ryzen is everything I hope at the right price my house will become intel free, this is nothing to do with hating intel its all about buying the best parts for me.

[ET3D]

It's a question that has different answers at different levels.

Personally, all I need is Bristol Ridge. Been waiting to put that in my HTPC for almost a year, but AMD doesn't want to sell it to me. I don't much care about Ryzen vs. Intel in this respect, but hopefully Zen based APU's, if and when they appear will be a decent upgrade over Bristol Ridge (if and when it appears .

For AMD fans, I think that Ryzen will be good enough even if it's not as good as Intel.

For consumers in general, I think that Ryzen needs to beat Intel CPU's in at least some games. And when I say 'Intel CPU's' I mean a Core i5. That to me will be the measure of AMD's success. If AMD wants to be more mainstream and not just relegated to fans, it needs to be at a 'no brainer' position. It doesn't need to be faster at everything, but it shouldn't be at the classic 'I'm only buying this because I don't want to spend more money on something faster' position.

For servers, performance / power matters.

[peterb]

No matter if Ryzen is equally competitive to Intel chips a few thousand sales to enthusiasts is not going to make a big difference to AMD, they need to get a good advertising/sales pitch like the annoying "Intel inside" ads/logos and get inside prebuilt systems from the likes of Apple, HP and other retailers.
Most folk who are not tech followers have never heard of AMD and if they know any CPU manufacturer then its going to be Intel, AMD need to change that to be competitive.

This. It's volume sales to the likes of HP and Dell that AMD need to be pitching too, for corporate sales, along with embedded systems. In these cases saving a few dollars to achieve an acceptable performance is crucial. Purchasers of commodities don't care what is in the box, so long as the box delivers. AMD needs high volume sales to system mfrs.

[jonah]

It just needs to be similar but better value.

[rainman]

A lot of folk appear to be in the camp which says "if it's close enough to Intel performance and cheaper then I'll buy AMD".

Being "close enough" isn't going to cut it. Just being cheaper and close enough isn't going to cut it either. There is another dimension which a lot of folk are missing out on. There is a segment of the market which won't stump £1000 on Intel's hottest CPU, but would happily spend double the highest AMD offering for only 25% more performance. That's the reality right now. Most people that have bought an i7 in the last five years probably could have already brought an AMD CPU simply because their main performance requirements came from either their GPU or an SSD, and not necessarily the CPU. Nevertheless, they spent way more than they needed to on a K Sku 4c/8t i7 than they really needed to, regardless of whatever AMD had on offer at half the price. So the question remains is "what remains for AMD do to get this market share?" because some folk will just buy intel and pay the premium for no real justification than bragging rights, or because their friend who knows about this stuff told them too.

AMD can and will undercut Intel to gain market share, but it won't be enough. It actually needs to be better (at least in some workloads) and not merely close enough, otherwise it's just the exact same scenario we have now.

We need AMD to be better in order to keep Intel honest, otherwise I'm going to bypass their £200 offering for Intel's £350 offering, which is exactly what I've been doing this last 5 years.

AMD need to be better or go home.

[Smudger]

For me, I've never gone for an i7, apart from in my laptop, cos that was the only way to get the screen I wanted, but if AMD's mid-range line beats the equivalent i5 and is cheaper, I would get that when it came to changing. I have no interest in the XTREME! versions, I think you're into the realms of diminishing returns there.

[Rad77]

I prefer to give my money to AMD, so in my case the answer would be that unless Intel is offering a significant price/performance advantage then Ryzen only has to be "good enough". I'm pretty sure there must be quite a few people with a similar bias (although I realise that's not going to affect the mass market sales which is presumably where success will live or die...)

It will be very nice to see AMD once again able to offer both a competitive price and equivalent performance right up through Intel's product range, rather than just at the budget level, as that has to help when pushing the brand to oem's and the wider public.

[virtuo]

I like squeezing performance out of chips, Ryzen needs to beat intel out of the box (by performance, not necessarily Ghz) and clock a nice lead with some gentle OC and cooling.

I doubt they'll pull it off at the top end, but I'm open to being pleasantly surprised.

[Corky34]

AMD can and will undercut Intel to gain market share, but it won't be enough. It actually needs to be better (at least in some workloads) and not merely close enough, otherwise it's just the exact same scenario we have now

^This^ Despite most people saying fast enough but cheaper is what they want that would just put AMD in the same place its been in for the last 5-10 years, what AMD need is for Ryzen to be faster better in at least some form, or metric, and by how much and how many of those metrics it beats Intel the better.

Its already been shown the market will spend more money on a faster better more expensive CPU despite it not necessarily being cost effective to do so, what AMD need are for people to read that Ryzen is better than the alternative.

[CAT-THE-FIFTH]

The main issue AMD has had with CPUs is the power consumption and the fact they have been selling largish chips for not much money.AMD does not need to beat Intel in single core performance,getting reasonably close is enough,as long as power consumption and core size goes down. This will be more important for the areas in which it is competing in - servers,supercomputers and laptops.

The whole "it has to beat XYZ at any cost" is a very enthusiast E-PEEN mindset,when in reality most of the computing devices people are buying are not powered by top end DESKTOP CPUs.

Plus TBH,I would be shocked if Ryzen could best Intel in single threaded performance,especially using GF as their foundry.

I know loads of people who work in the computing area,gamers,scientists,etc. I am yet to see anyone of them buy anything better than a £300ish CPU and a 4C/8T one at that. The reality,is that unlike graphics cards,people are less likely to be spending £400+ for most builds.

For instance,if a 4C/8T Ryzen CPU was around Haswell level single threaded performance for gaming and was around Core i5 6600K level pricing with a reasonable included cooler and cheaper motherboards most sane people would seriously consider the AMD chip since a £100 difference is better spent on the graphics card. If AMD can get to Broadwell level for gaming that would be an achievement IMHO.

If a 6C/12T one was around Core i7 6700K pricing,again it would probably be fine.

But then wait for the enthusiasts to cart out examples of idiotic games which they are willing to throw money at which are so poorly optimised and hardly anybody plays or poorly optimised and millions play on less than great hardware(WoW or WoT being some of those examples). Then they will wheel out stupid high voltage maximum overclocks arguments,etc even though not a single one of them can guarantee those clockspeeds for everybody.

I can even appreciate someone spending a £1000 on Titan X since even that might not be fast enough for 4K games as time progresses.

The E-PEENers for bragging rights will still pay £320 for a Core i5 6700K,so in the end I doubt AMD is going to convert such people,as such people will always buy Intel but want AMD to reduce the price of their Intel purchase for them.

Remember loads of enthusiasts still bought the Pentium 4 in droves,even the Athlon series chips for most of their lifespans were a better purchase.

I still remember back to the days of the Phenom II X4 955BE,when people were still pushing the more expensive Q9550 for hardly a performance difference(once pricing settled down after launch the Phenom II could be had a decent amount cheaper),even though at the time some of the AMD motherboards were better specified for a similar price.

In the end the rise of ARM based computers indicates,that having the fastest chips is not the only important metric when it comes to things nowadays. Power consumption and cost of manufacture are just as important.

[scaryjim]

... Despite most people saying fast enough but cheaper is what they want that would just put AMD in the same place its been in for the last 5-10 years ...

For the last 5 years, AMD hasn't been within 50% of Intel's flagship single threaded performance. At it's best in the last ten years it hasn't been within 10%. OTOH for some heavily threaded tasks the FX 8350 [i]outperformed[/]i the i7 3770k; performance depends on the metric you're looking at. Intel didn't need to beat AMD at every test in order to be seen in a more favourable light.

... what AMD need is for Ryzen to be faster in at least some form, or metric, and by how much and how many of those metrics it beats Intel the better.

FX 8350 was faster than the i7 3770k in some metrics. It was very close to it in a number of others. Typically, the ones that hammered lots of threads with integer tasks. Didn't do it any good, because in some areas Intel were so much faster. So I don't think Ryzen does need to be faster than Intel, not on a single metric. But it does need to be only marginally slower for everything. Don't give Intel a metric where they can say "look, we're 50% faster than AMD". That's been AMD's problem for the last decade - not that they were slower for everything, but that they were a lot slower for one thing.

[Corky34]

Maybe i should have just quoted what rainman said and left it at that as it seems I've done a bad job of saying good enough but cheaper isn't really going to cut it.

[CAT-THE-FIFTH]

Yeah,but lest we forget CPUs like the Phenom II X4 955BE which by the time pricing settled,was a reasonable bit cheaper than a Q9550,had a better stock cooler,and the AMD AM2+ and AM3 motherboards had some decent specifications too - you could get motherboards with SATA3,USB3,XFire support,etc for much less than the Intel equivalents. The AMD chip also had a much better stock cooler too.

What were the arguments which were carted out - maximum overclock arguments and it consumed more power,hence if you ran the CPUs at 100% 24/7 the Intel CPU would be cheaper longterm,etc. Nobody needed SATA3,etc.

Some will buy Intel whatever AMD has - whether they are around the same performance,or even ahead. Lest we forget all those enthusiasts who still bought Pentium 4 chips even when AMD had superior price comparable ones(yes I know just before the Athlon 64 was released,the Pentium 4 started to claw back some ground).

[scaryjim]

... it seems I've done a bad job of saying good enough but cheaper isn't really going to cut it.

*shrug* I think there's a difference between "good enough" and "doesn't have to beat Intel but needs to get close". You're absolutely right in that what they have now is "good enough but cheaper than Intel". I don't think anyone is saying that as long as they can manage an incremental improvement on Piledriver that'll be good enough - but there are areas in which they can make a huge improvement over Piledriver and still not quite catch Intel.

A better perspective might not be to compare them with Intel directly, but to ask how much faster they need to be than AMD's incumbent CPUs. If they can do a 4C/8T Zen at the same price as an FX 8350 that's 50% faster in single-threaded tasks, I reckon that would make people sit up and take notice - but that still might not be enough to beat Kaby Lake.

So can AMD get away with being "good enough"? Probably not. But do they need to be faster then Intel across the board? Probably not. If they can be nearly as fast as Intel, that'll be staggeringly faster than their incumbent chips. And I reckon that'll do it for them.

[preter_s]

Do we really need anything more powerful? Sure we do. Yeah. Right.

But if ARM has shown us anything, its great energy efficiency and great value (low cost) will conquer the market.

Intel has been milking the consumer suckers for a decade at least. With incremental BS 10% improvements each time while creaming consumers with sky high prices.

AMD, do what ARM does. Be fast enough (match Intel), but offer it at a much lower cost (force Intel to sweat by cutting prices) and much greater energy efficiency (Gaia will appreciate it).

A pincer movement from ARM and AMD will surely end the Intel hegemony.

[preter_s]

Given how the demos show that AMD can do at a TDP of 95W what Intel does at 140W, I am personally more interested in how that translates to OEM portable computing products.

OEM platform wins for high end laptops and ultraportable 2-in-1 devices could also be a profitable segment for AMD. Not just gaming PCs.

AMD doesn't have to be the poor cousin to Intel scraping the bottom of the barrel with budget PC consumers.