AMD Ryzen 5 1500X and Ryzen 5 1600X (14nm Zen)

[Ozaron]

Competition is great though for the average consumer looking for a PC its now become a minefield again.

The 2C/4T i3-7350K is £170 (plus cooler so effectively £190), the Ryzen 5 1400 is £159 including cooler. The i3 needs a Z270 board to overclock (cheapest £90), whereas the cheapest overclocking B350 board is £72.

So totals are:
i3 = £280 (£170 + £20 + £90)
r5 = £231.

Agree that the i3-7350K is the wrong choice for... almost anyone, but he was aiming at "average consumer" and I don't know an average consumer who puts both money and studying (+ minimal risk?) into overclocking their brand new CPU. Average consumers don't get a choice on which board their Pentium is slapped into either, usually.

That said I'm not so sure about "minefield" given that the lowest end Ryzens have 4 cores and the worst Kaby Lake chips still have great IPC and single thread. No matter what cheapo option a store throws at you, you get a good deal if it's cheap enough, and an average consumer can deal with that. Only bad deals right now are old FX processors and mobile Intel chips for anything other than low end netbooks / laptops, IMO.

[scaryjim]

... Basically the "more cores is better" slant will be taken by the sales folk ...

Don't see why it would - it hasn't for the last however many year AMD has offered more cores @ the same price as Intel. Sales people are far more likely to say "You need an Intel processor" since that's the mainstream line that's been peddled for almost twenty years now...

Most consumers don't need an i3 7350k. They don't need an R5 1500X. The highest most consumers *need* to go up the product stack is an AMD A8 APU or the cheapest Intel i3 - and the i3's worth is questionable now Intel have enabled HT on the Kaby Lake Pentiums. I'm still doing the bulk of my day to day computing on an A10 4600M, and I'm really not feeling a burning need to upgrade.

For the mainstream consumer, AMD still haven't released anything worth talking about. That's intentional - they always said that Zen would come enthusiast desktop first (after all, Bristol Ridge APUs are actually filling out the mainstream OEM space quite nicely at the minute). Mainstream will come with Zen APUs, which I hope will come in 4, 3, and 2 core variants, all with SMT. They should make things very interesting in the mainstream - Ryzen's already shown that it can keep most of its performance while dropping TDP significantly, and at 35W it should be an excellent proposition...

[CAT-THE-FIFTH]

Intel really needs to start dropping some prices now - their current pricing stack is not really looking that great TBH!!

[aniilv]

I'm holding my breath for next 2 years to see if intel will start dropping price or actually care to make anything good again since they have not needed to improve since 2500k. It's sad, really.

[watercooled]

No matter what cheapo option a store throws at you, you get a good deal if it's cheap enough, and an average consumer can deal with that. Only bad deals right now are old FX processors and mobile Intel chips for anything other than low end netbooks / laptops, IMO.

I think the biggest 'minefield' in CPUs is Intel's daft naming scheme, where they use the Celeron and Pentium lines to refer to both Core microarchitecture processors, and their Atom line - two vastly different processors which share the same brand names.

I'm holding my breath for next 2 years to see if intel will start dropping price or actually care to make anything good again since they have not needed to improve since 2500k. It's sad, really.

Pricing - you can put that down to competition. Products, not so much, at least not in the short-medium term. Anything Intel comes out with will have been in development for 5+ years - just look how long it took from us first hearing about Zen, the hiring of Jim Keller and so on to it actually appearing on the shelves, and therefore how long AMD had to do the best they could with Bulldozer which didn't perform or clock how they expected. You don't pull back on your long-term projects because of speculation which could come back to haunt you - even without AMD, Intel have still been up against fierce competition on other fronts such as POWER and ARM in the server space, and mainly ARM in the mobile space, and the core they design needs to compete across the lot.

[scaryjim]

I think the biggest 'minefield' in CPUs is Intel's daft naming scheme, where they use the Celeron and Pentium lines to refer to both Core microarchitecture processors, and their Atom line ...

Let's be impartial; it didn't take very long for AMD to slide the 'cat' cores into the A-series of APUs leading to exactly the same situation (so an A6 APU could be a dual construction core or a quad cat core, for instance)

[watercooled]

Very true - they should carry completely separate and distinctive branding IMO, and I don't understand how they can do the company brands any favours either - if someone who doesn't understand this difference (so probably most people outside of enthusiasts) ended up with a small core system and was expecting big-core performance, it could tarnish their opinion of the whole company.

[kompukare]

Very true - they should carry completely separate and distinctive branding IMO, and I don't understand how they can do the company brands any favours either - if someone who doesn't understand this difference (so probably most people outside of enthusiasts) ended up with a small core system and was expecting big-core performance, it could tarnish their opinion of the whole company.

Speaking of brands, AMD should try to do a good reference design (or do a google and partner with someone to do one) for when the APUs come out. Intel's Ultrabook (and the previous Centrino) designs helped their brand a lot and AMD don't even have to (wouldn't be able to) try to use it force OEMs to buy their wifi cards etc. since the don't make any. Otherwise someone will sell a Ryzen 7 laptop with a single channel of RAM and a 768P TN panel etc.

[scaryjim]

... Intel's Ultrabook designs helped their brand a lot ....

Problem with that is that Intel pumped quite a lot of money into the Ultrabook programme, and AMD don't have anything like the cash reserves Intel can draw on.

That's been AMD's big problem for decades now - the quality of the tech is almost irrelevant, as they simply can't get good buy in from OEMs...

[=assassin=]

I really want to like Ryzen, and it is *much* better than AMD's previous offerings, but because the most demanding things I run are games, I couldn't really justify ever upgrading to Ryzen as I'd be missing out on a chunk of performance, and it would also be less worthwhile of an upgrade when compared to my (admittedly aging) first-gen i7.

[watercooled]

I think it depends on a couple of things though really, both what games you're playing as some show little difference and others perform better on Ryzen, and what price range you're looking for - for example I'm not sure how good of a choice an i5 with its 4 non-SMT cores is when placed against the likes of the 1600X.

There aren't many games (aside from Fallout which probably won't change) where the 1600X falls considerably behind the i5, and a point a few people have raised is how many games are properly saturating every thread available on an i5. Plus game updates are still changing the comparisons, with games that were previously performing weirdly on Ryzen now brought back in line e.g. AoTS where the 1600X is now comfortably ahead of the i5. That's not future speculation, that's already happening after just a few weeks. There are a couple of interesting discussions on why this has happened so quickly, and it's not necessarily anything too difficult to avoid from now on e.g. updated compilers, avoiding prefetch instructions in code, etc. Pretty much exactly the sort of thing you expect with a brand new uArch. Basically, it's one to watch, and I don't think most of these 1600X reviews are done with the latest AGESA code from AMD which improves performance in some areas too, such as memory latency.

The decider would likely be a mix of what else you do with your PC, as a 1600X unquestionably beats any i5, and by a huge margin, once you can make use of the threads e.g. with encoding, rendering, etc. Another thing would be the longevity gamble - the i5 has already been falling further behind its i7 counterparts lately (quite different to a few years ago where SMT made no real difference for games), and that's the sort of area where Ryzen's ample thread headroom shows to be helping. I can't think of many situations where I'd be recommending or buying an i5 now.

But as always, horses for courses.

In its current form, its somewhere between a desktop and a HEDT chip and trades blows with what Intel has to offer in both segments. What it does well, it does *really* well, even thread for thread it frequently matches or beats X99 systems. And it's not like Bulldozer with it's 'yeah, but...' e.g. with power consumption or single thread performance vs Intel.

As one guy put it, make no mistake, Ryzen makes a great gaming processor. Outside of a few expected (see Nehalem launch) side-cases, it's within striking distance of the higher-clocked Skylake. Just it's not topping Skylake in this segment for the most part. But that was pretty much expected from my perspective, and the side-cases where performance is weird are becoming fewer, many are not relevant (e.g. both very high anyway) and are clearly not representative of future releases given how quickly developers have been able to release fixes for existing games after spending minimal time with Ryzen and its optimisation guides for instance.

As I think I've said before, it's nice that we have a real choice again. The answer to 'what's the best processor for me' is once again 'it depends...'

Edit: I'm really making a habit out of typing posts faaaaaar longer than the single-sentence replies I originally intended.

[scaryjim]

... I couldn't really justify ever upgrading to Ryzen as I'd be missing out on a chunk of performance, and it would also be less worthwhile of an upgrade when compared to my (admittedly aging) first-gen i7.

Which i7 do you have? Chances are that a Ryzen 5 would actually be a noticeable upgrade in single threaded tasks as well as multi-threaded, and from Intel unless you're happy losing half your threads your only real option would be an i7, which is going to set you back ~ £300. And unless you're running a GTX 1080 at 1080p, you'll find (as watercooled points out above) very little difference in most games...

... Edit: I'm really making a habit out of typing posts faaaaaar longer than the single-sentence replies I originally intended.

We'll make a Saracen out of you yet

[CAT-THE-FIFTH]

I wish review sites could try and at least get a rough gist of online games other than BF1 MP - something like WoW or PS2:

http://www.pcgameshardware.de/World-...marks-1204205/

It is one of the worst case scenarios for PD,but interestingly enough from the rough tests down there,SKL looks to be around 50% better IPC,which is approximately what AMD said the IPC uplift for Ryzen is.

[sweatyBallacks]

I've seen other benchmarks and the 1600/1600X basically match the 7600K right now, before game optimisations and the memory latency update (memory frequency update is still to come as well in May) and both lay waste to it and even the 7700K in applications that like more cores/threads. The 1600 is also cheaper than a 7600K!

AMD has really pulled a rabbit out of the hat with these CPUs, especially those two 1600s. Performance-per-dollar king.

[watercooled]

I wish review sites could try and at least get a rough gist of online games other than BF1 MP - something like WoW or PS2:

http://www.pcgameshardware.de/World-...marks-1204205/

It is one of the worst case scenarios for PD,but interestingly enough from the rough tests down there,SKL looks to be around 50% better IPC,which is approximately what AMD said the IPC uplift for Ryzen is.

A lot of places also rely on built-in benchmarks, which is understandable because of its consistency, but as a few people have found they're not necessarily representative of real gameplay, particularly for online games with lots of players running around, blowing stuff up. Of course, it's important for the less controlled environment of multiplayer to be taken into account as a margin of error.

[CAT-THE-FIFTH]

A lot of places also rely on built-in benchmarks, which is understandable because of its consistency, but as a few people have found they're not necessarily representative of real gameplay, particularly for online games with lots of players running around, blowing stuff up. Of course, it's important for the less controlled environment of multiplayer to be taken into account as a margin of error.

Well I suppose it will be a rough indication unless you have different PCs and enter the game at the same place,which does make testing a bit more complex!! However,I suspect we might see some more tests from users as time progresses so we can get a rough indication of where things are at.

Edit!!

Also regarding FO4 and SC2,they both seem to show large improvements going from Haswell to SKL/KL so there is something going on there which is more than sheer IPC. I suspect it might be memory bandwidth and latency,although Bethesda are infamous for things like the use of X87 in Skyrim when it launched and the community patching it before Bethesda did!