a look at (ry)Zen's IPC - from Excavator to KabyLake

[laminutederire]

Just now, zMeul said:

here's the thing tho, the i7s are cheaper (up to a point)

and the i5s are cheaper still

 

why would you pay premium for less performance? it doesn't make any sense

You pay half less for the i7 performance but less than double for i5 gaming performance. So instead of buying a 1000$+ CPU to work with it and game a bit better, you can go for the 500$ Dollar one which will get you comparable working performance. It's still nice it gets i5 performance for gaming. 

And i7 are certainly not cheaper. Only the i7 meant for gaming. But some people need something to work with first, and are okay with i5 gaming performance. Those same people are perfectly okay without a 1080ti or a titan as well.

[Godlygamer23]

12 minutes ago, laminutederire said:

That's an as lame argument since you're just extremely exigent. Loads of people are fine with 70-80 fps steady, why do you guys have to shit all over products which are perfectly fine for many people who would be quite okay with it.

It's not the ultimate gaming chip, but it's decent enough. As some reviewers point out, it's on par or better than i7 for many workstation workloads, and is on par with i5 and could maybe be better when amd fixes issues.

Since when i5 performance is bad? It's not optimal but it is in no case bad.

You have those 144 fps in e sports titles, and you have 70+ fps in all the other titles, so it's quite alright to be fair for someone who need the work while not having a second rig to game on, while still not paying twice the price.

If I use a machine that's primarily for gaming, and Intel's offering is cheaper, and provides better performance for the intended purpose, why would I choose Zen? I'm not saying that Zen is bad - in many areas it's quite decent, but its IPC is not particularly good, causing the clock speeds to have to be ramped up in order to compete. I use a 3570K for gaming, and at the moment, it's sufficient. 

 

But is optimization REALLY the problem here? Or is AMD just playing off the issues like they did with Bulldozer and derivatives of it? 

[zMeul]

Just now, laminutederire said:

You pay half less for the i7 performance but less than double for i5 gaming performance. So instead of buying a 1000$+ CPU to work with it and game a bit better, you can go for the 500$ Dollar one which will get you comparable working performance. It's still nice it gets i5 performance for gaming. 

And i7 are certainly not cheaper. Only the i7 meant for gaming. But some people need something to work with first, and are okay with i5 gaming performance. Those same people are perfectly okay without a 1080ti or a titan as well.

what?!

the use case in your discussion was games, right

it's already established it performs like an i5 in gaming, so why not buy an i5? it's cheaper? yes!

hell, for that saved difference you can go with a better video card, more RAM, larger storage, bigger monitor ... you name it

why would you pay more when the use case scenario can be achieved for less - the same, I repeat, the same is the marketing behind Ryzen: why pay for a Broadwell-E when you can achieve the same with less money?

 

it's valid to be the cheaper alternative only when AMD sais it, or when it actually makes sense!?

[Sauron]

12 hours ago, zMeul said:

Intel has nothing to fear from Zen - AMD's products for enterprise (Naples), based on the current IPC, have no reason to exist other than being a cheap alternative

A cheap alternative might be all it takes to force intel to do something, even if it's just lowering the price. Besides IPC difference or not we have seen the overall benchmarks and ryzen has been performing pretty well.

[laminutederire]

1 minute ago, Godlygamer23 said:

If I use a machine that's primarily for gaming, and Intel's offering is cheaper, and provides better performance for the intended purpose, why would I choose Zen? I'm not saying that Zen is bad - in many areas it's quite decent, but its IPC is not particularly good, causing the clock speeds to have to be ramped up in order to compete. I use a 3570K for gaming, and at the moment, it's sufficient. 

 

But is optimization REALLY the problem here? Or is AMD just playing off the issues like they did with Bulldozer and derivatives of it? 

That's the damn point though... some people can't afford multiple setups, actually a lot of people can't. For those people having 6900k performance for way less is huge, and having it perform like an i5 for gaming is well enough for them because they know that to have better they'd have to lose a lot on work related performance.

 

I never said optimization will make it godly. It remains that codes fairly optimized for Intel CPUs which have dominated the marker for a while,  so it is probable that it boosts their performance a bit. That's just a worsening factor.

[zMeul]

2 minutes ago, Sauron said:

A cheap alternative might be all it takes to force intel to do something, even if it's just lowering the price. Besides IPC difference or not we have seen the overall benchmarks and ryzen has been performing pretty well.

at the moment, Intel does not have products in that price bracket

what Intel will do is not to slice prices, but to offer new products to fit that bracket - i5s with HyperThreading would be a good start; highly clocked i7s with 16 PCIe lanes for X99 would be another step

[laminutederire]

1 minute ago, zMeul said:

what?!

the use case in your discussion was games, right

it's already established it performs like an i5 in gaming, so why not buy an i5? it's cheaper? yes!

hell, for that saved difference you can go with a better video card, more RAM, larger storage, bigger monitor ... you name it

why would you pay more when the use case scenario can be achieved for less - the same, I repeat, the same is the marketing behind Ryzen: why pay for a Broadwell-E when you can achieve the same with less money?

 

it's valid to be the cheaper alternative only when AMD sais it, or when it actually makes sense!?

Because as I repeat some people need the work performance horsepower and can't afford multiple builds, so it is okay for them to only have i5 performance in games because they have immensely improved work performance compared to i5s. To be fair, you almost could buy a z270 and a 7700k with an r7 1800x or 1700x with a motherboard and you'd still have payed less than for a 6900k and it's mobo...

So considering that you'd have the best of both worlds for not that much more.

[zMeul]

2 minutes ago, laminutederire said:

Because as I repeat some people need the work performance horsepower and can't afford multiple builds, so it is okay for them to only have i5 performance in games because they have immensely improved work performance compared to i5s. To be fair, you almost could buy a z270 and a 7700k with an r7 1800x or 1700x with a motherboard and you'd still have payed less than for a 6900k and it's mobo...

So considering that you'd have the best of both worlds for not that much more.

then your use case isn't gaming ...

and enough workloads can be GPU accelerated via OpenGL/CUDA/OpenCL

[Sauron]

4 minutes ago, zMeul said:

at the moment, Intel does not have products in that price bracket

what Intel will do is not to slice prices, but to offer new products to fit that bracket - i5s with HyperThreading would be a good start; highly clocked i7s with 16 PCIe lanes for X99 would be another step

A hyperthreaded i5 will trespass into the i7 market, forcing intel to either step up i7 features or cannibalize their own products. And either way, even just the existance of a cheaper hyperthreaded quad core is a small victory after intel decided to prevent xeons from working on mainstream chipsets.

[zMeul]

1 minute ago, Sauron said:

A hyperthreaded i5 will trespass into the i7 market, forcing intel to either step up i7 features or cannibalize their own products. And either way, even just the existance of a cheaper hyperthreaded quad core is a small victory after intel decided to prevent xeons from working on mainstream chipsets.

Opterons aren't mainstream either 

[Prysin]

and the trainwreck keep piling up. Told you guys. Told you.

[tlink]

1 hour ago, zMeul said:

then you'd have Zen exes and rest of the world exes

and then you'll have people who'll accuse AMD for specifically compiling for them - like they're now accusing Intel of 

of course people where trying to reach optimal performance on intel, but thats not necessarily a bad thing if there is healthy competition. thats what the criticism towards intel is more geared towards imo.

[Misanthrope]

34 minutes ago, Sauron said:

And either way, even just the existance of a cheaper hyperthreaded quad core is a small victory

If you mean SMT is so badly implemented right now it would be detrimental to performance yes, small is the word I'd use to describe it. 

 

And yes I know it took intel a while to get it right with hyperthreading as well, but time isn't a luxury AMD has right now.

[PCMasterDebater]

13 hours ago, PCGuy_5960 said:

I was spot on  This is EXACTLY how I thought Ryzen would stack up

Alright Captain Hindsight... They call that autobiographical memory.

 

Whatever floats your boat though.

[Humbug]

1 hour ago, Godlygamer23 said:

But is optimization REALLY the problem here? Or is AMD just playing off the issues like they did with Bulldozer and derivatives of it? 

How did AMD play off the issues for the bulldozer launch? Can't really remember. I do remember then that too that AMD was trying to emphasize high resolution gaming LOL.

 

But regardless this has been an infinitely better launch than bulldozer. There is no doubt here that AMD has produced a good allround CPU architecture that is powerful and efficient across loads of different scenarios. And even when it loses (like in gaming) the margin is not huge like it was when comparing bulldozer to sandy bridge back in 2011.

[Godlygamer23]

1 minute ago, Humbug said:

How did AMD play off the issues for the bulldozer launch? Can't really remember. I do remember then that too that AMD was trying to emphasize high resolution gaming LOL.

They played off the issues as just requiring optimization, when their architecture was just not good for the tasks at hand.

[Imglidinhere]

14 hours ago, ace_cheaply said:

Correct me if I'm wrong please, but is this saying that kabylake has 25-30% ipc advantage over haswell?  I don't believe that to be the case.

Thank you. This is completely garbage and 100% lopsided.  Sorry but, no... no.... this is dumb.

[Imglidinhere]

14 hours ago, zMeul said:

source: https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/

 

"The Slit", established member over Anand's forums has done some testing that feature more technical benchmarks that you don't usually see

go check the source for all the benches and results plus in depth technical details on the CPU and chipset (Zeppelin)

 

the tests were conducted at 3.5Ghz

mobo used ASUS CROSSHAIR VI HERO

results with ER mean extremes removed (highest and lowest)

 

let's talk IPC and how it compares to CPUs ranging from Excavator era to Intel's latest KabyLake:

 

back in 2015, AMD estimated a 40% IPC increase over Excavator, they achieved that and 6.6% over it - that's good

what's not so good is that Zen's IPC is practically lower than Haswell's, a microarchitecture that's already 4y old - AMD took 5y in development to produce a CPU that's slower than an already 4y old microarchitecture

 

the FMA set implementation in Zen seems very problematic as tests with Linpack and Himeno shows:

 

Himeno benchmark (evaluates performance of incompressible fluid analysis code - takes in measurements to precede major loops in solving the Poisson's equation solution using the Jacobi iteration method) shows particularly bad results:

 

and before you start blaming the compiler, here's what "The Slit" said:

GCC - GNU Compiler Collection

MSVC - Microsoft Visual C++ compiler

ICL - Intel C++ compiler

 

---

 

AMD needs to get their house in order, because after seeing what happened with this launch, I don't see a bright future for AMD on the market

Intel has nothing to fear from Zen - AMD's products for enterprise (Naples), based on the current IPC, have no reason to exist other than being a cheap alternative

The OP cherry-picked the results that display AMD's Zen at its worst? Fanboyism at its finest.  The performance in the actual damn article actually shows it nipping at the heels of Haswell across the fucking board.

 

Next time you wanna derail AMD's performance perhaps you should actually not link the article or post it in a reddit so the less intelligent users don't think otherwise about it.  Anyone who actually reads the article at the source shows the performance is stellar compared to excavator and shows that AMD is right where it needs to be.

[zMeul]

1 hour ago, tlink said:

of course people where trying to reach optimal performance on intel, but thats not necessarily a bad thing if there is healthy competition. thats what the criticism towards intel is more geared towards imo.

the "criticism" towards intel is misguided and fueled by fanboyism and AMD's PR machine

[Humbug]

Maybe this is my ignorance but when people talk about IPC isn't it a bit misleading because isn't IPC dependent on the type of workload?

For example there are certain workloads where the IPC of ryzen is even slightly better than a 6900k. But in other workloads it is lower. So what do people mean when they definitively talk about IPC. e.g. Even a processor like the original bulldozer wouldn't the IPC be totally different depending on whether you loaded it with floating point or integer math? From what I have seen you can make ryzen IPC look as good as modern intel, or worse than haswell depening on the workload.

[Space Reptile]

14 hours ago, zMeul said:

ASUS CROSSHAIR VI HERO

arent the MSI boards alot better? they just released a fixed bios that according to some gave a huge bump and also some where able to run 3000mhz mem 

[zMeul]

14 minutes ago, Imglidinhere said:

The OP cherry-picked the results that display AMD's Zen at its worst? Fanboyism at its finest.  The performance in the actual damn article actually shows it nipping at the heels of Haswell across the fucking board.

 

Next time you wanna derail AMD's performance perhaps you should actually not link the article or post it in a reddit so the less intelligent users don't think otherwise about it.  Anyone who actually reads the article at the source shows the performance is stellar compared to excavator and shows that AMD is right where it needs to be.

I don't know what you think I'm painting but you are wrong

 

the tests I picked are FMA related and all of them (Bullet, Himeno, NBody, Linpack) show the same thing - ZEN doesn't perform good

add these to those in the OP

 

and yes, if you go and read all the data and charts, it all adds up to what I said and shown - Zen barely reaches Haswell in IPC

[tlink]

11 minutes ago, zMeul said:

the "criticism" towards intel is misguided and fueled by fanboyism and AMD's PR machine

well feel free to prove that. just saying that it isn't true doesn't make it wrong. its very logical that programs optimize for the clearly superior product for that workload.

[zMeul]

2 minutes ago, Space Reptile said:

arent the MSI boards alot better? they just released a fixed bios that according to some gave a huge bump and also some where able to run 3000mhz mem 

why are you asking me?! this is AMD's problem

they should've sorted all this shit up before launch and not after - especially when they knew there were issues (see RAM freq problems and 2 DIMMs vs 4)

[Space Reptile]

Just now, zMeul said:

why are you asking me?! this is AMD's problem

they should've sorted all this shit up before launch and not after - especially when they knew there were issues (see RAM freq problems and 2 DIMMs vs 4)

just sayn , and yea its definetly amds problem , but there working on it atleast