AMD Ryzen (Summit Ridge) Benchmarks Thread (use new thread)

[inf64]

I've just showed that AT review shows the same 18% app boost over IB just as the other two do. I also said that it is imperative that Zen is 5% faster than IB which would put it at 1.18/1.05=1.12 or 12% lower IPC level than Skylake in apps (core for core) .

Performance in games varies a lot and since games are getting better MT support I doubt 16T Zen will have any problems matching Skylake provided the Turbo clocks are at the range I said they have to be and games used for comparison are not relying on ancient ST engines.

 

[KTE]

We are making a brave assumption that that sample could be running at 1Ghz.

Which is likely...

Or it's 33% slower than IVB

Sent from HTC 10
(Opinions are own)

 

[Dresdenboy]

so I did some digging and apparently AMD promised 20% IPC boost in Steamroller over Piledriver. Was this actually accurate? From what I can tell, this wasn't the case across the board at all. Not even close.

We are working under the assumption that AMD's 40% number is actually legit, but they have fluffed up the numbers before...

"Up to" statements can't be validated by "across the board" measurements. The Stilt's slide tells the whole story.

 

[Arachnotronic]

"Up to" statements can't be validated by "across the board" measurements. The Stilt's slide tells the whole story.

One has to wonder if that Zen figure is "up to" or "average"?

 

[Arachnotronic]

Which is likely...

Or it's 33% slower than IVB

Sent from HTC 10
(Opinions are own)

33% slower than IVB per clock just seems excessively pessimistic. Given the execution resources and key buffer sizes in Zen I would expect a much better per clock showing than that.

 

[Glo.]

I have to say. I was expecting much worse than that. I will be satisfied with quad core Zen APU with Vega, if it will really have Ivy Bridge or slightly higher IPC.

 

[Arachnotronic]

I have to say. I was expecting much worse than that. I will be satisfied with quad core Zen APU with Vega, if it will really have Ivy Bridge or slightly higher IPC.

IPC is only one term in the performance equation.

 

[Abwx]

One has to wonder if that Zen figure is "up to" or "average"?

Likely an average, if in FP it can be largely exceeded thanks to usually better instruction parralelism of FP code
it wont be the same story in INT, at 40% this would put Zen well above anything Intel in applications like 7 Zip, so in INT code there will be some variability, probably that the apps that can get 40% are the ones that use AVX2 for X264/265 encoding FI, indeed if you look at Hardware.fr comparison most of the throughput improvement in Intel CPU gens is concentrated in FP, with encoding as about the only INT code that is as much improved thanks to new ISA...

As to where it will exactly stand in respect of the comparison i think that AMD s demo was a willfull indication
that it s in the same ballpark as BDW at least in FP, in INT there s no reason that it should be different if we are to look at the diagrams, there s as much ALUs as in BDW, so in INT the difference should be of the same order than in FP, that is a handfull of percents advantage for a design or the other.

 

[bjt2]

Having more ports doesn't mean that you can have better performance, and the Blender test clearly shows that (with Zen having also double the FP computing capability): 2% more is negligible result.

P.S. Talking of IPC, as well.

My hypothesis is that ST is inferior and that AMD's better SMT makes up for the differences plus that 2% seen in blender test...

 

[Arachnotronic]

My hypothesis is that ST is inferior and that AMD's better SMT makes up for the differences plus that 2% seen in blender test...

AMD's better SMT? Do you really think AMD's first crack at it, after mocking Intel for using the technique back in 2011/2012, will be superior to a 4th gen implementation in Broadwell?

 

[DrMrLordX]

If they cribbed from IBM's implementation of SMT on POWER8/9, then possible so.

 

[Abwx]

AMD's better SMT? Do you really think AMD's first crack at it, after mocking Intel for using the technique back in 2011/2012, will be superior to a 4th gen implementation in Broadwell?

To hedge out BDW in Blender require that they had either better SMT scaling or better ST perf or comparable perfs for boths, it cant be otherwise...

 

[itsmydamnation]

AMD's better SMT? Do you really think AMD's first crack at it, after mocking Intel for using the technique back in 2011/2012, will be superior to a 4th gen implementation in Broadwell?

But its not , how do you think CMT works?????? the big expensive complex parts are shared in CMT. So really if you talk about a core that shares resources between threads this is their 5th design...........

 

[Arachnotronic]

But its not , how do you think CMT works?????? the big expensive complex parts are shared in CMT. So really if you talk about a core that shares resources between threads this is their 5th design...........

We'll see. Anyway, hoping that somebody gets their hands on an ES and leaks some benchmarks

 

[itsmydamnation]

We'll see. Anyway, hoping that somebody gets their hands on an ES and leaks some benchmarks

If bulldozer had of been a success by excavator im sure you would have seen SMT on CMT, The way the registers worked was at the same level as Sandy bridge, it would have been a logical next step, CMT core still suffer from the same stall's a non SMT core suffers from.

It would have been interesting to see if bulldozer was just a little wider (3 ALU's) and didn't have a terrible cache design.

 

[moonbogg]

People who are hoping for a quad core Zen with Ivy IPC, what exactly has stopped you from buying an IVY BRIDGE CPU???

 

[StrangerGuy]

People who are hoping for a quad core Zen with Ivy IPC, what exactly has stopped you from buying an IVY BRIDGE CPU???

You see, it's better to wait 5 years on an awful AMD platform to maybe save like a gigantic $100 in CPU costs for Zen because screw the greedy evil monopolistic Intel.

 

[Arachnotronic]

People who are hoping for a quad core Zen with Ivy IPC, what exactly has stopped you from buying an IVY BRIDGE CPU???

You win this thread, Bogg

 

[Alqoxzt]

People who are hoping for a quad core Zen with Ivy IPC, what exactly has stopped you from buying an IVY BRIDGE CPU???

price, oc potential.

 

[frozentundra123456]

People who are hoping for a quad core Zen with Ivy IPC, what exactly has stopped you from buying an IVY BRIDGE CPU???

The only reason is that one would expect a much better igpu than IB. Even Intel's igpu with skylake is better than IB. Now whether Zen will have a game changing igpu also remains to be seen, and personally, I am skeptical, but hope springs eternal. In general though, I agree with you. I just dont understand the "SB/IB level of performance (clockspeed plus ipc) is good enough to get me to buy zen", especially when new intel products are faster and you can still buy cheap used intel products with that level of performance.

 

[gammaray]

People who are hoping for a quad core Zen with Ivy IPC, what exactly has stopped you from buying an IVY BRIDGE CPU???

the smell

 

[AtenRa]

People who are hoping for a quad core Zen with Ivy IPC, what exactly has stopped you from buying an IVY BRIDGE CPU???

They were not for an upgrade back in 2012. Also, there are still people with Core 2 Quads, Nehalems, Socket 1155 CPUs and SandyBridge Dual/Quad Cores that are upgrading or are going to upgrade in 2017.

I have a Core i7 3770K @ 4.4GHz, if i dont get 6 Cores or more for the same price i got my Ivy back in 2012, then there is no reason to upgrade for only 10-20% higher IPC.

 

[DrMrLordX]

People who are hoping for a quad core Zen with Ivy IPC, what exactly has stopped you from buying an IVY BRIDGE CPU???

Sorry, nobody in their right mind is actually hoping for Zen to have the same overall IPC and throughput per core as Ivy Bridge.

The only reason I would even look at quad-core Zen (regardless of IPC) is to get it in an APU format. Raven Ridge etc.

 

[Arzachel]

People who are hoping for a quad core Zen with Ivy IPC, what exactly has stopped you from buying an IVY BRIDGE CPU???

If it's for a laptop then a lower tdp, more powerful igpu with better driver support. If it's for a desktop then price and not being tied to Intel's platform limitations.

 

[AtenRa]

The only reason I would even look at quad-core Zen (regardless of IPC) is to get it in an APU format. Raven Ridge etc.

Or for a cheaper Overclockably Quad Core CPU priced between Core i3 and Core i5 Kabylake. At a price of $180 and IF it can OC to 4-4.5GHz it will be a really nice alternative to locked Core i3-i5 KabyLake.