• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Fudzilla: New AMD Zen APU boasts up to 16 cores (plus Greenland GPU with HBM)

Page 40 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
ShintaiDK said:
So Zen more or less equal to Nehalem/Westmere in IPC if the slides also transform into reality.
Click to expand...

Lol, you had to come with your usual fud to try downplaying the thing, no wonder...

They are talking of Integer IPC, for the record and depending of the softs Haswell has 26-41% better Integer IPC than Piledriver, and the 40% are in respect of Excavator.

Other than that would be great if this thread is kept from the usual pollution by some thread crappers...
 
ShintaiDK said:
So Zen more or less equal to Nehalem/Westmere in IPC if the slides also transform into reality. And Skybridge cancelled as the next project on the list.
Click to expand...


If it actually is a %40 gain it's closer to Ivy Bridge. It'll be interesting to see if that holds. And it'll be interesting to see something new.
 
Ivy IPC is good, considering they will close the gap with Skylake with moar coars.

Lol at the poster thinking Ex*1.4 is Nehalem IPC. Really no clue about CPUs at all.
 
Fjodor2001 said:
How did you arrive at that conclusion?
Click to expand...

Just look at benches.

4.2Ghz PD scores 96 in CB15 ST. 3.73Ghz Nehalem scores 117. Factor~1.43 with normalized clocks.

40% is obviously some best case scenario. But lets use it.

So add a few % to PD for XV since SR added nothing in ST. And then 40% and you got Nehalem, in best case.
 
Yeah and Cinebench is what exactly?

Oh, the bench that doesnt even represent the performance you would expect of C4D (the actual software). What a joke.

If it wasnt as clear: CB is the Antutu of the renderers. Canned bench that can be optimized to death and doesnt reflect real user case scenarios of the mother software it represents.

EDIT: The stilt was right on the money regarding platforms: FM was long dead for the high perf CPU/APU space. APUs on AM4 would make really high end iGPs+competent CPUs for gaming possible.
 
Last edited:
Abwx said:
Could you provide us the single thread perf of your CPU in 7 ZIP, that s easy, there s a built in bench and the result is in MIPS, we can then compare to Slowspyder s results below and see if your myth hold :
Click to expand...

Could you link his post? Slowspyder and 3726 gives no result in the search.
 
ShintaiDK said:
Could you link his post? Slowspyder and 3726 gives no result in the search.
Click to expand...

He mailed me his numbers, other than that beware of using FP benches, moreover Intel optimised, when talking IPC.

I can as well use Povray to display that the FX has about SB FP IPC, why wouldnt it be more reliable than CB R15 given that its not apparently favouring any brand.?.
 
wvlsfaj.png


No HBM for 2016 APUs :/
 
Abwx said:
He mailed me his numbers, other than that beware of using FP benches, moreover Intel optimised, when talking IPC.

I can as well use Povray to display that the FX has about SB FP IPC, why wouldnt it be more reliable than CB R15 given that its not apparently favouring any brand.?.
Click to expand...

Right.....

You can look at this chart and ask yourself if you think its a good bench for actual core performance.
51137.png
 
Abwx said:
Lol, you had to come with your usual fud to try downplaying the thing, no wonder...

They are talking of Integer IPC, for the record and depending of the softs Haswell has 26-41% better Integer IPC than Piledriver, and the 40% are in respect of Excavator.

Other than that would be great if this thread is kept from the usual pollution by some thread crappers...
Click to expand...

Abwx said:
Could you provide us the single thread perf of your CPU in 7 ZIP, that s easy, there s a built in bench and the result is in MIPS, we can then compare to Slowspyder s results below and see if your myth hold :
Click to expand...

7- zip is a terrible indication of IPC.

Why?

Because its so branchy and latency/bandwidth focused that large cores absolutely suck at it.

The LZMA compression benchmark only measures a part of the performance of some real-world server applications (file server, backup, etc.). The reason why we keep using this benchmark is that it allows us to isolate the "hard to extract instruction level parallelism (ILP)" and "sensitive to memory parallelism and latency" integer performance. That is the kind of integer performance you need in most server applications.
This is more or less the worst-case scenario for "brawny" cores like Haswell or Power 8. Or in other words, it should be the best-case scenario for a less wide "energy optimized" ARM or Atom core, as the wide issue cores cannot achieve their full potential.
Click to expand...
66524.png

The Xeon E3-1230L and Atom C2750 run at similar clock speeds in this single threaded task (2.8GHz vs 2.6GHz), but you can see how much difference a wide complex architecture makes. The Haswell Core is able to run about twice as many instructions in parallel as the Silvermont core. Meanwhile the Silvermont core is about 45% more efficient clock for clock than the old Saltwell core of the Atom N2800. The Haswell core result clearly shows that well designed wide architectures remain quite capable in "high ILP" (Instruction Level Parallelism) code.
Click to expand...
66523.png
Decompression uses a rather exotic instruction mix and the progress made here is much smaller. The Haswell core is about 15% faster clock for clock than the old Harpertown core. Compared to the Silvermont core, the Haswell core is about 40% more efficient in this kind of software. The X-Gene core is about 10% slower than the Atom C2000.
Click to expand...
Multithreaded

66521.png


In most other programs the E3-1240 leads by a factor of ~3x. In 7-zip its around 2x.

http://www.anandtech.com/bench/product/836?vs=1229

7-zip is a really 'small-core friendly' test.
 
40% more IPC .... how would that stack against a Ivy Brige Core ? I just bought a Xeon 1240v2... its an amazing upgrade but this Zen announcement is kinda making me wonder if il keep it more than a year 🙂
 
That roadmap looks incomplete. If it is accurate, it seems like a disaster for AMD. If all they have from Zen is an FX without an igp, how much market can there be for that?

Does that mean the APUs are still on steamroller, not Zen, and no HBM? That is what the slide implies, since they specifically mention Zen for FX and DDR3/DDR4 for the APUs.
 
Shakabutt said:
40% more IPC .... how would that stack against a Ivy Brige Core ? I just bought a Xeon 1240v2... its an amazing upgrade but this Zen announcement is kinda making me wonder if il keep it more than a year 🙂
Click to expand...

If geekbench single-thread benches are any indication, Zen should be on par with Haswell overall. No idea how the improvements are distributed though.

Dropping CMT also means no more penalty on multi-threading.

This is assuming XV has a small improvement over Steamroller btw. Haswell's geekbench score is about 1.5x that of Steamroller (calculated using 4670k vs 7850k, single thread scores from HWbot normalized for frequency)
 
Last edited:
You must log in or register to reply here.

TRENDING THREADS

Back
Top