Associate
- Joined
- 29 May 2014
- Posts
- 390
assuming that the AMD graph is up to scale the height difference from bulldozer to excavator is 5.4 times smaller than the height difference from excavator to zen.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
*** AMD "Zen" thread (inc AM4/APU discussion) ***
- Thread starter Boomstick777
- Start date
Caporegime
If that chart is accurate to scale Zen has 9x higher IPC than Bulldozer, it obviously isn't.
Where did that slide come from?
Soldato
Slide originally from AMD’s 2015 Financial Analyst Day : http://www.anandtech.com/show/9231/amds-20162017-x86-roadmap-zen-is-in
I highly doubt the graph is actually to scale, just something to look more prettier for the investors.
I highly doubt the graph is actually to scale, just something to look more prettier for the investors.
It's 100% to reassure investors & is pretty meaningless except in showing the performance trend AMD is hoping for. The only way we are going to know anything for sure is when the actual chips land in reviewers hands, and that won't be for a while yet 
Soldato
They have, but the Carrizo design is a very low power one. The IPC gains cannot be scaled to higher power CPUs / APUs. So you cannot necessarily judge by that.
I'm taking it as 7.5% over Bulldozer and then a compound 40% over that. However, Zen will have a starting point of 8c/16t so that makes it very promising imo. Someone around here is very fond of pointing out that without knowing the frequencies that Zen runs at, it's impossible to extrapolate performance from it. And because they're probably exhausted from pointing it out, I'll say it for them this time. We don't know.
All we really have to go on for performance is a statement from AMD that it will be "competitive with Skylake" which from the context I took to mean performance, but could refer to price-performance. I am really hoping it's the former as well as the latter. AMD need it and frankly so do we.
Yep it's exhausting pointing it out because people just see "40% IPC increase" banded around and automatically assume that it means the cores will be 40% faster straight off the bat.
It's a lot easier to reach higher frequencies when you have a lower IPC design like Pentium 4 and Bulldozer, if AMD can release a 8c/16t cpu with high IPC, 4ghz+ clockspeed and it's not £500+ then I'll rightly take my hat off to them.
"Competitive with Skylake" means nothing really, you can argue that FX8350 is competitive with Skylake if you cherry pick applications/situations.
It's a lot easier to reach higher frequencies when you have a lower IPC design like Pentium 4 and Bulldozer, if AMD can release a 8c/16t cpu with high IPC, 4ghz+ clockspeed and it's not £500+ then I'll rightly take my hat off to them.
"Competitive with Skylake" means nothing really, you can argue that FX8350 is competitive with Skylake if you cherry pick applications/situations.
Soldato
Another point about current Excavator based APUs is they lack the L3 cache, so who knows how much better Excavator could perform if it was a full FX chip.
Come on AMD, leak some benches already
.
Come on AMD, leak some benches already
.
Soldato
Im hoping to upgrade my CPU/MOBO in Q2/3 2017.
Chances of Zen besting KabyLake?
Chances of Zen besting KabyLake?
Soldato
Performance? Price\Performance? Latter, likely, former, probably not. Waiting to be surprised though
Soldato
Just would really like an AMD system in my home gaming rig again. I like supporting the underdog. AMD make fantastic stuff and Intel are ripping the **** same as Nvidia due to lack of competition.
Having said that, Im a PC user first, fanboy second and will end up buying whatever is best option.
Regardless of its abilities, if AMD want to win CPU buyers over again they are going to have to come in well under Intel prices initially to get peoples trust back.
Soldato
Very low chance when it comes to single-threaded. Multi-threaded is going to depend on number of cores. On an equal number, it will be behind. But Zen starts at 8c/16t so I think there's a good chance you'll be able to get more cores at an equivalent price which means it theoretically could be better at a given price point in a lot of scenarios. We'll have a better idea in October, I guess.
That said, you mentioned gaming and I suspect it will be more than adequate for that.
Also remember we have no idea how good AMD's SMT implementation is yet. Intel have had it for over a decade and it hasn't gotten much better so we can assume it's probably very well optimised. AMD's might not be as good, to start with at least.
Soldato
Fair point. Hadn't considered that.
I actually like AMD's last approach - real cores with the less used function (FP) shared between them. But everyone else on the planet hated it. We'll see how they do with SMT.
Caporegime
I don't think the fundamental ideal of modules was necessarily bad, it was more that the implementation resulted in poor IPC in many cases, even compared to AMD's previous standard CPU cores. The design either just ran out of steam earlier than they expected (much like Intel's Pentium 4 design), or it wasn't an optimal design to begin with.
Soldato
My FX is doing the job nicely for now. If DX12 and other low level boosting techniques continue then the multi-core dogs may still cut the mustard for enough time to see a great choice when it comes to a leap forward in upgrade potential.