[Techpowerup] AMD "Zen" CPU Prototypes Tested, "Meet all Expectations"

[Gikaseixas]

AMD reportedly finished testing some of its first "Zen" micro-architecture CPU prototypes, and concluded that they "meet all expectations," with "no significant bottlenecks found" in its design. This should mean that AMD's "Zen" chips should be as competitive with Intel chips as it set them out to be. The company is planning to launch its first client CPUs based on the "Zen" micro-architecture in 2016, based on its swanky new AM4 socket, with DDR4 memory and integrated PCIe (a la APUs). Zen sees AMD revert to the large, monolithic core design, from its "Bulldozer" multi-core module design with a near doubling of number-crunching machinery per-core, compared to its preceding architecture.

http://www.techpowerup.com/217251/amd-zen-cpu-prototypes-tested-meet-all-expectations.html

 

[Tuna-Fish]

I find it very interesting how little is necessary to create a rumor that circulates all over the web.

This one started with: http://www.realworldtech.com/forum/?threadid=154302&curpostid=154823

Regarding Zen performance, a guy who worked for AMD (at least his linkedin profile says that) and who, as he claims, worked on designing L2 cache for Zen and K12 said that their focus was to be competitive against Intel. He no longer works there but apparently his old colleague who still works there said Zen chips have already been tested and so far "it has met all expectation" and they "haven't found any significant bottlenecks". Apparently they haven't finalized the specifications for the clocks and TDP, but their partners in server market are "very excited".
It's not much detail, but I think if there was a problem from having only 2 AGUs, it would count as a significant bottleneck.
Also this is my first post ever, I just usually lurk here and this is the first time I have something useful to add to the discussion. Please no bully.

RWT doesn't require registration, so no-one knows anything at all about the guy who posted this. You or I could have posted something we totally made up, and if it sounds even remotely technically competent, we could read it from the various tech sites within a week. Just a thought.

 

[Burpo]

Based on information from a linkedin profile..

No specs on speed or TDP.. What their expectations were is unknown..

 

[NTMBK]

You're missing a poll option for "I can't possibly have any idea".

Seriously, do we need to start hyping this chip already? It's not out for another year. Just wait for the benchmarks and calm the F down.

 

[frozentundra123456]

You're missing a poll option for "I can't possibly have any idea".

Seriously, do we need to start hyping this chip already? It's not out for another year. Just wait for the benchmarks and calm the F down.

Too late, it has already been hyped for months.

I refused the poll, because first we have no idea of the performance, and secondly there is no choice for X99 hex core, which at best is what I think Zen may have a shot to reach. But I am expecting it to be very workload dependent as well. Zen being competitive in highly threaded workloads, but less so in single thread performance (IPC x clockspeed). Much like Bulldozer, but more competitive in lightly threaded workloads.

 

[Gikaseixas]

Seriously, do we need to start hyping this chip already? It's not out for another year. Just wait for the benchmarks and calm the F down.

Whos hyping here?
Im calm bro, its just for fun.

 

[monstercameron]

Whos hyping here?
Im calm bro, its just for fun.

He is probably just worried about the hype train collapse a la fury/x.

 

[dark zero]

Well... A prediction...

Best Case
- ST on Broadwell level
- MT on Skylake level

Worst case
- ST on Ivy Bridge level
- MT on Broadwell level

Realistic case
- ST on Haswell level
- MT only 5% than Skylake level.

 

[ShintaiDK]

No surprise really since this screams of clickbait and nothing else. We all remember engineers dancing in the aisles as well.

And remember LinkedIn info can be faked easily. And some random forum posts only. Oh lord. We all remember the last AMD Zen slides that was leaked. All fake.

the RWT convo with lurker also shows he is pretty much just spree out random caca.

Seriously, do we need to start hyping this chip already? It's not out for another year. Just wait for the benchmarks and calm the F down.

Desperation and impatience is the fuel.

 

[eternalone]

If they try to sell these at Intel prices they better perform like intel chips, other that that sub $150 for a Octocore would be nice. Im happy with my 8320E overclocked. But if they overprice these like they did in the beginning with Bulldozer well I will personally wait to see what Intel does with their prices, hopefully this is what we need to see some price wars, and faster tech.

 

[waffleironhead]

No surprise really since this screams of clickbait and nothing else. We all remember engineers dancing in the aisles as well.

And remember LinkedIn info can be faked easily. And some random forum posts only. Oh lord. We all remember the last AMD Zen slides that was leaked. All fake.

the RWT convo with lurker also shows he is pretty much just spree out random caca.



Desperation and impatience is the fuel.

I've never heard of this "dancing in the aisles". Well except from every AMD related thread you post in.

 

[sm625]

Worst case
- ST on Ivy Bridge level

lol you must not have been following AMD for the last decade or so. The Worst case can get much, much worse than Ivy Bridge level ST performance. There is every possibility that ST performance will only exceed excavator by 10% or less. Tasks like super pi calculations and other legacy x86 instruction heavy code could be a drag on overall ST scores.

 

[AtenRa]

People dont believe AMD's 14nm FF ZEN could be competitive against Intel's 2014 14nm FF architecture in 2016 ??

 

[dark zero]

lol you must not have been following AMD for the last decade or so. The Worst case can get much, much worse than Ivy Bridge level ST performance. There is every possibility that ST performance will only exceed excavator by 10% or less. Tasks like super pi calculations and other legacy x86 instruction heavy code could be a drag on overall ST scores.

Keller was there making Zen and he was there when they did K8. So it will be no surprise if they improve dramatically their ST. But as not all is perfect, they won't improve MT as dramatically

 

[Raftina]

People dont believe AMD's 14nm FF ZEN could be competitive against Intel's 2014 14nm FF architecture in 2016 ??

AMD's 2011 32 nm Bulldozer architecture was competitive against Intel's 2008 45 nm Nehalem architecture. I would say the level of skepticism about the current, unsubstantiated, and LinkedIn profile extrapolated Zen claims is healthy.

 

[NTMBK]

He is probably just worried about the hype train collapse a la fury/x.

I'm sick of forum drama. I'd rather talk about the facts when we have them, not fight endlessly over nothing.

 

[AtenRa]

lol you must not have been following AMD for the last decade or so. The Worst case can get much, much worse than Ivy Bridge level ST performance. There is every possibility that ST performance will only exceed excavator by 10% or less. Tasks like super pi calculations and other legacy x86 instruction heavy code could be a drag on overall ST scores.

Dont think so,

 

[Tuna-Fish]

Dont think so,

While I think that Zen will improve much more than 10% over excavator.

IPC IS NOT PERFORMANCE

Performance is IPC * clock speed. IPC is not independent of clock speed, and in design they can be traded off against each other. That is, there are easy ways to increase IPC by sacrificing clock speed, and to increase clock speed by sacrificing IPC. This graph only shows the IPC. This means that this graph tells you almost nothing useful about the performance of Zen. All it tells you that they aimed for a more brainy, higher-IPC core than BD, which makes sense, because in the age of power-limited computing, higher IPC seems to give better performance than higher clock speeds.

 

[ShintaiDK]

I've never heard of this "dancing in the aisles". Well except from every AMD related thread you post in.

Google is your friend then.
http://www.theinquirer.net/inquirer/news/1016287/amd-to-offer-speed-boosts-on-barcelona

With Bulldozer there was no limit either. IPC increases!

Another source claimed the memory controller, long a bottleneck in K8 scaling, came in way better than expected.

A bit like Zen and no "bottlenecks"

 

[dark zero]

I'm sick of forum drama. I'd rather talk about the facts when we have them, not fight endlessly over nothing.

The fact then is that the whole tech industry is getting stuck to the point that they will be forced to evolve in order to advance

 

[Headfoot]

you're missing a poll option for "i can't possibly have any idea".

Seriously, do we need to start hyping this chip already? It's not out for another year. Just wait for the benchmarks and calm the f down.

+100000000000000000

 

[Fjodor2001]

Desperation and impatience is the fuel.

Naaahhh... nervousness and fear of monopoly-loss is the fuel for the Intel denial camp.

 

[AtenRa]

While I think that Zen will improve much more than 10% over excavator.

IPC IS NOT PERFORMANCE

Performance is IPC * clock speed. IPC is not independent of clock speed, and in design they can be traded off against each other. That is, there are easy ways to increase IPC by sacrificing clock speed, and to increase clock speed by sacrificing IPC. This graph only shows the IPC. This means that this graph tells you almost nothing useful about the performance of Zen. All it tells you that they aimed for a more brainy, higher-IPC core than BD, which makes sense, because in the age of power-limited computing, higher IPC seems to give better performance than higher clock speeds.

I dont expect quad core 8 thread desktop ZEN to be less than 4GHz.

 

[sandorski]

Time will tell. Would be nice if it traded blows with Intel's best, but even if they got closeish it would dramatically help AMD's competitiveness.

 

[moonbogg]

If Zen was only 10% slower than Intel's best but cost half as much, I'd buy Zen. Ok fine, no I wouldn't. I'm dying to see benchmarks. This is going to be one hell of a release day. I can't wait to read all about it over a massive cup of coffee and see the nukes start to drop in the forums.