ZEN ES Benchmark from french hardware Magazine

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

USER8000

Golden Member
Jun 23, 2012
1,542
780
136
Small technical point: 6900K has 3.2 Base clock, and turbo mode 3.7 GHz, with all-core turbo 3.5 GHz.

If you look at reviews and people running the chips,it tends to run at 3.5GHZ when running multi-threaded software and 3.7GHZ when running lightly threaded software. Look at the graphs I linked to in post20 - 4 of the games the magazine tested don't actually scale well with HT and realistically only tax upto 4 cores a decent amount. That means that Core i7 6900K chip is going to boosting closer to 3.7GHZ during those games,as half the cores and most of the virtual threads will be not doing much.
 
  • Like
Reactions: unseenmorbidity

jpiniero

Lifer
Oct 1, 2010
14,510
5,159
136
How do you know how much will Ryzen overclock? What if it will overclock over 4Ghz? We have already had a poster claiming that 4.3Ghz is max. "stable" OCing that can be done on current samples. Would that be good enough for you?

It would for me. I just seriously doubt it's going to get there any time soon.
 
  • Like
Reactions: ButtMagician

sm625

Diamond Member
May 6, 2011
8,172
137
106
That engineering sample needs a 13% gain just to match a 4790k in gaming. How likely is that? And even if it did, haswell is still 3 years old. This is looking like the i5 will still be the better value for gamers, same as it ever was,
 
  • Like
Reactions: Sweepr

inf64

Diamond Member
Mar 11, 2011
3,685
3,957
136
That engineering sample needs a 13% gain just to match a 4790k in gaming. How likely is that? And even if it did, haswell is still 3 years old. This is looking like the i5 will still be the better value for gamers, same as it ever was,

Same applies to 6900K then, right?
We know AMD will have 4C/8T parts too, how about we wait and see how those perform in games when clockspeed is way higher than what that poor ES ran at.
 

itsmydamnation

Platinum Member
Feb 6, 2011
2,743
3,075
136
That engineering sample needs a 13% gain just to match a 4790k in gaming. How likely is that? And even if it did, haswell is still 3 years old. This is looking like the i5 will still be the better value for gamers, same as it ever was,


look at clocks......
its stated that the Zen sample gets to 3.3ghz max. We know minimum base is 3.4ghz. So long Zen can turbo in light threads to 3.7ghz there is you 13% right there.
 

USER8000

Golden Member
Jun 23, 2012
1,542
780
136
Same applies to 6900K then, right?
We know AMD will have 4C/8T parts too, how about we wait and see how those perform in games when clockspeed is way higher than what that poor ES ran at.

Like I mentioned before 4 of the games tested don't really use SMT and only really used two to four cores.

The extra performance of the Core i7 6900K is easily explained by its 10% higher clockspeeds and the 10% higher performance of the Core i7 6700K over the Core i7 6900K is explained by its 10% higher clockspeed.
 

plopke

Senior member
Jan 26, 2010
238
74
101
I am confused by this thread off course a core i5 4/4 4/8 with high frequency is going to be better for gaming. a 6700K/7700 is king of PC-gaming atm and will very likely still be after Ryzen 8/16 launch.
It is not like we would advise anyone for just gaming on these forums to buy a 6800/6900? Neither will we for Ryzen 8/16?

What i hope for Ryzen is that 4/8 will become more main stream so developers will see 8 threads as mainstream. I personally am quiet impressed and happy with the results shown here.
 
Mar 10, 2006
11,715
2,012
126
What i hope for Ryzen is that 4/8 will become more main stream so developers will see 8 threads as mainstream. I personally am quiet impressed and happy with the results shown here.

8C/8T is already mainstream in the AMD APUs powering the consoles.
 
  • Like
Reactions: Headfoot

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
I take a 6900 for gaming only over a 7700 any day of the week if its the same price.
In 2 years its probably much better. And we tend to keep the cpu for like 6 years these days.
The days of tf2 and arma is over. And a bwe can run them fine anyway. What matters is eg bf1 64 man heavy fights wth tons of destruction. And a future bf5 and bf6.
The thought of entering bf5 or bf6 with a 7700 is not something i like.
 
  • Like
Reactions: KompuKare

mikk

Diamond Member
May 15, 2012
4,112
2,106
136
Like I mentioned before 4 of the games tested don't really use SMT and only really used two to four cores.

The extra performance of the Core i7 6900K is easily explained by its 10% higher clockspeeds and the 10% higher performance of the Core i7 6700K over the Core i7 6900K is explained by its 10% higher clockspeed.


You math won't work because it does not scale linearly with CPU clock. Battlefield 4 scales very well over 4 cores, this game alone has an impact of the overall game index to some degree. There is clearly a performance advantage in place for CPUs over 4 cores in the gaming index.
 
  • Like
Reactions: Sweepr

unseenmorbidity

Golden Member
Nov 27, 2016
1,395
967
96
8C/8T is already mainstream in the AMD APUs powering the consoles.
Which is why the i5 will soon be the budget cpu. It's already starting to bottleneck some games, when pared with powerful gpus.

What will happen a few years down the road? It won't be 'good enough' for games anymore. I5's will be relegated to budget/upgrade systems, just as the i3 is now.

If you want a futureproof system, then you at least need a cpu with 8 threads.
 
Last edited:
  • Like
Reactions: RussianSensation

USER8000

Golden Member
Jun 23, 2012
1,542
780
136
You math won't work because it does not scale linearly with CPU clock. Battlefield 4 scales very well over 4 cores, this game alone has an impact of the overall game index to some degree. There is clearly a performance advantage in place for CPUs over 4 cores in the gaming index.

Your math does not work since 4 of the 6 games,tested don't scale well with SMT or beyond 4 cores,so trying to twist the results by looking at only one of the six games is funny.

A Haswell Core i3 dual core,is faster in 4 of the 6 games than an FX8350. Nice,try but you and your mates are not really doing very well,are you??

Gaming_08.png


IMG0039213.png

GRIDAutosport_proz_amd.jpg


fc_proz.jpg


The socket 2011 CPUs also have much larger caches,etc which do help a bit,just like the Broadwell CPUs with L4 cache which nicely helped with performance.

First it is you and your mates on purpose ignoring this is a validation sample CPU not running at full clockspeeds,then secondly you ignore the Core i7 6900K on purpose,now you make excuses when it is evident the Core i7 6900K performance advantage is due to clockspeeds,and lastly you then make sure you ignore 4 of the 6 games tested not scaling that well with SMT,etc.

You and your mates are trying every trick you can to bury the results,but there is only one way AMD has managed to get to within 10% of a Core i7 6900K in those games,its via a massive jump in per core performance and not better SMT performance.

AMD touted close to Broadwell E IPC,and now an independent review of a lower clockspeed validation sample on a pre-production motherboard is hinting at the same from a well known magazine in Europe,but no its all some conspiracy,right??

Try your luck with someone else,mate!

This is getting to stupid levels now. Its almost like some don't want competition in the market for some absolutely weird reason. Some of us remember the good old days back 15 years ago. It was a great time for enthusiasts and it makes me wonder whether some of the younger folk have forgotten what proper competition did in the CPU space.
 
Last edited:

mikk

Diamond Member
May 15, 2012
4,112
2,106
136
Your math does not work

What math??? Your are the one that used some math to explain things, but it just won't work. Several peoples explained why a linearly scaling won't work. There are other bottlenecks to bear in mind. It can be clearly seen by the results in this test and dozens of other CPU reviews in the web. Try a little bit harder and you might understand why it won't work in this test. Core i5 against another Core i5 with 20%+ higher clock speed only gets <10% better performance in the gaming tests. Core/SMT scaling is another factor you have to consider if you want to compare the IPC, which is clearly a factor given that Broadwell-E with lower IPC and clock is above Skylake 4/4. Even if there is a small scaling in 2-3 games it will affect the overall index.


http://www.pcgameshardware.de/Far-Cry-4-Spiel-23145/Specials/Technik-Test-Benchmark-1143026/
http://www.pcgameshardware.de/Grid-...karten-Benchmarks-CPU-Kernskalierung-1126276/

10% advantage for 8 cores against 4 cores in Far Cry 4 and Grid Autosport.


http://www.pcgameshardware.de/Battl...ts/Battlefield-4-Test-Benchmarks-CPU-1095298/

36% advantage for 6 cores against 4 cores! No wonder Broadwell-E is well above Skylake 4/4 with a similar clock speed.
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
Call me cautiously optimistic, the news looks better than expected. I might just have to try one out!
 

SlickR12345

Senior member
Jan 9, 2010
542
44
91
www.clubvalenciacf.com
Guys can we not take every fake post and every fake image as fact? The fact is they probably did an estimation and lied that they somehow managed to get a real sample of the chip. There is no way someone could give them a chip in early beta development, that would be criminal and I doubt AMD would not know how many chips and where they are. It would be investigated, person would be arrested and chp returned. Since it would be technically stolen material, the paper would not be able to publish results that they got and I see no court that would not sign off on that.

So lets wait for the official benchmarks and real chips that AMD provides to the press and websites.
 
  • Like
Reactions: Zstream

inf64

Diamond Member
Mar 11, 2011
3,685
3,957
136
Guys can we not take every fake post and every fake image as fact? The fact is they probably did an estimation and lied that they somehow managed to get a real sample of the chip. There is no way someone could give them a chip in early beta development, that would be criminal and I doubt AMD would not know how many chips and where they are. It would be investigated, person would be arrested and chp returned. Since it would be technically stolen material, the paper would not be able to publish results that they got and I see no court that would not sign off on that.

So lets wait for the official benchmarks and real chips that AMD provides to the press and websites.

Early beta development? You do realize that NDA is going to be lifted in 14 days and launch is on CES or shortly after that? CanardPC is one of the most trusted sources in the IT journalism, they were the same people who first tested the K8 ES back in the day. How they obtained the ES/motherboard or whether they signed NDA (obviously not) is beside the point.
 

USER8000

Golden Member
Jun 23, 2012
1,542
780
136
The print magazine which leaked the results is very well known in France and generally in Europe - the people behind it leaked Athlon 64 scores before it was released and were the first to do it. They gain little by making stuff up as it would harm their reputation permanently.

But I get what you are saying too,that we should wait for reviews,as we can see final clockspeeds,etc,but the results look solid overall.
 

DeeJayBump

Member
Oct 9, 2008
60
63
91
The attempts to ignore, refute, dis-inform, FUD-ify and/or diminish the combination of New Horizon demos and this CPC preview are getting even more hilarious.

Stolen.FAKED. Estimations. It's all been a complete figment of our collective imaginations, yeah, that's the ticket.
 
Last edited:

.vodka

Golden Member
Dec 5, 2014
1,203
1,537
136
Guys can we not take every fake post and every fake image as fact? The fact is they probably did an estimation and lied that they somehow managed to get a real sample of the chip. There is no way someone could give them a chip in early beta development, that would be criminal and I doubt AMD would not know how many chips and where they are. It would be investigated, person would be arrested and chp returned. Since it would be technically stolen material, the paper would not be able to publish results that they got and I see no court that would not sign off on that.

So lets wait for the official benchmarks and real chips that AMD provides to the press and websites.

Did Canard PC fake this too back in 2003? 8 months before Athlon 64 launch. Hell, look at the numbers of that ES. A64 turned out to be quite the stunner when it was tweaked enough at release.

It's a major French PC magazine publication that has done this in the past... and are now doing this again. The very same guy, while we're at it. Yet, he's still here. If anything, this is as far as you're getting from a fake, since we're near launch. Reviews are going to be better than this. In the new horizon event it was stated that BASE clocks are going to be 3.4GHz or higher. This sample runs at 3.15/3.4 GHz... plus other platform tweaking that still has to be done (turbo, XFR, etc, those features weren't enabled in the new horizon demo which ran at a fixed 3.4GHz).

Anyway, Lisa said desktop SR (what has been leaked by Canard PC) launches in Q1 and CES is only two weeks away. Let's see what AMD has to show then.
 
Last edited:

clemsyn

Senior member
Aug 21, 2005
531
197
116
Will wait for release and see how reviews go before deciding to upgrade my Xeon 775 mod....