Leaked AMD Zen Engineering Sample Benchmarks?

Would definitely consider jumping ship as long as they can keep power consumption under control, was the only reason I didn't opt for one of their previous offerings.
 
If you wanted to show cpu advancement why on earth would you do it showing it on a gpu bound API ?
A lot of people are getting this wrong. AMD isn't showing anything. This was an engineering sample that got caught being used in a benchmark and leaked. This is a Zen SE; the Pollux SE, for example, was clocked at 800MHz which is only 63% of its released frequency of 1266 MHz. So expect the final Zen to have a significant boost in frequency from this engineering sample.
 
can you prove Skylake is 3% faster than haswell?.. =) I would like to see that both in high-level multi-thread apps and specially gaming scenarios, thanks in advance.

Sorry, this is kinda off-topic for this thread.

I'm not sure what your post was about, are you saying Skylake is more or less then 3% faster? To DukenukemX's credit he did say:
like 3% difference

Non-Gaming CPU benchmark: 5.7% average increase in performance between Haswell and Skylake:
The Intel 6th Gen Skylake Review: Core i7-6700K and i5-6600K Tested

Gaming benchmark shows 1.3% average decrease in performance between Haswell and Skylake:
The Intel 6th Gen Skylake Review: Core i7-6700K and i5-6600K Tested

Kyle's review of Skylake:
Gaming Benchmarks - Intel Skylake Core i7-6700K IPC & Overclocking Review
Metro: Last Light is certainly one of our newest and most popular gaming engines used. This one is also run at a higher resolution with higher image quality, so it should be more GPU limited than the others. Our "apples to apples" results show Skylake with a 3% increase over Haswell

Maybe he looked at one of these?

I believe the point of his post was that outside of a benchmark and/or running FRAPS, there's at best a small performance delta between the two. Seems like you're nitpicking someone's generalized comment. Perhaps start a thread in the Intel Processors section for discussion?
 
A lot of people are getting this wrong. AMD isn't showing anything. This was an engineering sample that got caught being used in a benchmark and leaked. This is a Zen SE; the Pollux SE, for example, was clocked at 800MHz which is only 63% of its released frequency of 1266 MHz. So expect the final Zen to have a significant boost in frequency from this engineering sample.

From what I have "heard" it is prolly not going to be much more then what the ES is showing, don't expect 4ghz from Zen.
 
This looks good based on these benchmarks seems to be at Haswell level IPC.

I was concerned it would show up with Ivy bridge level IPC.
 
can you prove Skylake is 3% faster than haswell?.. =) I would like to see that both in high-level multi-thread apps and specially gaming scenarios, thanks in advance. I'll be waiting for the proof of your claims.

but in the mean time.. I would reconsider that 3% claim with this video.


Whatever, either way there isn't enough reason for a Haswell K part owner to upgrade. There are some applications that do benefit more from the Skylake CPUs but not enough to upgrade. Maybe if the 6600K had hyperthreading or 6 cores then sure. I think we're starting to see Intels limit to edge more out of x86. And yea, my 3% remark is something I'm repeating cause I've heard it before many times.

BTW in your video the 8350 does really well in Witcher 3 walking around town. I found that interesting.
 
I don't think they're going to sell too many of these.

I think the AMD fans will buy them / upgrade but most everyone else will just stick with Intel since they make the best stuff, even if you do have to pay a bit more for it.
 
I don't think they're going to sell too many of these.

I think the AMD fans will buy them / upgrade but most everyone else will just stick with Intel since they make the best stuff, even if you do have to pay a bit more for it.
Probably right but I do see some of those Gaming system builders pushing out a few of these. If these can push CF/SLI better than the FX, as they should be able to, it would open up that platform a bit to AMD.
 
can you prove Skylake is 3% faster than haswell?.. =) I would like to see that both in high-level multi-thread apps and specially gaming scenarios, thanks in advance. I'll be waiting for the proof of your claims.

but in the mean time.. I would reconsider that 3% claim with this video.



Wow, at least according to their results, the FX 8350 was beating every processor but the 6600k, interesting.
 
I used to jump back and forth between AMD and intel, ATI and nVidia, now it's AMD intel/nVidia but have I been intel/nVidia since 2009 because AMD keeps releasing stuff 1-2 years slower than what I already own.
 
Last year I heard that top yield for ZEN was 3Ghz base with 3.5Ghz Turbo on the 8 core/16HT version
 
If I were a betting man 4 core version should be able to do 3.2Ghz base w/3.6Ghz Turbo.

What I really hope for is AMD tying the turbo mode to the temperature, just like the old i7 2600 series. So we could just grab the top processor, bolt on a huge cooler and run 3.5Ghz all day on the 8 core version.
 
Wow, at least according to their results, the FX 8350 was beating every processor but the 6600k, interesting.

Did you forgot to say only in the witcher 3? performing equal in Crysis 3? and being embarrassed outperformed in every other title?.
 
Did you forgot to say only in the witcher 3? performing equal in Crysis 3? and being embarrassed outperformed in every other title?.

I am sorry but, what does that have to do with anything? If you do not like those results you did see, tough, not my problem. :)
 
Based on my math a 3Ghz ZEN processor would get 62 FPS and still be slower than the i7 4790 processor in Ashes of the Singularity.

But if max turbo "3.5Ghz" can be sustained as I mentioned previously by just adding a better heatsink.... 72 FPS should be possible, and THAT could give INTEL a run for their money.

again this is ONLY applicable to Ashes of the Singularity
 
Look at it this way, AMD doubled the number of cores, but only squeezed out a 32% performance gain. Work the math, and IPC in AoS actually DROPS compared to the FX-8350.

Any benchmarks to show AoS scales extremely well on Intel when doubling threads?
This approach to finding out Zens secret sauce is very flawed. Show me any real world game where doubling of threads over 4 shows such a dramatic improvement that you can then estimate that CPU's IPC?
Your logic would make any 8+ thread Intel CPU appear to have its IPC crap as well.
 
Apparently the Oxide engine scales well even with 16 threads on a CPU ;), which it does seem to do that well.

Its one of the few engine/games that does this right now, most likely because the game takes advantage of the extra resources from the CPU.

If AMD matches Haswell with a 8 core Zen, err, they won't have an answer for Kaby Lake, Kaby Lake is likely not be the regular 8%-15% performance increases we have been seeing from Intel. From what I have heard it sounds to be more, more like the 10%-20% range.

But competition at the high end isn't everything, if AMD can get it power consumption down they will become a viable option for lower end laptop and desktops, cause OEM's, system builders can have an alternative to Intel with all other components staying roughly at the same price.
 
Last edited:
Apparently the Oxide engine scales well even with 16 threads on a CPU ;), which it does seem to do that well.

Its one of the few engine/games that does this right now, most likely because the game takes advantage of the extra resources from the CPU.

Well that should have been VERY easy for you to show. Benches please!
We are apparently extrapolating Zen's performance based on this, right?
As if somehow a game bench in which the doubling of cores = IPC performance metric.
LOL!
Surely you're just Taking the piss atm. (stupid emoticon "here")
 
I don't think they're going to sell too many of these.

I think the AMD fans will buy them / upgrade but most everyone else will just stick with Intel since they make the best stuff, even if you do have to pay a bit more for it.
People will buy what's best and if Zen is better then people will buy it. I can see companies and organizations buy Intel for this reason, but not end users. And if 32 core Zens are cheap and competitive, I can even see servers using them in favor of Intel.
 
Well that should have been VERY easy for you to show. Benches please!
We are apparently extrapolating Zen's performance based on this, right?
As if somehow a game bench in which the doubling of cores = IPC performance metric.
LOL!
Surely you're just Taking the piss atm. (stupid emoticon "here")
Just a quick " ashes of the singularity cpu thread scaling" search had this as one of the first results. Don't know if there are more, it's not that important to me.

Basically, if your CPU isn't waiting on the GPU, it scales pretty well. More with physical cores than hyperthreading, but still respectable either way.
 
Well that should have been VERY easy for you to show. Benches please!
We are apparently extrapolating Zen's performance based on this, right?
As if somehow a game bench in which the doubling of cores = IPC performance metric.
LOL!
Surely you're just Taking the piss atm. (stupid emoticon "here")


Yeah, who is talking piss when they can't use google?

broadwell_e_core_i7_6950x_ashes_of_the_singularity_beta_cpu_benchmark-100663335-orig.png


We can see even more scaling on a 10 core chip 20 threads.
 
I don't think they're going to sell too many of these.

I think the AMD fans will buy them / upgrade but most everyone else will just stick with Intel since they make the best stuff, even if you do have to pay a bit more for it.

Substantially more,
8320 $180, 8350 around $210, 3570k $310, 6600k $370
Current prices CAD
 
Also anyone thinking ES's are not indicative of final release CPU's they are fooling themselves, ES's are sent to OEM''s/system builders for testing and getting their systems ready and qualified for retail samples, so if ES's aren't indicative of the retail CPU's, then OEM's and system builders might as well not even bother with them, that being said, what we are seeing here, might not even be a ES, although with the time frame of when Zen is to release it seems about right and the serial numbers seem to match with the ES sample that was leaked earlier.
 
Last edited:
Yeah, who is talking piss when they can't use google?

broadwell_e_core_i7_6950x_ashes_of_the_singularity_beta_cpu_benchmark-100663335-orig.png


We can see even more scaling on a 10 core chip 20 threads.

Mind you, the scaling between cores is a little over half the theoretical. This seems to be the universal problem we see in all these multi-threaded engines. Going to be interesting to see if AMD can surpass the lowly 6700k this time around, if they're still going to trail this much in clock and IPC
 
I think they will catch up some what, a little bit with the clock speeds and IPC, but I don't think it will be enough to match Sky Lake let alone Kaby lake, this is a tough spot for AMD to be in, its been 10 years or so since they have been competitive, that is 4 generations, I think of Intel processors? If it doesn't happen within one generation... like it did in the past, between Intel and AMD, there are more trying times ahead.

The reason why Intel and AMD switched back and forth in the past, is they already had designs in the pipeline they could use. Intel had PIII which they fell back on, and AMD bought out Nexgen which they finally incorporated its tech in the Athlon line.

Taking the wrong direction for a generation is one thing, but coming out with Phenom, then Phenom II, then Bulldozer, that's 3 gens that were subsequently worse to Intel's offerings.

Lack of engineering IP and experienced gained in the now, does hurt in the future as you have pretty much stalled your forward progress, now if they were able to modestly build on their IP in the meantime we would have been able to see that in some way or form, which I think we will see in Zen.
 
Last edited:
I think they will catch up some what, a little bit with the clock speeds and IPC, but I don't think it will be enough to match Sky Lake let alone Kaby lake, this is a tough spot for AMD to be in, its been 10 years or so since they have been competitive, that is 4 generations, I think of Intel processors? If it doesn't happen within one generation... like it did in the past, between Intel and AMD, there are more trying times ahead.

The reason why Intel and AMD switched back and forth in the past, is they already had designs in the pipeline they could use. Intel had PIII which they fell back on, and AMD bought out Nexgen which they finally incorporated its tech in the Athlon line.

Yeah, the last time they were anywhere close was Nehalem versus Phenom II, and that was because it was the first generation for a lot of things they added, and the benefit of multi-threading was still limited to video conversions. STARS was a good core, just hampered by the fact that there wasn't enough TDP for everything to be native quad-core on 65nm. And of course that stupid recall :(
 
That's a huge win if it comes to pass.

Agreed.

AMD doesn't need to have the top dog CPU for it to be a success in the gaming community They just need it to perform above the threshold where the CPU is the bottleneck.

Current AMD designs are not doing this in all titles. If Zen does, it doesn't matter if it is single digits behind Intel's latest. I'd still buy it.
 
Sadly, that means that a price war would be temporary, AMD will die and Intel will just go back to rape us.

Personally, if Zen is good enough, I will buy it, instead of sitting down and wait for Intel to cut prices.


It will be interesting to see what happens.

Intel has been slowing down their Tick-Tock plan as of late, and laying off thousands of people, just as AMD is potentially getting back into the game.

Could be that Intel is finally making the determination that there is no future in x86, and they are refocusing their efforts and investments elsewhere.
 
These are big chips, so AMD will price them to make a profit IF THE PERFORMANCE IS THERE.

That's exactly my guess. If Zen performs at a competitive level, then AMD will price it to fit the market. There have been many times in the past where AMD hardware was priced higher than their competitor's products because they gained the performance crown. I'm recalling the $800+ A64 X2 4800+, as one example.
 
It will be interesting to see what happens.

Intel has been slowing down their Tick-Tock plan as of late, and laying off thousands of people, just as AMD is potentially getting back into the game.

Could be that Intel is finally making the determination that there is no future in x86, and they are refocusing their efforts and investments elsewhere.

To me, Intel hit a hard limit in shrinking and IPC, plus they haven't had any competition whatsoever from AMD in a while, so why rush to that metaphorical wall that is getting closer and closer to them?

They tried hard to compete with ARM, but sounds like they are giving up there, since IMHO, the rest of industry wants to escape Intel insane prices and did not abandon ARM as Intel hoped.

As you said, all AMD needs is a CPU that does not bottleneck games and to be honest, most games are really GPU bound.

All that Zen needs is to be as fast as current Intel's and way cooler/lower voltage consumption.

Again, I'm just an internet chair analyst, so I could be completely wrong. ¯\_(ツ)_/¯
 
Agreed.

AMD doesn't need to have the top dog CPU for it to be a success in the gaming community They just need it to perform above the threshold where the CPU is the bottleneck.

Current AMD designs are not doing this in all titles. If Zen does, it doesn't matter if it is single digits behind Intel's latest. I'd still buy it.


They don't need to compete in the top end, but they still need to have some semblance of competition where ever they end up pricing their chips, both in power usage and in performance.

The big uplift from Zen even if its not that competitive at the top end, its still a much smaller chip than the one it will be replacing which should help improve margins if not priced in the bargain bin section. Current AMD CPU's aren't doing anything at all right now outside of killing margins and not selling well.

AMD margins right now are around 30% but that is over all their products, but lets take that as the figure for now, how much would a 5% or 10% increase in margins look for AMD? I think we are talking about possible even 100's of millions of dollars or more per quarter. That surely will help stabilize their bottom line.
 
The big uplift from Zen even if its not that competitive at the top end, its still a much smaller chip than the one it will be replacing

I expect that the 8 core / 16 threaded Zen not to be that much smaller than the current 8 core. The reason is 1 core will be at least as many transistors as the current 2 core bulldozer module. Although with that said the 8 core / 16 threaded CPU will certainly be more expensive so I expect there to be more room for AMD turning a profit.
 
I expect it to be a lot smaller,

1) they aren't using automated processes anymore for their CPU's, which they did for Bulldozer, that right there can save up to 30% space or 30% more performance due to higher clocks and power, if they were on the same node,
2) smaller node

I expect Zen core for core will be around the same size as Intel's 14nm (within a few % points)
 
Back
Top