Official AMD Ryzen Benchmarks, Reviews, Prices, and Discussion

[Ryzenenthusiast]

Nice! Could you post a link to where you got that?

It supports the idea that the R7 1700 is made up of lower quality dies that obviously couldn't fit the targeted clockspeeds/voltage and 95w bin of the R7 1700x and 1800x, and as such runs perfectly well underclocked at 3.0/3.7GHz to fit the 65w bin.

This also supports the rumor/report that being lower quality dies, obviously requiring higher voltages to get to 4 GHz, 4GHZ R7 1700 were cooking the VRMs of lower end boards but not the monster VRM of the Crosshair VI Hero.

If true, the $70 extra for the R7 1700x is warranted if it can easily clock past 4GHz.



Excellent.

Man I can't wait for actual reviews.

This is the video; https://www.youtube.com/watch?v=UhKmeCdB914

To sum up the things he stated;

"Overclock is no problem at all. Though we can not say a certain frequency due to NDA, I can say that it is impressive. Overclockers will be happy."
"I have friends from AMD Engineering department due to my experience in the hardware industry. Zen 2 or Zen B will be even more competitive. So we can say that Intel shall brood on the future"
"With our overclocked 1800x sample,under Noctua cooler given by AMD, we have passed beyond the stock single thread performance of 7700k, in a specific bench, and the temps were great. We had no concern about temps during our run which passes the ST performance of 7700k."
"In some benchmarks AMD(!). It seems ironical yes but AMD is presenting a CPU performance that Intel can not keep up with even with their 10 core 6950x!"
"Breaking NDA won't be a problem since the scores are beyond fantastic. But we will stick with the tradition."
"With one click I can reach great OC's. So I won't really bother with the manual OC no more."
"Single thread score will be so great. According to this performance we can say that 7700K will be history, even for gaming, from now on" he said.
"Intel shall shake themself. Because Ryzen will be a great choice for Overclock enthusiast."

 

[Ryzenenthusiast]

Ahh he also stated that there is auto overclocking feature. "With one click I can reach great OC's. So I won't really bother with the manual OC no more."

 

[imported_jjj]

Maybe it wasn't all core OC?

 

[Ryzenenthusiast]

Maybe it wasn't all core OC?

"Single thread score will be so great. According to this performance we can say that 7700K will be history, even for gaming, from now on" he said.
"Intel shall shake themself if they do not want to bankrupt. Because Ryzen will be a great choice for Overclock enthusiasts too."

 

[richierich1212]

Maybe it wasn't all core OC?

Doesn't matter, R7 still has 7 more cores or 14 more threads to back it up lol.

 

[.vodka]

This is the video; https://www.youtube.com/watch?v=UhKmeCdB914

To sum up the things he stated;

"Overclock is no problem at all. Though we can not say a certain frequency due to NDA, I can say that it is impressive. Overclockers will be happy."
"I have friends from AMD Engineering department due to my experience in the hardware industry. Zen 2 or Zen B will be even more competitive. So we can say that Intel shall brood on the future"
"With our overclocked 1800x sample,under Noctua cooler given by AMD, we have passed beyond the stock single thread performance of 7700k, in a specific bench, and the temps were great. We had no concern about temps during our run which passes the ST performance of 7700k."
"In some benchmarks AMD(!). It seems ironical yes but AMD is presenting a CPU performance that Intel can not keep up with even with their 10 core 6950x!"
"Breaking NDA won't be a problem since the scores are beyond fantastic. But we will stick with the tradition."
"With one click I can reach great OC's. So I won't really bother with the manual OC no more."
"Single thread score will be so great. According to this performance we can say that 7700K will be history, even for gaming, from now on" he said.
"Intel shall shake themself. Because Ryzen will be a great choice for Overclock enthusiast."

Ahh he also stated that there is auto overclocking feature. "With one click I can reach great OC's. So I won't really bother with the manual OC no more."

"Single thread score will be so great. According to this performance we can say that 7700K will be history, even for gaming, from now on" he said.
"Intel shall shake themself. Because Ryzen will be a great choice for Overclock enthusiast."

Holy crap. You say they're one of the biggest publications in Turkey, I suppose they wouldn't burn their name by spreading false information. Oh well, hope that ends up being true in reviews on March 3.


If not, well, let's say that an 8C/16T CPU that can do 4GHz on all cores and provides the performance we already know it does, for the asking price, is a more than compelling product that already shakes up this long stagnated market.


If it does... damn. I mean this information backs this up

edit:

it's donanimhaber!

Here is the Alexa link for the website of the Turkish leak I mentioned before;
http://www.alexa.com/siteinfo/donanimhaber.com

• 34th most popular website in Turkey
• 1680th most popular website in the world

legit source.

 

[CatMerc]

1 core OC'd to 5GHz, 7 cores set to 3GHz.

For when you want to make a 7700K feel bad.

 

[imported_jjj]

Everything matters and it matters quite a bit here as other reports suggest that all cores don't quite go above 4GHz.
Plus i want to go air for less noise at idle (no top rad leaves the sound dampening foam in place and that makes a difference) so not quite pleased if i'll be forced to go water.

ST OC at 4.5GHz would be ok, for the few things that require ST perf.
For gaming, games will go many cores, no other way forward and you are not supposed to be CPU bound to begin with.

 

[Phynaz]

Holy crap. You say they're one of the biggest publications in Turkey, I suppose they wouldn't burn their name by spreading false information. Oh well, hope that ends up being true in reviews on March 3.

But yet they will burn their name by breaking their NDA.

 

[looncraz]

Yes. The L3 probabily clocks as the NB, like INTEL and BD. AMD promised 5x BW for the L3. The bus is 4X and the clock should be +25%. But what was the clock on BD? The NB clock, that was 2.4GHz. 2.4+25%=3GHz. On BD the NB should go at least at half frequency of the fastest core so if for Zen is the same, up to 6GHz the NB can also stay at default. But obviously the scaling will not be linear if you raise only CPU clocks...

Yes, and the LN2 run was done with a higher bus clock. Showing 137.78Mhz tells me that bus frequency stretching is occurring (which can be ~7%, IIRC)... which I'll get to in a bit. If the team were going for maximum performance, and the cache multiplier is user-configurable, that would have been a no-brainer to adjust. I'm not convinced they were strictly going for highest frequency... otherwise they would have pushed the chip until it died, taking CPU-z validation screenshots along the way (maybe they did this, though, and we just don't know...). This was definitely a benching run.


Back to bus clock stretching... I've seen this with the x4 845 and couldn't explain it because AMD has been pretty quiet about it... the multiplier stayed at 38x, the core always showed about 3.8Ghz, but the FSB would bounce around.

The bus clock 'stretch' is done to save power, so at lower frequencies you will have more stalls to save power, but at higher frequencies those stalls go away increasingly more as you fill the queues faster from the caches. With bus-based clocking this would result in the relative speed against the L3 changing, which would make it more likely that the CPU would clear out the L3 data queue faster at higher clock speeds when it is used - a positive feedback with higher core frequencies is the only result to expect from that.

 

[IEC]

But yet they will burn their name by breaking their NDA.

They specifically stated they are only releasing information that does not go against the embargo terms. Which is why no specific frequency or additional details were provided.

 

[Phynaz]

They specifically stated they are only releasing information that does not go against the embargo terms. Which is why no specific frequency or additional details were provided.

Just stating that you are under NDA breaks the NDA . You can't publish / say a thing that isn't made publicly available by AMD. So therefore just stating that you are running a Ryzen CPU is in violation.

 

[looncraz]

Just stating that you are under NDA breaks the NDA . You can't publish / say a thing that isn't made publicly available by AMD. So therefore just stating that you are running a Ryzen CPU is in violation.

They aren't under an NDA, they are under a reviewer's embargo. You can say you are under embargo.

Showing the board... fine. Showing the setup... fine. Showing the box... fine. Talking about it... even in general terms... gray area.

 

[PotatoWithEarsOnSide]

It's an embargo, not a NDA.

 

[Ryzenenthusiast]

Just stating that you are under NDA breaks the NDA . You can't publish / say a thing that isn't made publicly available by AMD. So therefore just stating that you are running a Ryzen CPU is in violation.

They just said "We can not disclose certain numbers due to NDA" couple of times. Everybody is saying "We can't talk because we have NDA" I do not see a violation. These guys are in the hardware news business for +15 years. So they know what they are doing.

 

[Phynaz]

They aren't under an NDA, they are under a reviewer's embargo. You can say you are under embargo.

Showing the board... fine. Showing the setup... fine. Showing the box... fine. Talking about it... even in general terms... gray area.

Ummm....They said they are under NDA. Anyone who has a legitimate Ryzen CPU only has it because they are under an NDA.

 

[IEC]

Ummm....They said they are under NDA. Anyone who has a legitimate Ryzen CPU only has it because they are under an NDA.

It's a review embargo. They may be using the term NDA interchangeably, but a true legally-binding NDA is very uncommon and is not the proper term to describe the agreement.

<-- Formerly subject to actual NDAs and non-compete agreements.

 

[zinfamous]

Ewww... D.C..

I don't plan on going that way for as long as possible.

well, I'm outside of the wall, at least.

 

[zinfamous]

Yeah, the data actually shows a 6% IPC deficit against Kaby Lake... which, frankly, is a miraculous result from AMD.

Uh, wow--so that would mean it is Skylake...only better because of multithreaded performance?

damn...

 

[Phynaz]

It's a review embargo. They may be using the term NDA interchangeably, but a true legally-binding NDA is very uncommon and is not the proper term to describe the agreement.

<-- Formerly subject to actual NDAs and non-compete agreements.

<-- Currently under various NDA's

 

[zinfamous]

Ummm....They said they are under NDA. Anyone who has a legitimate Ryzen CPU only has it because they are under an NDA.

call it a translation error if you must. If it is officially an embargo, and not an NDA, then it is an embargo with its own terms. These guys are possibly just using the wrong expression.

 

[zinfamous]

<-- Currently under various NDA's

^currently violating various NDA's.

 

[looncraz]

Uh, wow--so that would mean it is Skylake...only better because of multithreaded performance?

damn...

In a couple tests, that appears to be the case. In integer-heavy workloads, I expect AMD to take the lead with room to spare. AMD ALUs have been superior to Intel ALUs for some time - that's how they thought they could get away with having just two ALUs in Bulldozer. In theory, if they could have kept them 100% utilized, they would have only had momentary reductions in IPC and 20% higher IPC than Phenom II, catching them up to Intel in integer using Bulldozer.

The problem is that it was an impossible chore... whoever was in charge of the high-level design managed to miss that. I caught it within twenty minutes of reading the optimization guide and started yelling at JF-AMD about it. IPC can't increase if you move from an architecture which had an average ILP of 1.8 if your new architecture's peak ILP is 2.0 and your efficiency is ~75%... it goes down. A lot.

 

[XMan]

/crosses for fingers for Mini-ITX board support with full-speed M.2.

 

[bjt2]

Yes, and the LN2 run was done with a higher bus clock. Showing 137.78Mhz tells me that bus frequency stretching is occurring (which can be ~7%, IIRC)... which I'll get to in a bit. If the team were going for maximum performance, and the cache multiplier is user-configurable, that would have been a no-brainer to adjust. I'm not convinced they were strictly going for highest frequency... otherwise they would have pushed the chip until it died, taking CPU-z validation screenshots along the way (maybe they did this, though, and we just don't know...). This was definitely a benching run.


Back to bus clock stretching... I've seen this with the x4 845 and couldn't explain it because AMD has been pretty quiet about it... the multiplier stayed at 38x, the core always showed about 3.8Ghz, but the FSB would bounce around.

The bus clock 'stretch' is done to save power, so at lower frequencies you will have more stalls to save power, but at higher frequencies those stalls go away increasingly more as you fill the queues faster from the caches. With bus-based clocking this would result in the relative speed against the L3 changing, which would make it more likely that the CPU would clear out the L3 data queue faster at higher clock speeds when it is used - a positive feedback with higher core frequencies is the only result to expect from that.

According to AMD papers, the clocks stretch of 7% is performed, core per core, on the CPU clock.
Probabily CPUz has no mean to measure bus clock and infer it from CPU clock (that can actually measure or read in MSRs) and multiplier (also MSRs), so when the CPU clock is stretched, the multiplier is not varied (the stretch is done in other ways) and CPUz calculates a wrong varying bus clock...

EDIT: the clock stretch is a vdroop protection. The saving in power is due to the reduction of the voltage margin to cope with vdroop.