1st Ryzen CPU Review Leaked

S

SixFootDuo

Supreme [H]ardness
Joined
Oct 5, 2004
Messages
5,825
From a French Review Site that's been around since 2001. I guess everyone is saying this guy and the review is legit.

update: it's from an upcoming magazine article / review that has not been published yet. This seems to give more cred to the review as the lead out time on magazines can be up to 90 days.

http://wccftech.com/amd-ryzen-review-leaked/

Productivity seems to match and slightly beat the Intel 6900

Sadly, game performance doesn't look all that great.

If this turns out to be true, it's not anything I would be interested in. I can do better with Intel on price and performance, as I always suspect I would be able to do.
 
Hrmm well Operating System drivers has A LOT to do with CPU performance. And if this is truly some late model engineering sample then all we need is a good update to WIndows 10 etc... and I bet it does far better.

From the standpoint of gaming.... seriously and I mean seriously... does a few percent really matter given that AMD's own demo at New Horizon event showed the 490 coupled with the Zen at well over 60fps steady in 4K. We really really REALLY REALLY need to wait on official release and reviews before making these obsurd prejudices.
 
This is supposedly an un-optimized engineering sample chip.

Boost speeds could increase, OS driver support and general CPU optimizations.

I would hate to be the people at AMD that are in charge of pricing this chip. That's gotta be a horrible task to undertake. Get the pricing wrong, which they will and we are going to end up with a meh ... experience across the board.

Fantastic pricing could make this chip very awesome. Lower the price, then gain a lot more sales. Pricing it too high and people will be forced to consider Intel.
 
SixFootDuo said:
This is supposedly an un-optimized engineering sample chip.

Boost speeds could increase, OS driver support and general CPU optimizations.

I would hate to be the people at AMD that are in charge of pricing this chip. That's gotta be a horrible task to undertake. Get the pricing wrong, which they will and we are going to end up with a meh ... experience across the board.

Fantastic pricing could make this chip very awesome. Lower the price, then gain a lot more sales. Pricing it too high and people will be forced to consider Intel.
Click to expand...


Even this supposed real review which I think is another propaganda fake dumb shit fake news piece, the damn thing is running literal circles around my 3930K unbelievably so. I am going to get a Zen IF and only IF the pricing is right. You are dead smack on the money with your comment about getting the price right. Less than $500 and we may have a winner here. More and we will have a bomb loser.
 
If you run the math an uptick to about 3.4Ghz will change the outcome to around 102% while an uptick to 3.8Ghz is around 107%

AMD-Ryzen-Gaming-Performance-Benchmarks-Witchere-3-Arma-III-Far-Cry-4-Battlefield-4.jpg


This is still very good performance. Gaming relies more on clock speed and baseline IPC, shows that the Zen architecture is about Haswell level and the chip performs in that level which is good .
 
Why not clock it at 5Ghz now you are at it. I dont see your Haswell performance that half your posts have been about.

Anyway, notice Intel chips gain about 1.5% per 100Mhz. Good luck with a 5Ghz Ryzen to fight a stock 6700K. We all know you need 5Ghz Haswells to match a stock 6700K ;)
 
Last edited:
SixFootDuo said:
Productivity seems to match and slightly beat the Intel 6900
Click to expand...

Its quite far from a 6900K isn't it.

SixFootDuo said:
This is supposedly an un-optimized engineering sample chip.

Boost speeds could increase, OS driver support and general CPU optimizations.
Click to expand...

Yes and we are still waiting for the FX optimizations. Oh wait.

3.15Ghz baseclock, 3.4Ghz boost on all cores, 3.5Ghz single core boost.
 
Looks promising. Lisa said 3,4 GHz and then boost on top of that so clearly AMD has made massive improvements. Now it's all about how high it will clock, for gaming performance.
 
Didn't AMD say that the 8-core Ryzen will have a 3.4Ghz base?

And really? The chip is, like, 5% behind a 6700 in games and we are declaring the death of AMD?

In the end, pricing is key. I think $499 is too much. A penny over $399 and this chip fails.
 
Well I'd expect the 8 core part to be quite a bit cheaper then Intels comparable chip, which is selling for 1100€
Which would most likely still be out of the ballpark for practically any gamer

And also not very usefull for gaming anyway
But a lot more useful if you're working with you're rig

Way more interesting would be 4 and 6 core parts
And how high those would clock

Yield should also be better the less cores Ryzen has and therefore prize
 
Shintai said:
Why not clock it at 5Ghz now you are at it. I dont see your Haswell performance that half your posts have been about.

Anyway, notice Intel chips gain about 1.5% per 100Mhz. Good luck with a 5Ghz Ryzen to fight a stock 6700K. We all know you need 5Ghz Haswells to match a stock 6700K ;)
Click to expand...


http://www.hardocp.com/article/2015/08/05/intel_skylake_core_i76700k_ipc_overclocking_review/1

Kyle Bennett did this review, so if you are worried about authenticity, take it up with him.

1438184048QCHM79YbJA_3_5.png


1438184048QCHM79YbJA_6_1.png


So we start at similar clocks in CPU bound perameters, the epic 6700K is a whopping 1% faster than a 3 year old 4770K at similar clocks

1438184048QCHM79YbJA_6_2.png


4770K happened to beat the 6700K 1%

1438184048QCHM79YbJA_6_3.png


The 4770K is faster by 8%

1438184048QCHM79YbJA_6_4.png


The 6700K is faster by 2%

The long and short is, that outside Synthetics, a Haswell is still almost the same as Skylake in gaming, this is well know. I sold my 6600K setup to get a i7 4960X setup and pocket a bit as the 4960X combo was considerably less than what I sold my Skylake setup for.
 
OrangeKhrush said:
http://www.hardocp.com/article/2015/08/05/intel_skylake_core_i76700k_ipc_overclocking_review/1

Kyle Bennett did this review, so if you are worried about authenticity, take it up with him.

1438184048QCHM79YbJA_3_5.png


1438184048QCHM79YbJA_6_1.png


So we start at similar clocks in CPU bound perameters, the epic 6700K is a whopping 1% faster than a 3 year old 4770K at similar clocks

1438184048QCHM79YbJA_6_2.png


4770K happened to beat the 6700K 1%

1438184048QCHM79YbJA_6_3.png


The 4770K is faster by 8%

1438184048QCHM79YbJA_6_4.png


The 6700K is faster by 2%

The long and short is, that outside Synthetics, a Haswell is still almost the same as Skylake in gaming, this is well know. I sold my 6600K setup to get a i7 4960X setup and pocket a bit as the 4960X combo was considerably less than what I sold my Skylake setup for.
Click to expand...

You contradict yourself again just as you dont understand core IPC. Placing the bottleneck somewhere else. Explain the faster memory numbers. And 1866Mhz isn't the standard for Skylake. And for Ryzen its 2400Mhz.

I can also make Bobcat/Jaguar, Atom, K8, Ryzen, P4 and Skylake perform exactly the same if I place a strong enough bottleneck.

pcars-fps.gif


Ring a bell?
 
Last edited:
OrangeKhrush said:
http://www.hardocp.com/article/2015/08/05/intel_skylake_core_i76700k_ipc_overclocking_review/1

Kyle Bennett did this review, so if you are worried about authenticity, take it up with him.

1438184048QCHM79YbJA_3_5.png


1438184048QCHM79YbJA_6_1.png


So we start at similar clocks in CPU bound perameters, the epic 6700K is a whopping 1% faster than a 3 year old 4770K at similar clocks

1438184048QCHM79YbJA_6_2.png


4770K happened to beat the 6700K 1%

1438184048QCHM79YbJA_6_3.png


The 4770K is faster by 8%

1438184048QCHM79YbJA_6_4.png


The 6700K is faster by 2%

The long and short is, that outside Synthetics, a Haswell is still almost the same as Skylake in gaming, this is well know. I sold my 6600K setup to get a i7 4960X setup and pocket a bit as the 4960X combo was considerably less than what I sold my Skylake setup for.
Click to expand...


which numbers are you looking at because there is a major variation based on memory clock speeds here, that means Skylake needs faster memory/ more bandwidth to reach its peak performance. Hence why Purley platform is going to be needed for it to show its full potential since its going to give 50% more bandwidth because of the use of 6 channel memory vs what they are using right now quad channel.
 
Last edited:
Shintai said:
You contradict yourself again just as you dont understand core IPC. Placing the bottleneck somewhere else. Explain the faster memory numbers. And 1866Mhz isn't the standard for Skylake. And for Ryzen its 2400Mhz.

I can also make Bobcat/Jaguar, Atom, K8, Ryzen, P4 and Skylake perform exactly the same if I place a strong enough bottleneck.

pcars-fps.gif


Ring a bell?
Click to expand...

The memory at 2133 is still only 3% higher, then you have to add 2133mhz to the Haswell so it is all relative.

Ryzen ES is locked at 2133mhz.

razor1 said:
which numbers are you looking at because there is a major variation based on memory clock speeds here, that means Skylake needs faster memory/ more bandwidth to reach its peak performance. Hence why Purley platform is going to be needed for it to show its full potential since its going to give 50% more bandwidth because of the use of 6 channel memory vs what they are using right now quad channel.
Click to expand...

Again he showed 2133mhz and the variation is only 3%, but a Haswell running 2133mhz will also show the same uptick soa gain all relative.
 
OrangeKhrush said:
The memory at 2133 is still only 3% higher, then you have to add 2133mhz to the Haswell so it is all relative.

Ryzen ES is locked at 2133mhz.



Again he showed 2133mhz and the variation is only 3%, but a Haswell running 2133mhz will also show the same uptick soa gain all relative.
Click to expand...

AMD says it used 2400Mhz. Its not locked to 2133Mhz so prove it unless its just more of the fairy tale adventures of yours. DDR3 at 2133Mhz got very different timings than DDR4 at 2133Mhz.

And again, you keep avoiding the facts because its directly contradicts your posts.

Not to mention I even showed you a 5775C beating a 6700K with what, 20-30% clock for clock?

Guess what this is, from your own post:
13-14-14-31-2T vs 9-10-9-27-2T
 
Last edited:
OrangeKhrush said:
The memory at 2133 is still only 3% higher, then you have to add 2133mhz to the Haswell so it is all relative.

Ryzen ES is locked at 2133mhz.



Again he showed 2133mhz and the variation is only 3%, but a Haswell running 2133mhz will also show the same uptick soa gain all relative.
Click to expand...


I didn't see any mention of Haswell with higher memory clocks there did you?

And yeah Shintai is right, AMD did use 2400mhz for memory on the Ryzen presentations, so I have no idea why you are trying to even compare that to these......

Look its simple, just wait for the real numbers to come out, you trying to extrapolate things by pulling bits of information from here and there, and then trying to compare things that shouldn't be compared, is just hokey.
 
Shintai said:
AMD says it used 2400Mhz. Its not locked to 2133Mhz so prove it unless its just more of the fairy tale adventures of yours. DDR3 at 2133Mhz got very different timings than DDR4 at 2133Mhz.

And again, you keep avoiding the facts because its directly contradicts your posts.

Not to mention I even showed you a 5775C beating a 6700K with what, 20-30% clock for clock?

Guess what this is, from your own post:
13-14-14-31-2T vs 9-10-9-27-2T
Click to expand...


I will give you the contact details of a tech writer, tester and international overclocker, he will tell you that the Zen IMC on all ES chips are locked to 2133Mhz, even if you use 2400 it will just run natively at 2133.

I was just proving you wrong, a Haswell doesn't have to run at 5Ghz to match a Skylake and the test was done by Kyle, so if you have any issues with it go bitch to him about it.


DDR3 has lower latecy that DDR4 but less bandwidth so their is an offset, a DDR4 CPU should not be beat by a DDR3 CPU except it does. Again, got no need to get a 6700K with a 4790K
 
razor1 said:
I didn't see any mention of Haswell with higher memory clocks there did you?

And yeah Shintai is right, AMD did use 2400mhz for memory on the Ryzen presentations, so I have no idea why you are trying to even compare that to these......

Look its simple, just wait for the real numbers to come out, you trying to extrapolate things by pulling bits of information from here and there, and then trying to compare things that shouldn't be compared, is just hokey.
Click to expand...


No but it doesn't have to, DDR3 vs DDR4 should still benefit DDR4 system clock for clock yet the Haswell CPU can win in certain titles. DDR type is not really going to affect the outcome all that much.
 
No one stated Haswell has to clocked at 5ghz to match Skylake, the IPC difference between Haswell to Skylake is like 10-15% ish at most? Its just one gen difference.

Ok so you contacted the article author. 2133mhz looks fine, now where is Haswell benchmarks with 2133mhz on its memory? I haven't seen anything that you linked to in the [H] pics you have shown?

You are trying to extrapolate between apples and oranges and pineapples.
 
OrangeKhrush said:
I will give you the contact details of a tech writer, tester and international overclocker, he will tell you that the Zen IMC on all ES chips are locked to 2133Mhz, even if you use 2400 it will just run natively at 2133.

I was just proving you wrong, a Haswell doesn't have to run at 5Ghz to match a Skylake and the test was done by Kyle, so if you have any issues with it go bitch to him about it.


DDR3 has lower latecy that DDR4 but less bandwidth so their is an offset, a DDR4 CPU should not be beat by a DDR3 CPU except it does. Again, got no need to get a 6700K with a 4790K
Click to expand...

There is nothing wrong with Kyles review. There is however something wrong with your extrapolation like you keep doing.
 
razor1 said:
No one stated Haswell has to clocked at 5ghz to match Skylake, the IPC difference between Haswell to Skylake is like 10-15% ish at most? Its just one gen difference.

Ok so you contacted the article author. 2133mhz looks fine, now where is Haswell benchmarks with 2133mhz on its memory? I haven't seen anything that you linked to in the [H] pics you have shown?

You are trying to electroplate between apples and oranges and pineapples.
Click to expand...

Actually Shintai did but whatever.
 
he was being sarcastic lol, yeah it doesn't come across posts well, but that was pretty easy to see.
 
Shintai said:
There is nothing wrong with Kyles review. There is however something wrong with your extrapolation like you keep doing.
Click to expand...

Errrr no, there is something wrong with your rose tinted specs, under the same environment and clockspeed the 6700K doesn't need a 5Ghz 4770K to beat it, it does that clock for clock.

Ther is possibly less than 5% gaming difference between a Skylake and Haswell comparitive part. There is barely reason for anyone to upgrade from Haswell to Skylake.
 
OrangeKhrush said:
Errrr no, there is something wrong with your rose tinted specs, under the same environment and clockspeed the 6700K doesn't need a 5Ghz 4770K to beat it, it does that clock for clock.

Ther is possibly less than 5% gaming difference between a Skylake and Haswell comparitive part. There is barely reason for anyone to upgrade from Haswell to Skylake.
Click to expand...

Why dont you prove it. :)
 
Only the bottom of the latest page was scanned in the other post, the part containing the results. As it was already shared on over half the internet, I'll give you all a few precisions. CPC hardware 31 is centred on Intel's decline and management problems, and they finish their cover of that by doing a 6-pages preview of a Zen ES.
Test configuration (missing from latest page)
They used a Zen ES CPU named 2D315A2M88E4 ( 8c at 4.15/3.3/3.5 Ghz ) They used a motherboard with a Promontory SB and 16 gigs of DDR4-2400. They had to use a Radeon Fury X and a 3To harddrive because the motherboard wouldn't detect the 1080 and Crucial SSD.
They stated on twitter they did not hit more than 3.4 Ghz in practice, hinting at an older sample.
I won't just scan the whole ordeal though, they said already some sites just reposted their stuff and they are not too happy about it, I'll make some more comments on what the 6-pages report contains feel free to ask.
If you live in France/Belgium/Switzerland, go buy it now instead of reading it on the Internet ! It's 5€ and it's a great read all-round, especially their focus on Intel and their management problems, from their insiders.




I think the highlighted text should show you how fragile and unstable their setup was. :)
 
cageymaru said:
Only the bottom of the latest page was scanned in the other post, the part containing the results. As it was already shared on over half the internet, I'll give you all a few precisions. CPC hardware 31 is centred on Intel's decline and management problems, and they finish their cover of that by doing a 6-pages preview of a Zen ES.
Test configuration (missing from latest page)
They used a Zen ES CPU named 2D315A2M88E4 ( 8c at 4.15/3.3/3.5 Ghz ) They used a motherboard with a Promontory SB and 16 gigs of DDR4-2400. They had to use a Radeon Fury X and a 3To harddrive because the motherboard wouldn't detect the 1080 and Crucial SSD.
They stated on twitter they did not hit more than 3.4 Ghz in practice, hinting at an older sample.
I won't just scan the whole ordeal though, they said already some sites just reposted their stuff and they are not too happy about it, I'll make some more comments on what the 6-pages report contains feel free to ask.
If you live in France/Belgium/Switzerland, go buy it now instead of reading it on the Internet ! It's 5€ and it's a great read all-round, especially their focus on Intel and their management problems, from their insiders.




I think the highlighted text should show you how fragile and unstable their setup was. :)
Click to expand...

I was gonna post that right now. I think I'll buy the magazine copy online and upload the pages for the entirety of the article. I'm not seeing all six pages on the sites I'm checking.

Edit: Never mind, only physical copies are attainable.
 
OrangeKhrush said:
1438184048QCHM79YbJA_6_1.png


The difference between 341 and 337 in a completely CPU bound resolution is 1%
Click to expand...

Its not CPU bound is it. If it was it wouldn't change with faster memory.

Also you compare 2 setups with completely different timings.
13-14-14-31-2T vs 9-10-9-27-2T

Now try again, see if you get it right next time.
 
Shintai said:
Its not CPU bound is it. If it was it wouldn't change with faster memory.

Also you compare 2 setups with completely different timings.
13-14-14-31-2T vs 9-10-9-27-2T

Now try again, see if you get it right next time.
Click to expand...


640x480 resolution takes load off the GPU like 4K takes the load off the CPU and puts it all on the GPU, this is very old news, Kyle tested the two CPU's in an environment where there is no dependence on the GPU to distort the numbers.

Haswell runs on DDR3, timings don't alter performance that much and that has been proven to death already. DDR3 also runs lower latency but less bandwidth so there is a trade off. The test was 100% accurate.
 
OrangeKhrush said:
640x480 resolution takes load off the GPU like 4K takes the load off the CPU and puts it all on the GPU, this is very old news, Kyle tested the two CPU's in an environment where there is no dependence on the GPU to distort the numbers.

Haswell runs on DDR3, timings don't alter performance that much and that has been proven to death already. DDR3 also runs lower latency but less bandwidth so there is a trade off. The test was 100% accurate.
Click to expand...

Again, you clearly have no idea what a CPU bound limitation is. Just as you have no idea what IPC is or understanding about why you cant extrapolate data.

I already gave you a big hint earlier. Let me post it again since you didn't understand it.
upload_2016-12-24_17-31-33.png


A 3.3Ghz base, 3.7Ghz turbo Boradwell part beating the shit out a 4Ghz/4.4Ghz Haswell and a 4Ghz/4.2Ghz Skylake. And the 5775C uses the same memory as the 4790K.

And hold my horses there, isn't the 6700K 13% faster than the 4790K? OH MY!

The 4790K is clocked 19% higher than the 5775C and its 19% faster on top over the 4790K.

The big hint is memory bottleneck. Because we all know a 5775C isn't faster computational than any of the 2 others.
 
Last edited:
OrangeKhrush said:
http://www.guru3d.com/articles_pages/core_i7_6700k_processor_review_desktop_skylake,14.html

index.php


The difference between DDR4 2133 and 3200 is less than 2% .

index.php


Again the difference is less than 2% and the i7 4770K with 1600 is inside 3% of the score of a 6700K with much faster memory.

Your obsession with arguing baseless facts and naivity are bordering a common syndrome, it is called fanboy syndrome.
Click to expand...


What are you trying to show here? Memory bandwidth, IPC, changes, what? What are you trying to baseline before you make your extrapolation? Cause it all changes based on what programs what you are looking at and what you are trying to equalize before you extrapolate. Pointing to different reviews and trying to show something, doesn't show only one thing.
 
Since nobody cares about prj
razor1 said:
What are you trying to show here? Memory bandwidth, IPC, changes, what? What are you trying to baseline before you make your extrapolation? Cause it all changes based on what programs what you are looking at and what you are trying to equalize before you extrapolate. Pointing to different reviews and trying to show something, doesn't show only one thing.
Click to expand...

Proved to you that Memory speed makes no difference, there is a 4770 with 1600Mhz inside 4 percent of the score posted by a 6700K with 3200mhz DDR4. So in short that crap about the Skylake being choked is bull. The reality is that there is hardly a distance between a Haswell and Skylake in gaming.
 
You must log in or register to reply here.
Back
Top