Leaked AMD Ryzen Benchmarks?

I have already expressed my 2 concerns with it:
1. Version and build number just look... wrong. Are there any other benchmarks on this build?
2. AMD Eng Sample, not AMD Ryzen: . That one is obvious red flag. Oh, also, it has _Y instead of _N to indicate lack of iGPU. Name just does not align with AMD's ES naming (and yes, the CES computer with 3.6Ghz base clock did have AMD Eng Sample naming, so AMD does use it).

that IS the latest/current version of AOTS. (see pic below, from my install btw)
do you have access to a ryzen es? how do you know what they are all named?! internet "knowledge" or have you actually used one/have access?
you don't believe what you see, so its fake?

upload_2017-2-2_18-36-56.png


edit: these "I know more that you internet knowledge pissing contests" are getting really fucking old...
 
I have already expressed my 2 concerns with it:
1. Version and build number just look... wrong. Are there any other benchmarks on this build?
2. AMD Eng Sample, not AMD Ryzen: . That one is obvious red flag. Oh, also, it has _Y instead of _N to indicate lack of iGPU. Name just does not align with AMD's ES naming (and yes, the CES computer with 3.6Ghz base clock did have AMD Eng Sample naming, so AMD does use it).

Agreed, something is screwy. I just did a post at Toms:

http://www.tomshardware.com/forum/id-2986517/discussion-amd-zen/page-40.html#19246010

According to the closest results I can match to [where I note many times the Version of the benchmark and settings run at are different], I computed Ryzen IPC at...90% less then Haswell. Which would be worse then Bulldozer was. :/

So yeah, if anyone with a Titan X/Intel combo could run that specific version of the benchmark at the same settings AMD did, that would be great. Because honestly: Those results are crap, and Ryzen is DOA if those results are even remotely accurate.
 
that IS the latest/current version of AOTS. (see pic below, from my install btw)
do you have access to a ryzen es? how do you know what they are all named?! internet "knowledge" or have you actually used one/have access?
you don't believe what you see, so its fake?

View attachment 16115

edit: these "I know more that you internet knowledge pissing contests" are getting really fucking old...

I don't get why we then have AoS results from version 2.2. There a Beta version out or something? Also, AMD is using non-standard defaults again, as the AoS database results for the Crazy preset ALL have two differences: Terrain Shading Samples at 12 Million instead of 16 Million, and Half-Resolution Terrain set to On. That could be suppressing AMD's results a bit.

Hence why I'm confused here: Why are the versions and default settings in the AoS database different then the leaked benchmark? And more importantly: Are they responsible for the poor showing compared to corresponding Intel results?
 
idk, gog vs steam maybe?! they usually have independent version numbers for some reason. mine is gog.
and yeah i agree, the setting have to be exactly the same.
 
Agreed, something is screwy. I just did a post at Toms:

http://www.tomshardware.com/forum/id-2986517/discussion-amd-zen/page-40.html#19246010

According to the closest results I can match to [where I note many times the Version of the benchmark and settings run at are different], I computed Ryzen IPC at...90% less then Haswell. Which would be worse then Bulldozer was. :/

So yeah, if anyone with a Titan X/Intel combo could run that specific version of the benchmark at the same settings AMD did, that would be great. Because honestly: Those results are crap, and Ryzen is DOA if those results are even remotely accurate.

Yeah, I would not worry about that and the non provable results. Manufacturers are not going to start producing hardware that supports Ryzen if they know they will not sell well. There would be no reason whatsoever for these manufacturers and OEM's to invest in a DOA product especially since it was one produced by Jim Keller and team. We have nothing to worry about unless you are expecting a 7700K killer in all situations because then, you will be sorely disappointed.
 
do you have access to a ryzen es? how do you know what they are all named?! internet "knowledge" or have you actually used one/have access?
Everyone fucking knows how AMD names their engineering samples. CanardPC has shown that, a fucking photo from CES has shown that. Literally every other leak has shown that they name them "AMD Eng Sample *Identifier*".

Good to know about version number, only thing to figure out is why the hell database does not allow to see these results over at site directly.

Also, it is sort of weird to find people generally defending AMD jumping on defense of utterly terrible results for AMD CPU.
 
Everyone fucking knows how AMD names their engineering samples. CanardPC has shown that, a fucking photo from CES has shown that. Literally every other leak has shown that they name them "AMD Eng Sample *Identifier*".

Good to know about version number, only thing to figure out is why the hell database does not allow to see these results over at site directly.

Also, it is sort of weird to find people generally defending AMD jumping on defense of utterly terrible results for AMD CPU.
Kinda have to agree especially with the last part, usually you two are on the other side. Switch places so all is right with the world... lol
 
f0vgxw.png


1) 6800K 3.4Ghz - 3.6Ghz only shows base clock no OC numbers or turbo clocks.

2) Less terrain samples

3) Half terrain detail on

4) lower score than the Ryzen "leak"

These in game benches don't show overclocks which makes me question validity to begin with or at least validity to argue. A stock Hexcore Broadwell with similar clocks to Ryzen runs slower with taxing settings toned down.

This is very interesting.
 
had that issue with most. They would only count multi-OCs. I ran fsb so my multi at 20 always gave 4.0ghz even though I had 233fsb for 4.66ghz.
 
o107rt.jpg
[/IMG]
o107rt.jpg
[/IMG]
o107rt.jpg


AMD-Ryzen-3.63.9-GHz-Turbo-Clock.jpg


I am going to address the big pink elephant in the room, the official ES is called engineering sample and the code ends in an N not Y which is for iGPU string, the AOTS is fake, sorry Razor/Shintai.

Myth busted


Ok here lol

low-1080p-zen-1d-1140x555.png


This looks even worse doesn't it? Find some comparisons to Intel chips. This is an older one though..

This was well before (in August) the screenshot taken for the new spin of Zen CPU's that reached 3.6

So why are they reading the same serial? When a spin is done the serial changes usually a number or letter at the end changes.

I wanted to point this out a long time back, but that's how you know what is fake and what was not.

http://wccftech.com/amd-zen-es-benchmarks-leak-out/

Yeah so a chip that we saw in August, where AMD didn't get over 3.2 GHZ or trouble getting over that, now miraculously gets over 3.6 GHZ at stock clock, with no respin, which there was supposed to be a stepping after..... So which serial is wrong or write, cause we have enough info to pick from to figure out Zen's "performance"

Mind you WTF has another stepping serial number too so..... (3rd picture)

and for people that want to see how the comparison is

AMD-Zen-ES-AotS-Benchmarks-635x379.jpg


That is ouch.
 
Last edited:
I7 5930K @ 3.5 GTX1080 half terrain on 12million 62.3 CPU

I7 6800K @ 3.4 GTX1080 half terrain on 12million samples 67.9 CPU

"Ryzen" @3.6 TitanX half terrain off 16million samples 70.5 CPU.

Faster than a Haswell and Broadwell E hexcore with more taxing on the CPU.

How is that ouch?
 
How is that ouch?
Don't embarass yourself comparing apples and oranges. Find 1.50.xxxxx benches if you want to make a point. Hint: 7700k fucking stomps all over that Zen sample if it is real.
Yeah so a chip that we saw in August, where AMD didn't get over 3.2 GHZ or trouble getting over that, now miraculously gets over 3.6 GHZ at stock clock, with no respin, which there was supposed to be a stepping after..... So which serial is wrong or write, cause we have enough info to pick from to figure out Zen's "performance"
Actually, revision did change, from E4 to F3 to F4, so i do not understand you.

Anyways, version issue has been dealt with. Now, the only issue that remains for me: is the eng sample name legitimate? Because WhyCry does report on those names used: https://videocardz.com/65654/amd-ryzen-6-core-cpu-exists

And that is weird, because now we have 2 versions:
1) Someone else has seen it and faked the bench promtly.
2) It is legitimate... and Zen fucking sucks.

I am conflicted.
 
I7 5930K @ 3.5 GTX1080 half terrain on 12million 62.3 CPU

I7 6800K @ 3.4 GTX1080 half terrain on 12million samples 67.9 CPU

"Ryzen" @3.6 TitanX half terrain off 16million samples 70.5 CPU.

Faster than a Haswell and Broadwell E hexcore with more taxing on the CPU.

How is that ouch?


Because the ryzen which we now know is real if we are to believe the serial numbers, only had 4 million to 16 million and is getting crushed by a 4790, do your extrapolations of core frequency on that and see where it comes out to lol. This is why you don't sit around and extrapolate, just doesn't work that well.
 
Last edited:
Don't embarass yourself comparing apples and oranges. Find 1.50.xxxxx benches if you want to make a point. Hint: 7700k fucking stomps all over that Zen sample if it is real.

Actually, revision did change, from E4 to F3 to F4, so i do not understand you.

Anyways, version issue has been dealt with. Now, the only issue that remains for me: is the eng sample name legitimate? Because WhyCry does report on those names used: https://videocardz.com/65654/amd-ryzen-6-core-cpu-exists

And that is weird.
ah missed that good catch,

well so there are Y serials another stepping.
 
Because the ryzen which we now know is real if we are to believe the serial numbers, only had 4 million to 16 million and is getting crushed by a 4790, do you extrapolations of core frequency on that and see where it comes out to lol. This is why you don't sit around and extrapolate, just doesn't work that well.

What are you talking about? The numbers i poseted were official numbers from AOTS database the Ryzen is faster than both Haswell E and Broadwell E with 16 million vs 12 milliion and half terrain detail was turned off ie: beat them with more taxing settings.

http://www.ashesofthesingularity.co...b3c2-c71c916a4ad3/match-details/e92d4f79-77ce

I7 6950x with titan x 77.7 score also running less taxing settings. That is only 9.7% slower with 2 cores and 4 threads less.

I am actually starting to think AMD has a very good CPU here.
 
What are you talking about? The numbers i poseted were official numbers from AOTS database the Ryzen is faster than both Haswell E and Broadwell E with 16 million vs 12 milliion and half terrain detail was turned off ie: beat them with more taxing settings.

http://www.ashesofthesingularity.co...b3c2-c71c916a4ad3/match-details/e92d4f79-77ce

I7 6950x with titan x 77.7 score also running less taxing settings. That is only 9.7% slower with 2 cores and 4 threads less.

I am actually starting to think AMD has a very good CPU here.


OrangeKush, you didn't just start thinking AMD has a good CPU here, you imagined it for months now since your first post on this forum.

Ok didn't go through your numbers.

What are the revisions of the software and go from there.

Then now you already see the Y serial is real, its there.

So this screenshot from AOTS came out before the leaks of the serials did.

You made an assumption, which was incorrect about the serial number, now we have no idea who AMD_Fanboy is, but that is anyone's guess, so can't take that into anything you have stated.
 
OrangeKush, you didn't just start thinking AMD has a good CPU here, you imagined it for months now since your first post on this forum.

Ok didn't go through your numbers.

What are the revisions of the software and go from there.

Then now you already see the Y serial is real, its there.

So this screenshot from AOTS came out before the leaks of the serials did.

You made an assumption, which was incorrect about the serial number, now we have no idea who AMD_Fanboy is, but that is anyone's guess, so can't take that into anything you have stated.

I had more time to go through the bench scores of stock CPUs and i found that Zen may actually equal a 6900k like the CES benches showed.

I don't like the id inaccuracy, but delving into numbers seems that Zen runs with Broadwell in a bench that was not even equal to others. I think maybe 2-3 extra FPS to the CPU score with 12 milliion and half detail enabled.

Looks kind of exciting
 
I had more time to go through the bench scores of stock CPUs and i found that Zen may actually equal a 6900k like the CES benches showed.

I don't like the id inaccuracy, but delving into numbers seems that Zen runs with Broadwell in a bench that was not even equal to others. I think maybe 2-3 extra FPS to the CPU score with 12 milliion and half detail enabled.

Looks kind of exciting


You can't compare across different software versions either....

The scores change quite a bit. Just check the data base, you can sort by software version.

You have two major problems, fly by wire, which I explained before why AOTS isn't that good to use you can get an idea by don't sit here and try to give something too much credence, then you have the software version. Which yeah, you just can't get a good picture of what is going on.

The ones that we do have with equal software versions show a totally different picture. Albeit earlier lower clocked samples.

feEPCz1.jpg


Hmm see what I'm saying, this is a 6700k at 4 ghz, 4k similar settings, much higher frame rates. versions different though, can't really tell ya what is going on.

SOWMk99.jpg


Similar settings 6 core Intel, getting less than the 4 core,
 
Last edited:
You can't compare across different software versions either....

The scores change quite a bit. Just check the data base, you can sort by software version.

You have two major problems, fly by wire, which I explained before why AOTS isn't that good to use you can get an idea by don't sit here and try to give something too much credence, then you have the software version. Which yeah, you just can't get a good picture of what is going on.

The ones that we do have with equal software versions show a totally different picture. Albeit earlier lower clocked samples.

feEPCz1.jpg


Hmm see what I'm saying, this is a 6700k at 4 ghz, 4k similar settings, much higher frame rates. versions different though, can't really tell ya what is going on.

SOWMk99.jpg


Similar settings 6 core Intel, getting less than the 4 core,


1) software cannot be responsible for major difference

2) AOTS likes high clocked quads i7 4790 was well off a i7 4790k despite being the same chip binned higher.

3) AMD are yet to release a high clocked quad to draw inference.

4) mainstream quads beating octocore behemoths in gaming is more common than trolls on WCCFtech, gaming doesn't scale well with low clocked heavy threaded juggernaut SKU's
 
1) software cannot be responsible for major difference

2) AOTS likes high clocked quads i7 4790 was well off a i7 4790k despite being the same chip binned higher.

3) AMD are yet to release a high clocked quad to draw inference.

4) mainstream quads beating octocore behemoths in gaming is more common than trolls on WCCFtech, gaming doesn't scale well with low clocked heavy threaded juggernaut SKU's


Oh software can be responsible, that's why I stated look at the data base or just glance over to the old AOTS thread here, you will see many of us came to the same conclusion, the software version of AOTS make a HUGE difference.

The rest outside of 4 yeah possibly, but again assumption is the mother of all fuck ups.

4 What did I say about IPC and why AMD isn't showing it yet? You just came to the same conclusion......
 
Last edited:
Just because it hasn't been mentioned yet. 16 vs 32 gb of memory. No idea how much of a difference that could actually make. Just wanted to point it out.
 
This is just what I needed. People fighting over Another AOTS piece of shit bench that has results all over the place all the time. Hahahaha. Exciting times.
 
This is just what I needed. People fighting over Another AOTS piece of shit bench that has results all over the place all the time. Hahahaha. Exciting times.
Oh, come on. Having literally no real information on performance was getting a little too boring.
 
I have already expressed my 2 concerns with it:
1. Version and build number just look... wrong. Are there any other benchmarks on this build?
2. AMD Eng Sample, not AMD Ryzen: . That one is obvious red flag. Oh, also, it has _Y instead of _N to indicate lack of iGPU. Name just does not align with AMD's ES naming (and yes, the CES computer with 3.6Ghz base clock did have AMD Eng Sample naming, so AMD does use it).
Dammit man ur argument makes too much sense! Thx for wrecking the hype train on this 1 =(
 
I have already expressed my 2 concerns with it:
1. Version and build number just look... wrong. Are there any other benchmarks on this build?
2. AMD Eng Sample, not AMD Ryzen: . That one is obvious red flag. Oh, also, it has _Y instead of _N to indicate lack of iGPU. Name just does not align with AMD's ES naming (and yes, the CES computer with 3.6Ghz base clock did have AMD Eng Sample naming, so AMD does use it).

The name string is a fused setting and AMD Eng Sample is one of the standard choices for pre-release parts sent to external partners. So, at least this part of it is ok. Still, early sample part, so salt is in order.
 
Perhaps people are so skeptical of Ryzen's performance not because they're AMD haters, but because AMD has a well established track record of over promising and under delivering. Every AMD release in recent memory has not lived up to the marketing/hype/leaks. People are tired of it and tired of being taken for saps. So when someone comes out and says it's so great, there's lots of push-back. You can only cry wolf so many times.
 
Perhaps people are so skeptical of Ryzen's performance not because they're AMD haters, but because AMD has a well established track record of over promising and under delivering. Every AMD release in recent memory has not lived up to the marketing/hype/leaks. People are tired of it and tired of being taken for saps. So when someone comes out and says it's so great, there's lots of push-back. You can only cry wolf so many times.

Maybe it is because they lack certain knowledge on what a chip does and why it performs either well or worse in certain situations. But the cat was out of the bag after they released Bulldozer by then it was known that this chip lacked IPC to get near Intel (and could not get fixed). Hence no new cpu upgrades beside Piledriver variations on AM3+

There are 2 things that I saw that were truly marketing blunders at Bulldozer (over clocking hype) and some AMD people claiming the price of the product was justified because it could keep up in Winrar and 2 other flukes they discovered.

This event took place about 6 years ago , are these people all suicidal from that time or did most of them kill themselves already ?
 
The name string is a fused setting and AMD Eng Sample is one of the standard choices for pre-release parts sent to external partners. So, at least this part of it is ok. Still, early sample part, so salt is in order.
I totally agree. This another phony piece of trash from WCCFTECH. These clowns do not know how to do a credible forgery. So crude, so illogical. So unbelievable.
 
so much variance. Why don't people just wait and shit and see instead of baseless arguments. We see intel CPUs with scores all over, different software version making big differences. It took a while to let go of the last AOTS bench and now people are jumping all over this. This really does look fishy and not apples to apples. Pointless arguments back and forth.
 
Have you guys heard the latest? Intel is rolling out new cpus to counter ryzen launch. More info monday. It is from canard so it should be legit.
 
AOTS is a legit benchmark since it makes Ryzen look bad now. You know what.... this is getting ridiculous.
 
a47ho7.png
[/IMG]
a47ho7.png


AOTS shows the latest game update is 2.3 but the benches run on 2.2 is actually a newer update so it refutes the rubbish about 1.5.

jgqudi.png


i7 6950X Intels flagship CPU running lower stress setting than the "RYZEN" only gets 9.7% higher score on 2C/4T advantage and 32GB RAM which in a real world strategy game with thousands of units is actually very intensive.

1pji8m.png


The initial expectations of Haswell or faster but not quite Broadwell level is stacking up, But Ryzen is again able to run with an Intel Hex core of similar clocks and running less stressing settings.

35dawbp.png


And a Haswell Hexcore.

2mpxt3t.png


Lastly a 4/8 stock locked Devils Canyon Haswell, showing that AOTS is very limited in its scaling beyond 4 cores, though again similar performance to a Haswell E with the same IPC gives indications of genrational limits.

I was just looking up stock clocks as I wanted to see what Ryzen 3.6Ghz can do relative to Intel lower clocked CPU's and the results are rather impressive. AOTS bench is not an ingame bench, but like 3Dmark or Unigine or the old Farcry 2/Res Evil 5 is an ingame build preset stress test without any variables. The game takes a liking to higher clock speed, as soon as there is a higher frequency the scores escalate high, there are examples in the bench list of a 6800K that was overclocked and scored about 100+. Arma III is another game that shows clockspeed as a major performance boost when comparing a 5960X to a 7700K in Arma III it was shown that a 4.5Ghz 5960X is still a potent CPU for that title. I will wait for OC benches on a Ryzen to get a real score but the 7700K and 6700K scores are to high to assume base clocks, even though baseclocks are north of 4Ghz which helps.

As a gaming CPU enough benches are showing that Ryzen's 8 core flagship is extremely competitive with intels big core CPU's 6900/6800/5960/5930 and even the 6950X with some indication of overall CPU performance between Haswell and Broadwell. The results if true are actually very encouraging.
 
a47ho7.png
[/IMG]
a47ho7.png


AOTS shows the latest game update is 2.3 but the benches run on 2.2 is actually a newer update so it refutes the rubbish about 1.5.

jgqudi.png


i7 6950X Intels flagship CPU running lower stress setting than the "RYZEN" only gets 9.7% higher score on 2C/4T advantage and 32GB RAM which in a real world strategy game with thousands of units is actually very intensive.

1pji8m.png


The initial expectations of Haswell or faster but not quite Broadwell level is stacking up, But Ryzen is again able to run with an Intel Hex core of similar clocks and running less stressing settings.

35dawbp.png


And a Haswell Hexcore.

2mpxt3t.png


Lastly a 4/8 stock locked Devils Canyon Haswell, showing that AOTS is very limited in its scaling beyond 4 cores, though again similar performance to a Haswell E with the same IPC gives indications of genrational limits.

I was just looking up stock clocks as I wanted to see what Ryzen 3.6Ghz can do relative to Intel lower clocked CPU's and the results are rather impressive. AOTS bench is not an ingame bench, but like 3Dmark or Unigine or the old Farcry 2/Res Evil 5 is an ingame build preset stress test without any variables. The game takes a liking to higher clock speed, as soon as there is a higher frequency the scores escalate high, there are examples in the bench list of a 6800K that was overclocked and scored about 100+. Arma III is another game that shows clockspeed as a major performance boost when comparing a 5960X to a 7700K in Arma III it was shown that a 4.5Ghz 5960X is still a potent CPU for that title. I will wait for OC benches on a Ryzen to get a real score but the 7700K and 6700K scores are to high to assume base clocks, even though baseclocks are north of 4Ghz which helps.

As a gaming CPU enough benches are showing that Ryzen's 8 core flagship is extremely competitive with intels big core CPU's 6900/6800/5960/5930 and even the 6950X with some indication of overall CPU performance between Haswell and Broadwell. The results if true are actually very encouraging.

Don't slam your head against the wall. Hate for AMD goes a long way by some in this thread that they can't see past that lol.
 
Back
Top