AMD Ryzen benchmarks against 7700K and 6900K

[vanished]

20 minutes ago, samcool55 said:

These are fake, the AMD results are actually from an E5 2660.

Original source was from the Baidu forum but got removed very quickly.

So we don't know anything  

https://imgur.com/NhJRN8g

If that is true (which honestly we don't know... it's about as provable as the OP itself) then I'd just like to say a huge I told you so so everyone who jumped right to believing some random screenshot with literally nothing to back it up  

[Deli]

Does it qualify for "fake news" ?

[Space Reptile]

5 minutes ago, Ryan_Vickers said:

If that is true (which honestly we don't know... it's about as provable as the OP itself) then I'd just like to say a huge I told you so so everyone who jumped right to believing some random screenshot with literally nothing to back it up  

DARN IT , patrick was right , IT WAS A DELIDDED SANDY XEON 

 

[samcool55]

5 minutes ago, Ryan_Vickers said:

If that is true (which honestly we don't know... it's about as provable as the OP itself) then I'd just like to say a huge I told you so so everyone who jumped right to believing some random screenshot with literally nothing to back it up  

It's very likely because the results actually match an E5 2660, keeping a margin of error in mind

The E5 2660 an 8C/16T cpu that gets around 1k in cinebench R15 so maybe a cherry picked benchmark number and forcing all cores to max boost (3Ghz) could have been that 1.1k+ result

 

https://hwbot.org/benchmark/cinebench_-_r15/rankings?hardwareTypeId=processor_2840&cores=8#start=0#interval=20

 

Atm every piece of information tells us it's fake. The E5 2660 stuff, weird results, very dodgy source.

Enough reasons to ignore them and move on

[leadeater]

46 minutes ago, samcool55 said:

These are fake, the AMD results are actually from an E5 2660.

Original source was from the Baidu forum but got removed very quickly.

So we don't know anything  

https://imgur.com/NhJRN8g

 

26 minutes ago, Ryan_Vickers said:

If that is true (which honestly we don't know... it's about as provable as the OP itself) then I'd just like to say a huge I told you so so everyone who jumped right to believing some random screenshot with literally nothing to back it up  

It's not an E5 2660 or any variant/generation of the SKU. The E5 2660v2 scores 916, the E5 2660v3 scores 1593 and the E5 2660v4 scores 1774.

 

Dual E5 2660 maybe but highly unlikely, as that at a guess would be 800-900 per CPU, and it's not a E5 2660v2 BLK OC either.

 

Edit:

However I do agree with you about ignoring this information, even if it is actually a Zen it's not a finalized retail sample so meaningless.

[i_build_nanosuits]

17 hours ago, cozz said:

They've been leaked, we can finally see how well Ryzen will perform against it's opponents. 

 

 

 

The benchmarks are done in Cinebench R15 with Ryzen getting a total score of 1188 beating the 7700K which sits at 966

 

Cinebench R15 results:  

AMD Ryzen - 1188

Core i7-7700K - 966

Core i7-7700K @ 5GHz - 1083

Core i7-6900K - 1500

Core i7-6950X - 1800

 

 

 

 

 

source; TweakTown
 

wooohh woohhh...if that is real then it means Zen is GARBAGE...8 core 16 thread barely beat the intel quad-core that would mean they are still -50% in terms of per core performance which would point to either low clockspeed and/or low IPC...let's hope these are not real otherwise AMD is done with.

[samcool55]

14 minutes ago, leadeater said:

 

It's not an E5 2660 or any variant/generation of the SKU. The E5 2660v2 scores 916, the E5 2660v3 scores 1593 and the E5 2660v4 scores 1774.

 

Dual E5 2660 maybe but highly unlikely, as that at a guess would be 800-900 per CPU, and it's not a E5 2660v2 BLK OC either.

 

Edit:

However I do agree with you about ignoring this information, even if it is actually a Zen it's not a finalized retail sample so meaningless.

From where do you get those numbers?

 

https://hwbot.org/benchmark/cinebench_-_r15/rankings?hardwareTypeId=processor_2840&cores=8#start=0#interval=20

1010 for the E5 2660@2.9Ghz. With a bit of tweaking and cherry picking the result i can see it reaching the 1.1k+ 

[leadeater]

7 minutes ago, samcool55 said:

From where do you get those numbers?

 

https://hwbot.org/benchmark/cinebench_-_r15/rankings?hardwareTypeId=processor_2840&cores=8#start=0#interval=20

1010 for the E5 2660@2.9Ghz. With a bit of tweaking and cherry picking the result i can see it reaching the 1.1k+ 

From hwbot, there is no amount of cherry picking that will get an E5 2660 to a score of 1188 considering the world record reported by them is 1010. Even a E5 2660v2 OC would struggle to get 1188, Xeons just don't OC that well.

 

For what ever reason hwbot's E5 2660v2 single socket score is wrong but the dual socket score is 2182 meaning 1091 for single socket. Again reported as world record.

 

Edit:

An E5 2670 gets 1127 (WR)

[vanished]

Lets also all remember that they could have just grabbed a screenshot of Cinebench, photoshopped it and then taken a picture of the screen in 5 minutes as I did  

[Space Reptile]

5 minutes ago, Ryan_Vickers said:

Lets also all remember that they could have just grabbed a screenshot of Cinebench, photoshopped it and then taken a picture of the screen in 5 minutes as I did  

the fact that nobody noticed that the Chinese lettering is also at an angle suggesting it being on the same layer as the cb result ...

 

 

[samcool55]

1 minute ago, leadeater said:

From hwbot, there is no amount of cherry picking that will get an E5 2660 to a score of 1188 considering the world record reported by them is 1010. Even a E5 2660v2 OC would struggle to get 1188, Xeons just don't OC that well.

 

For what ever reason hwbot's E5 2660v2 single socket score is wrong but the dual socket score is 2182 meaning 1091 for single socket. Again reported as world records.

Are you sure? The V4 doesn't even have a score and the score for the V2 is 163...

https://hwbot.org/benchmark/cinebench_-_r15/rankings?hardwareTypeId=processor_2978&cores=10#start=0#interval=20

 

I know Xeon's don't OC well but i don't think the results on HWbot are world record worthy, it's easy to be first if there are only 1-2 results and both of them don't show any OC effort at all, i mean the E5 2660 can boost to 3Ghz, even an agressive boost by forcing all cores to boost frequency (3Ghz) would get you probably to number 1 because the current #1 ran the benchmark at 2.9Ghz and that's before any real OC is being done to it.

[Master Disaster]

3 minutes ago, samcool55 said:

Are you sure? The V4 doesn't even have a score and the score for the V2 is 163...

https://hwbot.org/benchmark/cinebench_-_r15/rankings?hardwareTypeId=processor_2978&cores=10#start=0#interval=20

 

I know Xeon's don't OC well but i don't think the results on HWbot are world record worthy, it's easy to be first if there are only 1-2 results and both of them don't show any OC effort at all, i mean the E5 2660 can boost to 3Ghz, even an agressive boost by forcing all cores to boost frequency (3Ghz) would get you probably to number 1 because the current #1 ran the benchmark at 2.9Ghz and that's before any real OC is being done to it.

WR is the highest recorded score, the fact you think its low doesn't change what it is.

[samcool55]

1 minute ago, Master Disaster said:

WR is the highest recorded score, the fact you think its low doesn't change what it is.

True, the only way to settle this is to try it ourselves and see what we end up with, but that's unlikely to happen

Unless someone with an E5 2660 is prepared to do a few hours of OC'ing but that's again, unlikely...

[Guest Kloaked]

17 hours ago, RKRiley said:

When it cuts off the cpu name in the picture, i find it quite hard to trust.

I find it hard to trust when TweakTown, WCCFTech and so on try to make these rumors look true.

[leadeater]

5 minutes ago, samcool55 said:

Are you sure? The V4 doesn't even have a score and the score for the V2 is 163...

https://hwbot.org/benchmark/cinebench_-_r15/rankings?hardwareTypeId=processor_2978&cores=10#start=0#interval=20

 

I know Xeon's don't OC well but i don't think the results on HWbot are world record worthy, it's easy to be first if there are only 1-2 results and both of them don't show any OC effort at all, i mean the E5 2660 can boost to 3Ghz, even an agressive boost by forcing all cores to boost frequency (3Ghz) would get you probably to number 1 because the current #1 ran the benchmark at 2.9Ghz and that's before any real OC is being done to it.

See my edit, check the 2670 scores. That's on a CPU that boosts to 3.3Ghz, the real problem is server motherboards just have zero OC options and Xeons have locked multipliers. So even if you chucked one in a gaming board you're limited to BLK OC which is terrible and unstable.

[Master Disaster]

1 minute ago, samcool55 said:

True, the only way to settle this is to try it ourselves and see what we end up with, but that's unlikely to happen

Unless someone with an E5 2660 is prepared to do a few hours of OC'ing but that's again, unlikely...

I'm sure Linus has a few of them, I mean he's got every other Intel server CPU on the planet

[samcool55]

Just now, Master Disaster said:

I'm sure Linus has a few of them, I mean he's got every other Intel server CPU on the planet

Probably, Luuke we need your help!

Or the other dude that does benchmarks, if he is still there at least (i'm not exactly up-to-date anymore about every LMG employee).

[Master Disaster]

2 minutes ago, leadeater said:

See my edit, check the 2670 scores. That's on a CPU that boosts to 3.3Ghz, the real problem is server motherboards just have zero OC options and Xeons have locked multipliers. So even if you chucked one in a gaming board you're limited to BLK OC which is terrible and unstable.

I was just thinking exactly that, a Z170 would bclk a Xeon but bclk is a bitch to get right.

 

I wonder if you could use the modded BIOS released a while back for H150 boards to OC a Xeon further than you'd normally be able to?

[leadeater]

6 minutes ago, Master Disaster said:

I was just thinking exactly that, a Z170 would bclk a Xeon but bclk is a bitch to get right.

 

I wonder if you could use the modded BIOS released a while back for H150 boards to OC a Xeon further than you'd normally be able to?

No idea, just brings back horrible memories of OC'ing my AMD Athlons and Pentium 4s. BLK frequencies, FSB ratios, memory speed issues.. ahhh NOOOOOOOooooo shoot me now. All just to play Rome: Total War with large unit counts and not get 10 FPS when flaming archers fire .

 

Edit:

Also @samcool55 sorry I did miss a source I used, http://www.cpu-monkey.com/en/cpu_benchmark-cinebench_r15_multi_core-8

[DXMember]

Saw that Cinebench screenshot earlier this week, my first thought was - could it be Dual Channel Vs. Quad Channel memory, or perhaps looser memory timings?

Because Cinebench easily benefits from faster memory sub-system

[Master Disaster]

6 minutes ago, leadeater said:

No idea, just brings back horrible memories of OC'ing my AMD Athlons and Pentium 4s. BLK frequencies, FSB ratios, memory speed issues.. ahhh NOOOOOOOooooo shoot me now. All just to play Rome: Total War with large unit counts and not get 10 FPS when flaming archers fire .

Don't remind me, I spent days getting my Q9650 to 3.5 stable. It was a nightmare back then.

 

Probably didn't help I only had a cheapo cooler at the time. £350 CPU £25 cooler

[porina]

4 minutes ago, DXMember said:

Saw that Cinebench screenshot earlier this week, my first thought was - could it be Dual Channel Vs. Quad Channel memory, or perhaps looser memory timings?

Because Cinebench easily benefits from faster memory sub-system

For competitive benchmarking maybe, but in the big picture I don't see it as memory sensitive. That or I've not had enough CPU to stress it enough to make it noticeable. I mean more than low single digit %.

 

From my previous notes testing with 6700k fixed at 4.2 GHz with HT on: going from 3000 dual channel to 2133 single channel (a factor of 2.8x) only made 1.5% difference. I didn't investigate timings.

[laminutederire]

10 hours ago, patrickjp93 said:

Or, since blender is OpenSource, AMD chose an image that played nicely with its L2 cache size but not Intel's, or, even better, MODIFIED SPECIAL SOURCE CODE FOR THE RYZEN TEST!

 

AMD faked the tests. Are you really so surprised?

They're always better. They're mine.

And after saying nonsense like that you want to give lessons to everyone and feel superior?

That's pathetic, stop trolling everyone, that's annoying. You know perfectly well NO ONE can prove you're wrong since the CPU hasn't been reviewed and no 3rd party benchmark which are remotely legitimate are found. Heck we barely know the prices of those things. We don't know anything, and you don't know either so stop lecturing.

Be consistent if you want to be taken seriously. You can give sources on how good intel is, we don't care since we gave nothing yet to compare it to besides biased information, since every company in the world is biased about its products and we all know it. I am sorry but that annoys me, since you repeat so often you're better than everyone, act like it and stop trolling. 

[Prysin]

1 hour ago, DXMember said:

Saw that Cinebench screenshot earlier this week, my first thought was - could it be Dual Channel Vs. Quad Channel memory, or perhaps looser memory timings?

Because Cinebench easily benefits from faster memory sub-system

from my testing with a 4790k and FX 8320. Cinebench doesnt get any notable boost going from 1600Mhz Dual channel to 2400MHz dual channel. Nor does it see any notable benefit going between 2133 CL 9 or 2133 CL 11.

[TechGod]

3 hours ago, i_build_nanosuits said:

wooohh woohhh...if that is real then it means Zen is GARBAGE...8 core 16 thread barely beat the intel quad-core that would mean they are still -50% in terms of per core performance which would point to either low clockspeed and/or low IPC...let's hope these are not real otherwise AMD is done with.

Thankfully these benchmarks aren't accurate. See above. 

 

I mean I won't be buying it cause I got a sweet rig with an Intel i5 6500 in so I'm very happy.