AMD Geekbenchs for engineering examples???

[TechGod]

Some new (somehow strange?) AM4 benchmark results on Geekbench
http://reddit.com/r/Amd/comments/4eaxcc/some_new_somehow_strange_am4_benchmark_results_on/

 

Score of 2500 single core and 8100 multi core for 32 bit Windows and a 2Ghz processor looks very nice. 

[DXMember]

that reddit thread pretty much ripped everything apart, there's nothing too exciting, it's still an excavator core

[DocSwag]

6 minutes ago, DXMember said:

that reddit thread pretty much ripped everything apart, there's nothing too exciting, it's still an excavator core

You sure about that? I don't know of any Excavator Core capable of 2500 single core on 2 ghz...

[The Benjamins]

If you scale that yo 6 cores the score would be 24000, and a 5820K (6 core) is 24000, but clocked at 4.1 vs 2.0

 

if you scale single core preformance from 2.0 to 4.1 you get a score of 5330.

 

But things dont scale like that but it still looks really good.

 

If you scale GHz to 4.1 and to 6 cores with a 80% efficiency factor it adds up to a 31700 multicore score. if its 70% efficient at scaling in GHz and core then the score is 24300.

[DXMember]

2 minutes ago, DocSwag said:

You sure about that? I don't know of any Excavator Core capable of 2500 single core on 2 ghz...

the reddit boiz came to a conclusion it's a Bristol Ridge based on cache sizes, they seem pretty confident in what they say

[DocSwag]

6 minutes ago, The Benjamins said:

If you scale that yo 6 cores the score would be 24000, and a 5820K (6 core) is 24000, but clocked at 4.1 vs 2.0

 

if you scale single core preformance from 2.0 to 4.1 you get a score of 5330.

 

But things dont scale like that but it still looks really good.

 

If you scale GHz to 4.1 and to 6 cores with a 80% efficacy factor it adds up to a 31700 multicore score.

Well cores don't exactly scale that well...

Single core scales better. If we put it at 4ghz that's 5000, or a little more than a 6700k.... Sounds good.

 

EDIT: Just realized these scores are on 32 bit, and most of the time 64 bit offers 20% ish more performance than 32 bit.... If we factor that in the single core at 4ghz is around 6200 ish.... holy crap.

[DocSwag]

Just now, DXMember said:

the reddit boiz came to a conclusion it's a Bristol Ridge based on cache sizes, they seem pretty confident in what they say

It doesn't make sense for Bristol Ridge to be that powerful though, since Carrizo isn't capable of anywhere near that and Bristol Ridge is only a little better than Carrizo.

Of course, now that you bring it up it is kinda suspicious... L2 cache is 2 MB which is basically Carrizo. It could very well be a mis reading if you ask me.

But that seems kinda unlikely too, so I'm not sure what to make of this. Of course, we don't know what AMD is planning for Zen in terms of cache.

[TheRandomness]

6 minutes ago, DocSwag said:

Well cores don't exactly scale that well...

Single core scales better. If we put it at 4ghz that's 5000, or a little more than a 6700k.... Sounds good.

 

EDIT: Just realized these scores are on 32 bit, and most of the time 64 bit offers 20% ish more performance than 32 bit.... If we factor that in the single core at 4ghz is around 6200 ish.... holy crap.

So this is what AM4 is showing in ES. AMD, please make a comeback ;~;

[DXMember]

1 minute ago, DocSwag said:

It doesn't make sense for Bristol Ridge to be that powerful though, since Carrizo isn't capable of anywhere near that and Bristol Ridge is only a little better than Carrizo.

Of course, now that you bring it up it is kinda suspicious... L2 cache is 2 MB which is basically Carrizo. It could very well be a mis reading if you ask me.

But that seems kinda unlikely too, so I'm not sure what to make of this. Of course, we don't know what AMD is planning for Zen in terms of cache.

not familiar with geekbech scores, when I read the post wasn't aware those scores where being crazy

are those scores crazy?

if it is Zen engineering sample then i don't think there's anything stopping AMD from purposefully misleading the readouts of both cache sizes and model/family numbers to disguise it

[TechGod]

10 minutes ago, DocSwag said:

Well cores don't exactly scale that well...

Single core scales better. If we put it at 4ghz that's 5000, or a little more than a 6700k.... Sounds good.

 

EDIT: Just realized these scores are on 32 bit, and most of the time 64 bit offers 20% ish more performance than 32 bit.... If we factor that in the single core at 4ghz is around 6200 ish.... holy crap.

That's the thing that impressed me! They were 32 bit scores!

[The Benjamins]

The Performance per GH/z looks like its a bit better then Haswell looks awesome to me.

 

http://browser.primatelabs.com/geekbench3/compare/6037057?baseline=6032076

[DocSwag]

This is a little of what I see:

I pulled up an fx 8800p thing to look, and one thing very suspicious is that the caches are the exact same.

I also looked at memory scores. This mysterious cpu consistently pulls out higher memory scores, leading me to believe it is probably ddr4 we are looking at. That narrows it to Bristol Ridge and Zen.

 

I looked at it compared to the fx 8800p in a few benchmarks, and the graphs are remarkably similar. Another thing I noticed is that the fx 8800p reports itself as 1.7 ghz sometimes with a 2000 single core score. We, however, know it isn't achieving that score at 1.7 ghz but with a turbo around 3 ghz. So this CPU may not be achieving 2500 at 2 ghz-it could be higher via turbos.

 

Due to the remarkably similar cache, my conclusion is, as @DXMember said, that this is probably Bristol Ridge we are looking at. It has more memory bandwidth, a sign of DDR4, but the score patterns are very similar to what Carrizo displays. Also, I forgot to mention that it's FPU score is quite a bit lower than it's integer, much like a bulldozer based CPU. As well, it's cache is super similar to Carrizo. The high single core score is probably from a turbo going up to around 3ghz ish.

 

Therefore, unfortunately, we probably aren't looking at Zen but Bristol Ridge. Still cool though, in my opinion.

[DocSwag]

NEW OBSERVATION

 

The family number it reports is 21. 21 is for Bulldozer. Zen is 23. Story over. It's not Zen. Probably Bristol Ridge

[DocSwag]

PARTY'S OVER GUYS SORRY IT AINT ZEN

It's probably Bristol Ridge.

[TechGod]

6 minutes ago, DocSwag said:

NEW OBSERVATION

 

The family number it reports is 21. 21 is for Bulldozer. Zen is 23. Story over. It's not Zen. Probably Bristol Ridge

If this new CPU is Bristol Ridge and still CMT AND still has this performance, I can't wait to see what a zen chip looks like...

[DocSwag]

18 minutes ago, TechGod said:

If this new CPU is Bristol Ridge and still CMT AND still has this performance, I can't wait to see what a zen chip looks like...

Still though, remember that the single core is still probably achieved with 3ghz on turbo. Kinda cool though that AMD has got a mobile chip to get a single core of 2500. I wanna see Bristol Ridge on desktop.

[Dabombinable]

32 minutes ago, DocSwag said:

NEW OBSERVATION

 

The family number it reports is 21. 21 is for Bulldozer. Zen is 23. Story over. It's not Zen. Probably Bristol Ridge

So AMD is still clinging to their poor CMT design? *sigh*

[DocSwag]

13 minutes ago, Dabombinable said:

So AMD is still clinging to their poor CMT design? *sigh*

No.

We knew Bristol Ridge would be CMT because it is Excavator. It is meant to be because it is like a prelude to Zen. It'll be the first CPUs to support AM4 and Zen will come later. Zen won't be CMT. We are pretty sure of that. Bristol Ridge is meant to be, in my opinion:

1. Something builders can buy from AMD for desktop that's better than stuff right now (so an upgrade to Kaveri).
2. Something builders can get now and upgrade to Zen with later.
3. Something to build hype for Zen

[Dabombinable]

5 minutes ago, DocSwag said:

No.

We knew Bristol Ridge would be CMT because it is Excavator. It is meant to be because it is like a prelude to Zen. It'll be the first CPUs to support AM4 and Zen will come later. Zen won't be CMT. We are pretty sure of that. Bristol Ridge is meant to be, in my opinion:

1. Something builders can buy from AMD for desktop that's better than stuff right now (so an upgrade to Kaveri).
2. Something builders can get now and upgrade to Zen with later.
3. Something to build hype for Zen

Still, they are clinging to CMT, and if they only use it their AM4 APU.....and don't make Zen APU....

[BuckGup]

1 hour ago, The Benjamins said:

If you scale that yo 6 cores the score would be 24000, and a 5820K (6 core) is 24000, but clocked at 4.1 vs 2.0

 

if you scale single core preformance from 2.0 to 4.1 you get a score of 5330.

 

But things dont scale like that but it still looks really good.

 

If you scale GHz to 4.1 and to 6 cores with a 80% efficiency factor it adds up to a 31700 multicore score. if its 70% efficient at scaling in GHz and core then the score is 24300.

Look at the iphone vs the tegra note. It is an octal core yet scores way less in benchmarks then the dual core iphone. Cores don't always scale perfectly and knowing AMD there might be trickery 

[TechGod]

37 minutes ago, Dabombinable said:

Still, they are clinging to CMT, and if they only use it their AM4 APU.....and don't make Zen APU....

You. Are. Unreal. 

[Doobeedoo]

Well, mixed feeling about this. Though, it will be quite exciting to see desktop Zen for sure. If IPC performance is truly amazing I'll be all over that 8-core

[shdowhunt60]

1 hour ago, Dabombinable said:

Still, they are clinging to CMT, and if they only use it their AM4 APU.....and don't make Zen APU....

I hope you're trolling. Because you cannot be implying that they're not making an Zen APU.

[Dabombinable]

8 minutes ago, shdowhunt60 said:

I hope you're trolling. Because you cannot be implying that they're not making an Zen APU.

My point was that if they reserve Bristol Ridge for APU (quite possible) and only use Zen for CPU, then I will not be buying a laptop with one of their APU. I can not support retarded decisions.

[shdowhunt60]

1 minute ago, Dabombinable said:

My point was that if they reserve Bristol Ridge for APU (quite possible) and only use Zen for CPU, then I will not be buying a laptop with one of their APU. I can not support retarded decisions.

Retarded because they're not giving you what you want immediately and right now? That's called entitlement. Which I HATE using that word, because it's so overused, but that's that this is by definition.

 

AMD cannot produce Zen based APU's. For one, the Zen core is STILL not ready despite them putting every other CPU project on hold to get Zen out the door as fast as possible (with evidence to suggest that this has even come to the detriment of Steamroller and Excavator's developement, taking resources from THOSE projects leaving them rather incomplete). And secondly, the process that Zen requires for production JUST started ramping up a couple months ago, and AMD is pooling all its resources now for Polaris.

 

The point is, Excavator is all that AMD has. They're not retarded for trying to sell it. That's what sane businesses do when they want to stay afloat and make it to the next product. That doesn't make them any more retarded as it makes you any less of a entitled manchild.