What's AMD doing on sep. 12th?

[BlueBlazer]

JF-AMD has said...

JF-AMD's word, saying....

This is getting tiresome....

 

[Vesku]

I like Lau808's way of representing the math:

Given a generous 25% IPC increase in BD core from Phenom II

Phenom II - 10
2 x Phenom II - 20

BD Core - 12.5
2 x BD Core - 25
BD Core as module - 12.5 * 1.8 - 22.5

So a 25% improvement in core results in a 12.5% gain of a module over 2 Phenom II cores.

Given that, a 4.2GHz turbo-ing 8150 = 4.2*1.125 = 4.725GHz Phenom II. In benchmarks that should get them on par with the 2500-2600K in low threaded apps and see them ahead in highly threaded apps.

That's making the assumption of same general performance of Phenom II core boosted by 25%. Seems the reality will be mixed but if leaked pricing is accurate AMD at least seems to think they've achieved ~90% of the above I listed for the 8150. Unless they are really looking to surprise everyone and they achieved ~110% of above. ;D

 

[Avalon]

Athlon II and Phenom II: K10.5. Llano: revised K10.5. Llano's version of K10.5 has 3-4% higher IPC than the one in the Athlon II and Phenom II. The Phenom II only has higher performance than the Athlon II in gaming and file compression, where the L3 cache comes into play.

Buh? Cache counts in a lot of places...

Athlon II v Phenom II
http://www.anandtech.com/bench/Product/105?vs=81

Phenom II beating Athlon II in the majority of tasks presented by 3-11%. Do you mean to say Llano has 3-4% higher IPC than Athlon II or Phenom II?

 

[lau808]

@russian how did u get 25% ipc increase being making 1 bd module = 2 phII cores? im just interested in the math as i estimate 10% being bringing bd just below phII.
my math (not nessesarily correct) would be
phII 2 cores = (10+10=20)
1 bd module(w/25%ipc increase) =(12.5*1.8= 22.5) or so 12.5% increase after module penalt

 

[LOL_Wut_Axel]

Buh? Cache counts in a lot of places...

Athlon II v Phenom II
http://www.anandtech.com/bench/Product/105?vs=81

Phenom II beating Athlon II in the majority of tasks presented by 3-11%. Do you mean to say Llano has 3-4% higher IPC than Athlon II or Phenom II?

I'm wondering if you observed what you linked. For one, Sysmark is a synthetic benchmark and therefore completely irrelevant. That leaves us with gaming and file compression, the two exact things in which you linked the Phenom II has an advantage of 5-10%. Even then, that doesn't mean higher IPC. Llano has 2x the L2 cache that the Athlon II X4 has. Given that, almost any advantage from the L3 cache in those limited applications is gone. Not only that, but IPC like I said is 3-4% higher. Overall it's around 2-3% faster than a Phenom II, but it loses a tiny bit in gaming and file compression due to the lack of L2 cache.

 

[GaiaHunter]

Bingo.
1 Module BD with 25% increased IPC over Phenom II @ 80% module penalty vs. 2 core design = 1.25x IPC * 0.8 penalty = 1.0x.

So HT adds like 20% more, if that, so the penalty for a 2c/4t is 60% of a true quad core.

That means in 4 threads an i3 2100 is 1xIPC*0.6=0.6x an i5 2500K or 1.5xIPC*0.6=0.9x phenom II.





What the hell? It seems an extra thread without module/HT penalty doesn't add 100% more performance sometimes.

And I mean, the concern about BD is mainly for lower thread count workloads right? It seems that the 60% penalty 2c/4t CPU compared to a true quad isn't as big as the math would imply, at least in these 2 cases.

And by the way what is the penalty of running 8 threads in a quad core?

 

[exar333]

Hardly any faster, if you say so pal. Aside from any review site that uses the most retarded ccc settings i've ever seen, I do know that many of my games had a 20-30 fps jump from 5870 to 6970, but if you feel like spreading crap, have at it. And without exception I could run everything with full detail, with far less loss of performance.

Lets keep the bullcrap going, tell us about how 7970 is going to be a disappointment and will barely outperform 5870...i'm expecting it. We have tons of threads with speculative bullcrap about the bulldozer, minuswell not stop there. Keep going. We need more intel fanboys and nvidia fanboys spreading garbage -- if that makes you feel better about your dated nvidia hardware.

Shens. Thats a rediculous claim, unless you game at 800x600...

 

[Terzo]

Just to be clear, the big announcement was video card related, not cpu related, correct? Or am I missing something?

 

[Vesku]

GaiaHunter, please post screencaps of things that utilize a full 4 cores to not sabotage the guesstimation.

Games are now just starting to find ways to use 3-4 threads. It seems your 0.9 estimation is pretty good for fully 4 threaded applications.

 

[Avalon]

I'm wondering if you observed what you linked. For one, Sysmark is a synthetic benchmark and therefore completely irrelevant. That leaves us with gaming and file compression, the two exact things in which you linked the Phenom II has an advantage of 5-10%. Even then, that doesn't mean higher IPC. Llano has 2x the L2 cache that the Athlon II X4 has. Given that, almost any advantage from the L3 cache in those limited applications is gone. Not only that, but IPC like I said is 3-4% higher. Overall it's around 2-3% faster than a Phenom II, but it loses a tiny bit in gaming and file compression due to the lack of L2 cache.

I merely asked you to clarify if you meant 3-4% higher than Athlon II or Phenom II, since they have different IPC due to cache. All I was looking for was this...

"Overall it's around 2-3% faster than a Phenom II."

Of course I read the bench. There were rendering benches in there as well as other items which showed a difference. On top of that, the benches are rather limited in scope. I don't see everything fitting into gaming, file compression, or synthetics, which is why I pointed it out and then asked you. I'm not trying to be rude here, and if I came across that way, I apologize.

 

[GaiaHunter]

GaiaHunter, please post screencaps of things that utilize a full 4 cores to not sabotage the guesstimation.

Games are now just starting to find ways to use 3-4 threads. It seems your 0.9 estimation is pretty good for fully 4 threaded applications.

Are you saying that Dragon Age origins isn't one of the games that is best optimized for 4 cores?

Shall we try with the i5 2500K?



Apparently the process that is so great at determining the HT penalty vs phenom II is crap to compare it vs the i5 2500K.

 

[Vesku]

2 cores * 1.2 HT = 2.4
2.4/4 = 60%
Meaning 2500K is 1 and 2/3rds faster at 4 threaded things than 2100 before multiplying in the extra L2 and L3 cache + 200MHz.

Seems the math still works out. Example, 2500K is a roughly 1 and 3/4s faster (which the math indicates the extra over the 1 and 2/3rd estimate is mostly due to the extra 200MHz) in 3dsmax r9 -SPECapc - Underwater.

Note: In addition to not being the best at fully utilizing 4 threads, Intel has made larger gains over Core 2 and Phenom II in gaming performance than many other areas .

 

[LOL_Wut_Axel]

I merely asked you to clarify if you meant 3-4% higher than Athlon II or Phenom II, since they have different IPC due to cache. All I was looking for was this...

"Overall it's around 2-3% faster than a Phenom II."

Of course I read the bench. There were rendering benches in there as well as other items which showed a difference. On top of that, the benches are rather limited in scope. I don't see everything fitting into gaming, file compression, or synthetics, which is why I pointed it out and then asked. I'm not trying to be rude here.

Alright. What I'm getting to: it has 4-5% higher IPC than Athlon II and Phenom II, but loses 1-2% or so to the Phenom II in some applications (the ones I mentioned) due to the lack of L3 cache. That doesn't mean IPC changes from either, mind you. Cache isn't something that's usually taken into account when talking IPC.

 

[PreferLinux]

So HT adds like 20% more, if that, so the penalty for a 2c/4t is 60% of a true quad core.

That means in 4 threads an i3 2100 is 1xIPC*0.6=0.6x an i5 2500K or 1.5xIPC*0.6=0.9x phenom II.


What the hell? It seems an extra thread without module/HT penalty doesn't add 100% more performance sometimes.

And I mean, the concern about BD is mainly for lower thread count workloads right? It seems that the 60% penalty 2c/4t CPU compared to a true quad isn't as big as the math would imply, at least in these 2 cases.

And by the way what is the penalty of running 8 threads in a quad core?

Only a very minimal penalty for the overhead of changing between threads. So generally doesn't matter.

 

[toyota]

so what the heck was the announcement???

 

[Silverforce11]

Just AMD showing off 28nm mobile gpus running dirt 3 on high settings fluidly outputting to a 1080p monitor.

No other details.

 

[BlueBlazer]

Just AMD showing off 28nm mobile gpus running dirt 3 on high settings fluidly outputting to a 1080p monitor.

This >> AMD demonstrates 28nm GPU.

 

[nanaki333]

Just to be clear, the big announcement was video card related, not cpu related, correct? Or am I missing something?

that and the world record overclock maybe. feels like a total letdown.

 

[Idontcare]

that and the world record overclock maybe. feels like a total letdown.

Yeah, my thoughts exactly.

None of it had anything to do with Fusion, which we've all been repeatedly assured is the future, and instead it had to do with the results of an OC'ing session that happened some 2 weeks ago and a demo of a laptop running a next-gen mobile graphics discreet card.

Whoopdee-effing-do

Intel tells the media to get ready for something big and they lay 3D xtor technology on us, AMD tells the media to get ready for something big and we get :\

 

[podspi]

Demo seems underwhelming, but to be fair, GCN is AMD's first GPU architecture that is really focused on GPU compute. So it was Fusion related, just... You couldn't really see that through the game. And wow are they pushing dirt 3...

 

[evident]

so is this cpu ever gonna be released?

 

[LOL_Wut_Axel]

so is this cpu ever gonna be released?

I'd say it's definitely near. Probably shipping for revenue by the end of this month or the beginning of October.

 

[nanaki333]

I'd say it's definitely near. Probably shipping by the end of this month or the beginning of October.

of THIS year?

 

[LOL_Wut_Axel]

of THIS year?

Yes. :sneaky:

 

[Vesku]

Yeah, my thoughts exactly.

None of it had anything to do with Fusion, which we've all been repeatedly assured is the future, and instead it had to do with the results of an OC'ing session that happened some 2 weeks ago and a demo of a laptop running a next-gen mobile graphics discreet card.

Whoopdee-effing-do

Intel tells the media to get ready for something big and they lay 3D xtor technology on us, AMD tells the media to get ready for something big and we get :\

Well if that was the low-mid priced discrete notebook gpu then it's a bit more interesting. But they didn't tell us that so it remains a meh attempt to steal some of Intel's thunder. Anyone know Intel's track record on forward looking info at IDF, the power info they gave for their 2013 mobile chips seems very rosy. They over-hyped the Atom line of cpus when it was in development and just being released but I don't know if the hype was started at an IDF session.