• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Bulldozer Review! Legit?

Page 9 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
frostedflakes said:
Think my X4 955 does better in Fritz at stock. :/

edit: Get 15.03x (7213 kn/s) at 3.2GHz
Click to expand...

My PIIx4 940BE @ 3.4ghz gets exactly the Cinebench render score that the FX-4110 gets. WTF? How do you make a chip that is different and yet performs the same as your previous gen?
 
Despoiler said:
My PIIx4 940BE @ 3.4ghz gets exactly the Cinebench render score that the FX-4110 gets. WTF? How do you make a chip that is different and yet performs the same as your previous gen?
Click to expand...

How about that its smaller in die and in reality an optimized dualcore as amd puts it.
 
richierich1212 said:
Do you believe everything on the internet? 😀
Click to expand...

No. I noticed the 4110 clock speed is @ 4.2ghz which is not what it should be per what I have seen.

grimpr said:
How about that its smaller in die and in reality an optimized dualcore as amd puts it.
Click to expand...

Right, we know they setup the architecture to be more die size efficient. That helps the consumer only in that if you buy by spec alone AMD will be more competitive on price.

Personally, I just want to see reliable benches.
 
Last edited:
Despoiler said:
No. I noticed the 4110 clock speed is @ 4.2ghz which is not what it should per what I have seen.




Right, we know they setup the architecture to be more die size efficient. That helps the consumer only in that if you buy by spec alone AMD will be more competitive on price.
Click to expand...

The Cinebench 11.5 ST scores with FX-4110 (China leak) and FX-8150 (PCM NL leak) @ 4.2GHz are exactly the same. It just means AMD pulled a Netburst on us.
 
grimpr said:
How about that its smaller in die and in reality an optimized dualcore as amd puts it.
Click to expand...
That's the funny thing, is the design even more space efficient? We don't have official numbers, but people have estimated Orochi's die size as ~300mm^2 based on die pictures. 2 modules and half the L2 and L3 cache should be roughly half that, so ~150mm^2 for the quad core FX-4110. Deneb is 258mm^2. Correct me if I'm wrong, but if they simply shrunk Deneb from 45nm to 32nm, area would be reduced by about half, right? So that would put it at about 129mm^2 on 32nm.

Orochi_die_size.png
 
grimpr said:
No it wont, ever, at the best it will be 10% slower as long as Blizzard,Maxon and others use Intels tools and not Microsofts or gnus.

attauf.png


http://www.agner.org/optimize/blog/read.php?i=49
Click to expand...

There is a patch that allows phenom to use SSE2, right? There has to be because a X4 competes fairly well with a Q6600 on cinebench 10.1. If this wasn't the case then the score would have been abysmal. And I'm sure it was at first. If Maxon doesn't implement FMA / XOP then cinebench will be replaced by something that does. I'm willing to bet that within a few months we're going to see some pretty well optimized BD code running. We'll see it because when you have 8 or 16 cores available at such a low cost, people are going to try to get the most out of them. Luckily for us, most of the best programmers are unwilling to blow gobs of cash.
 
frostedflakes said:
That's the funny thing, is the design even more space efficient? We don't have official numbers, but people have estimated Orochi's die size as ~300mm^2 based on die pictures. 2 modules and half the L2 and L3 cache should be roughly half that, so ~150mm^2 for the quad core FX-4110. Deneb is 258mm^2. Correct me if I'm wrong, but if they simply shrunk Deneb from 45nm to 32nm, area would be reduced by about half, right? So that would put it at about 129mm^2 on 32nm.

Orochi_die_size.png
Click to expand...

You're probably right, but a Deneb core couldnt support whats coming in the future or support what BD does, i still believe that the current implementation in zambezi is borked somewhere.
 
StrangerGuy said:
FX-4110 leaks from China, gathered from XS.org. Basically reconfirms the other leaks we have so far:

2iu2fed.jpg
Click to expand...

FX-4170 @ 4.22ghz vs. i7-960 @ 3.46 (max Turbo) vs. Core i7 860 @ 3.46ghz.

The FX-4170 still cannot beat a 3-year-old Nehalem core clocked at only 3.46ghz. This has fail written all over it unless this CPU has insane overclocking headroom or it just flat out hates Cinebench R11.5. I don't even want to imagine what a 2600k @ 4.7ghz will do to an FX-8150 @ 4.7ghz if the implied IPC per core here is true.

* CPU-Z reports the CPU as FX-4110, but Cinebench shows FX-4170. There is still hope these are fake benches 🙂

StrangerGuy said:
The Cinebench 11.5 ST scores with FX-4110 (China leak) and FX-8150 (PCM NL leak) @ 4.2GHz are exactly the same. It just means AMD pulled a Netburst on us.
Click to expand...

"The eight-core (four-module) FX-8120, clocked at 3.1GHz and boosted to 4GHz, will retail for $185 — and the 3.6GHz/4.2GHz FX-8150 will retail for $230."

At least the pricing looks A LOT better than the previous $220-230 rumors for the cheapest 8-core. I just can't believe AMD's brand new 8-core CPU is only going to compete with a 2500k. What are they going to do when Haswell arrives in 2013?
 
Last edited:
mosox said:
http://www.youtube.com/watch?v=-M1eX5Vvrg8
Click to expand...

Your argument makes no sense since Phenom II X6 convincingly beats 2500k in Cinebench R11.5 in stock or overclocked vs. overclocked states. If Bulldozer has at least equal IPC to Phenom II X6 and another 2 cores, it would be at least 30% faster (8 / 6 = 33% advantage in cores) than the scores posted by a 4.0ghz Phenom II X6 below.

The overclock speeds for each chip in our testing are as follows:

Athlon II X4 635 - 3.7GHz (stock 2.9GHz)
Phenom II X4 970 - 4.0GHz (stock 3.4GHz)
Phenom II X6 1100T - 4.0GHz (stock 3.3GHz)
Core i5 750 - 3.8GHz (stock 2.7GHz)
Core i7 930 - 3.9GHz (stock 2.8GHz)
Core i5 2400 - 3.8GHz (stock 3.1GHz)
Core i5 2500K - 4.4GHz (stock 3.3GHz)
Core i7 2600K - 4.8GHz (stock 3.4GHz)

intel_2425k_cine11.jpg


StrangerGuy said:
So BD performs worse than Phenom II clock for clock per core
Click to expand...

IPC doesn't matter...give me more cores!

r1151.jpg


See, I told you. 8 of my BD 3.11ghz cores are faster than 4 of your SB cores. Oh wait a second.
Core i7 860 sold for $229.99 at MicroCenter 2 years ago...
 
Last edited:
grimpr said:
No it wont, ever, at the best it will be 10% slower as long as Blizzard,Maxon and others use Intels tools and not Microsofts or gnus.

attauf.png


http://www.agner.org/optimize/blog/read.php?i=49
Click to expand...
mosox said:
http://www.youtube.com/watch?v=-M1eX5Vvrg8
Click to expand...
Maybe those software developers should use Microsoft compilers instead, which widens the gap even further as can be seen from Anandtech's review using Microsoft Studio and Excel. I'd say at least 29% and more. 😛
 
RussianSensation said:
Your argument makes no sense since Phenom II X6 convincingly beats 2500k in Cinebench R11.5 in stock or overclocked vs. overclocked states. If Bulldozer has at least equal IPC to Phenom II X6 and another 2 cores, it would be at least 30% faster (8 / 6 = 33% advantage in cores) than the scores posted by a 4.0ghz Phenom II X6 below.

The overclock speeds for each chip in our testing are as follows:

Athlon II X4 635 - 3.7GHz (stock 2.9GHz)
Phenom II X4 970 - 4.0GHz (stock 3.4GHz)
Phenom II X6 1100T - 4.0GHz (stock 3.3GHz)
Core i5 750 - 3.8GHz (stock 2.7GHz)
Core i7 930 - 3.9GHz (stock 2.8GHz)
Core i5 2400 - 3.8GHz (stock 3.1GHz)
Core i5 2500K - 4.4GHz (stock 3.3GHz)
Core i7 2600K - 4.8GHz (stock 3.4GHz)

intel_2425k_cine11.jpg




IPC doesn't matter...give me more cores!
Click to expand...

Why CB 11.5? CB10 was your favorite version for quite a while posts back, something changed in the meantime?
 
grimpr said:
I dont know about Excel, but VS2008 is nothing compared to the multicore support and capabilities of the newest version.

Maybe you should take a look at Linux benchmarks ,to widen your perspective.

http://openbenchmarking.org/
Click to expand...
Majority of desktop users are using Windows, not Linux. The other desktop operating system would be Mac OS/X which is based on Linux (for Apple machines), and I do know quite a number of people running "Hackintosh" machines as well. 😀
 
grimpr said:
Why CB 11.5? CB10 was your favorite version for quite a while posts back, something changed in the meantime?
Click to expand...

:sneaky:

The scores referenced in this thread (Post #191) for the FX-4170 included CB11.5, not version 10. This is why I referenced those graphs for comparison purposes. I can pull up CB10 from PureOverclock, but you already know that AMD does worse in that version.
 
Last edited:
AnandThenMan said:
Noooooooooooooo. I don't believe it, how shocking this news is to everyone. :hmm:
Click to expand...
I used to dabble around with Linux (usually Slackware and Red Hat) and FreeBSD, but those were mostly for server purposes (such as Samba, Cern, Apache, SquirrelMail, etc). I did use them at my previous workplace, though my replacement got stumped (he was a "Windows" guy). 😀
 
First slide is fron August

august1j.jpg


Second slide from Taiwan, colors have been changed

taiwan1.png


Third slide from Donanim again, must be latest from October
ABBYY OCR10 i7 2600K outperform 2500K

october1s.jpg



Any thoughts ??

Someone playing with slides and i believe that is AMD 😉

Edit: There is another slide, different from all the above (not published)

Edit 2: the second donanim slide could be fake
 
Last edited:
You must log in or register to reply here.

TRENDING THREADS

Back
Top