Bulldozer Information Thread.

[cadaveca]

I was wondering why I couldn't find you doing any of the CPU reviews.

Though without knowing the settings of CODEC, how much can anyone compare those "leaked" results to anything else?

Up to how many threads is Handbrake able to take advantage of anyway (if you happen to know)?

I do motherboard reviews only.

To me, this is unimportant. I suppose the they used the same settings, and same source files, as well as same ram speed and HDD config. That said, the percentile difference should be roughly equal. It is possible that Handbrake has been tweaked to take advantage of GPUs, too, there are many unknowns here.


However, I choose to to accept they used the same source file for each, and one finished faster than the other. If the price is the same, but one is faster, clearly one is the better option for your dollar?

I care more about usability than benchmark performance. My A8-3850 system, when paired with an SSD, seems just as fast as my Sandybridge rig for most of my daily uses. And it uses less power while doing so.

There seems to be one obvious thing everyone is missing, but I'm not gonna say anything myself.

 

[seronx]

2. It would be double if it had the same performance per core which probably does not, but I know what you meant.
What really bothers me for quite a long time is the way AMD chose to call it an 8 core CPU so everyone would assume it obliterates any 6 core CPU.
chew* himself made a lot of "rants" about AMD calling it an 8 core. This is the AMD's patent of a core:
http://img220.imageshack.us/img220/5570/25718801.png
Which I believe a BD core doesn't look like it. I'll also quote him:

The Module is the Core Architecture

While the cores that we have known in previous AMD architectures to be the integer cluster(GPR ALU cluster)

Windws OS will always see "cores"

The OS considers GPR ALU Clusters cores

There seems to be one obvious thing everyone is missing, but I'm not gonna say anything myself.

Cost?

980X/990X $1000
FX-8000 $270 and below

OK . BD has more IPC . How many ? I have no clue , How will it perform against the SB with the added IPC ? I do not know . But more is better .

Total Theoretical IPC for K15 is lower than Sandy Bridge but Total Actual IPC for K15 is higher than K10

 

[naoan]

Up to how many threads is Handbrake able to take advantage of anyway (if you happen to know)?

If they use x264 as the encoder (which they should), up to 128. http://mewiki.project357.com/wiki/X264_Settings#threads

 

[TheoneandonlyMrK]

Anyone else hammering the overtime at the min just in case it finally arrives im effin sick of work and approaching my second weekend in the place Feck.

come on AMD rep dude sort it, so i can blow some doe next month lol

 

[trickson]

Anyone else hammering the overtime at the min just in case it finally arrives im effin sick of work and approaching my second weekend in the place Feck.

come on AMD rep dude sort it, so i can blow some doe next month lol

I am ! I have 2 grand burning a hole in my pocket right now ! And till I get some solid information I have to wait !

 

[Covert_Death]

just got my ASRock 990FX extreme4 set up in my system. overclocks GREAT, easy as can be too...

im currently running as follows:
CPU: Phenom II x4 955 @ 4.2Ghz (210 x 20.0)
NB: @2520 Mhz
RAM: ddr3 1666


i will probably crank things up a bit more later but temperatures are great, i have air cooling (thermaltake spinQ) on my CPU and the rest is stock, good airflow through my case but again temps are really good for the speeds im currently running.

now i just need a bulldozer

 

[eidairaman1]

just got my ASRock 990FX extreme4 set up in my system. overclocks GREAT, easy as can be too...

im currently running as follows:
CPU: Phenom II x4 955 @ 4.2Ghz (210 x 20.0)
NB: @2520 Mhz
RAM: ddr3 1666


i will probably crank things up a bit more later but temperatures are great, i have air cooling (thermaltake spinQ) on my CPU and the rest is stock, good airflow through my case but again temps are really good for the speeds im currently running.

now i just need a bulldozer

Hmm wonder how my bros PII X2 555 Unlocked to X4 B55/955 would overclock with 970 Extreme 4, perhaps to that speed idk, rather not find out till i get an aftermarket heatsink on it

 

[billcat479]

I appreciate the "well, sorta", but the Phenom I was nowhere near an architectural change. The Phenom II wasn't either. This is AMD's first real architectural change in almost a decade.

I'm sure we all know how it turned out the last time AMD had a real architecture change too, but history has nothing to do with how Bulldozer will perform.

If you remember AMD's plans they made a cpu with long range plans from the start. It was built to expand upon and add seperate cores from the beginning so it was not any surprise of what AMD did. They had it worked out that way from the start.
To redesign the basic cpu would have taken them a long time that they didn't have and/or couldn't change because their long range plans were already in motion. It wasn't AMD's fault that Intel came out with a better design. It was because of AMD that forced Intel to redesign their cpus from the MHZ blast attack to follow AMD's line of thinking in making them more efficient per clock cycle and slow things down.
It was a drag for AMD that Intel has a lot more money and can spend more on their R&D with AMD already having done a lot of the ground work designs and so came out with the faster cpu's. AMD's goal was to make the true multi-core chip unline Intels tied together duo chip design. It's too bad AMD's design didn't work better but anyone can out do another if they spend enough to do it.
Look at what Ford did to Ferrari with the F40. It cost a ludicrous sum to build it with a lot of trial and errors along the way but they had a goal to beat Ferrari at Le-mans at that 24 hour race and they did it for a year or two and after that dropped out. In a way this is the way intel does it with AMD besides the nasty anti-competitive actions they did when AMD was putting out the best cpu's. That was a real crime and they let Intel off to easy on that one.
If AMD had Intel's funds who knows what we would have as AMD has been the design leader of trying new stuff with Intel following up with the same stuff for most of it but with a deeper pocket book. Like mentioned before Intel thought 64bits was a waste of time and I think they said the same with mult cores also but I'm not 100% sure about that point.
But they had to follow AMD's road-map and make it better no matter what the cost. They sure didn't like it when AMD was the real leader and had a short time as the performance leader as well as the design leader. Now AMD is trying a new design that is a bit different but I'm not sure how much different it really is from current cpu's. I haven't been following it that closely anymore. But every where you look you see AMD's name behind the new trends in computer innovation.
The only thing I've hard about Intel is their new 3D transistor which is great to see. But to follow it up with a radical new design is not their way of doing things.
In a way they are doing things like Japan does or how they used to do it. They don't invent, they take the tested new technology and just make it better.

 

[mastrdrver]

To me, this is unimportant. I suppose the they used the same settings, and same source files, as well as same ram speed and HDD config. That said, the percentile difference should be roughly equal. It is possible that Handbrake has been tweaked to take advantage of GPUs, too, there are many unknowns here.

This is really the question on my mind. Not so much with the results from the article as like you I assume they used the same settings. Question I have, without knowing those settings that AMD used, how is it possible to compare those results from yours or any other side to get a feel for how good/bad BD is?

 

[cadaveca]

This is really the question on my mind. Not so much with the results from the article as like you I assume they used the same settings. Question I have, without knowing those settings that AMD used, how is it possible to compare those results from yours or any other side to get a feel for how good/bad BD is?

Really, you cannot.

For me, because I run more benchmarks than are in my reviews, I just get a general feel. Memory performance seems better than Thuban, and the cores are clocking high, obviously. I am focused on memory performance more than core performance, as I have been saying for almost years now, so if they have solved that, and give better multicard scaling, then I'm happy. Pricing tells me alot too.



AMD did say 5 GHz on air, 6.6 GHz on LN2, then the rest on LHe?

That tells me that AMD's process is still a bit temperature sensitive. the core has 2 MB of L2 per module, so when they disabled the cores, they killed alot of the heat generation by removing nearly half the cache, and that explains why they got so much more scaling under LHe.

The socket has larger pins, and the CPUs might as well, and that tells me that thay are going to be feeding killer wattage through the chips while overclocking, and the ASUS Crosshair V Formula's extra CPU power plugs furthers that idea.

To me, it's looking positive. I don't expect much. FX8150 @ less than $300? FX 3170 @ $279, perhaps? that says something too.

In the end, all I am looking for is comparable game performance for the same dollar, and equal or lower power consumption. I want excellent memory bandwidth. I want IOMMU support.


I do not need the fastest CPU possible. I don't need world records. I need fast and affordable solutions, and a good user experience. I am confident AMD can deliver on those things, given what I've seen already.

I guess we'll find out more soon. I am broke right now anyway, so the longer it takes, the better for me anyway.

I could be wrong, and BD sucks. But I like to OC too, and Intel is a big disappointment in OC, so if a chip can give me a few weeks of tweaking fun rather than a few hours, then I'm in!

 

[repman244]

The Module is the Core Architecture

While the cores that we have known in previous AMD architectures to be the integer cluster(GPR ALU cluster)

Windws OS will always see "cores"

The OS considers GPR ALU Clusters cores

I know the OS will see them as cores that was not my point, windows can see them as 16 cores I don't really care about that. My point was that AMD's marketing is taking advantage of it to display it as an 8 core.
And like I quoted chew* some people may be "disappointed" with the multithread performance since it won't be that massive as having a true 8 core design.

 

[seronx]

I know the OS will see them as cores that was not my point, windows can see them as 16 cores I don't really care about that. My point was that AMD's marketing is taking advantage of it to display it as an 8 core.
And like I quoted chew* some people may be "disappointed" with the multithread performance since it won't be that massive as having a true 8 core design.

Multithreaded Performance is dedicated on to the program and OS

1. Final production silicon
2. Final processor microcode
3. Final system BIOS
4. Final OS optimizaitons
5. Final drivers
6. An app compiled with the latest flags
7. A person who understands the app and configures the test properly

Can you check them off

Then the best advice

Never trust any benchmark unless it is open source and compiled with a neutral compiler, such as Gnu or Microsoft.

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

 

[Damn_Smooth]

If you remember AMD's plans they made a cpu with long range plans from the start. It was built to expand upon and add seperate cores from the beginning so it was not any surprise of what AMD did. They had it worked out that way from the start.
To redesign the basic cpu would have taken them a long time that they didn't have and/or couldn't change because their long range plans were already in motion. It wasn't AMD's fault that Intel came out with a better design. It was because of AMD that forced Intel to redesign their cpus from the MHZ blast attack to follow AMD's line of thinking in making them more efficient per clock cycle and slow things down.
It was a drag for AMD that Intel has a lot more money and can spend more on their R&D with AMD already having done a lot of the ground work designs and so came out with the faster cpu's. AMD's goal was to make the true multi-core chip unline Intels tied together duo chip design. It's too bad AMD's design didn't work better but anyone can out do another if they spend enough to do it.
Look at what Ford did to Ferrari with the F40. It cost a ludicrous sum to build it with a lot of trial and errors along the way but they had a goal to beat Ferrari at Le-mans at that 24 hour race and they did it for a year or two and after that dropped out. In a way this is the way intel does it with AMD besides the nasty anti-competitive actions they did when AMD was putting out the best cpu's. That was a real crime and they let Intel off to easy on that one.
If AMD had Intel's funds who knows what we would have as AMD has been the design leader of trying new stuff with Intel following up with the same stuff for most of it but with a deeper pocket book. Like mentioned before Intel thought 64bits was a waste of time and I think they said the same with mult cores also but I'm not 100% sure about that point.
But they had to follow AMD's road-map and make it better no matter what the cost. They sure didn't like it when AMD was the real leader and had a short time as the performance leader as well as the design leader. Now AMD is trying a new design that is a bit different but I'm not sure how much different it really is from current cpu's. I haven't been following it that closely anymore. But every where you look you see AMD's name behind the new trends in computer innovation.
The only thing I've hard about Intel is their new 3D transistor which is great to see. But to follow it up with a radical new design is not their way of doing things.
In a way they are doing things like Japan does or how they used to do it. They don't invent, they take the tested new technology and just make it better.

I like your post, but I really don't see how it contradicts anything that I said. We really don't have anything to base Bulldozer's performance on other than speculation.

My speculation leads me to believe that Bulldozer will be a nice chip. I could be wrong though.

 

[repman244]

AMD did say 5 GHz on air, 6.6 GHz on LN2, then the rest on LHe?

Quoting chew*:

I think that any info other than what I or AMD have personally offered which is not much if any can be discarded until further notice.

There are still peices to the puzzle missing that I can assure you 99.9% don't have yet regarding CPU rev and bios support and agesa.

To sum up BD facts

BD is physically a 4 core 8 thread part.

It has no coldbug

Samples can bench at 5+ on stock cooler, 6 on phase change and 8.4 on lhe.

And most important of all SATA works

Like i said before the joke is on that guy we won't name, he was sent intentionally quite possilby the worst chip ever produced. 6.4 on ln2,
or he needs to learn how to OC, 6.4 is my validation speed on phase change......he who laughs last laughs best.............

In the last part he is refering to OBR

 

[cadaveca]

So, basically everything that I have been saying for the past 8 months or so still holds true.


Interesting.












Not.


I mean really, I ain't got no chips, no insider info, nothing, and I seem to know WTF is going on, so I don't get all the bullshit hype. When OBR posted shit, I said to ignore him, then he said he was full of it himself, and was trolling, so I guess we know nothing more than we did a year ago, other than..well...that it overclocks well under excessive cooling.


Yippie.

 

[bpgt64]

What really makes no sense to me, is when Intel Fan boys cheer on poor performance of AMD products. If you prefer Intel, AMD performing well only forces Intel to price more competitively(price wars and better prices for you), and vis versa. Competition only benefits the consumer.

 

[CDdude55]

What really makes no sense to me, is when Intel Fan boys cheer on poor performance of AMD products.

That's reality, and ive heard AMD fanboys talk about how they think Intel chips are overpriced because they ''offer similar performance to AMD chips''.

Logic and reasoning tend not to be a strong suit for people who worship certain companies.

 

[seronx]

That's reality, and ive heard AMD fanboys talk about how they think Intel chips are overpriced because they ''offer similar performance to AMD chips''.

Logic and reasoning tend not to be a strong suit for people who worship certain companies.

Intel chips are overpriced because they offer a marginal performance increase...5% to 10% wow....$200 vs $140....ya that price is totally representative

Everytime AMD releases a brand new chip it usually is beyond the marginal

Tick-Tock from Intel is a measure to keep away from the intimate tsunami called AMD

A8-3850 $140
i5 2500K $200

Set your ideologies straight

 

[CDdude55]

Intel chips are overpriced because they offer a marginal performance increase...5% to 10% wow....$200 vs $140....ya that price is totally representative

Everytime AMD releases a brand new chip it usually is beyond the marginal

Tick-Tock from Intel is a measure to keep away from the intimate tsunami called AMD

http://www.phoronix.com/data/img/results/amd_a83850_cpu/1.png

A8-3850 $140
i5 2500K $200

Set your ideologies straight

That benchmark is measuring on die GPU, so they tested the 2500K's GPU performance which definitely isn't as good as AMD's APU's. So you're overgeneralizing, i'm talking general CPU performance.

And the numbers speak for themselves:

Granted the AMD chips get frames that are more then playable, but again, performance warrants the higher cost with Intel to a lot of people.

Facts are facts, im running an AMD system and have a 990FX board and planning for BD, but that doesn't excuse what the numbers show.

 

[repman244]

But then again the games ran on a GTX 280 at 1680 x 1050 no AA/AF and medium settings...I know this is to show how much a certain CPU can bottleneck a card but aren't really real world numbers.
I do no imply that SB has the same performance as AMD's chips they are in doubt faster, I would only like to see numbers with "real world" settings.

 

[CDdude55]

But then again the games ran on a GTX 280 at 1680 x 1050 no AA/AF and medium settings...I know this is to show how much a certain CPU can bottleneck a card but aren't really real world numbers.
I do no imply that SB has the same performance as AMD's chips they are in doubt faster, I would only like to see numbers with "real world" settings.

True but that would sort of be missing the point in a CPU review for measuring CPU performance which is what i was trying to address.

Those benchmarks are still using real world software even if the system itself was built with CPU performance in mind. When you hit higher resolutions you start leaning on the GPU more so why do you want to see real world settings in a review meant to test certain parts?

If we were talking about testing games and other software such as benchmarks, then yes, real world settings and components would matter a ton.

 

[seronx]

That benchmark is measuring on die GPU, so they tested the 2500K's GPU performance which definitely isn't as good as AMD's APU's. So you're overgeneralizing, i'm talking general CPU performance.

And the numbers speak for themselves:

http://images.anandtech.com/graphs/graph4083/35038.png

http://images.anandtech.com/graphs/graph4083/35041.png

Granted the AMD chips get frames that are more then playable, but again, performance warrants the higher cost with Intel to a lot of people.

Facts are facts, im running an AMD system and have a 990FX board and planning for BD, but that doesn't excuse what the numbers show.

970BE vs i5 2500K

Crysis Warhead
82.6 vs 91.6

Fallout 3
87.5 vs 90.3

Intel i5 2500K + $15 off
Intel Core i5-2500K Sandy Bridge 3.3GHz (3.7GHz Tu...
$205.99

AMD Phenom II X4 970
AMD Phenom II X4 970 Black Edition Deneb 3.5GHz 4 ...
$139.99

$205.99 vs $139.99

33% increase of performance should come with this price in your expectations

Crysis Warhead
10%

Fallout 3
4%

A8-3850
AMD A8-3850 Llano 2.9GHz 4MB L2 Cache Socket FM1 1...
$139.99

The distance between Llano and i5 is much smaller the only reason i5 2500K gets any faster(2%-8%) is because of the L3 Cache which Llano doesn't have

 

[dumo]

Simple calculation based on Newegg price...

MIVE+2600K = $650
CH5 = $230
$420 difference which can be applied to FX8150 price ranges

The latest...http://www.xtremesystems.org/forums...-info-fans-!&p=4950616&viewfull=1#post4950616

 

[CDdude55]

970BE vs i5 2500K

Crysis Warhead
82.6 vs 91.6

Fallout 3
87.5 vs 90.3

Intel i5 2500K + $15 off
Intel Core i5-2500K Sandy Bridge 3.3GHz (3.7GHz Tu...
$205.99

AMD Phenom II X4 970
AMD Phenom II X4 970 Black Edition Deneb 3.5GHz 4 ...
$139.99

$205.99 vs $139.99

Sandy Bridge and AMD's Phenom II chips mainly separate in multi-tasking performance. Gaming wasn't an accurate representation of said performance difference.

Examples:

In most CPU oriented tasks Intel=better.

 

[seronx]

Sandy Bridge and AMD's Phenom II chips mainly separate in multi-tasking performance. Gaming wasn't an accurate representation of said performance difference.

Examples:

http://img.techpowerup.org/110915/Capture535353.png

http://img.techpowerup.org/110915/Capture465464.png

In most CPU oriented tasks Intel=better.

I would just say wait for AMD Zambezi...now you are getting to the point where you are hitting the architectural limit for AMD Deneb/Thuban

Those benchmarks heavily use the ALUs and AGUs which Phenom II can't use at the same time

Where in Video Games CPUs rarely require the max amount of ALU/AGUs(Notice most AMD chips are aimed at gamers)

So, if you do gimp, handbrake, mplayer, and 7-zip go right ahead

FX allows the ALUs/AGUs to be used at the sametime

Simple calculation based on Newegg price...

MIVE+2600K = $650
CH5 = $230
$420 difference which can be applied to FX8150 price ranges

The latest...http://www.xtremesystems.org/forums...-info-fans-!&p=4950616&viewfull=1#post4950616

I'm going to ignore shipping costs on this one

ASRock 990FX Pro
ASRock Fatal1ty 990FX Professional AM3+ AMD 990FX ... => $189.99
ASRock Z68 Pro
ASRock Z68 PROFESSIONAL GEN3 LGA 1155 Intel Z68 HD... => $234.99
i7 2600K
Intel Core i7-2600K Sandy Bridge 3.4GHz (3.8GHz Tu... => $314.99
FX-8150 Pre-order from ShopBLT
http://www.shopblt.com/cgi-bin/shop/shop.cgi?action=thispage&thispage=01100300U031_BLA5134P.shtml => $266.28(Newegg will most likely be $259.99 or $279.99 <-- I'll use the bigger number)


$189.99+ $279.99 = $470

$234.99 + $314.99 = $549.98