FX-Series Processors Clock Speeds 'Revealed'

[Benetanegia]

I read it already get out of town

The increase is like this

1000 for Phenom II per core

1000 for Zambezi per core

It is the same, the performance didn't drop like you think it would it actually increased

1150 for Zambezi per core

There is another increase but I would till benchmarks to see this increase

No I actually think the increase is more like this:

for Phenom 2 per
for Zambezi per plus when the is activated and a good deal of Turbo too.

Ok, ok I'm done with feeding the troll. Sorry guys.

 

[HalfAHertz]

@seronix I stopped reading your posts a long time ago. It's really nice to see someone who is very eager and interested in the technology but the majority of the things you post are total bull...half the time you pull numbers out of thin air and the other half you contradict yourself.
Please stop spamming the Bulldozer threads with made up numbers and rumors unless you can show something solid and believable.

Now what I would like to know is would it make any sense to combine CMT and HT, if at all possible? The thread management and logistics horror aside, 16 threads would be better than 8 if half the time you can't saturate the cores with information at a 100%.

 

[fullinfusion]

Thats not very nice^

This guy bullshit or not is giving more then others. If it's not an Intel fan bashing the BD *imo im thinking some are feeling threatened by amd this go around* but hey Who knows till the chip hits the reviews... Half the shit in these threads are BS any ways so to point fingers is some what bios don't you think?

 

[seronx]

No I actually think the increase is more like this:

for Phenom 2 per
for Zambezi per plus when the is activated and a good deal of Turbo too.

Ok, ok I'm done with feeding the troll. Sorry guys.

Same clocks
Phenom II with Zambezi has the same performance

But there is some unknowns with the Bulldozer architecture particularly the

There are four integer execution units per core.

(4 ALUs + 2AGUs + 1 Store/Load Unit) per core
2 ALUs + 2 AGLUs + 1 Store/Load
and what does this design mirror?

Sandy Bridge

I looked it up, in the k15h handbook in programming

What they changed is instead of:
4 Simple AGLUs like originally planned(8 IPC per module)
We get:
2 Complex ALUs + 2 Simple AGLUs + 1 Load Store Unit(8-10 IPC per module)

@seronix I stopped reading your posts a long time ago. It's really nice to see someone who is very eager and interested in the technology but the majority of the things you post are total bull...half the time you pull numbers out of thin air and the other half you contradict yourself.
Please stop spamming the Bulldozer threads with made up numbers and rumors unless you can show something solid and believable.

Well I never read your posts

Now what I would like to know is would it make any sense to combine CMT and HT, if at all possible? The thread management and logistics horror aside, 16 threads would be better than 8 if half the time you can't saturate the cores with information at a 100%.

Yes, IBM is going to do it for their new "undisclosed" processor

 

[cadaveca]

Thats not very nice^

This guy bullshit or not is giving more then others. If it's not an Intel fan bashing the BD *imo im thinking some are feeling threatened by amd this go around* but hey Who knows till the chip hits the reviews... Half the shit in these threads are BS any ways so to point fingers is some what bios don't you think?

Point is taht we should be able to contain our excitement and look at this with a clear head. FUD is FUD, not matter how enthusiastic you are about something.

All things surrounding actual performance are under NDA. Zambezi is in the wild, and those that have it can't talk about it.

So rather than getting all giddy like a preschool girl with Barbie dolls, and falsely rasing expectations, it seems most prudent to sit back adn watch the show.

There's nothing wrong with seronx's posts, except that alot of people can't seem to accept his posts as just a post, and are getting a bit excited. This causes arguments, which really serve no purpose but to create dissention among our members.

That's where the problem is..the arguments his posts spawn.

It'as doubly bad when really, nobody cares about Bulldozer no, really, because nobody can buy it. When we can go and pick it up, toss it in our boards, then we can discuss the ins and outs of it...doing such before having the product in hand can be nothing other than marketing.

 

[radaja]

In real world tasks there will be no differences between 1 core used to 2 cores used

1 core in a module has access to 100% of the resources
2 cores in a module has access to 100% of the resources

The idea of CMT is to make 2 cores use the same resources to increase throughput/speed

1 module provides 2x the resources 1 core needs

4 cores being used in any pattern or setup will have 100% access to all the resources it needs

Simply put you do not need to worry about the module as a whole

Everything that needs to be dedicated is dedicated and everything that needs to be shared is shared

You got it backwards

There is 800% resources

1 core can only access 100% of those resources

It is a hardware limitation

2 cores will use 200% of those

So, 1 core will completely use the stuff dedicated to it and only half the stuff shared with it there is only so much 1 core can do

The floating point is a dedicated entity shared between both cores, so it does not follow what we think of a normal FPU

That is why it is call a Flex FPU

1 core in a module has access to 100% of the resources in 1 module
2 cores in a module has access to 200% of the resources in 1 module

Module holds all the resources needed for 2 cores to run in it
There is no performance hit in this design

i dont quite get how you went from 100%(post 229) to 200%(post 233) in the parts i highlighted?
are you talking about two different things in these explanations?

 

[xenocide]

i dont quite get how you went from 100%(post 229) to 200%(post 233) in the parts i highlighted?
are you talking about two different things in these explanations?

Chalk it up to magic.

 

[Wile E]

i dont quite get how you went from 100%(post 229) to 200%(post 233) in the parts i highlighted?
are you talking about two different things in these explanations?

Faerie dust.

 

[seronx]

i dont quite get how you went from 100%(post 229) to 200%(post 233) in the parts i highlighted?
are you talking about two different things in these explanations?

A core can only access or get filled so much

Two different explainations

1 is talking about utilization and 2 is talking about the amount of utilization in a module scale because the module is pointless to talk about as the windows OS will schedule it not us

Core 1 will be utilized 100%
Core 2 will be utilized 100%

They won't be fighting for resources

If the module provides 2x that of a normal core that means 200%

Core 0 will use 100% of that 200%
Core 0 and 1 will use 200% of that 200%
Core 1 will use 100% of that 200%

CMT = Real cores

A lot of people are dissing this design thinking it is worse
but it isn't!

It is a lot better than what is expected

Chalk it up to magic.

 

[Damn_Smooth]

A core can only access or get filled so much

Two different explainations

1 is talking about utilization and 2 is talking about the amount of utilization in a module scale because the module is pointless to talk about as the windows OS will schedule it not us

Core 1 will be utilized 100%
Core 2 will be utilized 100%

They won't be fighting for resources

If the module provides 2x that of a normal core that means 200%

Core 0 will use 100% of that 200%
Core 0 and 1 will use 200% of that 200%
Core 1 will use 100% of that 200%

CMT = Real cores

A lot of people are dissing this design thinking it is worse
but it isn't!

It is a lot better than what is expected



http://familyfun.go.com/assets/cms/playtime/chalk-it-up-games-photo-260-FF0306DRIVEA01.jpg

I don't have anywhere near as much confidence as you do about this. But being the AMD fan that I am, I hope you get the last laugh.

 

[Pestilence]

Reading obr and his rants cracks me up. Posted Saturday about Llano

AMD had originally thought that the voltage will be 1.3 V and frequencies above 3 GHz, as we all know what happened - default voltage 1.4 V and 2.9 GHz frequency miserable for the highest model without Turbo. We caught this time, "the bullet" AMD itself ... with their old official slides. So, still think that 32nm production at AMD (Glof) is good ... nope, it's a complete meal. What does it mean for the Bulldozer? - AMD is in a really bad situation. To not live up to promise, and this year released a Bulldozer really had to compromise enough to offer the first models and their parameters. The market will really only waste. odpadek first is called the FX-8110 with a pitiful 2.8 GHz and 1.4 V default It will be a tragic performance chip is ridiculous. few good cores that are able to produce by October, will go to processors FX-8150 without a "P". It will be a very limited production, only a few chips for review. AMD with him at the last minute lifted the basic stroke but reduced the Turbo. The tension is high again, 1.4 V. Since the cores are not better enough, AMD had to discontinue models FX-8130 and FX-8120, they will be listed in the second wave - to be more and better cores, and somehow manages to zberchat manufacturing process. October will be great for fans of AMD's disappointment, when Bulldozer come the poor, the vast majority of children and wise guy enough to roll your eyes at the performance ... is executed.

 

[erocker]

Reading obr and his rants cracks me up. Posted Saturday about Llano

I'm wondering why people actually care what this guy says? Anyone can: Make a website, make stupid claims and opinions, fake benchmarks and spread it around the web. Whoopee!

*Here's a statement. AMD FX series will be released (in retail) at or before 8/23/2011 which is the release date for "something else".

 

[Pestilence]

I'm wondering why people actually care what this guy says? Anyone can: Make a website, make stupid claims and opinions, fake benchmarks and spread it around the web. Whoopee!

*Here's a statement. AMD FX series will be released (in retail) at or before 8/23/2011 which is the release date for "something else".

Gives us something to discuss E. Ser.. I plan on going with an 8150 and a Crosshair V on release day so i can test it head to head against sandy bridge.

 

[Pestilence]

It seems the original delay was because of poor yields

Analyst: Yield drove Globalfoundries change
Peter Clarke
6/21/2011 7:28 AM EDT

LONDON – Management changes recently announced at Globalfoundries Inc. are related to problems with 32-nm chip yield and the slowness of creating a foundry-like operation at Globalfoundries' Dresden facility in Germany, according to analysts at Nomura Equity Research.

Globalfoundries announced June 16 that Doug Grose, CEO and Chia Song Hwee, COO, would be stepping down from their posts and the appointment of Ajit Manocha as interim CEO. Globalfoundries itself cited "demands from customers for more capacity, faster technology development and greater agility," in its press release.

Yield problems at 32-nm were driving processor vendor Advanced Micro Devices Inc. (AMD) into the arms of rival Taiwan Semiconductor Manufacturing Co. Ltd. (Hsinchu, Taiwan) at the same time that Globalfoundries was struggling to reduce its reliance on AMD as lead customer, the analysts indicated in a note issued to clients. The transfer of the manufacturing operation out from AMD was the mechanism that created Globalfoundries in March 2009.

The note was issued mainly for the benefit of clients invested in AMD and concluded that the changes at Globalfoundries would have minimal impact on AMD, partly because AMD recently recut its deal so that it would only pay for good die coming off the 32-nm line.

The authors of the notes said the announced management changes reflect the past execution in changing the Dresden facility from a one-customer, few-product wafer fab into a traditional many-customer, many-product foundry.

"As it relates to AMD, we believe that the majority of the 32-nm issues have already been addressed. AMD is currently shipping the 32-nm Llano A-series APUs for production. And, while 32-nm yields are still not optimal, we do not foresee AMD materially under-shipping demand in the second half of 2011. We also note that AMD recently ratified its pricing arrangement with GF to provide downside protection by allowing AMD to pay only for good dies,” the note said.

Source :

http://www.eetimes.com/electronics-news/4217113/Analyst-chip-yield-Globalfoundries

 

[xenocide]

And, while 32-nm yields are still not optimal, we do not foresee AMD materially under-shipping demand in the second half of 2011.

Am I the only one that think's that's a rather douchey statement on their behalf? It almost sounds like they don't think AMD will sell well enough to be in short demand, but I could just be reading into these things unnecessarily.

 

[Damn_Smooth]

Am I the only one that think's that's a rather douchey statement on their behalf? It almost sounds like they don't think AMD will sell well enough to be in short demand, but I could just be reading into these things unnecessarily.

You could look at it worse and think that AMD just doesn't give a shit and is going to send them out no matter what.

I really don't know what to make of it though. I do know that the laptop I'll be getting for school will have Llano inside. They better not be sold out.

 

[Pestilence]

You could look at it worse and think that AMD just doesn't give a shit and is going to send them out no matter what.

I really don't know what to make of it though. I do know that the laptop I'll be getting for school will have Llano inside. They better not be sold out.

You don't need Llano to watch porn at college. Sandy Bridge will do just fine.

 

[Damn_Smooth]

You don't need Llano to watch porn at college. Sandy Bridge will do just fine.

I like my porn at 1080p thank you.

 

[Pestilence]

I like my porn at 1080p thank you.

Trust me bro. After class you'll be banging so many girls you won't have time for porn... Unless you're a fatty or ugly then porn will be your only friend.

 

[Neuromancer]

...

It'as doubly bad when really, nobody cares about Bulldozer no, really, because nobody can buy it. When we can go and pick it up, toss it in our boards, then we can discuss the ins and outs of it...doing such before having the product in hand can be nothing other than marketing.

So simple...

I love theorizing, but facts are facts. We have none and wont till it is released.

 

[Damn_Smooth]

Trust me bro. After class you'll be banging so many girls you won't have time for porn... Unless you're a fatty or ugly then porn will be your only friend.

 

[Thatguy]

Trust me bro. After class you'll be banging so many girls you won't have time for porn... Unless you're a fatty or ugly then porn will be your only friend.

I rather get a hooker, easy terms, agreed upon service for the investment.

 

[YautjaLord]

I rather get a hooker, easy terms, agreed upon service for the investment.

Like any one that doesn't have enough sources for long time commitment; much like me.

Back to topic. Seeing that AMD vs Intel trend is bad for my eyes + i don't know much about this ALU/FP/SSE stuff that much like you all, i tend to side with majority & say: til TechPowerUp gets the hands on actual CPU (be it 8-, 6- or 4-core Dozer) & pits it vs Sandy/Ivy Bridge as much as the speculation might sound valid, it's still speculation. Hard gaming/benchmarking numbers is what needed, period. If frequency-wise Dozer beats the f*** out of Sandy Bridge or vice versa (for example 3.1GHz FX-8120 vs 3.4GHz i7 2600K) in - say - SuperPi, wPrime, PCMark Vantage & 3DMark11's CPU benchies than this what counts. Til that happens - as much as the info sounds good it's only speculation. Hope this happens by end of August though: this delay is f***ed up.

 

[XoR]

Bulldozer will be fast CPU coz it will have high clocks for 1-4 thread apps and many cores for >4 thread apps

I wouln't hope for beating Intel in performance/clock ratio or performance/power_consumption though... Bulldozer will be one power hungry beast

ps. there were only one cpu family that was without confusion of how much cores it does really posess, legendary Pentium D

and it ran 5GHz with ease

all other multi-cores are just a scam...

 

[[H]@RD5TUFF]

Bulldozer will be fast CPU coz it will have high clocks for 1-4 thread apps and many cores for >4 thread apps

I wouln't hope for beating Intel in performance/clock ratio or performance/power_consumption though... Bulldozer will be one power hungry beast

ps. there were only one cpu family that was without confusion of how much cores it does really posess, legendary Pentium D
http://gadgets.softpedia.com/images...e/The-INTEL-Pentium-D-920-Dual-Core-CPU-1.jpg
and it ran 5GHz with ease

all other multi-cores are just a scam...

Now if only my UD5 would support one.