AMD Bulldozer Eng. Sample leaked, benched

[Damn_Smooth]

I guess, but it's equally unknown at this point. We can only look at past chips, but I think these CPUs are going to get a different process than previous AMD chips, even?

The only thing I can answer you with and be even somewhat honest about it, is that I don't know. I don't know if the process behind this is different, I don't know how these chips will perform in real life, hell, I don't even know a release date.

But I do know that AMD wouldn't invest the time and effort that they have, to take a step backwards. They do have stockholders that they are accountable to, even if there is no CEO, so taking backwards steps would probably end the company.

 

[trickson]

I do not know but it seems as if all AMD is really doing is pushing cores out . MORE CORES ='s better chips . I hope to see them taking the lead again but from the BM provided ( NOT reliable at all ) I would say they are just putting out more cores at the same performance as the Phenom . More of a lateral move to me .

 

[erocker]

Well first off there seems to be Bulldozers out . They are saying that they bench tested them and even gave some benchmarks of there sample , What are you talking about ?
The thread is titled AMD Bulldozer Eng. Sample leaked , Benched ! Am I seeing things here ? Don't flip the scrip on me son .

Um.. okay. So, someone got a hold of an early AMD BD engineering sample and it sucks. Where do we go from here? Make multiple posts claiming BS? Discuss it and everything to do with it? You don't seem to have much to contribute to the actual topic nor does anyone else. Why? There's no information. So I could just close the thread or let people be grown-up's and discuss various things involving CPU's that make for interesting and informative conversation. I'll do the latter.

I do not know but it seems as if all AMD is really doing is pushing cores out . MORE CORES ='s better chips . I hope to see them taking the lead again but from the BM provided ( NOT reliable at all ) I would say they are just putting out more cores at the same performance as the Phenom . More of a lateral move to me .

Well, you're wrong and basing your assumption of of some sample that we don't even know is real. As far as this part:

" would say they are just putting out more cores at the same performance as the Phenom"

I'll hold you to that. It makes absolutely zero sense, but we'll see.

 

[Dent1]

Well first off there seems to be Bulldozers out . They are saying that they bench tested them and even gave some benchmarks of there sample , What are you talking about ?
The thread is titledAMD Bulldozer Eng. Sample leaked , Benched ! Am I seeing things here ? Don't flip the scrip on me son .

Indeed the title does say that. Doesn't mean its true.

I'm sure Intel had samples in their test labs of a premature, defective and underperforming SandyBridge. The difference is it wasn't leaked.

Anyways, we have all already established many pages ago that the authenticity of the benchmarks can not be trusted.

To close, there is no official or confirmed unofficial authentic Bulldozer review or benchmark at present.

 

[trickson]

I really want to think that bulldozer is going to give Intel a run for the money . All I hope is that they will take the lead this will force lower prices in the market . I do not like to see things like this as it makes me think things will stay the same and prices will be even higher . This is all I am meaning to convey .

 

[cadaveca]

300w ? are you sure ?

Oh, look, here's OVER 400W going through an AMD CPU(Crosshair V Formula, 8-pin + 4-pin):

Oh, and look...a meter over the 8-pins. I guess some are really just not aware of what happens when you overclock to the extreme. 300W is NOTHING, and ASUS is more than aware of it(otherwise there'd be no need for both 4-pin and 8-pin)!!

 

[seronx]

Oh, look, here's OVER 400W going through an AMD CPU(Crosshair V Formula, 8-pin + 4-pin):

Oh, and look...a meter over the 8-pins. I guess some are really just not aware of what happens when you overclock to the extreme. 300W is NOTHING, and ASUS is more than aware of it(otherwise there'd be no need for both 4-pin and 8-pin)!!

And that is with the Crosshair V Formula version and not the Crosshair V Extreme version

(Extreme has 2x8 pin)

 

[cadaveca]

I didn't evne know there was an "Extreme" version planned.

I've had several peopel ask me...what's the difference between the Sabertooth, and the Crosshair, and why would I buy one over the other?

This is just one reason to choose one over the other.

 

[seronx]

I didn't evne know there was an "Extreme" version planned.

I've had several peopel ask me...what's the difference between the Sabertooth, and the Crosshair, and why would I buy one over the other?

This is just one reason to choose one over the other.

Sabertooth TUF 6+2 8pin , Formula 8+2 8pin+4pin , Extreme 8+2 2x8-pin


http://www.asus.com/Motherboards/AMD_AM3/Crosshair_IV_Extreme/

http://www.asus.com/Motherboards/AMD_AM3/Crosshair_IV_Formula/

If they did it before and didn't get punished it is most likely they will do it again

 

[cadaveca]

Huh. I'm gonna have to ask ASUS if I can try both out. My CPU is a power hog, and the coldest my pot has been has been -35c, in winter, as it's sitting in the garage collecting dust. These might be enough for me to pull it out again.

Interesting, very interesting indeed.

However, that link goes to the Crosshair IV, and I want Crosshair V Extreme.

ASUS has many products that catch my eye, like the ROG expander, OC panel, etc...and even the entry-level boards have alot of features I was NOT expecting.

 

[seronx]

Huh. I'm gonna have to ask ASUS if I can try both out. My CPU is a power hog, and the coldest my pot has been has been -35c, in winter, as it's sitting in the garage collecting dust. These might be enough for me to pull it out again.

Interesting, very interesting indeed.

However, that link goes to the Crosshair IV, and I want Crosshair V Extreme.

It usually takes 7-8 Months before the Crosshair Extreme is born

They need to get a mama crosshair and a daddy crosshair to make a perfect girl crosshair

But, I heard they got the mating session over with now it is just the waiting time

 

[Thatguy]

Oh, look, here's OVER 400W going through an AMD CPU(Crosshair V Formula, 8-pin + 4-pin):

Oh, and look...a meter over the 8-pins. I guess some are really just not aware of what happens when you overclock to the extreme. 300W is NOTHING, and ASUS is more than aware of it(otherwise there'd be no need for both 4-pin and 8-pin)!!

WTF am I supposed to do with a red X and no link ? and I explained to you that other things are on those connectors getting power. You want real cpu wattage, isolator socket and get power data out of the pins that the cpu gets power from. Never mind the fact that 80 amps through a few pins on the cpu sounds like its pushing the boundrys of established electrical engineering thoery.

 

[seronx]

WTF am I supposed to do with a red X and no link ? and I explained to you that other things are on those connectors getting power. You want real cpu wattage, isolator socket and get power data out of the pins that the cpu gets power from. Never mind the fact that 80 amps through a few pins on the cpu sounds like its pushing the boundrys of established electrical engineering thoery.

The highest peak he got is 1688 Watts 8pin+4pin for ya

 

[cadaveca]

WTF am I supposed to do with a red X and no link ? and I explained to you that other things are on those connectors getting power. You want real cpu wattage, isolator socket and get power data out of the pins that the cpu gets power from. Never mind the fact that 80 amps through a few pins on the cpu sounds like its pushing the boundrys of established electrical engineering thoery.

Here's the full link:

http://www.youtube.com/watch?v=w70T3h_re9A&feature=player_embedded

And, note that the 9-series baords with a black socket feature support for thicker pins to handle mroe current. And while you may have explained soemthing to me, I simply asked a very straight-forward question..."Like what?"

There's this amazing thing, called a continuity test...are you familiar with that? It allows sourcing power at any component and finding where that power comes from. I have yet to find anything other than CPU power to be powered from an ATX/EPS 8-pin, yet you are right..it IS possible(and i mention as much in my reviews, too)..but I have yet to find any examples in any of the boards I have tested so far.

 

[Thatguy]

Here's the full link:

http://www.youtube.com/watch?v=w70T3h_re9A&feature=player_embedded

And, note that the 9-series baords with a black socket feature support for thicker pins to handle mroe current. And while you may have explained soemthing to me, I simply asked a very straight-forward question..."Like what?"

There's this amazing thing, called a connectivity test...are you familiar with that? It allows sourcing power at any component and finding where that power comes from. I have yet to find anything other than CPU power to be powered from an ATX/EPS 8-pin, yet you are right..it IS possible(and i mention as much in my reviews, too)..but I have yet to find any examples in any of the boards I have tested so far.

I got a 6 core running at 4 ghz on a 450watt power supply with a 5770 video card. Max load the system is pulling 380 from the wall. not really seeing it. I don't have time right now to make a isolator socket with a current bridge.

 

[cadaveca]

I got a 6 core running at 4 ghz on a 450watt power supply with a 5770 video card. Max load the system is pulling 380 from the wall. not really seeing it. I don't have time right now to make a isolator socket with a current bridge.

You don't need to. Just because you are running less power through your CPU than I do for testing, doesn't mean it's not possible. If it wasn't possible, the CHVF would not have the extra 4-pin CPU power connector, and that's a fact.

Since you think this:

and it also power the pci express slots.

I'm about done explaining these things to you.

 

[Thatguy]

You don't need to. Just because you are running less power through your CPU than I do for testing, doesn't mean it's not possible. If it wasn't possible, the CHVF would not have the extra 4-pin CPU power connector, and that's a fact.

the problem is that the math doesn't add up. To consume 80amps would requires a significant drop in tempature to lower resistance to allow for more current.

 

[cadaveca]

the problem is that the math doesn't add up. To consume 80amps would requires a significant drop in tempature to lower resistance to allow for more current.

CPU's are semi-conductors, whose electrical properties vary according to temperature. Basic stuff.

CPUs take 80a all the time...more even...@ thier rated voltage. For example...80a @ 1.5v= 120W. Current CPUs run far less than 1.5v.

FYI, in case you didn't watch the video, the video shows LN2 cooling.

 

[Thatguy]

CPU's are semi-conductors, whose electrical properties vary according to temperature. Basic stuff.

CPUs take 80a all the time...more even...@ thier rated voltage. For example...80a @ 1.5v= 120W. Current CPUs run far less than 1.5v.

jesus I got 5v on the brain today, lol.

 

[cadaveca]

NO big deal. I very purposely have created testing for my reviews that is near infalliable. I test things no other reviewer does, show things no other reviewer does, and hold opinions most reviewers do not agre with, so I expect alot of shock and surprise from readers...but at the same time, my main purpose is trying to educate users on what thier PCs are really doing.

I take my job here seriously. If I am even slightly unsure about something, I say so.

I bin my CPUs for testing. I usually buy 10-15 CPUs, and keep the very worst sample for testing, so create worst-case scenarios. Of course, there's always a chance that there are other bits out there worse than what I find, but I very purposely get rid of what woudl be considered "cherry-picked" parts, as I do not beleive using top-level hand-binned parts serves any purpose.


I do try to break every product I get, too, but at the same time, I do try to maintain reasonable limits to my testing.


So I found a CPU that can take over 350W, at what would be modest clocks. This is specifically so I can push board VRMs to the limit. In my ASUS M5A97 EVO review, I managed to do just that, with OCP kicking in @ just over 280w.

 

[Thatguy]

NO big deal. I very purposely have created testing for my reviews that is near infalliable. I test things no other reviewer does, show things no other reviewer does, and hold opinions most reviewers do not agre with, so I expect alot of shock and surprise from readers...but at the same time, my main purpose is trying to educate users on what thier PCs are really doing.

I take my job here seriously. If I am even slightly unsure about something, I say so.

I bin my CPUs for testing. I usually buy 10-15 CPUs, and keep the very worst sample for testing, so create worst-case scenarios. Of course, there's always a chance that there are other bits out there worse than what I find, but I very purposely get rid of what woudl be considered "cherry-picked" parts, as I do not beleive using top-level hand-binned parts serves any purpose.


I do try to break every product I get, too, but at the same time, I do try to maintain reasonable limits to my testing.


So I found a CPU that can take over 350W, at what would be modest clocks. This is specifically so I can push board VRMs to the limit. In my ASUS M5A97 EVO review, I managed to do just that, with OCP kicking in @ just over 280w.

The best way to test real consumption would be to grab the socket pins. Yeah I have touch probe and current meters that might be workable but nothing with beat the accuracy of a current bridge to get the most solid numbers to within a very small window of error.

I don't have time to build anything like it, but I bet some large company has it.I also know I don't have a top bin part either.

 

[cadaveca]

Sure, and you are very right, and I know both Intel and AMD do have socket-based testers, to test VRM design, however, that would not really show VRM efficiency, merely the power the VRM produces.

I use the same CPU for testing in each platform(ie one CPU per platform), and things like ASUS's DIGI+ VRM will affect power output to the CPU, so it serves me better to measure @ the 8-pin, providing a metric based on the same CPU speed(which no matter the board, the CPU should consume the same power for the same clocks), and give numbers which can be comapred from board to board, highlighting the VRM, rather than the CPU. I am not the TPU CPU reviewer, so providing info specific to CPU power draw is not neccesary for me.

Testing via your proposed method would eliminate VRM efficiency numbers, and focus on the CPU, while I'm concerned about the VRM design. I coudl do both sets of testing, but I see no need for me to report specific CPU numbers when I'm not testing CPUs in my reviews.

 

[Thatguy]

Sure, and you are very right, and I know both Intel and AMD do have socket-based testers, to test VRM design, however, that would not really show VRM efficiency, merely the power the VRM produces.

I use the same CPU for testing in each platform(ie one CPU per platform), and things like ASUS's DIGI+ VRM will affect power output to the CPU, so it serves me better to measure @ the 8-pin, providing a metric based on the same CPU speed(which no matter the board, the CPU should consume the same power for the same clocks), and give numbers which can be comapred from board to board, highlighting the VRM, rather than the CPU. I am not the TPU CPU reviewer, so providing info specific to CPU power draw is not neccesary for me.

Testing via your proposed method would eliminate VRM efficiency numbers, and focus on the CPU, while I'm concerned about the VRM design. I coudl do both sets of testing, but I see no need for me to report specific CPU numbers when I'm not testing CPUs in my reviews.

VRM's can have alot of current shedding by way of thermal output. In fact I'd bet they could consume alot of power, espcially on a highly parrellel design where you ave lots of smal vrms. Sure its going to account for something, but certainly not everything. These boards your getting 280w on, just what are these boards equipped with feature wise ? are you subtracting addition power consumption or are you measuring the vrm to cpu pipeline only ?

 

[cadaveca]

I simply report power draw over the 8-pin, like is shown in the video above.

A bit of research into VRM design, and OEM requirements are that the 12V input to CPU VRM MUST be located within a maximum distance from the VRM itself, to eliminate EMI and line interference(which eliminates the possibility of the 24-pin connector providing input to the VRM). This is then translated in PSU power design, with specific requirements for both 12V inputs via 4-pin and 8-pin connections.

I focus on the 8-pin power draw, as this can highlight what users need to look for in power requirements for multi-rail PSUs.

I don't care about power loss from the VRMs...I think it can be assumed a fact that some of the power consumed via the 8-pin is lost via heat, which, technically, cannot be avoided with MOSFETs. I do not focus on any single part of the VRM, rather I look at the solution as a complete design, as difference in component choice, from high/low MOSFET pairs and triplets, to DRMOS compoenents, may introduce differences that aren't truly comparable unless doen in the method i have chosen. I do make a part of my reviews pointing out the individual components, but I do not feel users need more in-depth analysis than that.

You have to realize, that as it is, my testing takes 25-30 hours to complete, without my usage testing, where I typically use the board for several days, both for general web use, media playback, and gaming, typically done with other users on the forum here.

That said, I am earning just pennies for every hour I put in on some products, so I do have to limit the time I take for analysis. Even so, I do cover things like VRM power draw, which other review sites do not. I am not paid by the hour, so some caveats must be accepted. Really, in the end, I am volunteering my time to do reviews, and that time is ALOT every week, and other than the boards, every other part used in my testing is stuff i purchase myself. With that in mind, there is only so much I can do, although i do think I do more than some others do.

 

[Thatguy]

I simply report power draw over the 8-pin, like is shown in the video above.

A bit of research into VRM design, and OEM requirements are that the 12V input to CPU VRM MUST be located within a maximum distance from the VRM itself, to eliminate EMI and line interference(which eliminates the possibility of the 24-pin connector providing input to the VRM). This is then translated in PSU power design, with specific requirements for both 12V inputs via 4-pin and 8-pin connections.

I focus on the 8-pin power draw, as this can highlight what users need to look for in power requirements for multi-rail PSUs.

I don't care about power loss from the VRMs...I think it can be assumed a fact that some of the power consumed via the 8-pin is lost via heat, which, technically, cannot be avoided with MOSFETs. I do not focus on any single part of the VRM, rather I look at the solution as a complete design, as difference in component choice, from high/low MOSFET pairs and triplets, to DRMOS compoenents, may introduce differences that aren't truly comparable unless doen in the method i have chosen. I do make a part of my reviews pointing out the individual components, but I do not feel users need more in-depth analysis than that.

You have to realize, that as it is, my testing takes 25-30 hours to complete, without my usage testing, where I typically use the board for several days, both for general web use, media playback, and gaming, typically done with other users on the forum here.

That said, I am earning just pennies for every hour I put in on some products, so I do have to limit the time I take for analysis. Even so, I do cover things like VRM power draw, which other review sites do not. I am not paid by the hour, so some caveats must be accepted. Really, in the end, I am volunteering my time to do reviews, and that time is ALOT every week, and other than the boards, every other part used in my testing is stuff i purchase myself. With that in mind, there is only so much I can do, although i do think I do more than some others do.

good mosfest in the upper amperage ranges can easily shed 20% of the power they output by way of thermal loss.