Controversial Topic !

Super Nade · Jun 3, 2005

Hello all,

I've been hearing this Brainpower > JEDEC PCB for quite a bit. However, there seems to be no shred of technical evidence to support these "findings". Hence, I'm calling BS on this one. I believe BP is used because it's cheaper, not better. All this "needless" hype about PCB's is getting a bit old.

I believe there is no way to say BP > JEDEC unless a performance test is made using the same chips or those from the same batch (inaccurate but it may serve our purpose).

So, if you have a theory on why a particular non-standard PCB is better, lets hear it! It would be nice if reps from the RAM co's dropped in and provided their inputs.

I have read this review and it seem very sketchy:
http://www.legitreviews.com/article.php?aid=53

Please try NOT to limit yourself to a one liner

S-N

spydeymon · Jun 3, 2005

Good post id like to see where this thread goes

redwraith94 · Jun 3, 2005

What do you mean by brainpower? Is it some new kind of PCB, and if it is, how is it different?

Super Nade · Jun 3, 2005

redwraith94 said:
What do you mean by brainpower? Is it some new kind of PCB, and if it is, how is it different?

That is the myth I seek to destroy. Brain Power is a Taiwan based PCB company. Their PCB's are supposed to overclock better than the JEDEC PCB's. Apart from one sketchy write-up by legit reviews, I see no more proof.

http://brain-power.net/menus.php?name=Company&c_lang=english

WillysNut · Jun 3, 2005

I love posts that make me go out and learn new stuff. Thanks! I'm guessing you have JDEC based modules? From what I'm seeing in that article from Legit reviews, they realized a noticible difference in moderate and loose timings (in favor of BP) with modules all using the same Samsung TCCD IC's. Tight timings didn't realize much of a difference All modules had the same latencies.

What is your basis for saying it's sketchy? You are correct that it is cheaper to produce the BP design but it seems that there is more to it than cost alone. Whether or not you believe these numbers is your call but I'm certainly seeing a difference between BP and JDEC.

It's the nature of the industry. A standard is only a standard for so long. eventually improvements are made to the standard and we benefit from better performance and a lower out of pocket expense. Do not fear change!

redwraith94 · Jun 3, 2005

Wow, so after a little bit of research, I don't think that their pcb's are anything special.

1. The traces are made of copper, just like any normal pcb. Not silver

2. They said that they could go up to 12 layers, which may or may not be more than normal?

3. The only really novel idea that I saw, is their use of embedded resistors that are apparently manufactured along with the pcb itself, instead of pcb mounted 'normal' resistors, but even though this idea helps slightly with production costs, I would think that it would introduce more noise than a normal pcb mounted resistor. At best it wouldn't be any functionally better.

4. They don't seem very well organized...Shouldn't judge a company based on its website?...but their page is buggy, and their english is poor, which to me indicates that they don't have all that much money to throw around for R&D...a few years ago they only had 100 employees.

They have 6 patents on file w/ our patent office (US):
http://patft.uspto.gov/netacgi/nph-...&Refine=Refine+Search&Query=AN/"brain+power"+
Be warned our fair www.uspto.gov takes a short bit of forever to load.

Super Nade · Jun 3, 2005

@WillysNut

Mate,

I own BP based modules (PDP XBLK). From what I understand, most newer modules use BP PCB's. I find no inclination to believe that BP PCB's are far superior to JEDEC ones, without some serious testing. Now, that brings up the question of "What constitutes serious testing ?". First off, I would like to see some hardcore electrical specs on both PCB's? Surely, resistance, stray capacitance and contact channel length would be nice parameters for a newbie (like me) to comprehend. However, there may be other parameters as well. I want to know what those are and how they match up.

You can't just take two different modules, plug them in and say A > B or vice versa. That to me is a very shaky analysis and sounds half baked.

I read the legit reviews article. See for yourself, the tests are pretty ambivalent!
Have they taken into account binning? I guess not! What exact tests did they run? Many holes in their approach. I find that review unsatisfactory.

Super Nade · Jun 3, 2005

redwraith94 said:
Wow, so after a little bit of research, I don't think that their pcb's are anything special.

1. The traces are made of copper, just like any normal pcb. Not silver
Do JEDEC use Ag?

2. They said that they could go up to 12 layers, which may or may not be more than normal?

Any accont of EMI effects, stray capacitance issues?

3. The only really novel idea that I saw, is their use of embedded resistors that are apparently manufactured along with the pcb itself, instead of pcb mounted 'normal' resistors, but even though this idea helps slightly with production costs, I would think that it would introduce more noise than a normal pcb mounted resistor. At best it wouldn't be any functionally better.

Why would an embedded resistor be more noisy?

4. They don't seem very well organized...Shouldn't judge a company based on its website?...but their page is buggy, and their english is poor, which to me indicates that they don't have all that much money to throw around for R&D...a few years ago they only had 100 employees.

lol. They are Taiwan based and don't have to reach out to the consumer

They have 6 patents on file w/ our patent office (US):
http://patft.uspto.gov/netacgi/nph-...&Refine=Refine+Search&Query=AN/"brain+power"+
Thanks for the links!

WillysNut · Jun 3, 2005

Super Nade said:
@WillysNut
You can't just take two different modules, plug them in and say A > B or vice versa. That to me is a very shaky analysis and sounds half baked.

I read the legit reviews article. See for yourself, the tests are pretty ambivalent!
Have they taken into account binning? I guess not! What exact tests did they run? Many holes in their approach. I find that review unsatisfactory.

I think we are all suffering from a lack of concrete info. I just find it hard to dismiss a difference when most memory manufacturers are moving to BP. Cost is a main factor, but usually not enough alone to justify a company changing processes.

Super Nade · Jun 3, 2005

WillysNut said:
I think we are all suffering from a lack of concrete info. I just find it hard to dismiss a difference when most memory manufacturers are moving to BP. Cost is a main factor, but usually not enough alone to justify a company changing processes.

Agreed. You make a good point with cost. What irks me is people prescribing BP based modules as being the must have's. It is particularly worrying that Newbies are being mis-informed (albeit unintentionally). I fell for it too, untill I tried to dig up some facts and came up with none

WillysNut · Jun 3, 2005

The more I read the more I'm seeing a common thread...

1) Single stick Brainpower PCB is more overclockable than single stick JEDEC

2) Dual stick JEDEC performs better than dual stick Brainpower.

I can see where your concern lies. Everyone is jumping on the BP bandwagon and not getting the results they expected in a Dual Channel environment.

Super Nade · Jun 3, 2005

WillysNut said:
The more I read the more I'm seeing a common thread...

1) Single stick Brainpower PCB is more overclockable than single stick JEDEC

2) Dual stick JEDEC performs better than dual stick Brainpower.

I can see where your concern lies. Everyone is jumping on the BP bandwagon and not getting the results they expected in a Dual Channel environment.

Wow!

I didn't see the DC pattern. But do you have enough sample tests to see a pattern? If you do, then this is very interesting indeed!

redwraith94 · Jun 3, 2005

heheh, no jedec is copper too, I'm just saying that silver would probably be more oc'able due to less noise, and slightly better conductance.

I would think that the embedded resistor would be more noisy because the process and quality control ID think would be as good. Normal carbon resistors, or metal film are usually noisy themselves anyway. I don't have solid evidence that they are worse off, I just don't see them being made to the same quality as surface mount resistors. Even if they are worse off, it won't make much of a difference imo. The resistor thing is a profit saving accomplishment, it does not mean higher quality. I only brought it up because it is definitely different than JEDEC.

My over all point is that I don't personally see any wonderful differences. I will say that the emi interference between layers, and stray capacitence would matter most for pcb construction.

Maybe if they used another material than fiberglass that has a lower dielectric constant, then I would be impressed, that or they give out some solid reviews.

fldrice · Jun 3, 2005

WillysNut said:
The more I read the more I'm seeing a common thread...

1) Single stick Brainpower PCB is more overclockable than single stick JEDEC

2) Dual stick JEDEC performs better than dual stick Brainpower.

Hmm... this could explain some of the unusual results I had with some Winbond UTT sticks.

Supernade,
I agree with you on this one. It seems that BP PCB doesn't really increase OCing performance. I've seen "old skoo" Mushkin LVL II on JEDEC PCB top out @ 275 so I believe overclockability lies within the actual ICs and speed binning.

redwraith94 · Jun 3, 2005

So has anyone here actually used Brainpower's pcbs, and noticed any special differences?

Super Nade · Jun 3, 2005

Let me put it this way, majority of the high performance RAM being made today runs on the BP PCB. But, there is no way to attribute performance boosts to the PCB. That is the essense of my argument. Physically I cannot see any difference (JDEC PCB's are a bit bigger), but I have not pried open the RAM and looked for electrical differences.

Where are those bloody company Rep's when you need them

Audioaficionado · Jun 3, 2005

I just bought two 1GB sticks from OCZ. They were the OCZ Performance 2GB (2 x 1GB) 184-Pin DDR SDRAM Unbuffered DDR 400 OCZ4002048PFDC-K (3-3-3-7 2.6V). I don't have a clue as to which PCB they used but I hope they OC well in dual channel mode. They have been OCed as high as DDR490 in some reviews. I'll see what cards I drew when they arrive next week.

I think it's >90% what the ICs can do plus how they were speed binned and maybe <10% about the PCB.

Super Nade · Jun 3, 2005

Mate,

If you see a BUxxx in tiny letters on the PCB, its Brain Power. I don't thin its even 10% that the PCB plays a role.

Good Luck with your ram!

S-N

Audioaficionado · Jun 3, 2005

Now OCZ says these have ULNT or ultra low noise technogy.

Here's their blog on the subject: "OCZ PC-3200 1GB Unbuffered DDR memory modules featuring ULN2 technology have been designed to provide maximum bandwidth and performance in today's demanding PC applications."

I'll find out what they are if I can read around the heat spreaders. They use copper so at least they have a cool geek factor.

El<(')>Maxi · Jun 3, 2005

Super Nade said:
Where are those bloody company Rep's when you need them

What do you need them for? It's not like your gonna get solid info out of those guys half the time either

I think Willy might be onto something as I had some BP'd matched sticks that did not overclock as well in DC as a pair of old Mushkin LII.....Both BH-5

Nothing concrete but it does make you wonder.

Controversial Topic !

† SU(3) Moderator †

Member

Member

† SU(3) Moderator †

Senior Jeep Nut

Member

† SU(3) Moderator †

† SU(3) Moderator †

Senior Jeep Nut

† SU(3) Moderator †

Senior Jeep Nut

† SU(3) Moderator †

Member

Member

Member

† SU(3) Moderator †

Sparkomatic Moderator

† SU(3) Moderator †

Sparkomatic Moderator

Blank Senior Member

Similar threads