Is branded memory all it's cracked up to be?

I ask because I have two 512MB sticks of cheap generic PC3200 memory
in my machine, unmatched. Ran Memtest for 5 hours with zero errors.

So why the hype about branded?

Or am I just lucky?

D-Dan wrote:
I ask because I have two 512MB sticks of cheap generic PC3200 memory
in my machine, unmatched. Ran Memtest for 5 hours with zero errors.

So why the hype about branded?

Or am I just lucky?

I once ran MemTest86 (I don't know if that's the same as MemTest) for
4-5 hours with zero errors, but Gold Memory (www.goldmemory.cz) found
errors in ten minutes (MemTest86 eventually reported the same errors).
A www.realworldtech.com review said that both tests were good but that
Gold was better, but I've had other modules that tested fine with Gold
but failed with MemTest86, so I always test with both.

Do you mean branded modules or branded chips? Because many branded
modules, including some from Kingston, PNY, Centon, Spectek, K-byte,
Mustek, and Corsair, contain chips that lack the chip manufacturer's
normal brand markings because they're recycled or non-prime, and these
are the only kind of modules that have failed on me (Kingston, K-byte)
or couldn't be overclocked reliably at all, including Kingston (no-name
chips) and Mustek (Spectek chips, but a Spectec brand module with
Spectec chips overclocked well). Some chip brands, including Spectek,
are for recycled chips, but Elixir seems to be Nanya's brand for it's
less-supported chips, and I found them to be very overclockable, just
like PNY modules with Bravo chips, but I don't know anything about
Bravo.

D-Dan wrote:
I ask because I have two 512MB sticks of cheap generic PC3200 memory
in my machine, unmatched. Ran Memtest for 5 hours with zero errors.

So why the hype about branded?

Or am I just lucky?

I once ran MemTest86 (I don't know if that's the same as MemTest) for
4-5 hours with zero errors, but Gold Memory (www.goldmemory.cz) found
errors in ten minutes (MemTest86 eventually reported the same errors).
A www.realworldtech.com review said that both tests were good but that
Gold was better, but I've had other modules that tested fine with Gold
but failed with MemTest86, so I always test with both.

Do you mean branded modules or branded chips? Because many branded
modules, including some from Kingston, PNY, Centon, Spectek, K-byte,
Mustek, and Corsair, contain chips that lack the chip manufacturer's
normal brand markings because they're recycled or non-prime, and these
are the only kind of modules that have failed on me (Kingston, K-byte)
or couldn't be overclocked reliably at all, including Kingston (no-name
chips) and Mustek (Spectek chips, but a Spectec brand module with
Spectec chips overclocked well). Some chip brands, including Spectek,
are for recycled chips, but Elixir seems to be Nanya's brand for it's
less-supported chips, and I found them to be very overclockable, just
like PNY modules with Bravo chips, but I don't know anything about
Bravo.

On 28/9/05 1:25 pm, in article
, "larry moe 'n curly"
wrote:

D-Dan wrote:
I ask because I have two 512MB sticks of cheap generic PC3200 memory
in my machine, unmatched. Ran Memtest for 5 hours with zero errors.

So why the hype about branded?

Or am I just lucky?

I once ran MemTest86 (I don't know if that's the same as MemTest) for
4-5 hours with zero errors, but Gold Memory (www.goldmemory.cz) found
errors in ten minutes (MemTest86 eventually reported the same errors).
A www.realworldtech.com review said that both tests were good but that
Gold was better, but I've had other modules that tested fine with Gold
but failed with MemTest86, so I always test with both.

Do you mean branded modules or branded chips? Because many branded
modules, including some from Kingston, PNY, Centon, Spectek, K-byte,
Mustek, and Corsair, contain chips that lack the chip manufacturer's
normal brand markings because they're recycled or non-prime, and these
are the only kind of modules that have failed on me (Kingston, K-byte)
or couldn't be overclocked reliably at all, including Kingston (no-name
chips) and Mustek (Spectek chips, but a Spectec brand module with
Spectec chips overclocked well). Some chip brands, including Spectek,
are for recycled chips, but Elixir seems to be Nanya's brand for it's
less-supported chips, and I found them to be very overclockable, just
like PNY modules with Bravo chips, but I don't know anything about
Bravo.

I've done plenty of builds using quality, value and basic bulk buys of
memory and I've noticed that most of the chips on the memory are made by the
same company all that's different is a pretty cooling block they sometimes
clip to the outside.
--
Nik
http://www.pc-gear.co.uk

Nik Simms (Web Developer) wrote:

many branded
modules, including some from Kingston, PNY, Centon, Spectek, K-byte,
Mustek, and Corsair, contain chips that lack the chip manufacturer's
normal brand markings because they're recycled or non-prime, and these
are the only kind of modules that have failed on me

I've done plenty of builds using quality, value and basic bulk buys of
memory and I've noticed that most of the chips on the memory are made by the
same company all that's different is a pretty cooling block they sometimes
clip to the outside.

How do you tell when the chip packages don't list a real part number
(instead of just something like "32M x 8 - 7")? Do you use a memory
testing machine or a special utility? Because I've tried Everest,
CTspd, and CPU-Z, but apparently they only read the SPD, which often
wasn't programmed with correct information (or any information) about
the chip brands.

"D-Dan" wrote in
message ...
I ask because I have two 512MB sticks of cheap generic PC3200 memory
in my machine, unmatched. Ran Memtest for 5 hours with zero errors.

So why the hype about branded?

Or am I just lucky?



PC3200 *should* run as PC3200 regardless.
Sometimes more expensive, branded, memory is able to run at speeds much
higher than they are rated at. Also branded memory usually gives you a
guarantee of certain memory timings (CAS to RAS, CL, interleave, and
precharge timings) whereas generic memory makes no guarantee and sometimes
runs at very slow timings.
Just as an example, a while back there was a major run on Samsung branded
memory (still considered to be very good). Some people were able to get
PC3200 sticks to run at PC4200 speeds (or something like that). Also
consider that some *brands* are more hype than performance. A few years ago
OCZ got a bad rep. when they were found to be taking everyday generic sticks
of RAM, slapping OCZ heatsinks on them and selling them at premium prices. I
think they have since cleaned up their act, though (not sure).
--


"I don't cheat to survive. I cheat to LIVE!!"
- Alceryes

Well, for one, with unmatched sticks you CANNOT be running in dual channel
mode.

--
DaveW
__________

"D-Dan" wrote in
message ...
I ask because I have two 512MB sticks of cheap generic PC3200 memory
in my machine, unmatched. Ran Memtest for 5 hours with zero errors.

So why the hype about branded?

Or am I just lucky?

On Wed, 28 Sep 2005 16:15:49 -0700, "DaveW"
wrote:

Well, for one, with unmatched sticks you CANNOT be running in dual channel
mode.


Wouldn't it be good to know what the system is first, before
making such a sweeping assumption? I have at least 3 boxes
with "unmatched" (actually different brands) of memory,
running fine in dual channel mode.

Actually - my MB doesn't support dual channel - but from what I've
read - it doesn't make much difference.

For the record - ASUS A7N8X-X Athlon XP2400+ system. Dual channel
would be throttled by the CPU anyway.

On Fri, 30 Sep 2005 02:25:24 -0500,
lid
(D-Dan) wrote:

Actually - my MB doesn't support dual channel - but from what I've
read - it doesn't make much difference.

Depends on the board (chipset), usually that's true, only a
few (low single-digit percent) difference unless the board
has integrated video... in which case the dual-channel
memory makes for a very significant increase in (potential)
video performance.



For the record - ASUS A7N8X-X Athlon XP2400+ system. Dual channel
would be throttled by the CPU anyway.

? By FSB type (multi-data rate design) yes. Not really
CPU.