Wicked3d Uses 28NS EDO Ram...WHY???
Jeff Greeson
The Metabyte Wicked3D uses OKI 28ns RAM chips, which I hear are slower
than Silicon Magic's 25ns chips. Now why would a board manufactorer go with
28ns chips? Is it because of the costs? Or is there not a significant
performance/stability issue going with 28ns over 25?
I admit, I got sucked in by Metabyte's driver level stereo support and
extended resolutions. So far I enjoy the ability to play with these new
gems Wicked3D worked into their Voodoo2 board. I hope that they hold their
end of the deal and continue to make strong improvements in their code. It
is obvious that Wicked3D put a strong emphasis on their software, however it
looks like they cut too many a corner with their hardware. Not only do they
ship with 28ns ram, the pass-through cable included looks exactly like
Creative Lab's Dxr2's(dvd encoder) pass through. It is very thin, and lacks
the "girth" I've seen in Orchid R3DII's box.
So what am I to do? Should I overlook the fact that I'm being cut at
several corners in the hardware department? Or does Wicked3D's extended
drivers provide something that cannot be done in the future with boards,
overshadowing the cuts in hardware features?
At least I have 23 days left to make that decision.
Joe M
This has been covered in a previous post, but here goes again:
I too received a wicked board with 28 ns Oki. I'm not the slightest
bit worried, here's why. Oki is a technology partner with Silicon
Magic.
You could think of Oki as a manufacturing arm of SM.
More importantly, after testing many, many 3DFX products (Rush, voodoo)
with all kinds of memory chips, various speeds, vendors, I found that
one must not
take too literally the numbers on the package. Two years ago, SM 35 ns
were tops.
A year later, they were much slower even than their own chips made a
year
earlier. I tested Mosel Vitalec (the best EDO), Samsung, Etron Tech,
Oki,
SM, another Tiawanese brand with just initials which I can no longer
remember (the worst). I tested every speed rating from 28 ns SM, to 50
ns
of everyone, mostly I tested in the -30 to -35 range. The best was the
-35 Mosel Vitalec, better than SM -28, and even they were different
depending
on which date code I tested. Further tests on 2D boards, mostly by S3,
confirmed.
Point is, The lab test results varied all over the map, and they
typically outperformed
the advertised numbers. Those numbers, while not totally arbitrary,
just don't mean that much. The layout of the board, getting the right
termination
resistors (especially on CAS and MA 0-8) are far more important.
3DFX has a very large testing program dedicated
to tweaking their reference designs, both in house and contracted,
and they are a wealth of knowledge in knowing which chips, termination
values, layout guide lines, board charateristic impedance, board
topology
etc. are optimum.
Perhaps one of the board vendors like Creative Labs (Dylan, are you
listening?)
can back me up on this. Did I leave anything out?
Joe M
ernie
data just to be sure. Regardless of whether or not the memory is
equivalent (which is what I'm sure they will tell you), IMHO Metabyte
made a big marketing (notice I didn't say anything about performence)
blunder by switching to 28ns memory......if you advertise a product at
25ns, the consumer should expect to open the box and find the product as
advertised. I've been wavering as to whether or not to return the board,
since I doubt performance takes that much of a hit (but I'm no expert),
but each post I see about this topic hacks me off even more.......so
back it goes! There are less expensive boards out there, and people seem
quite happy with even the reference drivers for now (and these will
improve with time anyways).
Happy Monday!
ernie
>
Jeff Greeson
same in performance" and that OKI rates their chips "higher" than what they
can perform. However I do not want overclocked Memory chips! Yes you may
say, but what is the big deal? You have a voodoo2 card, and it works and
performs great. Well the deal is that I'm out $500 and it's stuck here with
28ns RAM Chips. I payed for what is on the back of the box. "25ns EDO
DRAM" and I will get "25ns EDO DRAM" I don't believe the story that Silicon
Magic's didn't have enought inverntory, and they had to go with the OKI RAM,
because they said that they have Silicon Magic boards there and that I can
exchange mine anytime during my 30 day "preview of 28ns ram chips workling
like 25ns chips" (ok maybe they didn't say that) Perhaps the profit margin
during their BIG SALE! was so low that they needed to find somewhere to
regain some ground lost.
Also ernie, I'm a Comet also. I'm glad to see someone that's into this
stuff up at UTD.
ernie wrote in message <35CF1EBB...@utdallas.edu>...
Dylan Rhodes [Creative]
Joe M wrote in message <35CF28FC...@uswest.com>...
>Perhaps one of the board vendors like Creative Labs (Dylan, are you
>listening?)
>can back me up on this. Did I leave anything out?
You're correct that Jeff has nothing to worry about. The Oki 28 chips are
bin sorted to 25ns.
-- Dylan
Dave
out on hooking my system up to diferent monitors with the same results
=( I think I give up, I'm not sure I'll keep the cards since they cost
quite a bit more then others out there. I have seen 12 meg CL for 199.00
with bundle. The stereo image in thier demo looks perfect with no
ghosting, I wonder whats up. Does Rendition version have this effect
too? The technology is so unbelievably cool but this is a major bug if
there are only one or two monitors in the world that dont cause this.
What really gets me is how MetaByte and H3D dont even acknowledge this
problem when I talked to them on the phone even though its on H3D web
site!
I benched a single card (had to actually take the SLI cable out) and the
results are pretty poor. At first I was getting 44 FPS in Quake2 but on
the third demo it dropped to 32! Wierd, Thats a huge drop in FPS from
85 in 1024X768 SLI. Its more thennthe resolution and must have to do
with the way the wrapper works. I was tempted to keep one card but not
after seeing that. It would have to be both or none at all. If I can get
the H3D working good I'll keep it all! CYA
Still no luck returning your Email BTW
ernie
25ns...huh? Maybe someone could explain that one to me. In any case, the person
I spoke with seemed nice (actually returned a phone mail message within 10
minutes), and said he'd speak to the necessary folks about posting something to
this newsgroup.
ernie
ernie
Don't want you to do Metabyte's job for them, but I saw this answer you gave
on an earlier post and I don't really understand it. The OKI chips are "bin
sorted" to 25ns, but are stamped at 28ns. If it walks like a duck and quacks
like a duck, why isn't is stamped "DUCK"? : )
Seeking enlightenment......ernie
PS Since you don't work for brand M, I won't hold it against you if you see
fit not to go into this further. The fact that you take the time to post to
non-CL questions puts you a few notches above most......thanks!
Dylan Rhodes [Creative]
ernie wrote in message <35CF50E9...@utdallas.edu>...
>Hi Dylan:
>
>Don't want you to do Metabyte's job for them, but I saw this answer you
gave
>on an earlier post and I don't really understand it. The OKI chips are "bin
>sorted" to 25ns, but are stamped at 28ns. If it walks like a duck and
quacks
>like a duck, why isn't is stamped "DUCK"? : )
I don't know (I'm not a real expert in this area), but my guess is that it
would be cost-prohibitive to take the bin-sorted memory, remove the old
stamping, and restamp them.
-- Dylan
Jarrod Smith
ernie wrote in message <35CF50E9...@utdallas.edu>...
>Hi Dylan:
>
>Don't want you to do Metabyte's job for them, but I saw this answer you
gave
>on an earlier post and I don't really understand it. The OKI chips are "bin
>sorted" to 25ns, but are stamped at 28ns. If it walks like a duck and
quacks
>like a duck, why isn't is stamped "DUCK"? : )
>
Ernie-
I understand your concern, but *if* it is in fact true that these chips have
been sorted, then you don't have anything to worry about. I don't know much
about RAM specs, but any type of electronics part is designed to a spec
which has built in tolerances. Some of the parts will be at the top of the
heap in terms of performance, some at the bottom. The manufacturer must go
through a statistical sampling and analysis of the lot, and then mark the
chips or whatever accordingly. This always leads to some of the parts being
underrated. It is very likely that this is what OKI does. They stamp all
the parts with 28ns knowing full well that some of them will perform better
than that. But there is a tradeoff between how much more they can sell the
faster chips for vs. how much it costs to test and locate the faster ones
out of a large batch of chips.
One area that I do have experience with is NiCad batteries. Some NiCads
just perform better than others, even if they were run off the same
production line in the same batch. It is time consuming (expensive) to test
each one, so some companies who make rechargeable battery packs do just this
to turn a profit. They buy unrated battery cells from Sanyo for cheap in
bulk, then they test every one of them, stamp them with their voltage and
charge capacity, put the batteries that performed similarly in packs with
one another, and then resell the really good packs at a huge profit, while
selling the less good packs at break-even or maybe even at a loss. I know,
batteries aren't RAM, but the concept is similar so I thought I'd relate the
story.
Hope you find satisfaction with your purchase, however it comes about :-)
Jarrod Smith (jsm...@scripps.edu)
The Scripps Research Institute
http://www.scripps.edu/~jsmith
Gamo
"The Voodoo2 CE is a cost-reduced version of the 3D Blaster Voodoo2 which we
sell in markets where the exchange rate makes the original 8MB and 12MB
models cost-prohibitive for some customers. The Americas division of
Creative Labs is selling the Voodoo2 CE in Canada and Latin America, but not
the United States.
The 3D Blaster Voodoo2 CE has a reduced software bundle. It also has a
lower clock speed of 83MHz, and it uses 28ns DRAM instead of the 25ns DRAM
found on our other Voodoo2 models."
Did you pay less for your Wicked3D card than people who bought cards with
25ns RAM? I think not so you better start complaining to Wicked 3D.
Gamo
-- Dylan
Jeff Greeson wrote in message <6qm1dv$5db$1...@news.3dfx.com>...
Ed Blum
Ummm maybe...
But as described here its not how I have seen it done.
Nor does it always have the desired results., basically they
(the chip manufacturer) run the chips through a automated
sorter that hammers them up in speed until the timing starts
to fail, back down one click in speed, test again and if it passes
it does to the automatic marking machine, as a 25 Nsec part, and
is marked as such.
My experience has been that the chips, even consumer product chips,
are ALWAYS marked with the Minimum speed in nanosecnds at which
they will reliably perform as provided by the original vendor.
The next level of "bin sorting" as described here is where the BOARD
manufacturer does the sorting themselves, and uses the chips which
pass whatever sorting test was decided was sufficient by the onsite
component/VQME/product engineer. These devices may not
neccessarily meet the full width Specification, but will "work" in the
application,
"somewhat good enuf"... Some chip manufacturers might do this for the
board manufacturer..or some garage shop might take a ton of low speed
marked chips, "bin sort" them to some portion of the entire spec, and
move them out for a cut rate price to whoever is looking for a low price.
Our fax machine churns out these offers all day long.
I doubt a reliable company like OKI would tolerate this inhouse as it
leaves them open to questionable business practices issues.
So if you got 28 Nsec ram on your card and the box says 25Nsec, no matter
what they claim.. you got ripped off, and your card might not meet the best
possible performance specs particularily if you run above the "default'
settings.
I would tell you how some card manufacturers set they're ACTUAL RAMDAC Speed
in
board EEPROM during test (it aint whats on the box kids).....but that would
REALLY
piss some of you off!!!!
heh
<remove spamguards before replying>
But my experience is it it works the other way.... a 25Nsec
device
Jeff Greeson wrote in message <6qnd7m$jck$1...@news.3dfx.com>...
> I'm fresh off the phone, and of course I got the "they're exactly the
>same in performance" and that OKI rates their chips "higher" than what they
>can perform. However I do not want overclocked Memory chips! Yes you may
>say, but what is the big deal? You have a voodoo2 card, and it works and
>performs great. Well the deal is that I'm out $500 and it's stuck here
with
>28ns RAM Chips. I payed for what is on the back of the box. "25ns EDO
>DRAM" and I will get "25ns EDO DRAM" I don't believe the story that
Silicon
>Magic's didn't have enought inverntory, and they had to go with the OKI
RAM,
>because they said that they have Silicon Magic boards there and that I can
>exchange mine anytime during my 30 day "preview of 28ns ram chips workling
>like 25ns chips" (ok maybe they didn't say that) Perhaps the profit margin
>during their BIG SALE! was so low that they needed to find somewhere to
>regain some ground lost.
>
ernie
in the coffin), Jarrod, Ed, for the responses. This thread's gotten long, so
here's my summary take on the information provided.
1. 28ns RAM is cheaper to buy, but each batch has some chips that are "better"
than others, i.e., will spec better than 28ns under certain testing conditions.
2. Bin sorting of the 28ns RAM to screen for chips performing "like" 25ns
(under certain testing conditions) was most likely done in this case by the
QA/testing division of the vendor (Metabyte or someone contracted by them, but
probably not OKI). These tests may not be entirely the same as those originally
used by OKI but, rather, may include tests more germane to the application (in
this case, a v2 video card).
3. It is possible (likely) that the 28ns RAM bin-sorted to perform to some level
"like" 25ns may not have the "reserve" capacity that RAM originally rated by the
manufacturer at 25ns would, i.e., may not be able to perform as long at higher
temps, currents, or whatever conditions have been used for testing.
4. Although you may never see a difference between 25ns and 28ns RAM on your
current system or using current applications, the product you have purchased
contains "inferior" components (by whatever slim margin) that are less
expensive.
5. If you were charged the same for your 28ns product as for the 25ns product,
someone made a few extra bucks on you.
Thanks to all who have posted....this has been extremely informative, both from
a technical and a consumer point of view. As I said in my original post, this is
probably more of a consumer/advertising/marketing issue than it is a
technical/performance issue (though by no means exclusively).
Have a nice day!
ernie
Joe M
>
> The next level of "bin sorting" as described here is where the BOARD
> manufacturer does the sorting themselves, and uses the chips which
> pass whatever sorting test was decided was sufficient by the onsite
> component/VQME/product engineer. These devices may not
> neccessarily meet the full width Specification, but will "work" in the
> application,
this is wrong. For them to work in the application, one would haveto bin sort
them while they are running the application, i.e. while they
are on the board. But these are surface mount parts, soldered
down. Therefore, they would not be bin sorted as you describe.
What you say here may be true at some company somewhere,
but in the context of a multimedia company making add in cards,
this makes no sense at all.
Here's how it really works: The sales reps from the chip vendors
meet with the engineering manager and tell him how wonderful their
chips are. Skeptical, the manager says, fine, give me samples and we will
test. The design engineer/test engineer/compatibility lab manager,
takes the samples and builds a few prototypes for testing. After testing
many boards with various chips, interesting facts become known.
Like a -35 Mosel Vitalec RAM is faster than a -28 Silicon Magic.
(for example). In this case, 3DFX would take a great deal of the responsibility
for this type of testing, because they want their board vendors
to be successful. You can bet 3DFX was heavily involved with Metabyte's
decision to put -28 Oki chips on their board.
> So if you got 28 Nsec ram on your card and the box says 25Nsec, no matter
> what they claim.. you got ripped off, and your card might not meet the best
> possible performance specs particularily if you run above the "default'
> settings.
No, read my post this thread.
>
>
> I would tell you how some card manufacturers set they're ACTUAL RAMDAC Speed
> in
> board EEPROM during test (it aint whats on the box kids).....but that would
> REALLY
> piss some of you off!!!!
what are you talking about? The RAMDAC speed is not "set".The RAMDAC speed
(pixel clock) is purely a function of the screen resolution,
vertical refresh period, and blanking interval (horizontal and vertical). The
RAMDAC speed is a variable, not a constant that is "set".
Very misleading post.
Joe M
Joe M
Joe M
> (under certain testing conditions) was most likely done in this case by the
> QA/testing division of the vendor (Metabyte or someone contracted by them, but
> probably not OKI).
Just for the record, this is not true. The board vendors would never bin sort the
memory chipsBEFORE they arrived in incoming inspection. This makes no sense
as I explained in my other post. Any bin sorting would be done by,
or under authority of, the chip vendor. The notion of "bin sorting by
application " is just plain wrong, doesn't happen.
> Have a nice day!
> ernie
You too, ernie.
Joe M
ernie
Thanks for the correction.....that's why I posted my summary, just to make sure I
understood.
Cheers.......ernie
Jeff Greeson
Anyways, here are my "experiences" with Wicked3D over the phone.
Call #1: "Hi I would like to arrange a board swap." Ok the person you will
need to talk to is the "C" Can I get your name.... [ON HOLD] Sorry "C" is
not in right now, may I give you his voice mail? "Yes Please" [voice mail
message explaining my discontent in the ram chips on my board including
phone number]
[2 hours pass]
Call #2: "Hello, I left a voice mail message with "C" and I haven't
received a call back yet." [different person] Oh I'm sorry he's not here at
the moment. Do you want his voice mail? "NO! I already went through his
voice mail and he never responds to it." OK, well if you could give me your
name and number, I will give this message to him. "Fine..."
[3 hours pass]
Call #3: "Hello I have called your company 2 times and have left numerous
messages with "C" to have him call me back regarding a board swap. This has
all been done long distance, and I'm very angry. Can I please speak with
"C" NOW?" Let me transfer you. [on hold] Well it seems like he's not
there, let me find him in the building. [on hold for 5 minuets] Well I got
up to him, but he ran off as soon as I could talk with him. We're very
understaffed, especially today (monday), and he has been doing pretty much
everything around here. I can have you talk to our technical support team
if that will help. "Fine..." Yes I would like to exchange my boards for a
board that has 25ns rated chips like your website and box says. "Ok all I
can do is take your name and number and I give this to my manager." {end}
[Today]
Call #4: "Can I please speak with "C"? Can I get your name...[on hold] He's
not here can I give you his voice mail? "Yes" [left name a phone number]
Ok well it looks like I've left 4 messages with this guy that's supposed to
be in charge of this kind of problem. And I haven't received 1 phone call
from them back. This is extremely frustrating. They will no get out this
easily. I will not go away. They cannot push me out of this. I WILL
PREVAIL!
Dylan Rhodes [Creative] <drh...@soundblaster.com> wrote in message
6qnns1$n4v$1...@news.3dfx.com...
>
>ernie wrote in message <35CF50E9...@utdallas.edu>...
>>Hi Dylan:
>>
>>Don't want you to do Metabyte's job for them, but I saw this answer you
>gave
>>on an earlier post and I don't really understand it. The OKI chips are
"bin
>>sorted" to 25ns, but are stamped at 28ns. If it walks like a duck and
>quacks
>>like a duck, why isn't is stamped "DUCK"? : )
>
Dylan Rhodes [Creative]
Gamo (probably unintentionally) quoted my post out of context, giving the
impression that I am in support of the position that a Wicked3D board with
memory chips marked "Oki 28" is inferior to a Wicked3D board with 25ns
Silicon Magic memory. In fact, the opposite is true; I've been doing my
best to convince Jeff that he has nothing to worry about.
You've also made the incorrect assumption that the chips labelled Oki 28
cost less for Metabyte to buy than do other approved memories. Your
"someone made a few extra bucks on you" statement is incorrect when applied
to the Wicked3D.
Let me put the 28 nanosecond concerns to rest for the Wicked3D owners.
Silicon Magic is a fabless design house. That means they design memories and
then have fabrication plants, or "fabs," build them. As a part of the deal
they had with OKI (the fab), they gave OKI a license to the design. Since it
was a silicon Magic design, Silicon Magic got to market with it first. A few
months after Silicon Magic got to
market with the part, OKI started to make the part and market it under its
brand.
The die (the actual silicon) is exactly the same for the Silicon Magic -25
and the OKI -28. the performance is exactly the same. OKI, however, chooses
to mark the part
more conservatively than Silicon Magic. In fact, the OKI part was the second
part to be qualified by 3DFX. It qualified easily because it is the same
physical design and the same physical process. 3Dfx was extremely
conservative in qualifying memory. In order to be placed on the AVL
(approved vendor list ) the memory had to be able to run at speed (>90MHz)
at margin-- that means 55C, 4.75 volts. That's one hot system!
The other point is that every memory vendor speed sorts: Silicon Magic speed
sorts, OKI speed sorts, Mosel speed sorts. Every vendor also has different
marking criteria. Some mark parts on the high side; others, like OKI are
more conservative. For example, even though the Silicon Magic parts are
marked as 100MHZ, that does not mean that very part will run at 100MHz.
So, let me just end this with one emphatic note: the silicon magic -25 and
the OKi -28 are the *same part*. Same design, same silicon, same fab, same
screening test, same 3Dfx qualification test -- just a different package.
Jeff, you'll gain nothing by insisting that Metabyte replace your card with
one with Silicon Magic memory. However, they'll have to take back a board
which they can't resell. This will increase their operating costs, and
increased operating costs are what cause vendors to raise prices or to cut
development budgets.
--
Dylan Rhodes: Product Marketing Manager, Graphics and Video
Creative Labs, Inc.
Joe M
> one with Silicon Magic memory. However, they'll have to take back a board
> which they can't resell. This will increase their operating costs, and
> increased operating costs are what cause vendors to raise prices or to cut
> development budgets.
>
> --
> Dylan Rhodes: Product Marketing Manager, Graphics and Video
> Creative Labs, Inc.
Great, great post Dylan. Maybe this message coming from one of the 'Saints'
of the newsgroup will make an impression. thanks for going into so much
detail.
Joe M
Jeff Greeson
with Wicked3D. The main factor in my decision was the fact that I had been
throughly rough housed by Wicked3D's support line. (see post: cutomers like
dirt) Once I had my foot in the door about these chips, I had to follow
through. I wish I had read this post 5 minuets earlier, but tracking
numbers have been swapped.
The whole issue wasn't clear with me. I kept hearing that the OKI chips
where bin sorted, hand picked from a pile among many other things. If I
knew that the OKI design was based on the Silicon Magic design, then I would
be somewhat happy. But still, it doesn't set my mind at peace knowing that
my board has OKI chips, while there is a majority of boards out there that
have SM chips.
I layed down $500 of my money expecting a 3dfx Voodoo2 board with
Silicon Magic 25ns RAM chips. Even the back on the box features a picture of
the card with Silicon Magic chips. Maybe they should have put a bunch of
Wicked3D stickers over the RAM chips before they took that picture. What
I'm still questioning is whether the OKI chips are cheaper than the SM
chips? I'm going on a hunch that the OKIs are cheaper, but may not be
cheaper in quality. If these OKI chips need special tweaking to work with
the Voodoo2, while the SM chips do not need this tweaking, then how can one
expect these chips to be a "clone" of what a SM board can do?
It probably all boils down to being able to sleep at night, knowing that
the Voodoo2 boards in my machine have 48 Silicon Magic stamped DRAM chips on
them. It's just that the knowledge I was searching for wasn't available
from Wicked3D. I got the generic answer and wanted the technical answer and
had insisted the technical answer 5 seperate times via long distance.
Through having W3D take a hit on not being able to resell my boards, I hope
Wicked3D learns that something needs to be done about their support lines.
Jeff Greeson
Administrator of The Realm of Mortal Kombat @ http://mk.hotweird.com
Dylan Rhodes [Creative] <drh...@soundblaster.com> wrote in message
6qq5cp$k7s$1...@news.3dfx.com...
ernie
this answer a mile or so up this thread when I originally asked the question (or
even on an earlier thread about 4 miles up this board), the issue might have met
an early death. : ) On the other hand, your posts help a lot of us, and it has
made for some good conversation.
I guess you just have to push the right buttons for someone to come clean with
the goods (though it was not intentional....I merely wanted to summarize the
thread). If I had known this, I probably would have kept the board (don't worry,
I didn't even open the antistatic bag when I saw the memory, so they can resell
it). As I mentioned to Joe M in an email, why is that Metabyte can't tell it's
customers this when they ask? I can appreciate it if they don't want to blindly
advertise something like this, as it might be misinterpreted (Hmmm, we just did
that, didn't we?). However, I phoned them four times, emailed once, seeking an
explanation. My tone was always cordial. Spoke with a real person twice (in tech
support), and none were forthcoming with the detailed information you posted. No
response from the email sent Friday. Since no one could provide an explanation,
I had no choice (because they gave me no information) but to return the board.
Maybe Metabyte will honor you as wicked Fireman of the Year : )........don't
beat me, couldn't resist.
Thanks again........ernie
Dylan Rhodes [Creative] wrote:
> Ernie,
>
> Gamo (probably unintentionally) quoted my post out of context, giving the
> impression that I am in support of the position that a Wicked3D board with
> memory chips marked "Oki 28" is inferior to a Wicked3D board with 25ns
> Silicon Magic memory. In fact, the opposite is true; I've been doing my
> best to convince Jeff that he has nothing to worry about.
>
> You've also made the incorrect assumption that the chips labelled Oki 28
> cost less for Metabyte to buy than do other approved memories. Your
> "someone made a few extra bucks on you" statement is incorrect when applied
> to the Wicked3D.
>
> Let me put the 28 nanosecond concerns to rest for the Wicked3D owners.....
>
>
Joe M
> numbers have been swapped.
>
> The whole issue wasn't clear with me. I kept hearing that the OKI chips
> where bin sorted, hand picked from a pile among many other things. If I
> knew that the OKI design was based on the Silicon Magic design, then I would
> be somewhat happy.
5 minutes earlier? Twice in the past 4 days, I have posted on this topicsaying,
in essance, the same thing Dylan has now said. Especially the part
about Oki being the manufacturing arm for SM. I suppose that now that Dylan
has confirmed, people will believe me.
> But still, it doesn't set my mind at peace knowing that
> my board has OKI chips, while there is a majority of boards out there that
> have SM chips.
Why?? Are Oki chips diseased?
> What
> I'm still questioning is whether the OKI chips are cheaper than the SM
> chips? I'm going on a hunch that the OKIs are cheaper, but may not be
> cheaper in quality.
No, no, no. that isn't it at all. Each company negotiates a deal with a
chipvendor/distributor, they don't buy from the chip house, they buy from
distributors. These chips are a commodity, so the price varies as a
function of market conditions, quantity, and how good of a negotiator
you are. It matters little if they are Samsung, Oki, Mosel Vitalec.
You are on the wrong track here. It's just like I said in earlier posts,
and now Dylan has said it: Oki=SM, price of the chips is not the issue.
> If these OKI chips need special tweaking to work with
> the Voodoo2, while the SM chips do not need this tweaking, then how can one
> expect these chips to be a "clone" of what a SM board can do?
What?? Why do you think Oki chips need special tweaking? You are onthe wrong
track. Did you read Dylan's post??
>
>
> I hope
> Wicked3D learns that something needs to be done about their support lines.
And I hope the inconvience you put yourself through for absolutelyno reason or
benefit to yourself whatsoever makes you use common sense
next time.
Joe M
Jeff Greeson
advice, however there is no need to emphasize that you were right by
insulting me. I may have caused them trouble, however they couldn't quite
convince me why I should be happy with OKI ram chips. I'll try to tie this
to an analogy. Chevrolet and GMC both make the Suburban line of trucks.
They's initially the same car. Except one has the ugly GMC logo in the
front and the back of the truck, while the other has a Chevrolet logo on the
front and says Chevrolet on the back. I'd rather have the Chevrolet than a
GMC, even though it may say something different on the outside, at least I
know I'm getting the same as anyone with a GMC. All I'm looking for is
peace of mind, and I'm willing to part with my voodoo2s for 2 days to
acheive that.
Joe M wrote in message <35D0E1CB...@uswest.net>...
>
>5 minutes earlier? Twice in the past 4 days, I have posted on this
topicsaying,
>in essance, the same thing Dylan has now said. Especially the part
>about Oki being the manufacturing arm for SM. I suppose that now that
Dylan
>has confirmed, people will believe me.
>
>Why?? Are Oki chips diseased?
>
Joe M
> I'm sorry you need to feel so defensive that I didn't take your eariler
> advice, however there is no need to emphasize that you were right by
> insulting me.
Jeff, I made a mistake, I'm sorry if I insulted you. I guess the reason we
docomputers/games in the first place is to feel good, and that goes for
the hardware we buy. even if it is true that the RAM issue got you worked
up for no reason, in your mind, there WAS a reason, and that is all that
matters. So, I sincerely hope you will not think badly of me cause
of the way I responded. You have a right to obtain happiness/peace
of mind no matter how that comes about. Please understand that part of my
frustration
is that I have been saying exactly what Dylan said, and no one took
notice. then Dylan speaks, and suddenly everyone understands.
Of course he is a guru, and I'm just some guy, so what do I know?
Anyway, sorry if I insulted you, and I take back anything offesive
I may have said.
regards,
Joe M
Jeff Greeson
to have reaffirming faith by someone in the industry. You saw how many
people posted to this thread, and how many misleading posts were made.
There's just something about knowing the information you just received is
backed by a position at a well known company. Until we all can become
knowledgable in every area of computers, we're going to have to rely on the
ones that do this for a living.
Jeff
Joe M wrote in message <35D0F02A...@uswest.net>...
a
It is true that OKI memory chips are labeled as 28ns parts. However,
the chips are bin sorted (a process of testing which qualifies a part). An
example of a bin sort may be as follows:
1) "Company X" makes 1,000 units of memory, and labels them with their
standard ink, which says "28ns".
2) "Company X" tests the memory. 990 units of memory pass a test to qualify
them at 28ns. (99% yield).
3) "Company X" tests the memory to run at (?) 27ns. 987 of these chips
pass, and 3 fail. The 3 are not actually bad. Since they passed the test
at 28ns, they still meet the specifications. However, they are sorted into
a different category.
4) "Company X" continues to test the memory at different speeds.
Eventually, they test the memory at 25ns. Let's say that 600 units pass.
These units are sorted and used as 25ns parts. A lot of these memory chips
may even be rated at FASTER than 25ns memory. If further bin sorting is
done, you may find a lot of "Company X"'s memory chips are capable of
operating at 23 or 22ns.
But why are these chips labeled as 28ns? One possibility is that being
so conservative on the minimum requirements of operations allows for a
higher yield rate. As you see in the example above, Company X maintains a
yield rate of 99% at 28ns. The yield drops to 60% when tested at 25ns. All
of the memory can be sold as 28ns, but only a portion of it can be sold as
25ns memory. Since 60% of the chips have fully qualified as 25ns parts,
they can be sold to companies needing 25ns memory. The only caveat is that
the ink says 28ns.
Why don't they ink the part AFTER they test? I can only speculate that
maybe inking the component after it's all been put together could reduce the
yield by exposing the part to hazardous conditions.
The memory on the Wicked3D featuring Voodoo2 product has gone through
thorough testing and has been proven to operate at the 25ns specification.
As such, we WILL NOT be issuing RMAs for the sole fact that the labeling on
the chip says 28ns. We guarantee that the Wicked3D featuring Voodoo2
product will operate with the same stellar performance, whether it's
equipped with OKI "28ns" memory, or Silicon Magic 25ns memory.
With regards to the issue with customer service, please send me an email
with as many details as possible, and I will get to the bottom of things for
you. Please accept my apologies for the inconvenience you have been caused.
Best regards,
William A. McClatchey
Member of Technical Staff
Metabyte, Inc.
Tony Cheung wrote in message <35D0D010...@netvigator.com>...
>I had ordered my wicked3d with the H3d glass on 8/2/98 but still had not
>received it in Hong Kong. I send me email to customer support but got a
>non-working tracking #
>(UPS) and now they don't response to my message anymore. Compounded with
the
>28ns memory issue, I think it is a mistake to buy anything from a small
company.
>
>Tony
>> Dylan Rhodes [Creative] <drh...@soundblaster.com> wrote in message
Joe M
> knowledgable in every area of computers, we're going to have to rely on the
> ones that do this for a living.
>
> Jeff
Yes, and didn't Dylan give a complete, well articulated response?
I feel very lucky that Dylan participates, I sure have learned
a lot from him.
Joe M
Gamo
Sorry to requoted your post without your permission. I didn't really mean to
give him impression that 28 is worse or better thans 25 ns RAM, but what I
really feel (in customer point of view) that when we buy a board and expect
to see 25ns memory we should get one with 25ns memory not 28ns memory on it.
Some customers do not has technical knowleadge to exactly know that 28ns =
25ns in some cases and not other. I requoted your post because I believe
that CL choose to do it better way by clearly indicating on the box that
this box contains cheaper or different parts than normal boxes. It's like
when you go out to buy a Porche but the seller give you a Ferrari and tell
you that they both have the same topspeed so it doesn't matter what car you
get. Umm, well......... Sorry agin Dylan if it is upset you and to Ernie as
well for wrong impression you might get from my ealier post.
Kind Regards,
Gamo
Tony Cheung
Tony Cheung
If the chip is bin sorted by the board manufacturer, what do the metabyte do with the
chips that fail the bin test ? return them to OKI ?
Tony
Maik
Memory on it instead of 25 ns ? LooooooooLLLLLL. You can't be serious Jeff ?!?
Why ? If your PC is slower than others, it's definetely not the 28 ns Mem Fault.
>
>
>
> So, let me just end this with one emphatic note: the silicon magic -25 and
> the OKi -28 are the *same part*. Same design, same silicon, same fab, same
> screening test, same 3Dfx qualification test -- just a different package.
>
>
Jorge Alberto Galiano
look the package of de CE product?
Thanks a lot, here we have too many "thief" who work as "Computer resellers"
Dave
Well I believe "C" was actually out sick when you were calling and the
secretary didnt know this. I have actually had nothing but the best
treatment with MetaByte personally.
Tony Cheung
What is your email address ?
I will send you the details of my order.
Something really funny happened. I received the invoice by air mail but still
no sign of my wicked3d board from UPS ??
Thanks.
Tony
a
Whoops! Sorry about that. wmc...@metabyte.com
Tony Cheung wrote in message <35D16BB9...@netvigator.com>...