• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

9500NP is now harder to mod, I believe

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Simply, adelphia83, no i dont. ati had too many working chips to just go and put all the malfunctioning ones on the 9500np boards. thats where the likeness to amd's 1700+ comes from.

amd's process is now soo good that they dont have enough "bad" parts to meet the demand of the 1700+'s, therefor they use the better parts - the reason we all love our 1700 dlt3c's. Strikingly similar, isnt it?

i you would have read my post a little more thoroughly, as well as Rocko[DPC]'s rude and mis-guided statements in and before this thread to me, then you would understand.

read his posts,moddable 9500np's are NOT all damaged or downgraded parts. would he listen? it sounds like you dont either, concerning this point. ask anybody who has a modded 9500np - has it suddenly failed for no reason? something that wasnt caused by them? the situation hasn't changed you know. There is nothing wrong with them.

Anyway, why bring this up now? this was months ago, and was finished months ago.
 
Last edited:
I certainly don't want to add to the arguement but I have seen many companies get an unexpected order or two for lower end products. You can bet that said company will fill that order as best they can regardless of what actually makes it into the box. JITI (Just in Time Inventory) means you gotta ship what you have now not later. no company can afford not to ship because the yields are too good -- it would force customers in market shortages to buy the competitiors product. Its best to make a little less and keep a customer than not get the customer to begin with.

I remember way back that ocasionally you'd see computers shipping with faster processors because that is what they had at the time. it still happens ocasionally.

If yields are especially good you can bet some of those good parts will end up in other bins.

Its all about the benjamins.

T
 
adelphia83 said:

Why would ATI "bin" the 9700's anyways unless they were actually defective? If demand was so high for the 9500np boards, wouldn't it be cheaper for them to just use 9500Pro's that either didn't make the grade, or speed bin those?
Um, you're not too bright are you? All flavors of the 9500 and 9700 series cards use the EXACT same GPU core -- the R300. That means that the R300 cores are binned appropriately to be put into either 9700 Pro, 9700 non-pro, 9500 pro and 9500 non-pro cards. It's all the same chip!! What makes them perform differently is clockspeed on the core itself, the memory trace layout on the PCB it's plugged into, and that single little resistor (along with a software flag built into the BIOS and/or the driver) that enables/disables the four extra pipes.

Let me say this again, just so you get it right: EVERY 9500 and 9700 CARD uses the EXACT SAME R300 core. Period. If you somehow have your head stuck far enough up in a secluded, warm and dark place to not know this, why don't you try looking at the markings on the core of each chip. Guess what? They're the same.
adelphia83 said:

Common sense would tell you it would be cheaper to produce a card with a 128bit memory bus, and disable 4 pipelines, than to manufacturer a card with 256 memory bus, and disable both half the memory bus, and the 4 pipelines.
Perhaps it makes sense to someone who knows absolutely zero about the work that goes into designing a GPU core, and knows nothing about the costs involved with multiple lithography processes. Anyone who actually DOES know about these costs also knows that what you're describing would be several orders of magnitude MORE expensive. Don't pretend to know things which you VERY obviously dont...
adelphia83 said:
Whoa.. This might be getting a little over your head now, so take a minute to process all that information. When you really stop and *think* about it, it DOES NOT make sense for ATI to "bin" the 9700 series into 9500's.
Sure it does. A single chip design. A single lithography process to QA. A single pinout configuration for VAR's to build the PCB's against. The only thing that has to be changed to make an R300 core into a 9500 Pro is to put it on a PCB that uses memory modules in a 128 bit trace layout... The only thing that has to be changed to make an R300 core into a 9500 Non-Pro is to use that same 128 bit trace layout board, and also to flip the resistor location to lock in the software instruction contained in the BIOS to kill the other four pipelines.

adelphia83 said:
But it also makes sense to have a card based on the 9700 to sell as a 9500np.. Why? so they could sell the cards w/ the messed up pipelines, and malfunctioning 256bit memory bus.
You have to be smarter than this... Oh wait, no you don't. On the 9500 non-pro cards that are equipped with 128mb of memory in the "L" shape, the 256 bit memory bus is already working -- the "hardmod" or "softmod" does absolutely ZERO to the memory bus, it ONLY TURNS ON the other four pipelines. This has been discussed AND PROVEN so many times that it makes my head spin as to how people somehow keep confusing it. We don't know why some modded 9500 cards get glitches, the best GUESS is about messed up pipelines. The honest TRUTH we that we dont' know, because NONE OF US WORK FOR ATI (wow, go figure that out!)

The fact of all this matter is that you were wrong on a whole bunch of counts, which ends up making your "opinion" worth essentially as much as a piece of used chewing gum found on the bottom of an outdoor picnic table.
 
Whoa buddy--- You don't think I KNEW these cards were based on the R300 core? Apparently you should reread my post before you go prejudging my intelligence. I never said anything about the cards having different cores.

As yes, while knowing that the cores are identical, upon ripping off the stock HS on several cards w/the R300, have.... what? Whoa-- R300 written on the chip! You're a genious!...

Just to state the obvious, the 9500np has a 128bit memory bus, four pixel pipelines, and runs at a stock speed of 275/270. The Pro version is practically identical, but the 8 total pixel pipelines are enabled, and the memory clock is increased to 275.

The 9700 is identical the the 9500np, except is utilizes all eight pipelines, and a 256bit memory bus.

The 9700pro is a 9700np but with increased clock speeds: 325/310

I think you misunderstood me. I'm not talking about designing whole new cards here. I'm not talking about physically altering the PCB traces in an attempt to create a different card.

When I said it would be cheaper to speed bin 9500Pro's, I meant that it would be cheaper for ATI to speed bin the CARDS w/ the 128bit memory bus, rather than binning 9700's that have the 256bit memory bus.

See what I'm getting at? Cost of card w/ 256bit memory bus is OBVIOUSLY (isn't it?) more expensive than the card w/128bit memory bus.

So why on earth, would ATI produce cards originally meant for 9700np (cards with 256bit trace layout), and sell them as 9500np? Perhaps the reason for this is that the 9700 didn't meet QC, maybe the card functioned fine with four pipes DISABLED, and possibly with the memory bus cut in half (again i'm not talking about physically altering this).

Your argument is that demand calls for more 9500np's that what is available, well why the hell not crank out a couple more 9500Pro's and speed bin those, with JUST the four pipelines disabled? Why produce a more expensive card (9700) to bin? That just does not make sense.

However it DOES make sense to take cards that do not meet QC, and bin them down to 9500np's... Hence why the L shaped PCB board exists as a 9500np.

You practically said this yourself:
"...The only thing that has to be changed to make an R300 core into a 9500 Pro is to put it on a PCB that uses memory modules in a 128 bit trace layout... The only thing that has to be changed to make an R300 core into a 9500 Non-Pro is to use that same 128 bit trace layout board, and also to flip the resistor location to lock in the software instruction contained in the BIOS to kill the other four pipelines."

While being able to create a 9500NP using a card w/ a 128 bit trace layout for the memory bus, all that would have to be done is to disable the 4 pipes. Don't you think this is a cheaper method than using a PCB layout w/ a 256bit memory bus?? Obviously there is a reason why this PCB layout was used for the 9500NP. For some reason those boards *MAY* not have met 9700 QC, but made the grade as a 9500np.

I still have yet to read *anything* in your post that I did not have prior knowledge about. Nothing you just wrote was in anyway proving my opinion to be wrong. And FYI the 256bit memory bus is *NOT* active in 9500np boards. Sure the trace layout is there, but the bandwidth is not.

So in the end, I believe you wasted your time writing that post. Your attempt to "inform" me of something I already knew about can be considered a failure.

I would appreciate future comments that are directed toward me to be subjective to the material that I had originally written, or comments that have to deal with the subject at hand.

[/B][/QUOTE]
 
Last edited:
adelphia83 said:
When I said it would be cheaper to speed bin 9500Pro's, I meant that it would be cheaper for ATI to speed bin the CARDS w/ the 128bit memory bus, rather than binning 9700's that have the 256bit memory bus.
no, because at the time there were NO 128bit cards around. Remember that the 9500po came out some time after the np
See what I'm getting at? Cost of card w/ 256bit memory bus is OBVIOUSLY (isn't it?) more expensive than the card w/128bit memory bus.
yes, but not if they dont have a 128bit card, which, at the time, they didnt.
So why on earth, would ATI produce cards originally meant for 9700np (cards with 256bit trace layout), and sell them as 9500np? Perhaps the reason for this is that the 9700 didn't meet QC, maybe the card functioned fine with four pipes DISABLED, and possibly with the memory bus cut in half (again i'm not talking about physically altering this).
The 256bus is perminatly enable. period. there are no if's and buts, its just enabled. thats how the mod was originally discovered - the 256bit 9500np's were somping the 128bit ones, being much faster than they should have been.
Your argument is that demand calls for more 9500np's that what is available, well why the hell not crank out a couple more 9500Pro's and speed bin those, with JUST the four pipelines disabled? Why produce a more expensive card (9700) to bin? That just does not make sense.
Because, at the time, its quicker and ultimatly cheaper to produce one type of board. Again, they did not have any 128bit 9500pro boards at that time.
However it DOES make sense to take cards that do not meet QC, and bin them down to 9500np's... Hence why the L shaped PCB board exists as a 9500np.
yes that does. but there are not enough bad cores to meet that demand alone. why can't you understand this?
I still have yet to read *anything* in your post that I did not have prior knowledge about. Nothing you just wrote was in anyway proving my opinion to be wrong. And FYI the 256bit memory bus is *NOT* active in 9500np boards. Sure the trace layout is there, but the bandwidth is not.
Acutually, FYI it is. i suggest you start doing some research. It's on, it works thats all there is to it.
So in the end, I believe you wasted your time writing that post. Your attempt to "inform" me of something I already knew about can be considered a failure.
only because you dont listen:rolleyes:
I would appreciate future comments that are directed toward me to be subjective to the material that I had originally written, or comments that have to deal with the subject at hand.
oh they do. your wrong. plain an simple.

[/QUOTE] [/B][/QUOTE]
 
God you're dense...

ATI doesn't speed-bin entire cards, ATI speed-bins the CHIPS that go on the card! The CHIP ITSELF is a 256-bit memory bus, 8-pixel pipeline FP24 superscalar 120-million processor unit by itself. THE CHIP ITSELF HAS ALL THESE FEATURES.

The BOARD is what decides if it uses 256 bits of memory bandwidth. The BOARD (actually, that single resistor and a software flag) itself is what decides if it uses all 8 pixel pipelines.

You obviously DO NOT KNOW what you're talking about, and as such my original claim stands: your opinion is worth as much as a pre-chewed piece of gum found on the underside of the local primary education picnic facility.
 
H%*#!!!! I am a gettin outta here!
I gots me's a 9500pro bebe and a it be hard modded soft modded, physically modded, extremely modded and mod modded, LMAO!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

So speed bin, de tune, retune, L shape mem, straight line mem, flashed & hacked bios, volt modded, or just break it, or what ever the heck ya all wantto do!

ADELPHIA83, YOU THE HEAT! I wouldn't waste another second trying to argue with your buddies about what you wanted to talk about.

I ain't sayin another word here!

"Stupid is, is what stupid does is what momma always said, and thats all I got to say about that!" (Forrest Gump)


Happy Trails,
Tweaker and RadMan:eek:
 
Back