Slashdot Mirror
How to change your Radeon 9500 into a 9700
Ian Bell writes "We have just posted a very difficult guide to turning your ATI Radeon 9500 into a 9700. But you have to have the correct 9500. A 9500 with 4 rendering pipelines, modified to enable all 8 pipelines, will effectively double the memory bus, if you have the extra 64 Meg of memory to attach it to. We will explain below which card to acquire for this awesome graphics card transformation. Check out how to do this yourself and get the power of a 9700 at half the price." Update: 01/19 18:33 GMT by T : And for those running Windows, Sanity writes "Aside from the hardware mod, there is a program called Riva Tuner that has, among other things, a software mod for unlocking those gates, plus overclocking to a full 9700 pro! Gives me more $$$ to spend on cool stuff."
Would it be illegal for people to modify these 9500's then sell them on somewhere like eBay for example? You could probably make a killing.
The DMCA is a Copyright Act. It makes circumvention of protected copyrighted works. What copy protection scheme does this mod allow us to circumvent? Of course, blaming the DMCA for everything is always acceptable.
Note that this only applies to the 9500 and not the 9500 PRO.
This hack has been crawling around the boards for a while and it seems fairly legit. The basic layout and architecture of the 9500 and 9700 are the same and this hack attempts to:
1) Re-enable the extra pixel pipelines that are present on the 9500 just not enable. A simple resister swap near the gpu is required.
2) Flash the bios of the 9500 with a 9700 bios image.
3) Overclock the 9500's core clock to compete with 9700.
4) (Optional) Add more memory.
The biggest problem I see is that the stock memory on the 9500 is of a cheaper variety and isn't rated for the frequency that a 9700 operates at.
So, even if you indeed have the skills/luck involved to pull off this cute hack, then you'll not necessarily be able to compete with a 9700.
My advice, go with the 9500 Pro. Out of the box it's only a step slower than the 9700 Pro and costs half the coin of a 9700/9700 Pro.
But if you've got a 9500 in your machine and some time/money to spare. Why not see if you can achieve great things with a minimum cost?
I have a friend that just did the software mod, and that alone nearly doubled his 3D mark. I would have to say with the ease and relative safety of the software mod, its probably best not to get greedy and kill your card. At least that my opinion. Either way my gForce 3 is starting to feel a little inferior.
"I am the Flail of God!" -Genghis Kahn
This looks like stealing. Does it look like stealing to anyone else?
Bringing up ethics on Slashdot? Prepare to be seriously flamed.
I would not go so far as to call it stealing, but I would question the ethics of it (just as you did). Many companies are surviving on razor-thin profit margins in the PC hardware sector and this kind of thing is going to hurt them if it's done by people who would have otherwise bought the more expensive card.
On the other hand, I think that it's likely that this procedure will result in a lot of incompetent people destroying their cards, so maybe it won't hurt ATI so badly in the long run.
Not that I don't agree with you in this case, but there have been cases, historically speaking, of companies releasing two models for marketing reasons and finding it cheaper to make a hard coded 'feature switch'. I recall a hard drive of olde that could be upgraded to twice the size. The engineers designed a drive. Marketing decided they needed one half the size too. It was cheaper and faster to do a mode switch then it was to pay the extra engineering and manufactering costs to build a second model of HDD.
:). As for me, I know my skills do not extend into the land of hot insterments of destruction, so I'll take a pass :)
See also the 'flippy discs' of the C-64 era. It was cheaper/easier for companies to use the double sided media they used on other systems, then produce new single sided floppies just for the c-64 market. You punched a second hole in the floppy, and turn it over.
Again, I don't argue that's what ATI is doing here. I personally agree with the person above who suggested they're probably using 'bin chips' that for some reason didn't make the cut for the 9700 boards.
So you pays your monies and you takes your chances. Mmmm I love the smell of newbies with solder suckers in the morning
On the whole, I find that I prefer Slashdot posts to twitter ones because I don't get limited to 140 chars before
Apparently there's a way to do this completely in software -- no hardware modifications required.
I have heard many stories whereby modifying a 9500 into a 9700 resulted in disaster -- sometimes the 9500 simply isn't good enough to perform at those levels. At least software gives you the option of reversing the damage.
People seem to have the ethos in computing of buying the cheapest junk with the highest specs possible, not realizing that they just supported junk instead of quality. This is how quality is ratcheted lower, and it becomes difficult or impossible to find anything decent. $15 PSUs that weigh maybe .5 lbs are frighteningly common, as are CAT5 cables thinner than a drinking straw, motherboards with %20 defect rates, and on-board audio that just crackles instead of recording.
I don't agree that "if it ain't broke, don't fuck with it," as tweaking and playing are both very natural and very educational: but don't return it. You broke it: you fix it. If you can't fix it: you buy one that can do what you wanted it to do in the first place. But don't fall into the pit of buyers remorse by getting a wall-mart, emachines, or other low-quality computer to save a few dollars, then chop it up to try and compensate for not buying something that could satisfy you. Buy and support the things that you want. And always, always do your research. If you could spend 3 hours finding out what the best available PSU for your system is, you could save 6 hours later on trying to cut it open and cool it.
(Which reminds me, my PSU is too loud. Where did my Dremel go?)
-C
This Sig is a mnemonic device designed to allow you to recognize this author in the future.
This reminds me of when the Radeon LE came out ($71) and by doing a couple simple things with some 3rd party software and the windows registry, you could make it perform the same as a Radeon DDR, which sold for about $139 at the time.
sPh
I just did this at home. FYI a Radeon 9700 (non pro) sells for about $225 on pricewatch. Once the bios is reflashed that card can be clocked up to the exact same memory and core speed as the Radeon 9700 Pro. I've been gaming on it for weeks and it's rock solid stable. It's very easy to turn your $225 card into a $380 card, and you don't have to solder anything. As always YMMV
You /. guys are no fun. Somebody comes out with a really nifty hack, and all you guys talk about is how dangerous it is. People who live on the edge like this generally know what they're doing and go in accepting the risk. Personally, I do this stuff occasionally (I OC'ed a 300A, joined the L1 bridges on an Athlon and modded my MP3 player) and I go in fully aware that I might be throwing $200 down the tube. That's okay, because I never try it unless I can afford to replace it if something goes wrong. If something doesn't go wrong, then I just saved a few bucks. So far, I'm ahead. The 300A and the Athlon are still running, but I killed the screen on my MP3 player (which gives me an excuse to get an iPod :)
A deep unwavering belief is a sure sign you're missing something...
You couldn't sell them as 9700's, but you could sell them as modified 9500's that run at 9700 speeds.
The problem, as everyone is pointing out, is that this is a very tricky hack, and that there's a very good chance that the 9500 you buy will not be physically capable of the hack. For the entrepreneur, that means that every 9500 worked on that doesn't result in a speed increase will have to be resold for well *below* cost, since it may be damaged and is no longer under warranty. Secondly, you have to figure out how many failed attempts you will have before you have a successful one (let's say the ratio is 5 to 1); and probably one of those 5 is totally unusable and unresellable. Now we can figure out your profit per "good" card. Add up the cost of your losses on the 5 bad cards, and add that amount to the cost of the one "good" card you are selling (which, BTW, you have to sell for a good discount below the 9700's price). Assuming you made a profit at that point, you then have to divide your profit by the total amount of hours you have invested in both modifying AND selling all of the cards.
By comparison, what's your opportunity cost? That is, what is the highest compensation alternatives you are giving up to spend time on this project (selling linux systems on eBay, delivering pizzas, etc.), and which is the better time investment?
Selling hacks and sophisticated hardware upgrades on eBay really only makes sense when the profit potential is significant enough for you to absorb all of the losses of failed attempts (not in this case, because the 9700 effectively sets a price cap on the project).