OK jongens, veel succes toegewenst
1. Preface
My choice of exactly this model of the Radeon and talking in general of all Iti products was determined by the following:
·Ration of price/output
·Support of the libraries Direct X 8 (almost total)
·The up-to-day output (capacity) required by the majority of modern games.
·Margin (reserve) of the same output for the nearest 6 12 months.
·Real universality possibility of use DVI + TV-out!!!
·Quality of the 2D (Ive got the 19 SM959NF display), by the way, the 3D is not bad either)
·Price (139 147 for the OEM, too mean and stingy to pay down the nose). Two remarks from the moment of purchase the prices have gone down; prices in outside Moscow are more expensive than in the capital.
·Appreciation of ATI for their persistence in the market competition. Few of the kind are left; but thanks to their existence and competition we have more chances (quality and low price).
One more thing! The negative sides drivers! It is everybodys opinion that things are getting better. So decided to neglect such a trifle.
My dream did come true! I was able to purchase the original ATi-card, but have to use slightly reduced frequencies 270/230.
I have installed it, tried and remain happy. The happiness was total wish I were faster in 3D in some games. I wanted to set decent resolution on the 19 display 1280 and more, but the card wasnt fast enough. Of course, couldnt help thinking about buying a faster piece of ATi, but it contradicted the point about reasonable price. So I, being an radio ham with considerable experience, started nursing th idea of accelerating it (for the sake of fun). I mean, of course, power supply increase for the kernel and memory, plus additional boosting of cooling in forms of radiators and coolers.
The rest of the video-card Radeon 7500 modification shouldnt be taking as an instruction, but mainly as a reference material which can be used for accelerating it, increasing up power supply and installation of extra cooling. If, inspired by the article, you decide to take practical steps, to upgrade the video car and damage it or other accessory, the author and the editors of the site will not be claimed responsible.
The card updating
To start from the very beginning.
I have tried to explore the E-net in order to find the necessary information. But having failed to discover anything of significance, came across an article by Alexey Petrov The way to organize Radeon 8500 LE in this site and examined my card. As a result I found 5 power supply stabilizers: 3 separate for the kernel (blue on the photo), 1,35 B (orange), 1,9 B (purple): one for the memory 3,5 (yellow) and one common (red) for the memory and the kernel. Measured all voltage (the result are given above) and recorded them.
These values made it possible to have stable performance at the frequencies 326 the kernel, 261 (522) memory. I would like to mention that after updating the voltages were selected during prolonged testing and were chosen for accelerating as optimal not maximal. Further increase of voltage brought negative effects (artifacts, and colored spots appeared), or the increase failed to bring tangible results.
Starting with the memory. I didnt feel like going inside with a soldering iron. So, following the authors advice I carefully and gently scratched the resister (blue) with a needle in order to increase its resistance. According to preliminary evaluation it should have boosted it up from 3,5 V to 3.8 3.9 V (drawing 2). Scratched and missed! It was more than 4 V. The memory worked fine, but to heated more than it should. That is why I had to scratch the second resister (the red one) in order to compensate. (Take apologies for a poor quality of the photos).
As a result I had 3.87 V, stopped there arrived to a conclusion that it wasnt the best way for me. Somebody might be satisfied, but not me, being the master of soldering. I decide to solder the rest of the resistors parallel in order in decrease resistance. I will remind that to increase the voltage parallely should be connected the resistor which is ground (body) terminated. According to my judgment I had to solder from 6 to 10 kOm,. I wish I were more exact, but simply had no possibility, as you understand. The memory began to function in a stable manner at frequency 270 MHz, but it all led to an extremely interesting effect, which tried to fight till got tired of it. I could set any frequencies 300, 400 etcBut as far as increase of speed was concerned, it brought no result. People write a lot about the problem, but fail to find any solution. It looks like boosting up the frequency is blocked at some value.
The rest was easier. Stabilizer 1.55B - solder 680 Om, and get 2B (drawing 2). The increase of this voltage enabled the kernel to function in a stable way at the kernel frequency 335.24 MHz.
It made it possible to boost up the kernel frequency by 2,25 MHz. Further increase of voltage cause artifacts and appearance of colored squares. This is typical for memory failures, not the kernel. So dont quite understand whether it feeds come circuits from the kernel to the memory or not?
The next stabilizer 1.9 B solder 2. 7 kOm (orange), get 2.15B.
The current increase of voltage enable the kernel to function in a stable manner at frequency of 346, 6 MHz. It carried through 5 circles in 3Dmark2001, test Dragothic High Detail even at 351 MHz, but then artifact appeared. It could have happened due to insufficient cooling? I will elaborate on this later.
Time came to deal with the kernel and memory stabilizer, which did me an ill turn. When its voltage was increase from 2,6 V to 2,8 V by soldering the 6.2 kOm resistor, the kernel started to act insensibly. So I had to reduce the frequency to 337.5 MHz. Finally, I gave up the attempt to introduce this modification.
In the end I had the following results:
·The kernel 236/347 (+6.5%)
·Memory 261/270 (+3.5%) due to incomprehensible blocking in the chip or BIOS.
Some words about cooling
I used a medium size processor Celeron cooler for the kernel and two halves from the processor Pentium MMX radiator for the memory. Glued everything with the glue ALSIL 5, used fine dust-paper to take off 0.8 1 mm off the kernels chip. By the way, the chip is very concave as all radiators are. So advise you to polish it on glass + fine dust-paper. During intensive operation the memory is warm, but the kernel behind the car is hot; a finger can stand it; the front part of the cooler is lukewarm.
?
Testing results
The testing was done on a PS with the following configuration:
CPUIntel Pentium 4 1400@2050 (14 x 146)
MainboardMS16547 (SIS645)
Memory 256Mb DDR HYUNDAI PC2100
Video ATi Radeon 7500 64 Mb
Monitor 19 Samsung SM 959NF
Operational
system Windows XP, Direct X 8.1, drivers ATi 6018.
The card operation frequency
Regular 270/230
Acceleration without card modification 326/261
Acceleration with card modification 347/268
The rating consequently increased by 7.5% and 10.2%. No comments. The rating depends on too many things.
Max Payne
Coming to real tests.
In Max Payne, when the test PCGH Final Scene ¹ 1 (VGA Demo) was performed in the mode 1024 x 768, three-line filtration without smoothing, all options of the graphic images were at their maximum, 32 bits, the picture looked like this.
Memory increase was consequently 9% and 10,5 %. It is not much and can be explain by an intensive use of memory.
Another interesting fact. In MaxPayne despite of many opinions in the forum at site (dont remember exactly where) the anisotropic filtration does work. When it is boosted 3 4 frames up, the FPS drops.
Quality
In Quake III Arena, when the demo version Quaver is run in the mode 1024 x 768, all all options of the graphic images were at their maximum, 34 bits, the AA4x4 and the anisotropic filtration are activated.
The image looks like this:
Acceleration increase is consequently 18% and 22.5 %. There is a direct dependence on the card frequency.
Conclusions
I cant say I am extremely happy with acceleration of my Radeon 7500. The reason being the memory acceleration. The kernel 270/347 (+28,5%) is fine, but the memory 230/270 (+17.5%) could be better. Somebody might have more luck in solving the blocking problem. Keep in touch. Be happy to get opinions in the forum. Have lots of OverClocking!!!
