With the apparent FXlop, those looking for a powerful but cheap video solution have been looking even more closely at these Radeon 9500 fixes, or will be.
Someone wrote me on the subject:
Hey Ed… quick note I thought may be applicable. I’ve tracked the modding results of about 30 cards, (NOT from the same “batch”) and have found that only about 7 worked successfully. That makes a 23% SUCCESS in my limited sampling.
I’m really hoping someone would get this word out to the “masses” before more people waste their money. A lot of people who think their card modded successfully just because they don’t see artifacts are often wrong. It seems that sometimes the damaged pipelines just don’t enable.
A sure fire way to test this (save performance of course) is the 3dmark2k1 multi-texture fill rate test. If the mod was successful, the multi-texturing fill rate should be in excess of 2100 Mtexels/sec. (Correction: The original piece said multi-texturing, I changed it to single after a little erroneous research, thanks to those who pointed this out–Ed)
A lot of people are wasting their money on this without knowing the truth, and it seems for some reason people are reluctant to report it.
This particular person went 0-for-4 himself.
XBit Labs reported a 30% failure rate based on someone testing a batch. This person thinks it’s more like a 70% failure rate.
That’s obviously a huge difference, and a decisive one for those who might want to try this. These 128Mb Radeon 9500s cost a bit over $150; so it’s not a cheap experiment.
So . . . .
If, and only if, you’ve actually tried doing this (I’m overwhelmed with emails as is), could you please tell me:
1) Did you succeed or not, and what methods did you use to do this?
2) If you succeeded, could you please send a screenshot (or weblink to your 3DMark 2001 results at futuremark.com) for single-texturing fill rate to confirm success?