Originally posted by Estuansis:

---

My favorite solution is to put a napkin or something over the screen, line a pencil tip up with the pixel, and flick the butt-end of the pencil so it jabs the dead pixel through the napkin. Try it a few times and see what happens. The dead pixel should flicker back to life. A lot of the time this will fix a dead pixel permanently. It worked great when I got my Dell 24". I never had to RMA it :)

---

Originally posted by sammorris:

---

3dmark03 means nothing to me. 3dmark06 is the only relevant test really. Scoring 7660 isn't bad, my file server with its X1900 scores 5000 ish with the CPU stock, and 6700 with it at 3.15.
Russ: Would people necessarily come back to you if something went wrong with your builds? Just a thought, people may want a second opinion or to have a go at the issue themselves.

---

Originally posted by Sophocles:

---

Quote:

---

3dmark03 means nothing to me. 3dmark06 is the only relevant test really.

---

3dmark03 is an aging benchmark. Another point to consider is that it is also possible to obtain a higher score by lowering graphics resolution.

---

Originally posted by sammorris:

---

As 3dmarks age, 06 included, they become excessively CPU-reliant, and ultimately become a benchmark of CPU performance, not games performance. Whether you play top end titles or not is irrelevant, all the games you use have to be as old as 3dmark03 for it to be a valid test. With your current CPU, your maximum limit in 3dmark06 standard for the CPU would probably be about 13,000, so you shouldn't be hitting the CPU bottleneck yet. The reason your overclock didn't net you much performance was that while you upped the core speed by 10% or so, memory and shader clocks only got 3-4%, which is why you only got 4-5% extra performance.

---

Originally posted by Estuansis:

---

Yeah a new video card once in a while definitely can't hurt. The 9500GT isn't exactly my first choice but it's definitely a solid card for someone like you Russ. The 7600GT was solid for a really long time too and I even had games like Oblivion and FEAR running quite decently on it. The 9500GT should be a relative powerhouse in comparison :)

EDIT: Sam is right though. I know 3DMark 03 has always had badly skewed results(13,000 Pts on my X850XT) but I can't see such a large increase from OCing.

I now no longer trust Guru3D. They showed a 9600GT getting a 9700 3DMark 06 with an E8400. LOL Mine got over 10,000 with my 7750BE.

---

Originally posted by Estuansis:

---

I've seen weirder from 03 so it's not entirely impossible for it to happen. I had an X850XT that outscored my X1800XT on nearly the same hardware.

3DMark 06 will give you more accurate results.

---

Originally posted by sammorris:

---

I think you're missing the point. 3dmark03 was not at all adapted for modern hardware, so its results are completely arbitrary and unreliable. The reason the variance in 3dmark06 is small is because that's the ACTUAL performance difference you get.
3dmark03's result is so much higher because it's completely wrong.
If I swapped out a CPU for one 10% more powerful and got double the performance figures (which is in essence, what you've got) there's so little accuracy, what's the point of a tangible result? Surely you can see a measure of performance has to be accurate to at least some degree.

---

Originally posted by sammorris:

---

I didn't think the first i5s would be quad cores, but seemingly my suspicions about that being false were justified. (I don't see how Quads can be overlooked at all these days, especially with cheap ones like the 9350 out there, even if they are pointless)
The Core i5 is quite an exciting development, especially seeing that data, though it isn't very specific. Let's see how cheap the boards are. If they can produce a £70 board, and the CPU is £160 ish, we've got a serious contender.

---

Originally posted by sammorris:

---

Yup, and this one's affordable, and better yet, the boards for it won't need a BIOS flash to run the CPUs it's designed to work with...

---