True, but it's unlikely any of those power supplies has cables long enough to route behind the motherboard tray. As I stated in the write up I do have an old power supply with the aux power.
On the bright side I have over $400 in tools to make whatever power supply cables I want in the future. It's just the fixed costs and minimum quantities that were high, the actual cable it self is about $1 in materials.
True but even so I don't see how just taking a power supply that already has a aux lead and making it longer with a some some wires, a wire stripper, some flux, some heat shrinks and like a soldering iron which is what I would do is expensive or any big deal. Thats almost like something normal.
Oh wow, that's interesting. Could you maybe run some memory bandwidth tests? Sandra, Everest, AIDA64, whichever you prefer, I'm wondering if the bad scaling is due to lack of bandwidth or your driver version.
AIDA64 gives about 1GB/s for read write and copy. Sandra gives 794MB/s for its memory bandwidth test.
Older driver scales better, but overall performance is lower. I'm guessing this is a result of the Quadro not having the game profiles of the Geforce cards. In 3dmark ECC made a difference of a ~9000 score vs a ~9300 score, but in Quake 3 the difference was next to nothing although no-ECC did give slightly faster frame rates.
Warlord wrote:
True but even so I don't see how just taking a power supply that already has a aux lead and making it longer with a some some wires, a wire stripper, some flux, some heat shrinks and like a soldering iron which is what I would do is expensive or any big deal. Thats almost like something normal.
You're assuming that I have those things and not taking them into account in the fixed costs. By your method I would have forgone the crimping tools and pin extractor, but added in a soldering station, solder, flux, and heat shrinks. Given that I would buy tools that will be useful for other jobs in the future (i.e. I don't need a Fluke 87V to test continuity, but that's what I bought since I'm likely to do other electronics work in the future so there's "$370" of my cable) that soldering station would no doubt have cost more than the crimping tools and connectors. If not it would be very close to the same cost.
At least they are scaling positively now, its probably a combination of drivers and Quadro features then, like you say.
EDIT: though it should still be maxing out 800x600 at around 200 fps, so I still find it weird. Maybe try another fast, non-Quadro card at 800x600 and see how much it gets?
You're assuming that I have those things and not taking them into account in the fixed costs. By your method I would have forgone the crimping tools and pin extractor, but added in a soldering station, solder, flux, and heat shrinks. Given that I would buy tools that will be useful for other jobs in the future (i.e. I don't need a Fluke 87V to test continuity, but that's what I bought since I'm likely to do other electronics work in the future so there's "$370" of my cable) that soldering station would no doubt have cost more than the crimping tools and connectors. If not it would be very close to the same cost.
no you are very wrong, Soldering station is way overkill for that job. This is like a 25-30 dollar max job if I had no tools. It doesn't require special tools pretty much any old soldering iron can do that job even 10 dollar ones. I know from experience it's not an assumption. You are right you don't need to test continuity on such a simple job where you can physically see if something is connected or not. If i twist 2 wires together apply some flux and cover them with solder I don't need to test continuity to that I can physically see its good. But a continuity test doesn't require high end gear even the cheapest multi meters I have seen at like home depot will test for continuity. Pin extrators, like ok that is nice to have but a lot of people have used things like clothes pins its not a hard thing to extract that type of pin. But like I said I never would of done it that way anyways that's like the hardest unnecessary way to do it. Cut the cable in half add a patch and u are done. Im not trying to bash you i just found it odd someone saying something cost 500 dollars when I have 30 years of experience soldering something like that.
Quake 3 scores from my 8500DV. A few notes: On the SMP test the first run would always show a higher frame rate (1280x720 showed 139fps for example) but this would be an out lier so I dropped it. Secondly, while the 1920x1080 results look decent these singular number doesn't really tell the full story. I can tell that this card is not cut out for 1080p as frame times varied wildly. Catalyst version is 6.11.
For some reason the Tualatin won't POST with the All-in-Wonder 8500 (non-DV), so I couldn't test it. The card works fine in the Coppermine system.
Warlord wrote:
Im not trying to bash you i just found it odd someone saying something cost 500 dollars when I have 30 years of experience soldering something like that.
Understood, if I thought you were I wouldn't even bother replying. Anyhow, this is just a different mindset to accounting and I like telling it this way since it typically gets a great reaction from people even after I give the break down. You're absolutely right that this job wouldn't need an expensive soldering station, but in the past I've run into scenarios where the $10 iron just doesn't cut it, so from my point of view there's no sense in wasting $10 when it could go towards a better tool to do more jobs. It was the same with the multimeter, there are much easier ways to do verification (and in your example just looking at it usually works), but every other week or so I find some way to use it to take the guess work out of something.
Blzut3 wrote:Quake 3 scores from my 8500DV. A few notes: On the SMP test the first run would always show a higher frame rate (1280x720 showe […] Show full quote
Quake 3 scores from my 8500DV. A few notes: On the SMP test the first run would always show a higher frame rate (1280x720 showed 139fps for example) but this would be an out lier so I dropped it. Secondly, while the 1920x1080 results look decent these singular number doesn't really tell the full story. I can tell that this card is not cut out for 1080p as frame times varied wildly. Catalyst version is 6.11.
For some reason the Tualatin won't POST with the All-in-Wonder 8500 (non-DV), so I couldn't test it. The card works fine in the Coppermine system.
I take it your 8500DV has 64MB? In any case, these results are exactly what I saw with a 9100 in the quake3 smp tests, so they pretty much confirm its a driver "issue" with the Quadro.
