Cheap Processor Upgrade? - Page 22

Because without volt mods the computer will not boot :P

Looks like vid4 + vid1 or vid4 + vid0 is the sweet spot here. I just attempted vid4 and vid2, temp flew up to over 90, prime failed in 6 minutes. Second time I tried the 9300 got pissed and just locked up with temp at 90 and fans blowing. Methinks my problems are temp related...

I think you miss read what I said, I meant that the reduction in your CPU's life will be so small you will be buying a new computer because the OC'd one is obsolete. I would hope I know what I'm talking about when it comes to OC'ing since I've been doing it since I replaced the crystal in my pentium pro, and I currently have they fastest Go 6800 on record (at futuremark and laptopvideo2go.com).

lucidchaos, keep it up. I think you'll have the world record soon, man! I wonder if the XPS2 has a better CPU heatsink than the 9100?

Any post that talks about a CPU that can "die from the strain of overclocking" is... well, I won't get into it. Who cares if you have the fastest 6800 Go on record; all overclocking a video card is is moving some sliders up and down. =\

Actually it's very easy to kill a cpu if up the voltage to high, even a microsecond of intense heat will ruin the architecture. And yeah it would be real easy to OC the 6800, but I doubt they would get scores that high since I had to volt mod the card and modify the heatsink substantially.

No, the heatsink in the XPS2/9300 is really just a glorified Northbridge heat sink, but that's really all you need for the standard P-M. An OC'd P-M on the otherhand, well I would definately use i9kfangui or set the bios to high for the fans and put better thermal paste on the CPU like AS5.

Before going to bed I set the multiplier to 19 and grounded vid4 and vid0, had it going stable all night, was delighted when I got up and didn't see any errors So 2.53 is stable... now to address that extra 133 mhz...

btw it failed prime in 5-6 mins with vid4 and vid0 grounded @ 2.66. So far I have had the best luck with vid4 and vid1, I just went back to that configuration and am going to try to reproduce my earlier results. My dinner table is a mess

vid4 and vid0 grounded translates to what voltages? or i guess in your case, which voltage? (speedstep disabled)

what temps idle/load are you seeing?

nice!

you may run into trouble at 2.66 GHz; the CT-479's cooler (though it is janky, it still has a better fan/heatsink/open air space) usually tops out around 2.7 GHz. Not even the magical properties of AS5 will keep it from being in tight spaces.

I just applied some AS5. Load temps WERE 77c @ 2.53ghz @ 1.596v w/ no thermal, things are now looking stable at 77c @ 2.66 @ 1.660.

I have been running prime for about 40 minutes now which is much longer than it would run earlier.

I'll let it go for a couple more hours to test it, then I can see if any lower than 1.660v will work.

Ziddey, the beautiful thing about this particular volt mod is that speedstep does not need to be disabled. With my configuration, there is not a single instance where the voltage delivered is less than the voltage required for speedstep. When I am at an idle load (28 degrees c), the voltage is 1.020, which is only 0.032v higher than default @ 6x multiplier. Therefore, the drain of this when idle will be very similar to the drain stock. My max temp just hit 78, nothing to worry about imo.

TSFroggy, are people using the pin mod to overclock, or are they using stock 533 chips or stock 400 chips? Are they using volt mods? Remember, by using the pin mod we are essentially unlocking the multiplier. If you are seeing people running pinmodded 7x5's on ct-479s, please let me know where!

Btw congrats on being the first to run a 745 @ 2.4 w/o volt mods. Seeing that done successfully inspired me to try the 2.0 mod, and well so far... things are looking good!

And one more thing for anyone who is contemplating modding an xps2 - take a look at these benchmarks of the ct-479 adaptor (pay close attention to the gameplay scores @ 2.6ghz) - this thing eats the A64 FX55 for breakfast!

http://www.pcper.com/article.php?aid=133&type=expert

I don't know if unlocking the multiplier is exactly what we're doing (though I see what you mean by that :P), but I just mentioned that the cooler of the CT-479 is going to be somewhat more efficient that one in a laptop, and that the CT-479 heatsink allows most P-Ms to top out around 2.7-2.8 GHz, but if AS5 is letting you run Prime stable, more power to you. Can you take some pictures of the cooling inside and outside the 9300?

Btw it's cool to be an inspiration I kind of want to try this with a 765 :P

I tried lowering my voltage to no avail; looks like 1.660 is the lowest I can take it. Been running prime for an hour now, and i'll let it go another hour before I re-assemble the whole thing.

TSFroggy, you can see pictures of the assembly in this thread:
http://notebookforums.com/showthread.php?t=80879

Nothing special, just a heatpipe with a heatsink smaller than my pinky finger.

Temps still at 78 under prime95 load.

maybe voltmod will push my 2.4 ghz to stable. ;(

does the voltage record correctly in RMClock, or do you have to add

You have to figure out what the voltage is yourself using the voltage table on intel's whitepaper.

Can you link me please? I know it's in this thread somewhere but don't know exactly where.

Add another 1.6 -> 2.13 success to the chart.

I left the thermal pad on the cooler. I did put a small amount of white paste between it and the new CPU though as the old CPU had taken most of the black goo in the center off.

I figure if it's not stable at these slightly higher temps saving an extra couple of degrees isn't going to make it stable enough anyway.

Intel's pentium m datasheet: http://download.intel.com/design/mob...s/30526201.pdf

There are two diagrams I found very helpful, one is the voltage chart on page 18, another is the pinout on page 35. One more trick I did was making a spreadsheet with a copy of the table in it, and then appending the volt mod values to the side. This way, if I select a voltage in RMClock, I know the corresponding value with my volt mods. Example:

An update on progress: After experementing with different pin configurations (and not much luck on any of them), I decided vid2 and vid4 was the way to go. Since it was late, rather than testing it again before putting it all back together, I just re-assembled it. I was eager to do a doom 3 timedemo, so I booted into windows, and it crashed a few seconds into the demo. After a litle further experementation I determined that something either wasn't seated correctly or something. It wasn't getting hot (at least not as reported by i8kfangui).

I disassembled it again am tweaking it now. I kinda have it set up like desktop computer now, so I can actually use it while testing it.



Now a question for the overclocking gurus: when prime runs for an hour and a half or so and then errors, is that usually a heat issue or a voltage issue. The chip CAN run very hot, but I don't think I want to feed that much voltage through the chip unless I absolutely have to.

Another alternative would be to call up dell spare parts and order a xps2 heat sink and heatpipe assembly... but that will only work if heat is my issue.

Does anyone have any suggestions?

CPU instability when overclocking is a combination of heat and voltage. You need more voltage. There's not a lot that can be done to significantly reduce the heat and that is all that would help.

You can't separate those two causes by what Prime does. They are really the same thing. Voltage improves signal within and around the CPU. Heat impedes that improvement. It's a tradeoff/balance thing.

Ok thanks, that's what I suspected (but i was hoping to hear otherwise )

The new game plan is to run prime on several voltages and see which is the most stable. I realize now that I probably can't run 100% stable at 2.66. However, I am rock solid at 2.53, and at a cool, comfortable 68 degrees C.

I want to tell everyone thanks for all the useful advices, now I have my new Sony fs 115s (fs 550 in USA) running at 2266 Mhz !!!!

some quick stress tests where positive and now I'm running it by 6 hours without problems. But I'm at the limit as undervolting make it instable in the long run.

Mine temps are 8-10c higher than with the stock 1733 cpu at 50-55C at idle and 68-69 under load.

When I first tried it I forgot to attach the Fan cable and it ran smoothly until 85C.

Ciao

Olaf