Overclocking a mobile Core Duo with 533 FSB to 667!

[smilingcrow]

I was just sniffing around - according to This review the ASUS N4L-VM board might be a good choice for overclocking chips with 533 fsb. no Vcore adjustment but shouldn't really be necessary.(

This board has been shown to have a limited range for over-clocking even though it has more settings than the MSI speedster. The article clearly states that the board has no PCIe lock which alone would explain why this is the case. It uses the 945GM chipset as opposed to the 945GT of the Speedster, so there may still be hope for the PCIe lock on the later.

[nzimmers]

ahh - no PCI lock..... that's a terrible thing...what were they thinking!

I've noticed more and more core duo's on ebay lately, they are getting downright cheap, wish I had picked up another MSI board back a few months.

The MSI Axis 945GM (barebone at newegg) uses the MSI 945GM1 mini-itx motherboard....which might have the similar overclocking options as the 945gt but probably not. Still....it's an interesting little motherboard.

you can still get the MSI 945GT board here at least. - edit nevermind they are out of stock

[buzzlightyear]

MSI board has problem reading the memory timing from my ram sticks too. Sometimes it reads 4,4,4,10, sometimes 5,5,5,12. I think MSI needs to update the bios. But I don't think that is going to happen since they have just update the bios to 1.4 and this mobo is not a main-stream big-seller.

My ram are HP OEM'ed ram (got them cheap - like $20/1G after rebate). That may be the problem. I will play with the settings more.

ahh, that might have something to do with it -

Buzz - I just checked something.... I am running the exact same memory in both my server (MSI socket M Board) and my main desktop system (conroe based Asus MB).... both have the same kingston ddr2 667 ram (bought at different times) but the timings are different

MSI board has 5,5,5,15
Asus board has 4,4,4,12

I'll check it by swapping it out (in both systems the timings are set on auto by spid) maybe the MSI board is not detecting the correct settings on your ram.




I'm curious about the RMClock utility - does the full version allow for FSB changes or is it only useful for voltages?

[nzimmers]

just got the T1350 (has the x14 multiplier) and have had some strange experiences right off the start on the MSI 945GT speedster

The CPU is definitely controlling what range of bus speed overclock is available. Any cpu that is a 533FSB is going to give you a 133-165 range, and any chip that is 667FSB will give you 166-199 range regardless what divider you choose in the memory settings.

I tried the 400fsb setting, got the same result, 133-165 range and best I got was 160 mhz bus

With the T1350, the bios will accept 533mhz fsb or 400mhz fsb but setting it at 667 fsb messes it up and it sets to 400mhz fsb when I try it.

best bus speed I have gotten so far is 160mhz (cpu clocked around 2.2Ghz) boots at 165mhz but won't make it through the IDE detection phase above 160mhz =( I am thinking this might be a limitation of the CPU and I would have to bump the Vcore up. I know my memory will do 199mhz without a problem so it's not t2.hat.

I think the yonah core doesn't do too well without bumping the Vcore, so I am thinking of trying a core 2 duo and test that out to see what I come up with.

I have not yet tried using clock gen, but will later on today.

so far, the best result I have gotten is this:

T1350 @ 2.24Ghz
Bus: 160Mhz
FSB : 639mhz
memory: 320mhz

[nzimmers]

tried clockgen, and have had some success - it allows me to set the bus speed up to 199 but computer locks above the 170mhz range

so I'm feeling pretty sure now that the yonah core solo, and core duo have a tough time in the 2.3ghz+ range without a vcore bump (which is right around where the top core duo maxes out) and it makes sense.

edit: yup 169fsb was the only one that was stable, ran orthos for a few hours and can still run fanless =) T1350 @ 2.36Ghz - temps are fine

right now though I am sticking to the 160 fsb with the cpu at 2.24Ghz - since I am using this as a server. speed wise, it was a decent upgrade for $20, the extra 400mhz is definately noticable

[smilingcrow]

so I'm feeling pretty sure now that the yonah core solo, and core duo have a tough time in the 2.3ghz+ range without a vcore bump (which is right around where the top core duo maxes out) and it makes sense.

From personal experience and on reading reviews I don’t think that Core Duo is limited to such a low speed at stock Vcore; I think 2.5 – 2.8GHz is achievable.
Core Solo chips are failed Core Duos I seem to remember and of all the mobile Core chips that have passed through my hands the only Core Solo that I had was the weakest chip in terms of under-volting by a noticeable margin.

The FSB limitation seems to be either a chipset or BIOS issue. Although it may well be that if the PCI or PCIe clocks aren’t locked then this would cause problems by itself. VesaM pointed me to a link that suggests that the PCI buss is locked so who knows what is holding these chips back!

MSI board has problem reading the memory timing from my ram sticks too. Sometimes it reads 4,4,4,10, sometimes 5,5,5,12. I think MSI needs to update the bios. But I don't think that is going to happen since they have just update the bios to 1.4 and this mobo is not a main-stream big-seller.

I’m having doubts about building another S479 system for this very reason. Looking at the power data for Penryn desktop chips I might build a G33 based system with an E4400 for now and upgrade to Penryn down the line. With a PicoPSU the system will have a fairly low power draw. It’s just a matter of finding a G33 motherboard with the right features now.

[nzimmers]

MSI board has problem reading the memory timing from my ram sticks too. Sometimes it reads 4,4,4,10, sometimes 5,5,5,12. I think MSI needs to update the bios. But I don't think that is going to happen since they have just update the bios to 1.4 and this mobo is not a main-stream big-seller.

I’m having doubts about building another S479 system for this very reason. Looking at the power data for Penryn desktop chips I might build a G33 based system with an E4400 for now and upgrade to Penryn down the line. With a PicoPSU the system will have a fairly low power draw. It’s just a matter of finding a G33 motherboard with the right features now.

I tested all the ram I have - everything was fine - I was stupid and forgot that ram timings differ depending on the FSB setting (doh!)

overall, I'd say that my problems with the T1350 are pretty much attributable to the CPU - the board runs fine at 199mhz bus when I use the T1200, but with the T1350 I just don't have a cpu that take that high of a clock.

anyone wanna lend me a core2duo merom to test with?

Oh yeah Penryn all grown up...gonna be a looker!

numbers are attractive - especially since the TDP is lower than merom. Bt the nice thing about core duo/merom is they are getting cheap on ebay - a new penryn system with the obligatory to of the line Aopen MOTD MB will take a bite out of the wallet.

[smilingcrow]

overall, I'd say that my problems with the T1350 are pretty much attributable to the CPU - the board runs fine at 199mhz bus when I use the T1200, but with the T1350 I just don't have a cpu that take that high of a clock.

I’ve now accepted that 133MHz chips max out at 165 and 166MHz chips max out at 199 as far as the BIOS is concerned.

Oh yeah Penryn all grown up...gonna be a looker!
numbers are attractive - especially since the TDP is lower than merom. Bt the nice thing about core duo/merom is they are getting cheap on ebay - a new penryn system with the obligatory to of the line Aopen MOTD MB will take a bite out of the wallet.

Agreed. I’m not planning on a Penryn mobile system but I am considering using a Core 2 Duo (E4400) for now but in a G33 chipset board so that it will take Penryn in the future if needed. A G33 board with an E4400 will cost roughly £110 - 120 whereas a S479 CPU/Mobo varies from £70 - 140; at least those are the prices I’m looking at.

I like the idea of building a system to last as I’ve been messing around with various systems for over a year now which is not my usual way. The G33 gives me the ability to move to Penryn and quad core if needed, although I can only see the need for that if I start video editing again. Choices, choices.

[nzimmers]

I like the idea of building a system to last as I’ve been messing around with various systems for over a year now which is not my usual way. The G33 gives me the ability to move to Penryn and quad core if needed, although I can only see the need for that if I start video editing again. Choices, choices.

If you have any current MB's capable of running a conroe cpu, you may want to check to see if they will support penryn. My year old Asus P5B (965) MB and many will support it with a bios upgrade.

As far as video editing goes...ugh that's an ugly monster to feed! I do batch conversion sometimes and the yonah server gets that job.

[smilingcrow]

If you have any current MB's capable of running a conroe cpu, you may want to check to see if they will support penryn. My year old Asus P5B (965) MB and many will support it with a bios upgrade.
As far as video editing goes...ugh that's an ugly monster to feed! I do batch conversion sometimes and the yonah server gets that job.

I only have the one PC, it’s an AM2 system and it has issues.
I looked at my options and bought a new Gigabyte S479 board yesterday as I figured that I really don’t need more than a low end Core Duo right now. The price swayed me as it was £44 including postage and the optional S/PDif I/O bracket. With T2300s going for ~£25 it’s an inexpensive and low power system.

I was seduced by the TDP figures for the G33 chipset but the only review I could find showed the power consumption to be poor. A review at Tomshardware shows the power consumption at idle for three G33 systems as being in the range 76 to 88W which seems high for an IGP system using an E2160. I tested a Gigabyte 965P S3 with a Nvidia 7300GS and E4400 and an almost identical spec to the G33 systems and achieved 66.5W at idle. Admittedly this was with a mild undervolt but that hardly makes up for my system being at least 9.5W less at idle with the disadvantage of having a discrete PCIe card!

I’m tempted to buy a Gigabyte GA-G33M-S2 and an E4400 to see how it compares to the Tomshardware review and my new S479 system. It’s a bit of a sickness that I have I think.
One thing I’m noticing is that motherboard power efficiency can vary a lot these days depending seemingly on the capacitors that they use. It seems as if certain boards are designed to be efficient at high loads but can subsequently be inefficient at the sort of low power draws that interest me. The Tomshardware article states this about the Gigabyte GA-G33M-S2H: “The more powerful regulators have the disadvantage of requiring more energy when running under low load, though: this motherboard was the one that consumed the most power.â€