Is CPUZ lying to me ??

[Arthalen]

As much as I've seen that this program is the defacto choice in displaying information about your processor and its characteristics, I can't help feeling it's actually lying to me

Yes you heard me. The hardware specs are below. Not much but fun to play with - Have managed to get the NVRAID working on a pair of wildly different PATA drives on the Primary channel.

http://img134.imageshack.us/my.php?image=uz14fu.png

Anyways, CPUZ is telling me that the processor I have has the x86-64 instruction set, making it a 64-Bit Sempron, when I remember getting the processor before these were mainstream (just before).

Also, on the memory page, its telling me that as I push the fsb bus up in my overclock attempts, its actually syncing the memory alongside, yet in the bios its telling me (as I'm right in thinking I should be doing, I hope) that I've got my memory set to manual clocking and set to 200Mhz.

Its a bit confusing.

Last Question: anyone know if there is a 'hidden' way of altering the Vcore voltage on the mobo ? I think this may stop me going any higher than the 240FSB clock I've reached with a previous clock...

The hardware:

Sempron 2600 (Eeck I know)
Asus Skt754 K8N4-E Deluxe Motherboard
1x 512MB Kingston Value Ram DDR400
Jeantech 400Watt PSU
Stock HSF


Your probably wondering WTF at this setup, but hey, I like pushing stuff and its works wonders compared to a P3 700mhz with 128Mb Ram

 

[Mul]

nope, you have the newer Sempron, which has the highly clockable Palermo core and 64bit function

Mul

 

[Topitoff]

Keep going these can go a long way on stock volts:

 

[Arthalen]

Without sounding lazy, where can I research the stepping I have ???

I've heard the Palermo was a good clocker, but even this 'high end' board seems to be clocking limited. No Vcore adjustment and my PSU isn't upto much either.....

Anyone had any kind of sucess with the K8N4-E Deluxe ?

 

[ajgoodfellow]

When you set your ram to 200MHz in the BIOS you're actually setting it to use the 1:1 divider. When your HTT is at its stock speed of 200MHz, your ram is also at 200MHz

You don't want your HyperTransport speed to go over 1GHz. Check that your HT multiplier is set to 4x up to 250MHz HTT, 3x to 333MHz HTT, etc. This is sometimes shown as HT frequencies of 1GHz, 800MHz, 600MHz, etc

To try clocking it higher reduce your HT multiplier to 3x and set your ram to 166MHz

 

[Arthalen]

Current Status: Last night I managed to reach 247Mhz HTT .... so close to hitting 2Ghz on the chip and 1Ghz HTT!

I'll post CPUZ shots tonight when I get home, if anyone can get excited enough about a Sempron overclock

When you set your ram to 200MHz in the BIOS you're actually setting it to use the 1:1 divider. When your HTT is at its stock speed of 200MHz, your ram is also at 200MHz

You don't want your HyperTransport speed to go over 1GHz. Check that your HT multiplier is set to 4x up to 250MHz HTT, 3x to 333MHz HTT, etc. This is sometimes shown as HT frequencies of 1GHz, 800MHz, 600MHz, etc

To try clocking it higher reduce your HT multiplier to 3x and set your ram to 166MHz

And, lo, a new life was born, a new idea was formed, and elightenment was reached by those in the dark.....

AJgoodfeelow - your a stormer! Of course!! I should have realized the 200mhz for RAM in the BIOS was acting as a mem divider - muwahahahahha. I'll drop it down tonight when I get home from work and see what I can get out of this chip.

Must admit I knew about the HTT multiplier, and knew to drop it down once I got above 250mhz.

 

[pcoltrane]

I wouldn't be suprised if your sempron was capable of at least 2.4ghz. It's going to be the board that holds you back with such a low multiplier on the chip. I don't know how well those K8N-4s overclock, but you'll need to squeze every drop from it to get the best out of the chip.

Initially i wouldn't worry about your total HTT, just take the multi right down to 2x or 3x and see how far the htt freq will go. once you've pushed the board/chip as high as they'll go , try upping the multi again.

In an ideal world you'd be able to lower the multi on your chip in order to test the max htt of your board, unfortunately that's not possible with the 2600 semprons. I don't think it's going to be an issue for you though as we know most semprons do at least 2.3/2.4ghz so get your ram on a low divider to get it out the equation and push the htt up until it falls over.

NF4 boards seem to do anything between 240 and 350+ HTT. Asus generally aren't the best overclockers (but can be nice and stable). I get 285 out of my A8N-E so hopefully you have some way to go yet

Marc

 

[MarkB2k1]

I've a sempron 3100+ non 64-bit, it's currently running as in my sig. 9*285mhz, 3xHTT and RAM set to 266mhz DDR in the BIOS.

This is with 1.55v and an AC Freezer 64 Pro the temps max out about 45 - 50C.

 

[Arthalen]

Update -

1) Yep, confirmed it has 64bit support., d/l and tested with WinXP x64 Trial. Recognised as a 64 bit system, so might have some fun there!!

2) The mem divider has worked an absolute treat for the Overclocking. I've attached a picture of the max overclock I have - so far.

Hitting the wall at 247FSB, I dropped the Mem Divider from 1:1 (200Mhz) to 6:5 (166mhz) and carried on without a problem. Over 250mhz, I dropped the HTT multi to 3x as a precaution, and kept on going! Nearing 300FSB, I dropped the Mem Divider again to 3:2 (133mhz) and have managed to go over to 306FSB as CPUZ shows. SuperPi won't run, but I've had no lockups at 306FSB while working away in windows.

Now, If I try and go any higher, it won't boot (although it does post). I know I could drop the Mem Divider again to 2:1 (100mhz), but I don't think its the memory now.

Any suggestions on how to proceed ? Drop the HTT Multi to 2x ? The only voltage modification I can make is to VDimm, which is set at 2.75 currently.

Btw, I know the timings are very loose for memory, I'll look at tightening them up (if I can on KVR) once I've gone as far as I can with this chip.

http://img140.imageshack.us/img140/5813/max39by.png

 

[Minstadave]

At over 300HTT you're likely to be hitting the max of your chipset, which you can't really get around, you can check this by dropping the multiplier and see if you can run at a higher HTT. If you can then your CPU needs more Vcore, if its still unstable you're stuck really, you could try better chipset cooling or a Vchipset modd.

Just one point, if your rig can't pass Superpi 1M then its too unstable for even windows use, I'd recommend backing off and testing stability with Prime95.

 

[Arthalen]

Unfortunately, I had a good play over the weekend, but wasn't able to get it stable over 285mhz FSB at 3x HTT with mem running at 228Mhz (CPU/10 Divider, according to CPUZ). This has got my chip running at just under 2.3Ghz - not bad for a chip that started out at 1.6Ghz.

Anyways, there is absolutely, definitly, no where to alter the processor voltage (Vcore) on this motherboard. RTFM, and no where does it mention Vcore, despite Asus website saying that the board features 'Precision Tweaker' and alterable Vcore. If only there was, because temperatures are low, voltage is stable and I know the board will post at 306FSB and the memory can go as high as 247mhz before causing a problem....

Such a shame... I know this chip can go further .... Anyone give me any advice at all to try and go further ?

Current Specs: Sempron 2600+ at 2.28Ghz, Stock HSF, 512Mb KVR400 at 228Mhz, Asus K8N4-E Deluxe Mobo at 285Mhz HTT, Asus Geforce 6600, 2x 40GB Western Digitals in RAID 0, Thermaltake Soprano & Jeantech 400Watt PSU

 

[turbotoaster]

If ya have a digital camera, take photos off all your bios screens and well have a look to see if we can see voltage adjustment

 

[Minstadave]

Unfortunately, I had a good play over the weekend, but wasn't able to get it stable over 285mhz FSB at 3x HTT with mem running at 228Mhz (CPU/10 Divider, according to CPUZ). This has got my chip running at just under 2.3Ghz - not bad for a chip that started out at 1.6Ghz.

Anyways, there is absolutely, definitly, no where to alter the processor voltage (Vcore) on this motherboard. RTFM, and no where does it mention Vcore, despite Asus website saying that the board features 'Precision Tweaker' and alterable Vcore. If only there was, because temperatures are low, voltage is stable and I know the board will post at 306FSB and the memory can go as high as 247mhz before causing a problem....

Such a shame... I know this chip can go further .... Anyone give me any advice at all to try and go further ?

Current Specs: Sempron 2600+ at 2.28Ghz, Stock HSF, 512Mb KVR400 at 228Mhz, Asus K8N4-E Deluxe Mobo at 285Mhz HTT, Asus Geforce 6600, 2x 40GB Western Digitals in RAID 0, Thermaltake Soprano & Jeantech 400Watt PSU

You might want to see if there's a newer bios that allows Vcore adjustment.

 

[Arthalen]

TurboToaster - I'll try, but it will be dodgy camera phone shots of the BIOS...

MinstaDave - There is only a Beta BIOS that is newer than the one I'm flashed too with no details of what its there for. Forgive me in thinking that Beta testing for Asus when they don't even post details of the Beta BIOS is a little..... lame.

I did google this problem up, and found a lot of other user's complaining at the lack of Vcore adjustment ... Why make a 'High End' 'Deluxe' board for 754 and not include Vcore adjustment when DDR VDimm is there and a maximum possible FSB of 400Mhz is available ?!?!

 

[vincentvega14]

Keep going these can go a long way on stock volts:

That really isnt a bad time for a 256k cpu/single channel memory

 

[Topitoff]

That really isnt a bad time for a 256k cpu/single channel memory

I think it might have something to do with the DFI mobo and TCCD memory at 1T.

I put the cpu next into an Asrock K8NF4G-SATA2; again @ 312 x8 2.5GHz; but had to run memory at 207 (133 divider) 2-2-2-6 1T - this time only 38.2s super-pi score.

 

[Defcon5]

I think it might have something to do with the DFI mobo and TCCD memory at 1T.

I put the cpu next into an Asrock K8NF4G-SATA2; again @ 312 x8 2.5GHz; but had to run memory at 207 (133 divider) 2-2-2-6 1T - this time only 38.2s super-pi score.

Yeah pi likes bandwith.