Cpu Freq Vs. Dram Clock Vs. Cas Vs. ?

[Radagast]

I have been having trouble getting my system stable at stock specs (please see my related posts: Memtest86+ V1.55 Crashes With Corrupted Display, Also, random crashes in Windows XP - system specs are in the first post - & Overclocking For Stability?, Please help with my Soyo KT600 MB).

 

After one users' suggestion, I was able to run my system stable (i.e., Memtest86+ no longer crashed) by reducing the CPU frequency. I started at 100MHz (1/2 speed) and then moved up by standard increments until 166MHz. However, at this speed, my CPU is being treated as an Athlon XP 2500+, rather than the XP 3200+ that I paid for, my FSB is being run at 166/334Mhz & my PCI at 33MHz, rather than the 200/400MHz FSB & 100MHz PCI I expected from the Soyo MB I paid for. Surprisingly, the GeIL memory is still being run at 2.4-4-4-8 & 200/400Mhz as set in the SPD (although I'm sure it's not being utilized at that speed). Note, all of these run settings are as reported by SiSoftware Sandra Lite 2005.SR1.

 

The downgrade to the PCI bus really hurts, because I have a Maxtor 6Y160P0 Ultra Series 160GB UATA/133 HD (8MB Cache, 7200RPM, 9.3ms), which should be able to utilize the UDMA 133 PCI controller, provided by my Soyo MB, to the fullest! :angry2:

 

Can anyone advise how to get my system running closer to stock (or better ) without reviving my instability problems? Which specs should I concentrate on pushing: CPU Freq, DRAM clock, CAS, or what? What's the best procedure for pushing these specs? I don't have any experience with overclocking, so any help would be appreciated. Thanks! B:)

 

BTW, please don't reply with posts about how my Soyo MB or GeIL DRAM (or whatever) sucks, because I can't afford to replace any components. I am on a very low budget. I got the best I could for what I could afford. I have not been able to prove that any of the components are defective, so I cannot RMA any of them. I just need some help in getting the most out of what I've got.

[Enyce2k9]

im to lazy to read wat u typed, im responding towards ur first sentence..

-try upping the vcore if o'cd.

-try loosing the timings on ram

-make sure all temps are normal

-scan ur pc for any malicious files

-update xp if possible

-reinstall xp as alternative (full format)

-make sure components are functioning correct

[Radagast]

im to lazy to read wat u typed, im responding towards ur first sentence..

-try upping the vcore if o'cd.

-try loosing the timings on ram

-make sure all temps are normal

-scan ur pc for any malicious files

-update xp if possible

-reinstall xp as alternative (full format)

-make sure components are functioning correct

463003[/snapback]

 

Thanks for the response. I'll respond to your points in reverse order:

 

- That's what I'm trying to do, is make sure components are functioning correct.

- This is already a fresh install of XP on a new (blank) system.

- I have all updates applied, except for SP2 (I'm not ready for that headache yet).

- I run scans with BitDefender regularly. I also run SpyBot S&D. System is clean.

- I have a DigitalDoc 5, and I run MBM5 & Via HW Guardian, & all temps are pretty normal (CPU runs between 40C & 50C - usually at ~43C).

- That was part of my question. How do I loosen timings on RAM?

- I am not overclocking - I am underclocking at the moment, and I want to get back to stock (or as close as possible).

[Nodgiles]

- That was part of my question.

Edited April 15, 2005 by Nodgiles

[evanin]

My Geil Value ran runs 2.5-4-4-8 at stock voltages at pc3200. I can tighten them alittle with voltage but then again it is just value ram. I would say check temps on your CPU and raise the voltage alittle to try to get to stock. Either the Ram, Mobo, or Cpu maybe be defective.

[Mad Cow]

I had that random crashing problem as u described in your other post about a year ago. I can't remember exactly what caused it, but I do know that it had something to do with my overclock. What I usually end up doing when I run into these kind of problems, is to one-by-one, start replacing each part in your system. You said you've built many systems before, so just scavange some of their parts (that you KNOW work) and individually start replacing everything. I have a funny feeling that your PSU may be at fault, motherboard monitors are sometimes off, try another PSU to start with. If that doesn't work, reinstall the old one and substitute another part. (try RAM next, then GFX card, then try running your HD without the PCI card - just go with ata straight to the mobo, and remove the card) If none of these solve the problem do another reformat (I know, I know... u just did one, but just do it). Your last two culprits would be CPU and mobo. just hope it doesnt come to that, cause that's a ton of work to figure out.

[RnDestroyer]

try upping ur vcore sometimes they have problems at like lower stockvcore settings with some boards and they automaticly underclock them so up the volts to wutever stock is and a little bit more over that and run it see wut happens had to do that on my barton while back msi board gave me problems like u present

[kewlboxer2]

PCI/AGP should be set at 33/66 not 100... 100 mhz is bad.

[Radagast]

Hey, thanks for all the responses! I'll respond in order:

 

Nodgiles: Thanks! That's what I was thinking I should try first. I assume you mean to put everthing else back to stock before trying this, right?

 

evanin: My CPU temp (in MBM5 & Via SmartGuardian) usually hang around 41C, sometimes going to 43C when very active. VCore is pretty steady at 1.64/1.65V (stock setting). I did raise my AGP voltage to 1.6V to try and make my ATI Radeon 9600XT run a bit more stable (using 4X AGP setting, rather than "ideal" 8X). I also raised my DRAM voltage to 2.8V (max setting in my BIOS), to try and make the GeIL DRAM run more stable (it didn't seem to have any effect - Memtest86+ still crashed with stock timings throughout). How do I determine if the Ram, Mobo, or Cpu are defective?

 

Mad Cow: I am not overclocking. At the moment I am underclocking (166MHz), but I don't like that. Unfortunately, most of my old systems are in storage in California. Here in Illinois, I only have my father's Dell Dimension 4100 (PIII 933MHz & PC133 SDRAM) and a rebuilt Compaq (now with DFI AK74-EC MB, T-Bird 950MHz & PC133 SDRAM), neither of which is appropriate for component swapping with my problem system (Athlon XP 3200+ & GeIL PC3200 DRAM). I have my biggest PSU (Enermax EG375P-VE 370W) in my problem system. I can't imagine putting the Enermax EG301P-VB 300W PSU from the rebuilt Compaq (or the unknown PSU from the Dell) helping my problem. As I mentioned in the other thread, I have tried swapping in a known good ATI Radeon 7200, with no discernible results. And as you can see, I don't have anything appropriate for swap testing the MB, CPU or RAM. I did do a reformat reinstall of WIndows XP after my initial problems, but the behaviour was the same afterwards. As you say, it seems to come down to CPU or mobo and I don't know how to go about proving either of those defective.

 

RnDestroyer: I didn't fully understand your post, but I am currently running the stock Vcore of 1.65V. I really don't want to burn out my CPU. What is a safe setting for that (note, I don't want to overclock)?

 

kewlboxer2: Why is AGP/PCI at 100 MHz bad? I am only using modern components (HDD, DVD, sound & video are only components on PCI/AGP busses) that are rated to handle the high-speed bus (in fact my HDD is rated for 133MHz PCI, and my Radeon is rated for 8X AGP (although currently running at 4X). I see the faster PCI bus as one of the most important performance gains of building this "modern" system (at least Athlon XP level), so I would not want to underclock that. As I mentioned, gaming is not a priority for me. I want good performance for database access, general system work, software development, and web browsing & development; all of which require good HDD throughput. I believe that modern mobo's have been engineered to handle a 100MHz PCI bus well, unlike the first pushed PCI busses, a few years back, which did not compensate properly for increased noise.

[oralpain]

kewlboxer2: Why is AGP/PCI at 100 MHz bad?  I am only using modern components (HDD, DVD, sound & video are only components on PCI/AGP busses) that are rated to handle the high-speed bus (in fact my HDD is rated for 133MHz PCI, and my Radeon is rated for 8X AGP (although currently running at 4X).  I see the faster PCI bus as one of the most important performance gains of building this "modern" system (at least Athlon XP level), so I would not want to underclock that.  As I mentioned, gaming is not a priority for me.  I want good performance for database access, general system work, software development, and web browsing & development; all of which require good HDD throughput.  I believe that modern mobo's have been engineered to handle a 100MHz PCI bus well, unlike the first pushed PCI busses, a few years back, which did not compensate properly for increased noise.

 

No hardrive connects directly to the PCI bus and no hardrive runs at 133MHz. Even if you could overclock the bus the hardrive is on, you would see no perfromance benifit (and probabaly alot of data corruptiona nd instability).

 

The standard PCI slots in desktop computers run at 33.3MHz as always.

 

AGP is speced to 66.67MHz as always.

 

You are drastically overclocking your AGP/PCI bus and I am amazed you computer even turns on.

 

PCI-E runs at 100Mhz, not PCI, and not AGP. PCI-X can run at 33, 66, 100, or 133MHz.

 

No system is made to run the standard PCI bus past 33MHz with the exception of some old and often unstable 486 systems using the short lived 40Mhz system bus. No standard PCI card is made to operate at past 33Mhz.

 

Also, the PATA ports on most any motherbaord made in the last decade are NOT connected to the PCI bus they are directly connected to the chipset.

Edited April 16, 2005 by oralpain

[Radagast]

kewlboxer2 & oralpain: I'm sorry, I was confusing the MHz clock speed of the bus and the MBps transfer rate. I checked my Sandra reports for both the current underclocked mode and the previous stock configurations. My PCI bus was running at 33MHz in both cases (1/5 multiplier for underclocked config & 1/6 multiplier for stock). Likewise the AGP bus was running at 66MHz for both configs (using a 1/2 multiplier in both configs - I don't know what that multiplier relates to - it's not relative to the CPU clock which is 166MHz underclocked & 200 MHz stock).

 

I'm sorry I was thinking the ATA or DMA device/controller specification (e.g. UDMA133) was referring to the clock speed, but it refers to the data transfer rate, which is UDMA133 (or UATA133) for my disk drive.

 

BTW, my HDD is connected by by an Ultra-ATA cable (80-pin) to the onboard UDMA133 controller on my Soyo motherboard. That controller is part of the Via VT8237 (Southbridge), which internally connects to the PCI bus.

 

BTW, I remember those old 40MHz PCI busses - that was what I was referring to when I said "the first pushed PCI busses, a few years back, which did not compensate properly for increased noise".

 

Most importantly, I now realize that the PCI bus multiplier is being automatically adjusted to keep the PCI bus at 33MHZ, so that my controller can still produce a 133MBps data tranfer rate, regardless of where the CPU clock is set, at least for the standard frequencies. I believe that if the 1MHz frequency adjustment in the Soyo mobo is used, then that affects the PCI/AGP bus speeds, which can then cause problems, as you mentioned. However, I believe that feature requires an unlocked CPU, which I do not have.

 

Thank you for helping me get cleared up on that. That is one big relief! B:) Now, what about all my other questions?

Edited April 17, 2005 by Radagast

[Guest Silenc3]

All I read was Soyo MB and unstable a Stock.

 

So IMO that says it all.