Sign in with
Sign up | Sign in
Your question

SATA "bottleneck"

Last response: in CPUs
Share
June 15, 2006 3:25:29 PM

Switched from Parallel ATA to SATA and performance is sensibly increased. However, over FSB 225 the disk is not detected. Anybody knows how to get around this ???

More about : sata bottleneck

June 15, 2006 3:30:10 PM

you have to be much more specific with explanation if you want to expect apropriate answer. what is your full hardware configuration and BIOS settings?
June 15, 2006 3:39:00 PM

Quote:
you have to be much more specific with explanation if you want to expect apropriate answer. what is your full hardware configuration and BIOS settings?


Athlon64 3000+
RAM @ 183/200
Asus A8V-MX
Maxtor 6V080E0 (SATAII but is unstable when run in this mode)
Related resources
June 15, 2006 4:10:56 PM

The bus on which the integrated SATA ports are is'nt frequency locked for the A8V-MX, only the SATA ports connected to the Promise controller found on the Deluxe version have that feature, or so I've read on This Thread
June 15, 2006 4:34:12 PM

I agree with Sid. You would need the ability to lock the SATA.
June 15, 2006 6:05:36 PM

Quote:
I agree with Sid. You would need the ability to lock the SATA.


Is't it possible that the disk itself does not respond? It's not totally locked if I can set my FSB at 225!
a c 126 à CPUs
June 15, 2006 7:01:22 PM

Quote:
The bus on which the integrated SATA ports are is'nt frequency locked for the A8V-MX, only the SATA ports connected to the Promise controller found on the Deluxe version have that feature, or so I've read on This Thread

this is the problem for you.....
dont feel bad...i have NO LOCKS(PCI/AGP/SATA/NOTHING) at all and 4 sata drives....if i try 225 i will kill all my drives(corupt all data on them)....

however i do not think that board has a promise controller onboard(said above too)....is that the MATX board right? with all 4 sata ports on the VIA SATA/Raid controller...

the only thing i could see working is a PCI sata controller(as long as your pci bus is locked).....but IMO its not worth it in your case.....why not just stick at 220? its still a useful OC

I hope this helps
June 15, 2006 7:03:12 PM

Quote:
Is't it possible that the disk itself does not respond? It's not totally locked if I can set my FSB at 225!

The problem is that the SATA bus clock is not locked at all, so moving your system clock (whatever its called on that mobo - system clock, HTT, FSB, whatever) also overclocks the SATA bus. In your case, your HDD and/or SATA controller has an oc limit of 25mhz (I only got about 10mhz). If you're lucky, moving it back down below 225 will let you boot up again. In my case (not an Asus, a Soyo) it wouldn't boot and I had to reformat & reinstall.

Mike.
June 15, 2006 7:05:29 PM

That is what we are trying to say "You SHOULD lock the SATA ports". Most will stop/have issues at ? frequency while overclocking. The reason for the ? is that is an unknown per controller/MB.

Alot of people while attempting an overclock will lock those to a certain freq so as to not have issues.

Sid beleives that your MB does not have the capability to LOCK the SATA. If so you are limited in your capabilites to OC.
June 15, 2006 8:33:40 PM

It's a feature (or lack of thereoff) of your motherboard chipset that is causing the problems you are experiencing. That's the good old PCI lock bug.

At stock settings, everything is fine since all the frequencies are within specs, as you increase the bus (be it the HT or FSB) all the other bus that are using it as a reference point see an increase in their operating frequencies as well.

In your case, a 225MHz HT clock is ~12% over specs, that's still within the tolerance margin of the PCI bus (36~37MHz or so), as you increase it over 38MHz, either the devices or the controller itself start failing, that's the wall you're hitting right now.

Due to that design flaw, the best you can hope for is a 12% OC unless you get a motherboard that was designed with overclocking in mind.

What we refer to as a PCI (or AGP) lock is a workaround that keep the bus frequencies separated from one an other by using dividers, this is something manufacturers integrate into their motherboards during conception so there is no way to force it on.

You can always try to swap your SATA HDD for IDE and hope that the PCI lock is working for that particular controller.
June 15, 2006 9:01:22 PM

Sid,

In this case his only option for OC is to stay at ~220 and open up the multiplier (if available/able) if I remember right? What increment does your board support for the HT 1, 5, 10?

I have not OC'd in quite a while "too much going on and rarely get to play anymore".

** Real Bummer **
June 15, 2006 9:19:42 PM

Quote:
It's a feature (or lack of thereoff) of your motherboard chipset that is causing the problems you are experiencing. That's the good old PCI lock bug.

At stock settings, everything is fine since all the frequencies are within specs, as you increase the bus (be it the HT or FSB) all the other bus that are using it as a reference point see an increase in their operating frequencies as well.

In your case, a 225MHz HT clock is ~12% over specs, that's still within the tolerance margin of the PCI bus (36~37MHz or so), as you increase it over 38MHz, either the devices or the controller itself start failing, that's the wall you're hitting right now.

Due to that design flaw, the best you can hope for is a 12% OC unless you get a motherboard that was designed with overclocking in mind.

What we refer to as a PCI (or AGP) lock is a workaround that keep the bus frequencies separated from one an other by using dividers, this is something manufacturers integrate into their motherboards during conception so there is no way to force it on.

You can always try to swap your SATA HDD for IDE and hope that the PCI lock is working for that particular controller.


Looks like I have to hold my FSB 225 tight!
June 15, 2006 9:26:10 PM

m25,

What is the boards increments? 5mhz?

Did not reseach the board just going by what sid said.
June 15, 2006 9:36:00 PM

Multipliers on A64 CPUs are locked upward so yeah, no other options than either be happy with his current OC, using IDE HDDs instead of SATA (they're supposed to be good up to 270MHz) or getting an OC friendly mobo.

My mobo have similar limitations, SATA ports 1 & 2 crap out even on a mild OC, fortunately, SATA 3 & 4 are fine. The Deluxe version of the A8V behave the same as my K8N Neo2, the two "extra" SATA ports it have over the MX version will work just fine.
June 15, 2006 9:39:52 PM

my asus mobo (see sig) doesnt have a sata lock, it has pci thou, and my fsb is at 270, i think i had ur problem before, it was cause i wasnt adding enuf voltage to my cpu that it wasnt detecting my hdd's, i just added 1 increment and everything was stable
June 15, 2006 9:41:26 PM

Sid,

I have the Neo2 FISR from MSI for my P4 3.0C and it is a nice board.

C'mon Sid you use the dynamic overclocking feature right ;) 

Just kidding I know better.

I tried it when I first got the board just to check it out and it was kinda weak!
June 15, 2006 9:57:14 PM

I actually laughed when I first read the manual and saw that "feature" :lol: 
June 15, 2006 10:14:18 PM

I have a new build coming up so I will get to play again! I actually may go with the "BadAxe" Intel board used for the Conroe testing/OCing. 1mhz increments for nearly everything gives a whole lot of leeway for a good solid OC.

I might even be putting together a merom HTPC for a client (side business). That should be fun. Have a limited form factor and limited heat disipation so need to stay cool!
June 16, 2006 5:06:10 AM

Quote:
m25,

What is the boards increments? 5mhz?

Did not reseach the board just going by what sid said.


It has 1 MHz increments and it locks exactly at 224
June 16, 2006 5:11:23 AM

Quote:
Multipliers on A64 CPUs are locked upward so yeah, no other options than either be happy with his current OC, using IDE HDDs instead of SATA (they're supposed to be good up to 270MHz) or getting an OC friendly mobo.

My mobo have similar limitations, SATA ports 1 & 2 crap out even on a mild OC, fortunately, SATA 3 & 4 are fine. The Deluxe version of the A8V behave the same as my K8N Neo2, the two "extra" SATA ports it have over the MX version will work just fine.


The MX has 4 SATA ports too! Tou are saying that just bt changing the connector position can improve overclocking?
June 17, 2006 5:01:27 PM

Quote:
It's a feature (or lack of thereoff) of your motherboard chipset that is causing the problems you are experiencing. That's the good old PCI lock bug.

At stock settings, everything is fine since all the frequencies are within specs, as you increase the bus (be it the HT or FSB) all the other bus that are using it as a reference point see an increase in their operating frequencies as well.

In your case, a 225MHz HT clock is ~12% over specs, that's still within the tolerance margin of the PCI bus (36~37MHz or so), as you increase it over 38MHz, either the devices or the controller itself start failing, that's the wall you're hitting right now.

Due to that design flaw, the best you can hope for is a 12% OC unless you get a motherboard that was designed with overclocking in mind.

What we refer to as a PCI (or AGP) lock is a workaround that keep the bus frequencies separated from one an other by using dividers, this is something manufacturers integrate into their motherboards during conception so there is no way to force it on.

You can always try to swap your SATA HDD for IDE and hope that the PCI lock is working for that particular controller.


I just locked the AGP/PCI to 66.6/33.3 and it just didn't boot at all, black screen and monitor off so I had to reset the defaults with the jumpers.
My board CAN lock PCI but I don't know why it doesn't work.
Will a PCI controller solve my problem or what else?
June 17, 2006 8:04:14 PM

Try 67, that's the "traditional" lock frequency.
June 17, 2006 9:22:20 PM

Quote:
Try 67, that's the "traditional" lock frequency.

Guess I ause I only got the options of 66/33, 75/37 (or something like this) and 80/40
!