XP 1700+ Tbred A, Overclocking!! Help Needed!!

[vikash4900]

Hello Friends,

I have the following hardware:

XP 1700+ Tbred "A" OEM
CoolerMaster DP5-6I31C Cooler (ALL Aluminium)
Jetway KT333 Mobo
Kingston 512MB PC2100 DDR Ram (2x256)
40gb Seagate Barracuda 4 7200rpm
ATI Radeon 8500LE 64mb
Onboard Audio
300w Power Supply
Full tower ATX Case with 2x80mm exhaust and 1x80mm intake fan.


I have tried almost every thing, but i am unable to get the cpu to work properly over 1750mhz.

the highest frequencies that worked stably are :

9.0 x 194fsb x 194ram = 1746 mhz
10.0 x 173fsb x 173ram = 1730 mhz
10.5 x 166fsb x 166ram = 1743 mhz
11.0 x 158fsb x 158ram = 1738 mhz
11.5 x 151fsb x 151ram = 1736 mhz
12.0 x 145fsb x 179ram = 1740 mhz
12.5 x 139fsb x 172ram = 1737 mhz

With the above results, it is clear that neither the board nor the ram are the limiting factor.

all these results i got with the vcore : 1.625v and Vdimm : 2.7v.

Increasing the voltages beyond these does not help at all....i've tried vcore upto 1.8v....

I've got the highest 3D Mark 2001 Result with 9.0x194fsbx194ram:

The Default benchmark gave a result of 8551 3D Marks.


Where am i getting stuck? why can't i cross this 1750mhz barrier? Are my 3dmark scores ok?

Please help me. I want to be able to reach 1.9ghz like others have done....


Vikash

 

[OS]

there's no such thing as a guaranteed overclock. I have a retail 1700+ tbred a also, bought less than two weeks ago. It doesn't overclock worth sh*t. I can only get 1667 mHz@1.7V rock solid stable. Anything significantly above that starts producing latent errors.

 

[0roo0roo]

your m/b lets you change multiplier?

 

[vikash4900]

No, not the motherboard. Actually, the xp tbreds come with unlocked multipliers.....so i can cahnge the multipliers through the mobo jumpers.

 

[boggsie]

Same results here from both of the thoroughbread A's that I got last week from newegg.

All of the B's were regularly hitting +2Ghz and I get stuck with TWO A's that won't reliably go above 1740Mhz, even with 2v and water cooling.

It is the luck of the draw.

 

[Blooz1]

I must be lucky (considering that I wanted a "B"!). My "A" that I received Friday does 1.92Ghz very reliably..still wish I had gotten the "B", though!

 

[Jeff7181]

that PC2100 DDR doesn't help anything ... especially at higher FSB's... You need to run some torture tests to figure out where it's getting errors

 

[human2k]

194FSB, that is impressive on kingston pc2100.

 

[KF]

When the core voltage does virtually nothing to increase the OC, you have topped out at a given temp. When a chip gets near its top, it gets very sensitive to a few degrees of temperature, and that temperature is way below its maximum temperature spec. Considering what happened to the fellow with water cooling, I wouldn't get any hopes up, but for max OC, you need an extreme cooler which would drop the temp 5-10 degrees. However, you can usually drop the temp 5-10 degrees by opening the case and directing an ordinary room fan into it to see if bettering cooling does anything worth while. (It is not really possible to have the temp inside the case the same as the outside temperature by ordinary means.) You can get a digital indoor/outdoor thermometer to check case temps.

The HS you have is very nice. It is probably about as good as an extruded chunk of aluminum which will fit can do. Coolermaster does a good job on this type of design, IMO (and they are cheap at Newegg). The fan speed (5000) is about as high as you would want to go to keep the racket reasonable. Copper insets for HSs do not seem to do much for temperatures (2 degrees?) but they may even out the temp over the surface of the CPU, which could help OCing. Copper heat sinks with lots of thin fins can drop the temp considerably (the well designed ones are often mentioned), but they generally need very high CFM fans to achieve this.

The only bargain HS-fan I know of that may be better is Coolermaster HAC-V81 ($15 shipped). It has an 80mm variable speed fan and the HS has a copper insert. Although it is loud at max speed, the dominant noise is rushing air, something which you have to accept from a high CFM fan of any kind. The fan whine, which is the annoying part, also gets very much louder, but since the wind noise is predominant, I don't find it hard to take. A 7000rpm 60mm fan is incomparably worse, beyond what I can tolerate. The sound volume goes down to a fraction when I drop the speed from 5000 to 4000, but the max load temp goes up about 4 degrees.

 

[xylem]

I'm running mine on an Asus A7V266-E, vcore set to 1.75 (bios readout 1.74, Sandra readout 1.84... ), using a Swiftech MC462A heatsink with a relatively-low-rpm fan. I'm running the cpu @ 12.5 x 153 = 1912.5 mhz. I have not tried higher, since I can't seem to get higher mults. to work, and I don't want to risk hdd corruption from higher fsb.

I got this cpu from Newegg recently... it's a T-bred A, and the stepping, if I remember correctly, is AIUGA.

I have to agree with KF here... the likely primary culprit in your scenario is unsatisfactory cooling. The core on the T-bred A is so small that, with air cooling, the thermal conductivity of the heatsink material should have a large impact on core temperature. What temperatures are you measuring @ different Vcore settings?

 

[jiffylube1024]

Yikes! My 1700+ A does 1.83 GHz so easily it's not even funny! 1.9 GHz is stable too. It can post at 2 GHz but will crash mid-boot.

Perhaps your PSU is maxed out and the voltage rails are unstable - monitor your 3.3/5/12V rails in a program like MBM5 (particularly the 5V, which the CPU runs on). Make sure they don't fluctuate too much or drop below their respective rated levels. If you jump all over the place, you probably need a better PSU.

Also, monitor your actual Vcore, again in a program like MBM5. Perhaps your motherboard undervolts, and you're actually only at 1.575V or something...

Also, monitor your temperatures (see above for what program ). I've found that once temps go above ~50-55C, the Tbred A will get unstable.

I run my 1700+ at 1.83 GHz pretty much all day at 1.65V (1.68V actual). Idle temp is 37-41C, Load temp is 46C.



WHOAH!!! Just realized something - if your post is correct, you're running RAM frequencies of 151MHz and above. Your PC2100 memory is only rated at 133 MHZ. If that is correct, then you will have to upgrade to better memory. PC2700 (166MHz) or PC3200 (200 MHz) is recommended for some of those FSB's (194ram? Is that possible on PC2100?)

 

[vikash4900]

Originally posted by: KF
When the core voltage does virtually nothing to increase the OC, you have topped out at a given temp. When a chip gets near its top, it gets very sensitive to a few degrees of temperature, and that temperature is way below its maximum temperature spec. Considering what happened to the fellow with water cooling, I wouldn't get any hopes up, but for max OC, you need an extreme cooler which would drop the temp 5-10 degrees. However, you can usually drop the temp 5-10 degrees by opening the case and directing an ordinary room fan into it to see if bettering cooling does anything worth while. (It is not really possible to have the temp inside the case the same as the outside temperature by ordinary means.)

Thanks KF for the insight on case temp...

I too think that its the cooling which is at fault...but since i don't have any better cooler at hand.....i tried the following things:

1) cleaned off the thermal pad and put ASII
2) removed the 60x60x15mm 27cfm cooler master fan from the heatsink and mounted a thermaltake 60x60x25mm 36cfm fan.

And tried overclocking again.... but to no effect !!

My temps though are not to high: 11x133fsb = 1463mhz (1.49 vcore, nominal speed) 32C - 35C idle and 40c to 44C full load.
12x145fsb = 1745mhz (1.61 vcore) 35C - 38C idle and 45c to 49C full load.


I think i will try a different method, i'll leave the case open in front of an Air Conditioner, and then try to overclock!!

We'll find out if heat is the culprit here....

 

[vikash4900]

Originally posted by: jiffylube1024
Perhaps your PSU is maxed out and the voltage rails are unstable - monitor your 3.3/5/12V rails in a program like MBM5 (particularly the 5V, which the CPU runs on). Make sure they don't fluctuate too much or drop below their respective rated levels. If you jump all over the place, you probably need a better PSU.

Also, monitor your actual Vcore, again in a program like MBM5. Perhaps your motherboard undervolts, and you're actually only at 1.575V or something...

Also, monitor your temperatures (see above for what program ). I've found that once temps go above ~50-55C, the Tbred A will get unstable.

I run my 1700+ at 1.83 GHz pretty much all day at 1.65V (1.68V actual). Idle temp is 37-41C, Load temp is 46C.

WHOAH!!! Just realized something - if your post is correct, you're running RAM frequencies of 151MHz and above. Your PC2100 memory is only rated at 133 MHZ. If that is correct, then you will have to upgrade to better memory. PC2700 (166MHz) or PC3200 (200 MHz) is recommended for some of those FSB's (194ram? Is that possible on PC2100?)

Yes, your inputs on Power Supply might be the reason, but i have no option here.....we dont get any better powersupplies here....this ECS Make 300w PSU is the best one available here the price is US$8 only!!

My Max temps at full load do reach 50C.

And, as for the kingston Ram, it is rated at 133mhz Cas 2, but when i run it higher than 150mhz, i am running it at cas 2.5. Probably, thats why it can reach 194fsb!!

 

[0roo0roo]

i thought new cpus run 12v?